1 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2 Martin Jambor <mjambor@suse.cz>
5 * tree.h (expr_type_first_operand_type_p): Declare
6 * tree.c (expr_type_first_operand_type_p): New function.
7 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
8 (ipa_value_from_jfunc): Adjust declaration.
9 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
10 Use it as result type for arithmetics, unless it is NULL in which case
12 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
13 ipa_get_jf_pass_through_result.
14 (propagate_vals_across_pass_through): Likewise.
15 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
16 is to propagate_vals_across_pass_through.
17 (propagate_constants_across_call): Pass PARM_TYPE to
18 propagate_scalar_across_jump_function.
19 (find_more_scalar_values_for_callers_subset): Pass parameter type to
21 (cgraph_edge_brings_all_scalars_for_node): Likewise.
22 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
23 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
24 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
25 target_type, pass it to ipa_value_from_jfunc.
26 (update_indirect_edges_after_inlining): Pass parameter type to
27 try_make_edge_direct_simple_call.
29 2017-11-28 Jeff Law <law@redhat.com>
31 * gimple-ssa-evrp-analyze.c
32 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
33 refine ranges if scev_initialized_p returns true.
34 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
36 2017-11-28 Julia Koval <julia.koval@intel.com>
38 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
39 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
40 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
41 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
42 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
43 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
44 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
45 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
46 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
47 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
48 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
49 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
50 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
51 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
52 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
53 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
54 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
55 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
56 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
57 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
58 __builtin_ia32_vpshrd_v2di_mask): New builtins.
59 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
61 2017-11-28 Julia Koval <julia.koval@intel.com>
63 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
64 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
65 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
66 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
67 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
68 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
69 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
70 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
71 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
72 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
73 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
74 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
75 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
76 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
77 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
78 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
79 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
80 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
81 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
82 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
83 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
84 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
85 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
86 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
87 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
88 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
89 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
90 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
91 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
92 __builtin_ia32_vpshld_v2di_mask): New builtins.
93 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
94 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
96 2017-11-28 Richard Biener <rguenther@suse.de>
98 PR tree-optimization/80776
99 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
101 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
103 (evrp_range_analyzer::record_ranges_from_incoming_edges):
104 If the incoming edge is an effective fallthru because the other
105 edge only reaches a __builtin_unreachable () then record ranges
106 derived from the controlling condition in SSA info.
107 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
108 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
110 2017-11-28 Olivier Hainque <hainque@adacore.com>
112 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
113 on mingw build hosts.
115 2017-11-28 Jakub Jelinek <jakub@redhat.com>
118 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
120 2017-11-28 Richard Biener <rguenther@suse.de>
123 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
124 copies generated from memcpy use a character array as reference
127 2017-11-28 Julia Koval <julia.koval@intel.com>
128 Sebastian Peryt <sebastian.peryt@intel.com>
130 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
131 c-family/c-cilkplus.o, c-family/array-notation-common.o,
132 cilk-common.o, cilk.h, cilk-common.c): Remove.
134 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
135 * builtins.c (is_builtin_name): Remove cilkplus condition.
136 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
137 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
138 cilk-builtins.def, cilkplus.def): Remove.
139 * cif-code.def (CILK_SPAWN): Remove.
140 * cilk-builtins.def: Delete.
141 * cilk-common.c: Ditto.
143 * cilkplus.def: Ditto.
144 * config/darwin.h (fcilkplus): Delete.
145 * cppbuiltin.c: Ditto.
146 * doc/extend.texi: Remove cilkplus doc.
147 * doc/generic.texi: Ditto.
148 * doc/invoke.texi: Ditto.
149 * doc/passes.texi: Ditto.
150 * gcc.c (fcilkplus): Remove.
151 * gengtype.c (cilk.h): Remove.
152 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
154 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
156 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
158 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
159 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
161 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
162 inline_read_section): Ditto.
163 * ipa-inline-analysis.c (cilk.h): Remove.
164 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
165 * lto-wrapper.c (merge_and_complain, append_compiler_options,
166 append_linker_options): Remove condition for fcilkplus.
167 * lto/lto-lang.c (cilk.h): Remove.
168 (lto_init): Remove condition for fcilkplus.
169 * omp-expand.c (expand_cilk_for_call): Delete.
170 (expand_omp_taskreg, expand_omp_for_static_chunk,
171 expand_omp_for): Remove cilkplus
173 (expand_cilk_for): Delete.
174 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
175 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
176 execute_lower_omp, diagnose_sb_0): Ditto.
177 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
178 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
179 * tree-nested.c: Ditto.
180 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
181 (dump_generic_node): Ditto.
182 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
183 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
185 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
187 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
189 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
190 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
191 for vector mode and !TARGET_SIMD.
193 2017-11-28 Jakub Jelinek <jakub@redhat.com>
195 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
197 * tree.h (SWITCH_LABELS): Remove.
198 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
199 assert SWITCH_BODY is non-NULL.
200 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
202 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
204 PR tree-optimization/80788
205 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
206 has TREE_OVERFLOW set, call drop_tree_overflow.
208 2017-11-28 Richard Biener <rguenther@suse.de>
210 PR tree-optimization/83158
211 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
214 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
217 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
218 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
220 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
222 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
223 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
225 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
228 * varasm.c (compare_constant): Take the mode of the constants into
229 account when comparing floating point constants.
231 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
233 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
234 from explicit instantiation of debug_helper.
235 * vec.h (DEFINE_DEBUG_VEC): Ditto.
237 2017-11-27 Richard Biener <rguenther@suse.de>
239 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
242 2017-11-27 Richard Biener <rguenther@suse.de>
244 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
245 (type_hash_canon): Also clear int_cst_hash_table entry for
247 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
249 2017-11-27 Tamar Christina <tamar.christina@arm.com>
251 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
252 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
253 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
255 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
257 * hash-map.h (gt_cleare_cache): Avoid UB.
259 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
261 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
262 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
263 (decide_unroll_runtime_iterations): Likewise.
264 (decide_unroll_stupid): Likewise.
266 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
269 * config/i386/i386.md: Add a loop with incssp.
271 2017-11-27 Martin Jambor <mjambor@suse.cz>
273 PR tree-optimization/81248
274 * tree-sra.c (splice_param_accesses): Remove size check.
275 (decide_one_param_reduction): Fix size check.
276 * gimple-pretty-print.c (dump_profile): Silence warning.
277 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
279 2017-11-27 Jakub Jelinek <jakub@redhat.com>
282 * dbxout.c (lastlineno): New variable.
283 (dbx_debug_hooks): Use dbxout_switch_text_section as
284 switch_text_section debug hook.
285 (dbxout_function_end): Switch to current_function_section
286 rather than function_section. If crtl->has_bb_partition,
287 output just one N_FUN, depending on in_cold_section_p.
288 (dbxout_source_line): Remember last lineno in lastlineno.
289 (dbxout_switch_text_section): New function.
290 (dbxout_function_decl): Adjust dbxout_block caller.
291 (dbx_block_with_cold_children): New function.
292 (dbxout_block): Return true if any LBRAC/RBRAC have been
293 emitted. Use dbx_block_with_cold_children at depth == 0
294 in second partition. Add PARENT_BLOCKNUM argument, pass
295 it optionally adjusted to children. Output LBRAC/RBRAC
296 around recursive call only if the block is in the current
297 partition, if not and anything was output, emit empty
299 * final.c (final_scan_insn): Compute cold_function_name
300 before calling switch_text_section debug hook. Call
301 that hook even if dwarf2out_do_frame if not emitting
305 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
308 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
310 PR rtl-optimization/82488
311 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
313 2017-11-26 Julia Koval <julia.koval@intel.com>
315 * config/i386/i386.c (processor_target_table): Add skylake_cost for
317 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
320 2017-11-26 Julia Koval <julia.koval@intel.com>
322 * config/i386/driver-i386.c (host_detect_local_cpu):
323 Detect skylake-avx512.
325 2017-11-26 Julia Koval <julia.koval@intel.com>
327 * config.gcc: Add -march=cannonlake.
328 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
329 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
330 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
331 (PTA_CANNONLAKE): New.
332 (processor_target_table): Add cannonlake.
333 (ix86_option_override_internal): Ditto.
334 (fold_builtin_cpu): Ditto.
335 (get_builtin_code_for_version): Handle cannonlake.
336 (M_INTEL_COREI7_CANNONLAKE): New.
337 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
338 * doc/invoke.texi: Add -march=cannonlake.
340 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
342 * plugin.c (add_new_plugin): Use platform-specific library extensions.
343 (try_init_one_plugin): Alternative implementation for MinGW.
344 * Makefile.in (plugin_implib): New.
345 (gengtype-lex.c): Fix broken AIX workaround.
346 * configure: Regenerate.
347 * doc/plugins.texi: Document support for MinGW.
349 2017-11-25 Jakub Jelinek <jakub@redhat.com>
351 PR rtl-optimization/81553
352 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
353 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
354 is a shift where C1 has different mode than the whole shift, use C1's
355 mode for MULT rather than the shift's mode.
358 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
359 builtins not enabled in the currently selected ISA.
361 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
363 PR tree-optimization/71026
364 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
365 (insert_reciprocals): Change to insert reciprocals before a division
366 by a square and to insert the square of a reciprocal.
367 (execute_cse_reciprocals_1): Change to consider division by a square.
368 (register_division_in): Add importance parameter.
370 2017-11-24 Richard Biener <rguenther@suse.de>
372 PR tree-optimization/82402
373 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
374 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
376 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
378 * match.pd (0-ptr): New transformation.
380 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
383 * ipa-inline.c (inline_small_functions): Set current badnes correctly
384 when skipping checking.
386 2017-11-24 Richard Biener <rguenther@suse.de>
388 PR tree-optimization/83128
389 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
390 (vn_reference_lookup_3): Likewise.
392 2017-11-24 Jakub Jelinek <jakub@redhat.com>
395 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
396 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
399 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
400 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
401 check for the latter with gimple_call_builtin_p. Do not handle
402 BUILT_IN_STPNCPY_CHK which is not a pass through call.
404 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
406 * config/arm/arm_neon.h: Fix pragma GCC push_options before
409 2017-11-23 Julia Koval <julia.koval@intel.com>
411 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
412 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
413 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
414 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
415 _mm512_maskz_expandloadu_epi16): New intrinsics.
416 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
417 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
418 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
419 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
420 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
421 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
422 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
423 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
424 _mm256_maskz_expandloadu_epi8): New intrinsics.
425 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
426 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
427 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
428 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
429 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
430 * config/i386/sse.md (VI248_VLBW): New iterator.
431 (expand<mode>_mask, expand<mode>_maskz): New patterns.
433 2017-11-23 Julia Koval <julia.koval@intel.com>
435 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
436 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
437 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
438 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
439 _mm512_mask_compressstoreu_epi16): New.
440 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
441 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
442 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
443 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
444 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
445 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
446 _mm256_mask_compressstoreu_epi8): New.
447 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
448 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
449 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
450 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
451 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
452 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
453 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
454 __builtin_ia32_compresshi128_mask,
455 __builtin_ia32_compressstoreuqi512_mask,
456 __builtin_ia32_compressstoreuhi512_mask,
457 __builtin_ia32_compressstoreuqi256_mask,
458 __builtin_ia32_compressstoreuqi128_mask,
459 __builtin_ia32_compressstoreuhi256_mask,
460 __builtin_ia32_compressstoreuhi128_mask): New builtins.
461 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
463 (ix86_expand_special_args_builtin): Handle new types.
464 (s4fma_expand): Handle new builtin array.
465 * config/i386/immintrin.h: Include new headers.
466 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
467 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
469 2017-11-23 Jakub Jelinek <jakub@redhat.com>
471 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
472 fixes. Declare temp and g variables at the top in order to avoid
473 {} in most of the cases.
475 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
477 * match.pd (ptr-0): New transformation.
479 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
481 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
482 (struct addr_mode_cost_table): New.
483 (struct tune_params): Add field addr_mode_costs.
484 * config/arm/arm.c (generic_addr_mode_costs): New.
485 (arm_slowmul_tune): Initialise addr_mode_costs field.
486 (arm_fastmul_tune): Likewise.
487 (arm_strongarm_tune): Likewise.
488 (arm_xscale_tune): Likewise.
489 (arm_9e_tune): Likewise.
490 (arm_marvell_pj4_tune): Likewise.
491 (arm_v6t2_tune): Likewise.
492 (arm_cortex_tune): Likewise.
493 (arm_cortex_a8_tune): Likewise.
494 (arm_cortex_a7_tune): Likewise.
495 (arm_cortex_a15_tune): Likewise.
496 (arm_cortex_a35_tune): Likewise.
497 (arm_cortex_a53_tune): Likewise.
498 (arm_cortex_a57_tune): Likewise.
499 (arm_exynosm1_tune): Likewise.
500 (arm_xgene1_tune): Likewise.
501 (arm_cortex_a5_tune): Likewise.
502 (arm_cortex_a9_tune): Likewise.
503 (arm_cortex_a12_tune): Likewise.
504 (arm_cortex_a73_tune): Likewise.
505 (arm_v7m_tune): Likewise.
506 (arm_cortex_m7_tune): Likewise.
507 (arm_v6m_tune): Likewise.
508 (arm_fa726te_tune): Likewise.
509 (arm_mem_costs): Use table lookup to calculate cost of addressing
512 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
514 * config/arm/arm.c (arm_mem_costs): New function.
515 (arm_rtx_costs_internal): Use arm_mem_costs.
517 2017-11-23 Mark Wielaard <mark@klomp.org>
519 * dwarf2out.c (init_sections_and_labels): Use generation to create
520 unique ranges_section_label and ranges_base_label. Return generation.
521 (output_rnglists): Add generation argument. Use generation to create
522 unique ranges labels.
523 (dwarf2out_finish): Get generation from init_sections_and_labels
524 and pass generation to output_rnglists.
526 2017-11-23 Mike Stump <mikestump@comcast.net>
527 Eric Botcazou <ebotcazou@adacore.com>
529 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
530 * cfgloop.h (struct loop): Add unroll field.
531 * function.h (struct function): Add has_unroll bitfield.
532 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
533 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
534 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
535 (pass_rtl_unroll_loops::gate): Likewise.
536 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
537 for which loop->unroll==1.
538 (decide_unroll_constant_iterations): Use note for consistency and
539 take loop->unroll into account. Return early if loop->unroll is set.
540 Fix thinko in existing test.
541 (decide_unroll_runtime_iterations): Use note for consistency and
542 take loop->unroll into account.
543 (decide_unroll_stupid): Likewise.
544 * lto-streamer-in.c (input_cfg): Read loop->unroll.
545 * lto-streamer-out.c (output_cfg): Write loop->unroll.
546 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
548 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
549 (print_loop): Print loop->unroll if set.
550 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
551 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
552 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
554 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
555 loop->unroll is set and smaller than the trip count. Otherwise bypass
556 entirely the heuristics if loop->unroll is set. Remove dead note.
557 Fix off-by-one bug in other note.
558 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
559 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
561 (tree_unroll_loops_completely): Make static.
562 (pass_complete_unroll::execute): Use correct type for variable.
563 (pass_complete_unrolli::execute): Fix formatting.
564 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
566 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
568 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
569 enable when TARGET_PREFER_AVX128 is set.
571 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
573 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
575 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
576 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
578 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
580 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
581 frequencies. Fix estimation of aggregate parameters.
583 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
585 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
588 2017-11-23 Tom de Vries <tom@codesourcery.com>
590 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
592 2017-11-23 Tom de Vries <tom@codesourcery.com>
594 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
597 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
600 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
601 sibcall_value_pcrel_fdpic): Use local variable instead of
603 (calli_tbr_rel): Add missing operand 2.
604 (call_valuei_tbr_rel): Add missing operand 3.
606 2017-11-23 Jakub Jelinek <jakub@redhat.com>
609 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
610 bitpos/bitsize covering the whole destination, use store_expr only if
611 the complex mode is the same. Otherwise, use expand_normal and if
612 it returns CONCAT, subreg each part separately instead of trying to
613 subreg the whole result.
615 2017-11-23 Richard Biener <rguenther@suse.de>
617 PR tree-optimization/23094
618 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
619 come here from walking over backedges in the first iteration.
620 (vn_reference_lookup_3): Skip clobbers that store the same value.
622 2017-11-23 Richard Biener <rguenther@suse.de>
624 PR tree-optimization/81403
625 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
626 a block we need a leader relative to.
627 (phi_translate_1): For nary processing require a leader from
628 get_representative_for given we run expression simplification
629 using match-and-simplify. Remove previous fix.
631 2017-11-22 Jeff Law <law@redhat.com>
633 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
634 Use new method allocate_value_range rather than accessing the
635 vrp_value_range_pool data member directly.
636 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
637 to use extract_range_from_stmt method to avoid need for
638 extract_range_from_assignment method.
639 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
640 method rather than setting values_propgated data member directly.
641 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
642 and values propagated data members and extract_range_from_assignment
643 method. Reorder private data members to conform to standards.
644 Add new methods set_lattice_propagation_complete and
645 allocate_value_range.
647 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
649 PR rtl-optimization/83030
650 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
651 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
652 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
653 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
654 CROSSING_JUMP_P flag is already set before setting it.
655 * cfgrtl.c (fixup_partition_crossing): Likewise.
656 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
659 2017-11-22 Jakub Jelinek <jakub@redhat.com>
661 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
662 Handle the case where both arguments are using gen_const_vec_series.
664 2017-11-22 David Malcolm <dmalcolm@redhat.com>
667 * pretty-print.c (pp_format): Move quoting implementation to
668 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
669 to pass address of "quote" local.
670 (pp_begin_quote): New function.
671 (pp_end_quote): New function.
672 * pretty-print.h (printer_fn): Convert penultimate param from bool
674 (pp_begin_quote): New decl.
675 (pp_end_quote): New decl.
676 * tree-diagnostic.c (default_tree_printer): Convert penultimate
677 param from bool to bool *.
678 * tree-diagnostic.h (default_tree_printer): Likewise.
680 2017-11-22 Jeff Law <law@redhat.com>
682 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
683 Initialize vr_values.
684 (evrp_range_analyzer::try_find_new_range): Call methods attached to
685 vr_values via vr_values class instance rather than delegators.
686 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
687 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
688 (evrp_range_analyzer::push_value_range): Likewise.
689 (evrp_range_analyzer::pop_value_range): Likewise.
690 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
691 most delegators. Those remaining are exposed as public interfaces.
692 Make vr_values a pointer and private.
693 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
695 (evrp_range_analyzer::get_vr_value): New method.
696 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
697 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
698 (class evrp_dom_walker): Attach evrp_folder class, initialize
699 it in the ctor. Remove temporary delegators.
700 (evrp_dom_walker::before_dom_children): Call methods in attached
701 evrp_range_analyzer class via class instance pointer. Use
702 free value_range_constant_singleton to remove need for
703 op_with_constant_singleton_value delegator method. Do not
704 create a vrp_prop class instance for every call! Narrow
705 scope of a couple variables.
706 (evrp_dom_walker::cleanup): Call methods in attached
707 evrp_range_analyzer class via class instance pointer.
708 * vr-values.h (class vr_values): Privatize many methods and
711 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
713 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
714 * cfgexpand.c (expand_debug_expr): Remove handling for them.
715 * expr.c (expand_expr_real_2): Likewise.
716 * fold-const.c (const_unop): Likewise.
717 * optabs-tree.c (optab_for_tree_code): Likewise.
718 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
719 * tree-inline.c (estimate_operator_cost): Likewise.
720 * tree-pretty-print.c (dump_generic_node): Likewise.
721 (op_code_prio): Likewise.
722 (op_symbol_code): Likewise.
723 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
724 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
725 * internal-fn.c (direct_internal_fn_optab): New function.
726 (direct_internal_fn_array, direct_internal_fn_supported_p
727 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
728 * fold-const-call.c (fold_const_reduction): New function.
729 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
731 * tree-vect-loop.c: Include internal-fn.h.
732 (reduction_code_for_scalar_code): Rename to...
733 (reduction_fn_for_scalar_code): ...this and return an internal
735 (vect_model_reduction_cost): Take an internal_fn rather than
737 (vect_create_epilog_for_reduction): Likewise. Build calls rather
739 (vectorizable_reduction): Use internal functions rather than tree
740 codes for the reduction operation. Update calls to the functions
742 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
743 Use calls to internal functions rather than REDUC tree codes.
744 * config/aarch64/aarch64-simd.md: Update comment accordingly.
746 2017-11-22 Olivier Hainque <hainque@adacore.com>
748 * config/vxworks.c (vxworks_override_options): Pick default
749 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
750 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
751 DWARF_GNAT_ENCODINGS_DEFAULT.
752 * config/vxworksae.h: Likewise.
754 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
756 PR tree-optimization/83104
757 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
760 2017-11-22 Marek Polacek <polacek@redhat.com>
761 H.J. Lu <hongjiu.lu@intel.com>
762 Jason Merrill <jason@redhat.com>
767 * calls.c (initialize_argument_information): Call
768 warn_parameter_passing_abi target hook.
769 (store_one_arg): Use 0 for empty record size. Don't push 0 size
771 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
772 * common.opt: Update -fabi-version description.
773 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
774 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
776 (ix86_is_empty_record): New function.
777 (ix86_warn_parameter_passing_abi): New function.
778 (TARGET_EMPTY_RECORD_P): Redefine.
779 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
780 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
781 * doc/tm.texi: Regenerated.
782 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
783 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
784 * dwarf2out.c (get_ultimate_context): Move to tree.c.
785 * explow.c (hard_function_value): Call arg_int_size_in_bytes
786 instead of int_size_in_bytes.
787 * expr.c (copy_blkmode_to_reg): Likewise.
788 * function.c (aggregate_value_p): Return 0 for empty types.
789 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
790 (locate_and_pad_parm): Call arg size_in_bytes instead
792 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
793 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
794 * target.def (empty_record_p, warn_parameter_passing_abi): New target
796 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
797 (std_gimplify_va_arg_expr): Skip empty records. Call
798 arg_size_in_bytes instead size_in_bytes.
799 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
800 * tree-core.h (tree_type_common): Add empty_flag.
801 (tree_decl_common): Update comments.
802 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
804 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
805 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
807 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
808 * tree.c (default_is_empty_type): New function.
809 (default_is_empty_record): New function.
810 (arg_int_size_in_bytes): New function.
811 (arg_size_in_bytes): New function.
812 (get_ultimate_context): New function.
813 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
814 TRANSLATION_UNIT_WARN_EMPTY_P.
815 (default_is_empty_record, arg_int_size_in_bytes,
816 arg_size_in_bytes, get_ultimate_context): Declare.
818 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
820 * config/arm/arm.c (cmse_clear_registers): New function.
821 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
822 code by call to cmse_clear_registers.
823 (cmse_nonsecure_entry_clear_before_return): Likewise.
825 2017-11-22 Tamar Christina <tamar.christina@arm.com>
827 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
828 (vdot_s32, vdotq_s32): New.
829 (vdot_lane_u32, vdotq_lane_u32): New.
830 (vdot_lane_s32, vdotq_lane_s32): New.
833 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
836 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
837 for unsigned values with fewer HWIs than the precision.
838 (test_overflow): New function.
839 (wide_int_cc_tests): Call it.
841 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
843 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
844 has the same mode class as Pmode.
846 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
848 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
849 padding_bits_to_clear_ptr.
850 (cmse_nonsecure_entry_clear_before_return): Likewise.
852 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
854 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
855 auto_sbitap instead of integer bitfield to control register needing
858 2017-11-22 Jakub Jelinek <jakub@redhat.com>
860 PR tree-optimization/83044
861 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
862 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
863 that need the upper bound. Subtract offset from
864 get_addr_base_and_unit_offset only if positive and subtract it
865 before division by eltsize rather than after it.
868 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
869 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
870 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
874 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
875 expand_binop with *mul_widen_optab, make sure at least one of the
876 operands doesn't have VOIDmode.
879 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
881 PR rtl-optimization/82044
882 PR tree-optimization/82042
883 * dse.c (record_store): Check for overflow.
884 (check_mem_read_rtx): Properly check for overflow if width == -1, call
885 add_wild_read instead of clear_rhs_from_active_local_stores on
886 overflow and log it into dump_file.
888 2017-11-22 Richard Biener <rguenther@suse.de>
890 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
891 fake edges to exit when looking for a place to insert.
892 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
894 (insert_into_preds_of_block): Commit edge insertion immediately,
895 assert that doesn't require new BBs.
896 (fini_pre): Release expressions.
897 (pass_pre::execute): Shuffle things around a bit, if the fn
898 is too large do not compute AVAIL either as this is really the
901 2017-11-22 Richard Biener <rguenther@suse.de>
903 PR tree-optimization/83089
904 * tree-if-conv.c (pass_if_conversion::execute): If anything
905 changed reset SCEV and free the number of iteration estimates.
907 2017-11-21 Martin Sebor <msebor@redhat.com>
909 PR tree-optimization/82945
910 * calls.h (warn_nonstring_bound): Remove unused function.
912 2017-11-21 Martin Sebor <msebor@redhat.com>
914 PR tree-optimization/82945
915 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
916 * calls.h (maybe_warn_nonstring_arg): Declare new function.
917 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
919 (initialize_argument_information): Call maybe_warn_nonstring_arg.
920 * calls.h (get_attr_nonstring_decl): Declare new function.
921 * doc/extend.texi (attribute nonstring): Update.
922 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
923 get_attr_nonstring_decl and handle it.
924 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
925 detection of nul-termination.
926 (strlen_to_stridx): Change to a pointer.
927 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
928 (pass_strlen::execute): Same.
930 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
932 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
933 for the new option -mprefer-vector-width=[none|128|256|512].
934 * config/i386/i386.c (ix86_target_string): remove old style options
935 -mprefer-avx256 and make -mprefer-avx128 as alias.
936 (ix86_option_override_internal): Apply defaults for the
937 -mprefer-vector-width=[128|256] option.
938 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
939 Implement macros to work with -mprefer-vector-width=.
940 * config/i386/i386.opt: Implemented option
941 -mprefer-vector-width=[none|128|256|512].
942 * doc/invoke.texi: Documentation for
943 -mprefer-vector-width=[none|128|256|512].
945 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
947 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
949 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
951 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
953 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
955 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
957 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
958 * expr.c (expand_expr_real_2): Likewise.
959 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
960 fold_binary_loc): Likewise.
961 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
962 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
963 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
964 MINUS_EXPR transformations.
965 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
966 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
967 * tree-inline.c (estimate_operator_cost): Likewise.
968 * tree-pretty-print.c (dump_generic_node, op_code_prio,
969 op_symbol_code): Likewise.
970 * tree-vect-stmts.c (vectorizable_operation): Likewise.
971 * vr-values.c (extract_range_from_binary_expr): Likewise.
972 * varasm.c (initializer_constant_valid_p_1): Likewise.
973 * tree.def: New tree code POINTER_DIFF_EXPR.
975 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
977 * config/i386/i386.md (*bswap<mode>2_movbe): Add
978 integer suffix to movbe mnemonic.
979 (*bswaphi2_movbe): Ditto.
980 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
982 2017-11-21 Jakub Jelinek <jakub@redhat.com>
985 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
986 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
987 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
989 2017-11-21 Martin Liska <mliska@suse.cz>
991 * tree-inline.c (expand_call_inline): Remove not needed
994 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
995 Jakub Jelinek <jakub@redhat.com>
998 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
999 Remove static keyword from f variable.
1001 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1003 PR tree-optimization/83086
1004 * gimple-ssa-store-merging.c
1005 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1006 rather than n.base_addr.
1008 2017-11-21 Martin Liska <mliska@suse.cz>
1010 PR rtl-optimization/82044
1011 PR tree-optimization/82042
1012 * dse.c (check_mem_read_rtx): Check for overflow.
1014 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1016 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1019 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1021 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1022 and its corresponding call arg location note.
1024 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1025 Andrew Burgess <andrew.burgess@embecosm.com>
1027 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1029 (arc_return_slot_offset): Likewise.
1030 (arc_eh_return_address_location): New declaration.
1031 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1032 (MUST_SAVE_REGISTER): Add exception handler case.
1033 (MUST_SAVE_RETURN_ADDR): Likewise.
1034 (arc_frame_pointer_required): Likewise.
1035 (arc_frame_pointer_needed): New function.
1036 (arc_compute_frame_size): Changed.
1037 (arc_expand_prologue): Likewise.
1038 (arc_expand_epilogue): Likewise.
1039 (arc_initial_elimination_offset): Likewise.
1040 (arc_return_slot_offset): Delete.
1041 (arc_eh_return_address_location): New function.
1042 (arc_builtin_setjmp_frame_value): Likewise.
1043 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1044 (EH_RETURN_STACKADJ_RTX): Define.
1045 (EH_RETURN_HANDLER_RTX): Likewise.
1046 * config/arc/arc.md (eh_return): Delete.
1048 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1050 * print-tree.h (debug_vec_tree): Remove prototype.
1051 * gdbinit.in (pvt): Remove macro.
1053 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1055 PR tree-optimization/83047
1056 * gimple-ssa-store-merging.c
1057 (imm_store_chain_info::output_merged_store): If the loads with the
1058 same vuse are in different basic blocks, for load_gsi pick a load
1059 location that is dominated by the other loads.
1062 * config/i386/i386.c (ix86_memmodel_check): Start
1063 -Winvalid-memory-model diagnostics with lowercase letter.
1066 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1067 set in_cold_section_p to the partition containing loc_list->first.
1068 When seeing loc_list->last_before_switch node, update secname and
1069 perform range_across_switch second partition handling only after that.
1072 * run-rtl-passes.c: Include debug.h.
1073 (run_rtl_passes): Call debug_hooks->assembly_start.
1074 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1078 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1079 OPTAB_DIRECT in calls to expand_simple_binop.
1081 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1084 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1086 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1089 * spellcheck.h (best_match::blithely_get_best_candidate): New
1092 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1094 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1095 with lower case letter.
1097 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1099 * config/i386/i386.md (bswaphi2): New expander.
1100 (*bswaphi2_movbe): New insn pattern.
1101 (bswaphi -> rorhi pepehole2): New peephole pattern.
1103 2017-11-20 Jeff Law <law@redhat.com>
1105 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1106 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1107 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1108 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1109 gimple-ssa-evrp-analyze.h.
1111 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1112 set BB_VISITED here.
1113 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1115 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1117 * tree-predcom.c: Add general comment on Store-Store chains.
1118 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1120 (get_chain_last_ref_at): Rename into...
1121 (get_chain_last_write_at): ...this.
1122 (get_chain_last_write_before_load): New function.
1123 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1124 CT_STORE_STORE when write reference is added.
1125 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1126 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1127 (initialize_root_vars_store_elim_1): Ditto.
1128 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1129 definition is created.
1130 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1131 chain by replacing it with dominant stored value.
1133 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1135 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1137 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1139 * vr-values.c (extract_range_from_binary_expr): Use a full range
1142 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1144 * config/arm/arm.md (R4_REGNUM): Define constant.
1145 (nonsecure_call_internal): Remove r4 clobber.
1146 (nonsecure_call_value_internal): Likewise.
1147 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1148 clobber and resequence match_operands.
1149 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1150 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1151 (nonsecure_call_value_reg_thumb2): Likewise.
1153 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1155 PR tree-optimization/78821
1156 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1157 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1158 to the address of the base rather than the base itself.
1159 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1160 (find_bswap_or_nop_finalize): New function.
1161 (find_bswap_or_nop): Use it.
1162 (bswap_replace): Return a tree rather than bool, change first
1163 argument from gimple * to gimple_stmt_iterator, allow inserting
1164 into an empty sequence, allow ins_stmt to be NULL - then emit
1165 all stmts into gsi. Fix up MEM_REF address gimplification.
1166 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1167 (struct store_immediate_info): Add N and INS_STMT non-static
1169 (store_immediate_info::store_immediate_info): Initialize them
1170 from newly added ctor args.
1171 (merged_store_group::apply_stores): Formatting fixes. Sort by
1173 (stmts_may_clobber_ref_p): For stores call also
1174 refs_anti_dependent_p.
1175 (gather_bswap_load_refs): New function.
1176 (imm_store_chain_info::try_coalesce_bswap): New method.
1177 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1178 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1179 (imm_store_chain_info::output_merged_store): Fail if number of
1180 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1181 and NOP_EXPR rhs_code.
1182 (pass_store_merging::process_store): Compute n and ins_stmt, if
1183 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1184 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1186 (pass_store_merging::execute): Calculate dominators.
1188 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1189 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1190 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1191 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1192 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1193 class pass_optimize_bswap, bswap_replace,
1194 pass_optimize_bswap::execute): Moved to ...
1195 * gimple-ssa-store-merging.c: ... this file.
1196 Include optabs-tree.h.
1197 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1198 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1199 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1200 anonymous namespace, remove static keywords.
1201 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1202 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1204 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1207 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1209 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1211 * vec.h (debug_helper): New function.
1212 (DEFINE_DEBUG_VEC): New macro.
1213 * hash-set.h (debug_helper): New function.
1214 (DEFINE_DEBUG_HASH_SET): New macro.
1215 * cfg.c (debug_slim (edge)): New function.
1216 Call DEFINE_DEBUG_VEC for edges.
1217 Call DEFINE_DEBUG_HASH_SET for edges.
1218 * cfghooks.c (debug_slim (basic_block)): New function.
1219 Call DEFINE_DEBUG_VEC for basic blocks.
1220 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1221 * print-tree.c (debug_slim): New function to handle trees.
1222 Call DEFINE_DEBUG_VEC for trees.
1223 Call DEFINE_DEBUG_HASH_SET for trees.
1224 (debug (vec<tree, va_gc>) &): Remove.
1225 (debug (<vec<tree, va_gc>) *): Remove.
1226 * print-rtl.c (debug_slim): New function to handle const_rtx.
1227 Call DEFINE_DEBUG_VEC for rtx_def.
1228 Call DEFINE_DEBUG_VEC for rtx_insn.
1229 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1230 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1231 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1232 (debug (vec<rtx_insn *> *ptr): Remove.
1233 (debug_insn_vector): Remove.
1234 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1236 2017-11-20 Tom de Vries <tom@codesourcery.com>
1238 PR rtl-optimization/82020
1239 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1240 IF_THEN_ELSE condition.
1242 2017-11-19 Jeff Law <law@redhat.com>
1244 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1245 of degenerates resulting from ignoring an edge.
1247 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1250 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1252 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1255 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1256 for uninitialied counts.
1258 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1261 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1263 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1266 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1268 2017-11-19 Tom de Vries <tom@codesourcery.com>
1270 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1273 2017-11-19 Tom de Vries <tom@codesourcery.com>
1275 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1278 2017-11-19 Tom de Vries <tom@codesourcery.com>
1280 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1283 2017-11-19 Tom de Vries <tom@codesourcery.com>
1285 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1287 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1289 2017-11-19 Tom de Vries <tom@codesourcery.com>
1291 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1293 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1294 semicolon after MCORE_EXPORT_NAME call.
1296 2017-11-19 Tom de Vries <tom@codesourcery.com>
1299 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1300 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1302 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1304 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1305 (cgraph_edge::clone): Cleanup updating of profile.
1306 * ipa-cp.c (update_profiling_info): Likewise.
1307 * ipa-inline-transform.c (inline_transform): Likewise.
1308 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1309 * ipa-split.c (execute_split_functions): Do not split when function
1311 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1312 * profile-count.c (profile_count::dump): Add global0.
1313 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1315 (profile_count::to_sreal_scale): Likewise.
1316 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1317 (profile_count::combine_with_ipa_count): New function.
1318 * profile-count.h (profile_guessed_global0adjusted): New.
1319 (profile_count::adjusted_zero): New.
1320 (profile_count::global0adjusted): New.
1321 (profile_count::combine_with_ipa_count): New.
1322 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1323 correct profile of return block of split functions.
1324 (copy_cfg_body): Remove unused profile_count.
1325 (copy_body): Likewise.
1326 (expand_call_inline): Update.
1327 (tree_function_versioning): Update.
1329 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1331 * hash-set.h (hash_set::empty): New.
1332 * tree-ssa-threadbackward.h: Delete.
1333 * tree-ssa-threadbackward.c (class thread_jumps): New.
1334 Move max_threaded_paths into class.
1335 (fsm_find_thread_path): Remove arguments that are now in class.
1336 (profitable_jump_thread_path): Rename to...
1337 (thread_jumps::profitable_jump_thread_path): ...this.
1338 (convert_and_register_jump_thread_path): Rename to...
1339 (thread_jumps::convert_and_register_current_path): ...this.
1340 (check_subpath_and_update_thread_path): Rename to...
1341 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1342 (register_jump_thread_path_if_profitable): Rename to...
1343 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1344 (handle_phi): Rename to...
1345 (thread_jumps::handle_phi): ...this.
1346 (handle_assignment): Rename to...
1347 (thread_jumps::handle_assignment): ...this.
1348 (fsm_find_control_statement_thread_paths): Rename to...
1349 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1350 (find_jump_threads_backwards): Rename to...
1351 (thread_jumps::find_jump_threads_backwards): ...this.
1352 Initialize path local data.
1353 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1354 from within thread_jumps class.
1355 (pass_early_thread_jumps::execute): Same.
1357 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1359 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1361 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1363 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1364 to canonical location.
1366 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1369 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1371 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1373 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1375 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1376 rcrt1.o%s/grcrt1.o%s for -static-pie.
1378 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1380 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1381 ix86_shift_rotate_cost): Break out from ...
1382 (ix86_rtx_costs): ... here.
1383 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1386 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1388 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1389 when entry block was promoted unlikely.
1390 (estimate_bb_frequencies): Increase frequency scale.
1391 * profile-count.h (profile_count): Export precision info.
1393 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1395 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1396 disturbin profile of entry block.
1398 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1400 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1403 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1405 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1406 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1408 2017-11-17 Jeff Law <law@redhat.com>
1410 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1411 from evrp_dom_walker class. Various methods moved into new class.
1412 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1413 (evrp_range_analyzer::enter): New method.
1414 (evrp_range_analyzer::leave): New method.
1415 (evrp_dom_walker): Remove delegators no longer needed by this class.
1416 Replace vr_values data member with evrp_range_analyzer
1418 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1419 method extracted from evrp_dom_walker::before_dom_children.
1420 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1421 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1423 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1424 Add private copy constructor and move assignment operators.
1425 Privatize methods and class data where trivially possible.
1426 (evrp_dom_walker::cleanup): New function, extracted from
1427 execute_early_vrp. Simplify access to class data.
1429 * vr-values.h (get_output_for_vrp): Prototype.
1430 * vr-values.c (get_output_for_vrp): New function extracted from
1431 vrp_visit_assignment_or_call and extract_range_from_stmt.
1432 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1434 2017-11-17 Luis Machado <luis.machado@linaro.org>
1436 * config/aarch64/aarch64.c
1437 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1438 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1439 tune_params::AUTOPREFETCHER_WEAK.
1441 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1444 * config/arm/arm.c (arm_valid_target_attribute_rec):
1445 Parse "arch=" and "+<ext>".
1446 (arm_valid_target_attribute_tree): Re-init global options.
1447 (arm_option_override): Make non-static.
1448 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1449 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1450 (__ARM_FEATURE_CRC32): Support undef.
1451 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1452 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1454 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1456 * gdbinit.in (break-on-diagnostic): New command.
1458 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1460 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1461 return to indirect jump for EH return if control-flow
1462 protection is enabled. Change explicit 'false' argument in
1463 pro_epilogue_adjust_stack with a value of
1465 * config/i386/i386.md (simple_return_indirect_internal):
1466 Remove SImode restriction to support 64-bit.
1468 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1470 * combine.c (added_notes_insn): New.
1471 (try_combine): Handle added_notes_insn like added_links_insn.
1472 Rewrite return value code.
1473 (distribute_notes): Set added_notes_insn to the earliest insn we added
1476 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1478 PR rtl-optimization/82621
1479 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1480 dest of one of those SETs is unused.
1482 2017-11-17 Richard Biener <rguenther@suse.de>
1485 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1486 * tree-pretty-print.c (dump_generic_node): Handle
1487 annot_expr_parallel_kind.
1488 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1489 * gimplify.c (gimple_boolify): Likewise.
1491 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1493 * config.gcc (extra_headers): Add cet.h for x86 targets.
1494 * config/i386/cet.h: New file.
1495 * doc/install.texi: Add --enable-cet/--disable-cet.
1497 2017-11-17 Richard Biener <rguenther@suse.de>
1499 PR tree-optimization/83017
1500 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1501 (gen_parallel_loop): Properly count iterations.
1502 (parallelize_loops): Handle loop->can_be_parallel independent
1503 of flag_loop_parallelize_all. Make static profitability test match
1505 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1506 * invoke.texi (parloops-min-per-thread): Document.
1508 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1510 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1511 upstreaming review comments.
1513 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1515 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1516 with fast unaligned access.
1517 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1519 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1521 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1522 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1524 2017-11-17 Richard Biener <rguenther@suse.de>
1526 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1527 folding of references.
1529 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1532 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1533 of replacing call to strncmp with corresponding call to strcmp when
1536 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1538 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1539 option prefer-avx256 for skylake-avx512 configuration.
1540 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1541 (get_builtin_code_for_version): Ditto.
1543 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1544 Monk Chiang <sh.chiang04@gmail.com>
1546 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1547 (FIXED_REGISTERS): Reserve more register numbers.
1548 (CALL_USED_REGISTERS): Likewise.
1549 (REG_ALLOC_ORDER): Likewise.
1550 (REG_CLASS_CONTENTS): Likewise.
1551 (REGISTER_NAMES): Likewise.
1553 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1554 Kito Cheng <kito.cheng@gmail.com>
1556 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1558 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1560 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1562 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1563 parameter list for vec_splats.
1565 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1567 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1569 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1571 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1574 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1575 with the required version of automake 1.11.6.
1577 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1579 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1581 (DU_C2_3_power9): Correct reservation combinations.
1582 (FP_DIV_power9, VEC_DIV_power9): New.
1583 (power9-alu): Split out rotate/shift...
1584 (power9-rot): ...to here, correct dispatch resource.
1585 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1587 (power9-fp): Correct latency.
1588 (power9-sdiv): Add div/sqrt resource.
1589 (power9-ddiv): Correct latency, add div/sqrt resource.
1590 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1591 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1593 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1595 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1597 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1598 switch statement mapping KF built-ins to TF built-ins if we don't
1599 have the proper ISA 3.0 assembler support.
1601 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1603 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1604 (lower_emutls_function_body): Do not compute it.
1606 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1608 * ipa-split.c (split_bb_info): Turn time to sreal.
1609 (split_point): Likewise.
1610 (dump_split_point): Likewise.
1611 (fine_split_points): Likewise.
1612 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1614 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1616 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1619 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1621 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1623 (expensive_function_p): Remove useless assert.
1624 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1626 2017-11-16 Martin Sebor <msebor@redhat.com>
1628 PR tree-optimization/82588
1629 PR tree-optimization/82583
1630 * tree-vrp.c (check_array_ref): Handle flexible array members,
1631 string literals, and inner indices.
1632 (search_for_addr_array): Add detail to diagnostics.
1634 2017-11-16 Nathan Sidwell <nathan@acm.org>
1638 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1639 (SET_DECL_ASSEMBLER_NAME): Forward to
1640 overwrite_decl_assembler_name.
1641 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1642 (overwrite_decl_assembler_name): Declare.
1643 * tree.c (overwrite_decl_assembler_name): New.
1644 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1645 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1646 (LANG_HOOKS_INITIALIZER): Add it.
1647 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1648 * langhooks.c (lhd_set_decl_assembler_name): Use
1649 SET_DECL_ASSEMBLER_NAME.
1650 (lhd_overwrite_decl_assembler_name): Default implementation.
1652 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1653 Jackson Woodruff <jackson.woodruff@arm.com>
1655 PR tree-optimization/71026
1656 * match.pd: Canonicalize constant multiplies in division.
1658 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1660 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1661 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1662 Remove OPT_fomit_frame_pointer entry.
1663 * common/config/arc/arc-common.c: Likewise.
1664 * common/config/arm/arm-common.c: Likewise.
1665 * common/config/avr/avr-common.c: Likewise.
1666 * common/config/c6x/c6x-common.c: Likewise.
1667 * common/config/cr16/cr16-common.c: Likewise.
1668 * common/config/cris/cris-common.c: Likewise.
1669 * common/config/epiphany/epiphany-common.c: Likewise.
1670 * common/config/fr30/fr30-common.c: Likewise.
1671 * common/config/frv/frv-common.c: Likewise.
1672 * common/config/ia64/ia64-common.c: Likewise.
1673 * common/config/iq2000/iq2000-common.c: Likewise.
1674 * common/config/lm32/lm32-common.c: Likewise.
1675 * common/config/m32r/m32r-common.c: Likewise.
1676 * common/config/mcore/mcore-common.c: Likewise.
1677 * common/config/microblaze/microblaze-common.c: Likewise.
1678 * common/config/mips/mips-common.c: Likewise.
1679 * common/config/mmix/mmix-common.c: Likewise.
1680 * common/config/mn10300/mn10300-common.c: Likewise.
1681 * common/config/nios2/nios2-common.c: Likewise.
1682 * common/config/pa/pa-common.c: Likewise.
1683 * common/config/pdp11/pdp11-common.c: Likewise.
1684 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1685 * common/config/riscv/riscv-common.c: Likewise.
1686 * common/config/rs6000/rs6000-common.c: Likewise.
1687 * common/config/rx/rx-common.c: Likewise.
1688 * common/config/s390/s390-common.c: Likewise.
1689 * common/config/sh/sh-common.c: Likewise.
1690 * common/config/sparc/sparc-common.c: Likewise.
1691 * common/config/tilegx/tilegx-common.c: Likewise.
1692 * common/config/tilepro/tilepro-common.c: Likewise.
1693 * common/config/v850/v850-common.c: Likewise.
1694 * common/config/visium/visium-common.c: Likewise.
1695 * common/config/xstormy16/xstormy16-common.c: Likewise.
1696 * common/config/xtensa/xtensa-common.c: Likewise.
1697 * invoke.texi (-fomit-frame-pointer): Update documentation.
1699 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1701 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1703 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1705 * cfg.c (scale_bbs_frequencies_int,
1706 cale_bbs_frequencies_gcov_type): Remove.
1707 * cfg.h (scale_bbs_frequencies_int,
1708 cale_bbs_frequencies_gcov_type): Remove.
1710 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1712 * tree-ssa-loop-manip.c
1713 (scale_dominated_blocks_in_loop): Update to profile counts.
1714 (tree_transform_and_unroll_loop): Likewise.
1716 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1718 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1719 scale_bbs_frequencies_int.
1721 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1723 * final.c (compute_alignments): Use counts rather than frequencies.
1725 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1727 * cfgloopanal.c: Include sreal.h
1728 (average_num_loop_insns): Use counts and sreal for accounting.
1730 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1732 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1735 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1736 Kito Cheng <kito.cheng@gmail.com>
1738 * config/nds32/constraints.md: Provide more constraints.
1739 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1740 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1741 support constraints usage.
1743 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1745 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1747 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1748 Kito Cheng <kito.cheng@gmail.com>
1750 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1751 * config/nds32/nds32.opt: Refine the layout.
1752 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1753 TARGET_EXT_STRING): Support new options.
1754 * config/nds32/nds32.h: Likewise.
1755 * config/nds32/nds32.md: Likewise.
1756 * config/nds32/nds32-predicates.c: Likewise.
1757 * config/nds32/constraints.md: Likewise.
1758 * common/config/nds32/nds32-common.c: Likewise.
1760 2017-11-16 Julia Koval <julia.koval@intel.com>
1763 * config/i386/gfniintrin.h: Add sse check.
1764 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1766 2017-11-16 Julia Koval <julia.koval@intel.com>
1768 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1769 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1770 (ix86_handle_option): Handle -mavx512vbmi2.
1771 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1772 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1773 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1774 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1775 (ix86_valid_target_attribute_inner_p): Ditto.
1776 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1777 * config/i386/i386.opt (mavx512vbmi2): New option.
1778 * doc/invoke.texi: Add new option.
1780 2017-11-16 Julia Koval <julia.koval@intel.com>
1782 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1783 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1784 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1785 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1786 _mm512_gf2p8mul_epi8): New intrinsics.
1787 * config/i386/i386-builtin-types.def
1788 (V64QI_FTYPE_V64QI_V64QI): New type.
1789 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1790 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1791 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1792 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1793 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1794 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1796 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1798 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1799 explicitly as a stream of bytes.
1801 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1803 * config/rs6000/altivec.h (vec_xst_be): New #define.
1804 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1805 and externalize from *altivec_vperm_<mode>_internal.
1806 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1808 (XL_BE_V8HI): Likewise.
1809 (XL_BE_V4SI): Likewise.
1810 (XL_BE_V4SI): Likewise.
1811 (XL_BE_V2DI): Likewise.
1812 (XL_BE_V4SF): Likewise.
1813 (XL_BE_V2DF): Likewise.
1814 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1815 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1816 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1817 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1818 VSX_BUILTIN_VEC_XST_BE.
1819 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1820 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1822 (altivec_init_builtins): Replace conditional calls to def_builtin
1823 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1824 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1825 with unconditional calls. Remove calls to def_builtin for
1826 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1827 __builtin_vec_xst_be.
1828 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1829 to define_expand, and add alternate RTL generation for P8.
1830 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1831 vsx_ld_elemrev_v8hi.
1832 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1833 add alternate RTL generation for P8.
1834 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1835 vsx_ld_elemrev_v16qi.
1836 (vsx_st_elemrev_v8hi): Convert define_insn
1837 to define_expand, and add alternate RTL generation for P8.
1838 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1839 vsx_st_elemrev_v8hi.
1840 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1841 add alternate RTL generation for P8.
1842 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1843 vsx_st_elemrev_v16qi.
1845 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1848 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1849 TARGET_AVX512ER check.
1850 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1851 neither -mzeroupper nor -mno-zeroupper is used and
1852 TARGET_EMIT_VZEROUPPER is set.
1853 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1854 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1856 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1858 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1859 folding of vector compares.
1860 (fold_build_vec_cmp): New helper function.
1861 (fold_compare_helper): New helper function.
1862 (builtin_function_type): Add compare builtins to the list of functions
1863 having unsigned arguments. Cosmetic updates to comment indentation.
1864 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1865 the not+eq combination.
1867 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1869 PR tree-optimization/82726
1870 PR tree-optimization/70754
1871 * tree-predcom.c (order_drefs_by_pos): New function.
1872 (combine_chains): Move code setting has_max_use_after to...
1873 (try_combine_chains): ...here. New parameter. Sort combined chains
1874 according to position information.
1875 (tree_predictive_commoning_loop): Update call to above function.
1876 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1878 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1880 PR tree-optimization/82726
1882 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1884 PR tree-optimization/70754
1885 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1886 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1887 combined stmt before it if not NULL.
1888 (combine_chains): Process refs reversely and compute dominance point
1892 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1894 PR tree-optimization/79663
1895 * tree-predcom.c (combine_chains): Process refs in reverse order
1896 only for ZERO length chains, and add explaining comment.
1898 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1900 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1901 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1902 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1904 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1906 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1908 2017-11-15 Martin Liska <mliska@suse.cz>
1910 * tree-cfg.c (pass_warn_function_return::execute):
1911 Compare warn_return_type for greater than zero.
1913 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1917 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1918 to return true on Xeon and not on Xeon Phi.
1919 (ix86_check_avx256_register): Changed to ...
1920 (ix86_check_avx_upper_register): ... this. Add extra check for
1921 VALID_AVX512F_REG_OR_XI_MODE.
1922 (ix86_avx_u128_mode_needed): Changed
1923 ix86_check_avx256_register to ix86_check_avx_upper_register.
1924 (ix86_check_avx256_stores): Changed to ...
1925 (ix86_check_avx_upper_stores): ... this. Changed
1926 ix86_check_avx256_register to ix86_check_avx_upper_register.
1927 (ix86_avx_u128_mode_after): Changed
1928 avx_reg256_found to avx_upper_reg_found. Changed
1929 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1930 (ix86_avx_u128_mode_entry): Changed
1931 ix86_check_avx256_register to ix86_check_avx_upper_register.
1932 (ix86_avx_u128_mode_exit): Ditto.
1933 * config/i386/i386.h: (host_detect_local_cpu): New define.
1935 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1937 * config/arm/xgene1.md (xgene1): Split into automatons
1938 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1939 (xgene1_f_load): Adjust reservations and/or types.
1940 (xgene1_f_store): Likewise.
1941 (xgene1_load_pair): Likewise.
1942 (xgene1_store_pair): Likewise.
1943 (xgene1_fp_load1): Likewise.
1944 (xgene1_load1): Likewise.
1945 (xgene1_store1): Likewise.
1946 (xgene1_move): Likewise.
1947 (xgene1_alu): Likewise.
1948 (xgene1_simd): Likewise.
1949 (xgene1_bfm): Likewise.
1950 (xgene1_neon_load1): Likewise.
1951 (xgene1_neon_store1): Likewise.
1952 (xgene1_neon_logic): Likewise.
1953 (xgene1_neon_st1): Likewise.
1954 (xgene1_neon_ld1r): Likewise.
1955 (xgene1_alu_cond): Added.
1956 (xgene1_shift_reg): Likwise.
1957 (xgene1_bfx): Likewise.
1958 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1960 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1963 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1965 (can_widen_mult_without_libcall): New function.
1966 (expand_mul_overflow): If only checking unsigned mul overflow,
1967 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1968 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1969 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1970 (expand_DIVMOD): Formatting fix.
1971 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1972 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1973 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1975 PR tree-optimization/82977
1976 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1977 constructed temporary to strlen_to_stridx.put.
1979 2017-11-15 Martin Liska <mliska@suse.cz>
1981 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1982 * configure: Regenerate.
1984 2017-11-15 Martin Liska <mliska@suse.cz>
1987 * config/sh/sh-mem.cc: Use proper probability for
1990 2017-11-14 Jeff Law <law@redhat.com>
1992 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1993 the red zone for stack_clash_protection_final_dynamic_probe targets
1994 when the total dynamic stack size is zero bytes.
1996 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1997 blocks is post order.
1999 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2001 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2002 * final.c (rest_of_clean_state): Set it for the
2003 -fcompare-debug dump.
2004 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2005 class when TDF_COMPARE_DEBUG is set.
2007 * dwarf2out.c (gen_producer_string): Discard
2010 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2013 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2014 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2015 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2016 (__TGMATH_REAL_2_3): Remove macros.
2017 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2018 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2020 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2022 (remquo): Define using __TGMATH_REAL_3.
2024 2017-11-14 Jeff Law <law@redhat.com>
2026 * vr-values.c: New file with contents extracted from tree-vrp.c.
2027 * Makefile.in (OBJS): Add vr-values.o
2028 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2029 (set_value_range, set_and_canonicalize_value_range): Likewise.
2030 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2031 (value_range_constant_singleton, symbolic_range_p): Likewise.
2032 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2033 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2034 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2035 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2036 (find_case_label_range, find_case_label_index): Likewise.
2037 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2038 (range_int_cst_singleton_p, value_inside_range): Likewise.
2039 (get_single_symbol): Likewise.
2040 (switch_update): Move structure definition here.
2041 (to_remove_edges, to_update_switch_stmts): Provide externs.
2042 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2043 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2044 (vrp_val_is_min, set_value_range): Likewise.
2045 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2046 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2047 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2048 (range_is_nonnull, range_int_cst_p): Likewwise.
2049 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2050 (get_single_symbol, operand_less_p): Likewise
2051 (compare_values_warnv, compare_values): Likewise.
2052 (value_inside_range, value_range_constant_singleton): Likewise.
2053 (zero_nonzero_bitgs_from_vr): Likewise.
2054 (extract_range_from_binary_expr_1): Likewise.
2055 (overflow_comparison_p): Likewise.
2056 (to_remove_edges, to_update_switch_stmts): Likewise.
2057 (find_case_label-index, find_case_label_range): Likewise.
2058 (switch_update, set_value_range_to_nonnegative): Remove.
2059 (set_value_range_to_truthvalue): Likewise.
2060 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2061 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2062 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2063 (find_case_label_ranges, test_for_singularity): Likewise.
2064 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2065 (x_vr_values): Move to its remaining use site.
2067 2017-11-10 Jeff Law <law@redhat.com>
2069 * vr-values.h (VR_INITIALIZER): Move #define here.
2070 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2071 * Makefile.in (OBJS): Add tree-evrp.o
2072 * tree-vrp.h (assert_info): Move structure definition here.
2073 (set_value_range_to_varying): Prototype.
2074 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2075 (infer_value_range, register_edge_assert_for): Likewise.
2076 (stmt_interesting_for_vrp): Likewise.
2077 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2078 (set_value_range_to_varying): No longer static.
2079 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2080 (infer_value_range, register_edge_assert_for): Likewise.
2082 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2084 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2085 generate the XXBRD instruction.
2087 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2088 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2089 as being compatible if -mabi=ieeelongdouble.
2090 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2091 to setup float128 built-ins with hardware support.
2092 (BU_FLOAT128_HW_2): Likewise.
2093 (BU_FLOAT128_HW_3): Likewise.
2094 (BU_FLOAT128_HW_VSX_1): Likewise.
2095 (BU_FLOAT128_HW_VSX_2): Likewise.
2096 (scalar_extract_expq): Change float128 built-in functions to
2097 accommodate having both KFmode and TFmode functions. Use the
2098 KFmode variant as the default.
2099 (scalar_extract_sigq): Likewise.
2100 (scalar_test_neg_qp): Likewise.
2101 (scalar_insert_exp_q): Likewise.
2102 (scalar_insert_exp_qp): Likewise.
2103 (scalar_test_data_class_qp): Likewise.
2104 (sqrtf128_round_to_odd): Delete processing the round to odd
2105 built-in functions as special built-in functions, and define them
2106 as float128 built-ins. Use the KFmode variant as the default.
2107 (truncf128_round_to_odd): Likewise.
2108 (addf128_round_to_odd): Likewise.
2109 (subf128_round_to_odd): Likewise.
2110 (mulf128_round_to_odd): Likewise.
2111 (divf128_round_to_odd): Likewise.
2112 (fmaf128_round_to_odd): Likewise.
2113 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2114 support for KFmode and TFmode xststdcqp calls.
2115 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2116 point, switch the built-in handlers for the get/set float128
2117 exponent, get float128 mantissa, float128 test built-ins, and the
2118 float128 round to odd built-in functions. Eliminate creating the
2119 float128 round to odd built-in functions as special built-ins.
2120 (rs6000_init_builtins): Eliminate special creation of the float128
2121 round to odd built-in functions.
2122 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2123 function insns to support both TFmode and KFmode variants.
2124 (xsxsigqp_<mode>): Likewise.
2125 (xsiexpqpf_<mode>): Likewise.
2126 (xsiexpqp_<mode>): Likewise.
2127 (xststdcqp_<mode>): Likewise.
2128 (xststdcnegqp_<mode>): Likewise.
2129 (xststdcqp_<mode>): Likewise.
2131 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2133 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2134 unused path_in_freq_ptr parameter.
2135 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2137 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2139 * ipa-inline.c (edge_badness): Dump sreal frequency.
2140 (compute_inlined_call_time): Match natural implementaiton ...
2141 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2142 forgotten division by CGRAPH_FREQ_BASE.
2144 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2146 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2147 Solaris 11. Update comment.
2148 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2150 * config/sol2.h (STARTFILE_SPEC): Likewise.
2151 * configure: Regenerate.
2153 2017-11-14 Carl Love <cel@us.ibm.com>
2155 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2156 le_ and be_ prefixes to swap* variables. Remove
2157 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2160 2017-11-14 Jason Merrill <jason@redhat.com>
2162 Support GTY((cache)) on hash_map.
2163 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2164 (ggc_cache_remove): Override it instead of ggc_mx.
2165 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2166 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2167 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2168 (simple_cache_map_traits): Override maybe_mx.
2169 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2170 (hash_map): Friend gt_cleare_cache.
2171 (gt_cleare_cache): New.
2172 * tree.h (tree_cache_traits): New hash_map traits class.
2173 (tree_cache_map): New typedef.
2175 2017-11-14 Richard Biener <rguenther@suse.de>
2177 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2178 paramter and handling.
2179 (cleanup_control_flow_bb): Likewise.
2180 (cleanup_control_flow_pre): New helper performing a DFS walk
2181 to call cleanup_control_flow_bb in PRE order.
2182 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2183 via cleanup_control_flow_pre.
2185 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2187 * config/aarch64/aarch64-simd.md
2188 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2189 (*aarch64_simd_bsl<mode>_alt): Likewise.
2190 (aarch64_simd_bsldi_internal): New.
2191 (aarch64_simd_bsldi_alt): Likewise.
2193 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2195 * tracer.c (better_p): Do not compare frequencies.
2196 * reg-stack.c (better_edge): Likewise.
2197 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2200 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2202 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2203 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2204 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2205 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2206 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2207 * predict.c (maybe_hot_frequency_p): Inline to ...
2208 (maybe_hot_count_p): ... here; rewrite to counts.
2209 (counts_to_freqs): Rename to ...
2210 (update_max_bb_count): ... this one.
2211 (expensive_function_p): Use counts.
2212 (estimate_bb_frequencies): Update.
2213 (rebuild_frequencies): Update.
2214 * predict.h (counts_to_freqs): Rename to ...
2215 (update_max_bb_count): ... this one.
2216 * profile.c (compute_branch_probabilities): Add debug info
2217 * tree-inline.c (expand_call_inline): Update debug info.
2218 (optimize_inline_calls): Use update_max_bb_count..
2219 (tree_function_versioning): Use update_max_bb_count..
2220 * value-prof.c (gimple_value_profile_transformations):
2221 Do not use update_max_bb_count.
2223 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2225 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2226 always use frequencies.
2228 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2230 * bb-reorder.c: Remove frequencies from comments.
2231 (better_edge_p): Use profile counts.
2232 (find_traces): Dump profile counts.
2233 (rotate_loop): Use profile counts.
2234 (find_traces_1_round): Likewise.
2235 (connect_better_edge_p): Use counts instead of probabilities for
2237 (copy_bb_p): Drop early check for non-0 frequency.
2238 (sanitize_hot_paths): Update comments.
2240 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2242 * ipa-split.c (struct split_point): Add count.
2243 (consider_split): Do not compute incoming frequency; compute incoming
2244 count and store it to split_point.
2245 (split_function): Set count of the call to split part correctly.
2247 2017-11-13 Carl Love <cel@us.ibm.com>
2249 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2251 2017-11-13 Tom Tromey <tom@tromey.com>
2253 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2255 2017-11-13 Carl Love <cel@us.ibm.com>
2257 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2258 Add support for builtins:
2259 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2260 vector {un,}signed {char,int,short},
2261 vector {un,}signed {char,int,short}) arguments.
2262 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2263 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2264 Add BU_P9V_AV_2 expansions for the builtins.
2265 * config/rs6000/altivec.h (vec_first_match_index,
2266 vec_first_mismatch_index, vec_first_match_or_eos_index,
2267 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2268 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2269 new extern declaration.
2270 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2271 * config/rs6000/vsx.md (first_match_index_<mode>,
2272 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2273 first_mismatch_or_eos_index_<mode>): Add define expand.
2274 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2275 * doc/extend.texi: Update the built-in documenation file for the new
2278 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2280 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2281 into the min/max operations for _Float<N> and _Float<N>X types.
2283 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2286 * dwarf2out.c (do_eh_frame): New static variable.
2287 (dwarf2out_begin_prologue): Set it.
2288 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2290 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2292 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2294 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2296 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2299 * debug.h (dwarf2out_do_eh_frame): Declare.
2300 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2301 (dwarf2out_do_frame): Use it.
2302 (dwarf2out_do_cfi_asm): Likewise.
2303 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2304 (dwarf2out_assembly_start): Likewise.
2305 (dwarf2out_begin_prologue): Fix comment.
2306 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2307 if the target needs either debug or unwind DWARF2 info.
2308 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2309 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2310 -ftrapping-math, -ftrapv and -fwrapv.
2312 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2314 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2315 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2316 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2317 (estimate_edge_size_and_time): Likewise.
2318 (ipa_merge_fn_summary_after_inlining): Likewise.
2319 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2320 (compute_uninlined_call_time): Use sreal_frequency.
2321 (compute_inlined_call_time): Likewise.
2322 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2323 * profile-count.c: Include sreal.h.
2324 (profile_count::to_sreal_scale): New.
2325 * profile-count.h: Forward declare sreal.
2326 (profile_count::to_sreal_scale): Declare.
2328 2017-11-13 Nathan Sidwell <nathan@acm.org>
2330 * diagnostic.c (maybe_line_and_column): New.
2331 (diagnostic_get_location_text): Use it.
2332 (diagnostic_report_current_module): Likewise.
2333 (test_diagnostic_get_location_text): Add tests.
2335 2017-11-13 Luis Machado <luis.machado@linaro.org>
2337 * doc/md.texi (Specifying processor pipeline description): Fix
2338 incorrect latency for the div instruction example.
2340 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2342 PR tree-optimization/78821
2343 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2344 that bit_not_p is the same.
2345 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2346 (split_group): Count precisely bit_not_p bits in each statement.
2347 (invert_op): New function.
2348 (imm_store_chain_info::output_merged_store): Use invert_op to
2349 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2350 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2352 2017-11-13 Martin Liska <mliska@suse.cz>
2354 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2355 (struct source_info): Likewise.
2356 (add_branch_counts): Likewise.
2357 (add_line_counts): Likewise.
2358 (function_summary): Likewise.
2359 (output_intermediate_line): Likewise.
2360 (generate_results): Likewise.
2362 2017-11-13 Martin Liska <mliska@suse.cz>
2364 * gcov.c (struct block_info): Remove typedef for block_t.
2365 (struct line_info): Likewise.
2366 (line_info::has_block): Likewise.
2367 (EXIT_BLOCK): Likewise.
2368 (unblock): Likewise.
2369 (circuit): Likewise.
2370 (get_cycles_count): Likewise.
2371 (process_file): Likewise.
2372 (read_graph_file): Likewise.
2373 (solve_flow_graph): Likewise.
2374 (find_exception_blocks): Likewise.
2375 (add_line_counts): Likewise.
2376 (accumulate_line_info): Likewise.
2377 (output_line_details): Likewise.
2379 2017-11-13 Martin Liska <mliska@suse.cz>
2381 * gcov.c (struct arc_info): Remove typedef for arc_t.
2382 (struct line_info): Likewise.
2383 (add_branch_counts): Likewise.
2384 (output_branch_count): Likewise.
2385 (function_info::~function_info): Likewise.
2386 (circuit): Likewise.
2387 (output_intermediate_line): Likewise.
2388 (read_graph_file): Likewise.
2389 (solve_flow_graph): Likewise.
2390 (find_exception_blocks): Likewise.
2391 (add_line_counts): Likewise.
2392 (accumulate_line_info): Likewise.
2393 (output_line_details): Likewise.
2394 (output_function_details): Likewise.
2396 2017-11-13 Martin Liska <mliska@suse.cz>
2398 * gcov.c (struct function_info): Remove typedef for function_t.
2399 (struct source_info): Likewise.
2400 (source_info::get_functions_at_location): Likewise.
2401 (solve_flow_graph): Likewise.
2402 (find_exception_blocks): Likewise.
2403 (add_line_counts): Likewise.
2404 (output_intermediate_file): Likewise.
2405 (process_file): Likewise.
2406 (generate_results): Likewise.
2407 (release_structures): Likewise.
2408 (read_graph_file): Likewise.
2409 (read_count_file): Likewise.
2410 (accumulate_line_counts): Likewise.
2411 (output_lines): Likewise.
2413 2017-11-13 Martin Liska <mliska@suse.cz>
2415 * gcov.c (function_info::function_info): Remove num_counts
2416 and add vector<gcov_type>.
2417 (function_info::~function_info): Use the vector.
2418 (process_file): Likewise.
2419 (read_graph_file): Likewise.
2420 (read_count_file): Likewise.
2421 (solve_flow_graph): Likewise.
2423 2017-11-13 Martin Liska <mliska@suse.cz>
2425 * gcov.c (function_info::is_artificial): New function.
2426 (process_file): Erase all artificial early.
2427 (generate_results): Skip as all artificial are already
2430 2017-11-13 Martin Liska <mliska@suse.cz>
2432 * gcov.c (read_graph_file): Store to global vector of functions.
2433 (read_count_file): Iterate the vector.
2434 (process_file): Likewise.
2435 (generate_results): Likewise.
2436 (release_structures): Likewise.
2438 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2440 PR tree-optimization/82954
2441 * gimple-ssa-store-merging.c
2442 (imm_store_chain_info::coalesce_immediate_stores): If
2443 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2445 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2447 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2448 Upddate call to ENDIAN_LANE_N.
2449 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2450 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2451 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2452 and use aarch64_endian_lane_rtx.
2453 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2455 2017-11-12 Tom de Vries <tom@codesourcery.com>
2457 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2459 2017-11-12 Tom de Vries <tom@codesourcery.com>
2461 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2463 2017-11-12 Tom de Vries <tom@codesourcery.com>
2465 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2467 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2468 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2469 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2471 2017-11-11 Martin Sebor <msebor@redhat.com>
2474 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2477 * prefic.c (translate_name): Replace strncpy with memcpy to
2478 avoid -Wstringop-truncation.
2480 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2482 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2484 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2486 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2488 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2490 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2493 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2495 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2496 enable generating XXBRH if the value is in a vector register.
2497 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2498 value is in a vector register.
2499 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2500 register to register bswap64's instead of doing the GPR sequence
2501 used on previous machines.
2502 (bswapdi2_xxbrd): New insn.
2503 (bswapdi2_reg): Disallow on ISA 3.0.
2504 (register to register bswap64 splitter): Do not split the insn on
2505 ISA 3.0 systems that use XXBRD.
2507 2017-11-10 Martin Sebor <msebor@redhat.com>
2510 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2512 (find_subframework_file): Same.
2514 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2516 * auto-profile.c (afdo_indirect_call): Drop frequency.
2517 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2518 (cgraph_node::create_edge): Drop frequency argument.
2519 (cgraph_node::create_indirect_edge): Drop frequency argument.
2520 (cgraph_edge::make_speculative): Drop frequency arguments.
2521 (cgraph_edge::resolve_speculation): Do not update frequencies
2522 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2523 (cgraph_node::dump): Check consistency in IPA mode.
2524 (cgraph_edge::maybe_hot_p): Use IPA counter.
2525 (cgraph_edge::verify_count_and_frequency): Rename to ...
2526 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2527 (cgraph_node::verify_node): Update.
2528 * cgraph.h (struct cgraph_edge): Drop frequency.
2529 (cgraph_edge::frequency): New function.
2530 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2532 (cgraph_edge::rebuild_edges): Likewise.
2533 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2534 (duplicate_thunk_for_node): Do not pass frequency.
2535 (cgraph_node::create_clone): Scale only counts.
2536 (cgraph_node::create_virtual_clone): Do not pass frequency.
2537 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2538 (cgraph_node::create_version_clone): Do not pass frequency.
2539 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2540 (cgraph_node::expand_thunk): Do not pass frequency.
2541 (cgraph_node::create_wrapper): Do not pass frequency.
2542 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2544 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2545 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2546 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2547 (gather_caller_stats): Use frequency function.
2548 (ipcp_cloning_candidate_p): Use frequency function.
2549 (ipcp_propagate_stage): Use frequency function.
2550 (get_info_about_necessary_edges): Use frequency function.
2551 (update_profiling_info): Update only IPA profile.
2552 (update_specialized_profile): Use frequency functoin.
2553 (perhaps_add_new_callers): Update only IPA profile.
2554 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2555 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2556 (dump_ipa_call_summary): Use frequency function.
2557 (estimate_edge_size_and_time): Use frequency function.
2558 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2559 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2560 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2561 (update_noncloned_counts): ... ths one; scale counts only.
2562 (clone_inlined_nodes): Do not scale frequency.
2563 (inline_call): Do not pass frequency.
2564 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2565 (compute_inlined_call_time): Use IPA profile.
2566 (want_inline_small_function_p): Use IPA profile.
2567 (want_inline_self_recursive_call_p): Use IPA profile.
2568 (edge_badness): Use IPA profile.
2569 (lookup_recursive_calls): Use IPA profile.
2570 (recursive_inlining): Do not pass frequency.
2571 (resolve_noninline_speculation): Do not update frequency.
2572 (inline_small_functions): Collect max of IPA profile.
2573 (dump_overall_stats): Dump IPA porfile.
2574 (dump_inline_stats): Dump IPA porfile.
2575 (ipa_inline): Collect IPA stats.
2576 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2577 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2578 (ipa_propagate_frequency): Use frequency function.
2579 (ipa_profile): Cleanup.
2580 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2581 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2582 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2583 (input_node): Do not stream frequency.
2584 (input_edge): Do not stream frequency.
2585 (merge_profile_summaries): Scale only IPA profiles.
2586 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2587 * predict.c (drop_profile): Do not recompute frequency.
2588 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2589 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2590 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2591 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2592 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2593 * tree-inline.c (copy_bb): Do not scale frequency.
2594 (expand_call_inline): Do not scale frequency.
2595 (tree_function_versioning): Do not scale frequency.
2596 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2598 2017-11-10 Julia Koval <julia.koval@intel.com>
2600 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2601 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2602 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2603 (_mm256_maskz_gf2p8affine_epi64_epi8)
2604 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2605 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2606 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2607 (__builtin_ia32_vgf2p8affineqb_v32qi)
2608 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2609 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2611 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2615 (arm_option_override): Refactor.
2616 (arm_option_reconfigure_globals): New.
2617 (arm_options_perform_arch_sanity_checks): New.
2618 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2620 (arm_options_perform_arch_sanity_checks): Likewise
2622 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2624 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2625 (power9-qpmul): New.
2626 * rs6000/rs6000.md ("type" attr): Add qmul.
2627 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2628 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2629 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2631 2017-11-10 Martin Sebor <msebor@redhat.com>
2634 * builtins.c (compute_objsize): Handle arrays that
2635 compute_builtin_object_size likes to fail for. Make extern.
2636 * builtins.h (compute_objsize): Declare.
2637 (check_strncpy_sizes): New function.
2638 (expand_builtin_strncpy): Call check_strncpy_sizes.
2639 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2640 -Wstringop-truncation.
2641 (gimple_fold_builtin_strncat): Same.
2642 * gimple.c (gimple_build_call_from_tree): Set call location.
2643 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2644 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2645 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2646 (handle_builtin_strlen): Use strlen_to_stridx.
2647 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2649 Use strlen_to_stridx.
2650 (pass_strlen::execute): Release strlen_to_stridx.
2651 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2652 (-Wstringop-truncation): Document new option.
2654 2017-11-10 Martin Liska <mliska@suse.cz>
2656 PR gcov-profile/82702
2657 * gcov.c (main): Handle intermediate files in a different
2659 (get_gcov_intermediate_filename): New function.
2660 (output_gcov_file): Remove support of intermediate files.
2661 (generate_results): Allocate intermediate file.
2662 (release_structures): Clean-up properly fn_end.
2663 (output_intermediate_file): Start iterating with line 1.
2665 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2667 PR tree-optimization/82929
2668 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2669 ops_swapped_p non-static data member.
2670 (store_immediate_info::store_immediate_info): Clear it.
2671 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2672 ops set ops_swapped_p.
2673 (count_multiple_uses): Handle ops_swapped_p.
2675 2017-11-10 Martin Liska <mliska@suse.cz>
2677 * coverage.c (coverage_init): Stream information about
2678 support of has_unexecuted_blocks.
2679 * doc/gcov.texi: Document that.
2680 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2681 * gcov.c (read_graph_file): Likewise.
2682 (output_line_beginning): Fix a small issue with
2685 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2687 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2688 reference of trivial component.
2690 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2693 * gimple-ssa-store-merging.c
2694 (pass_store_merging::terminate_all_aliasing_chains): For
2695 gimple_store_p stmts also call refs_output_dependent_p.
2697 PR rtl-optimization/82913
2698 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2701 2017-11-09 Jeff Law <law@redhat.com>
2703 * vr-values.h: New file with vr_values class.
2704 * tree-vrp.c: Include vr-values.h
2705 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2706 data objects into the vr_values class.
2707 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2708 (get_value_range): Make it a member function within vr_values class.
2709 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2710 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2711 (op_with_constant_singleton_value_range): Likewise.
2712 (extract_range_for_var_from_comparison_expr): Likewise.
2713 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2714 (extract_range_from_binary_expr): Likewise.
2715 (extract_range_from_unary_expr): Likewise.
2716 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2717 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2718 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2719 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2720 (compare_name_with_value, compare_names): Likewise.
2721 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2722 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2723 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2724 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2725 (extract_range_from_phi_node): Likewise.
2726 (simplify_truth_ops_using_ranges): Likewise.
2727 (simplify_div_or_mod_using_ranges): Likewise.
2728 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2729 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2730 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2731 (simplify_float_conversion_using_ranges): Likewise.
2732 (simplify_internal_call_using_ranges): Likewise.
2733 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2734 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2735 poitner via x_vr_values for calls into gimple folder.
2736 (vrp_initialize_lattice): Make this the vr_values ctor.
2737 (vrp_free_lattice): Make this the vr_values dtor.
2738 (set_vr_value): New function.
2739 (class vrp_prop): Add vr_values data member. Add various member
2740 functions as well as member functions that delegate to vr_values.
2741 (check_array_ref): Make a member function within vrp_prop class.
2742 (search_for_addr_array, vrp_initialize): Likewise.
2743 (vrp_finalize): Likewise. Revamp to avoid direct access to
2744 vr_value, values_propagated, etc.
2745 (check_array_bounds): Extract vrp_prop class instance pointer from
2746 walk info structure. Use it to call member functions.
2747 (check_all_array_refs): Make a member function within vrp_prop class.
2748 Smuggle class instance pointer via walk info structure.
2749 (x_vr_values): New local static.
2750 (vrp_valueize): Use x_vr_values to get class instance.
2751 (vr_valueize_1): Likewise.
2752 (class vrp_folder): Add vr_values data member. Add various member
2753 functions as well as member functions that delegate to vr_values.
2754 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2755 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2756 class instance from vr_values. Use it to call member functions.
2757 (vrp_dom_walker): Add vr_values data member.
2758 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2759 instance via x_vr_values.
2760 (identify_jump_threads): Accept vr_values as argument. Store
2761 it into the walker structure.
2762 (evrp_dom_walker): Add vr_values class data member. Add various
2764 (evrp_dom_walker::try_find_new_range): Use vr_values data
2765 member to access the memory allocator.
2766 (evrp_dom_walker::before_dom_children): Store vr_values class
2767 instance into the vrp_folder class.
2768 (evrp_dom_walker::push_value_range): Rework to avoid direct
2769 access to num_vr_values and vr_value.
2770 (evrp_dom_walker::pop_value_range): Likewise.
2771 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2772 Use vr_values to get to dump_all_value_ranges member function.
2773 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2774 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2775 Pass vr_values class instance down to identify_jump_threads.
2776 Remove call to vrp_free_lattice.
2777 (debug_all_value_ranges): Remove.
2779 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2780 (vrp_folder): Likewise.
2782 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2783 Get it from the existing bitmap instead.
2784 (vrp_intersect_ranges_1): Likewise.
2786 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2788 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2790 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2791 set bit_not_p to it.
2792 (imm_store_chain_info::coalesce_immediate_stores): Break group
2793 if bit_not_p is different.
2794 (count_multiple_uses, split_group,
2795 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2796 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2797 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2798 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2799 to store_immediate_info ctor.
2801 2017-11-09 Jim Wilson <jimw@sifive.com>
2803 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2804 (scan_prog_file): Likewise.
2806 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2808 * bb-reorder.c (max_entry_frequency): Remove.
2809 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2810 connect_traces, push_to_next_round_p): Remove prototypes.
2811 (find_traces_1_round): Use counts only.
2812 (push_to_next_round_p): Likewise.
2813 (find_traces): Likewise.
2814 (rotate_loop): Likewise.
2815 (find_traces_1_round): Likewise.
2816 (connect_traces): Likewise.
2817 (edge_order): Likewise.
2819 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2821 * config/arm/arm.c (output_return_instruction): Add comments to
2822 indicate requirement for cmse_nonsecure_entry return to account
2823 for the size of clearing instruction output here.
2824 (thumb_exit): Likewise.
2825 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2826 return in hardfloat mode.
2828 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2830 * config/rs6000/rs6000.c (machine_function): Add a bool,
2831 "toc_is_wrapped_separately".
2832 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2833 if it wasn't explicitly set or unset, we are optimizing for speed, and
2834 doing separate shrink-wrapping.
2835 (rs6000_get_separate_components): Enable the TOC component if
2836 saving the TOC register in the prologue.
2837 (rs6000_components_for_bb): Handle the TOC component.
2838 (rs6000_emit_prologue_components): Store the TOC register where needed.
2839 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2840 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2842 2017-11-09 Martin Jambor <mjambor@suse.cz>
2844 * ipa-param-manipulation.c: New file.
2845 * ipa-param-manipulation.h: Likewise.
2846 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2847 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2848 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2849 (ipa_parm_adjustment): Likewise.
2850 (ipa_parm_adjustment_vec): Likewise.
2851 (ipa_get_vector_of_formal_parms): Moved declaration to
2852 ipa-param-manipulation.h.
2853 (ipa_get_vector_of_formal_parm_types): Likewise.
2854 (ipa_modify_formal_parameters): Likewise.
2855 (ipa_modify_call_arguments): Likewise.
2856 (ipa_combine_adjustments): Likewise.
2857 (ipa_dump_param_adjustments): Likewise.
2858 (ipa_modify_expr): Likewise.
2859 (ipa_get_adjustment_candidate): Likewise.
2860 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2861 ipa-param-manipulation.c.
2862 (ipa_get_vector_of_formal_parm_types): Likewise.
2863 (ipa_modify_formal_parameters): Likewise.
2864 (ipa_modify_call_arguments): Likewise.
2865 (ipa_modify_expr): Likewise.
2866 (get_ssa_base_param): Likewise.
2867 (ipa_get_adjustment_candidate): Likewise.
2868 (index_in_adjustments_multiple_times_p): Likewise.
2869 (ipa_combine_adjustments): Likewise.
2870 (ipa_dump_param_adjustments): Likewise.
2871 * tree-sra.c: Also include ipa-param-manipulation.h
2872 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2875 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2876 Alan Hayward <alan.hayward@arm.com>
2877 David Sherwood <david.sherwood@arm.com>
2879 * doc/sourcebuild.texi (vect_masked_store): Document.
2881 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2882 Alan Hayward <alan.hayward@arm.com>
2883 David Sherwood <david.sherwood@arm.com>
2885 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2887 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2888 Alan Hayward <alan.hayward@arm.com>
2889 David Sherwood <david.sherwood@arm.com>
2891 * doc/sourcebuild.texi (vect_variable_length): Document.
2893 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2894 Alan Hayward <alan.hayward@arm.com>
2895 David Sherwood <david.sherwood@arm.com>
2897 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2899 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2900 Alan Hayward <alan.hayward@arm.com>
2901 David Sherwood <david.sherwood@arm.com>
2903 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2905 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2906 Alan Hayward <alan.hayward@arm.com>
2907 David Sherwood <david.sherwood@arm.com>
2909 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2910 previously undocumented selectors.
2911 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2913 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2915 * doc/rtl.texi (const_vector): Say that elements can be
2916 const_wide_ints too.
2917 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2918 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2919 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2920 * optabs.c (expand_vector_broadcast): Likewise.
2922 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2923 Alan Hayward <alan.hayward@arm.com>
2924 David Sherwood <david.sherwood@arm.com>
2926 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2927 scaled index even if the unscaled address was invalid.
2928 Don't increase the complexity of using a scale in that case.
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/rtl.texi: Rewrite the subreg rules so that they partition
2935 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2936 UNITS_PER_WORD bytes.
2937 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2938 based on REGMODE_NATURAL_SIZE of the inner mode.
2939 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2940 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2941 * expmed.c (lowpart_bit_field_p): Divide the value up into
2942 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2943 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2944 whether something is likely to occupy more than one register.
2946 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2949 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2950 profile_count::adjust_for_ipa_scaling.
2951 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2952 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2954 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2956 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2958 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2959 (split_group): Add total_orig and total_new arguments, estimate the
2960 number of statements related to the store group without store merging
2961 and with store merging.
2962 (imm_store_chain_info::output_merged_store): Adjust split_group
2963 callers, punt if estimated number of statements with store merging
2964 is not smaller than estimated number of statements without it.
2966 (handled_load): Remove has_single_use checks.
2967 (pass_store_merging::process_store): Likewise.
2969 2017-11-09 Richard Biener <rguenther@suse.de>
2971 PR tree-optimization/82902
2972 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2974 2017-11-09 Martin Liska <mliska@suse.cz>
2977 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2980 2017-11-09 Martin Liska <mliska@suse.cz>
2982 PR tree-optimization/82669
2983 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2985 2017-11-09 Martin Liska <mliska@suse.cz>
2987 PR gcov-profile/48463
2988 * coverage.c (coverage_begin_function): Output also end locus
2989 of a function and information whether the function is
2991 * gcov-dump.c (tag_function): Parse and print the information.
2992 * gcov.c (INCLUDE_MAP): Add include.
2993 (INCLUDE_SET): Likewise.
2994 (struct line_info): Move earlier in the source file because
2995 of vector<line_info> in function_info structure.
2996 (line_info::line_info): Likewise.
2997 (line_info::has_block): Likewise.
2998 (struct source_info): Add new member index.
2999 (source_info::get_functions_at_location): New function.
3000 (function_info::group_line_p): New function.
3001 (output_intermediate_line): New function.
3002 (output_intermediate_file): Use the mentioned function.
3003 (struct function_start): New.
3004 (struct function_start_pair_hash): Likewise.
3005 (process_file): Add code that identifies group functions.
3006 Assign lines either to global or function scope.
3007 (generate_results): Skip artificial functions.
3008 (find_source): Assign index for each source file.
3009 (read_graph_file): Read new flag artificial and end_line.
3010 (add_line_counts): Assign it either to global of function scope.
3011 (accumulate_line_counts): Isolate core of the function to
3012 accumulate_line_info and call it for both function and global
3014 (accumulate_line_info): New function.
3015 (output_line_beginning): Fix GNU coding style.
3016 (print_source_line): New function.
3017 (output_line_details): Likewise.
3018 (output_function_details): Likewise.
3019 (output_lines): Iterate both source (global) scope and function
3021 (struct function_line_start_cmp): New class.
3022 * doc/gcov.texi: Reflect changes in documentation.
3024 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3027 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3028 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3029 and similarly for not instead of neg.
3031 2017-11-08 Andi Kleen <ak@linux.intel.com>
3033 * config/i386/i386.opt: Add -mforce-indirect-call.
3034 * config/i386/predicates.md: Check for flag_force_indirect_call.
3035 * doc/invoke.texi: Document -mforce-indirect-call
3037 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3039 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3041 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3042 config/riscv/riscv.c (predict.h): New include.
3043 (riscv_slow_unaligned_access_p): No longer static.
3044 (riscv_block_move_straight): Add require.
3045 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3047 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3050 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3051 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3052 nonimmediate_operand predicate for operand 1 instead of
3055 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3057 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3059 * config/aarch64/constraints.md (Uml): New constraint.
3060 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3063 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3065 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3066 of two vec_duplicates into a vec_concat.
3068 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3070 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3071 Simplify vec_merge of vec_duplicate and vec_concat.
3072 * config/aarch64/constraints.md (Utq): New constraint.
3073 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3076 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3078 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3079 Simplify vec_merge of vec_duplicate and const_vector.
3080 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3082 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3083 mode iterator. Update predicate on operand 1 to
3084 handle non-const_vec constants. Delete constraints.
3085 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3087 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3089 PR tree-optimization/78821
3090 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3092 (store_operand_info::store_operand_info): Initialize it to false.
3093 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3094 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3095 store in the group, and if chain_info is non-NULL, to ignore altogether
3097 (compatible_load_p): Fail if bit_not_p does not match.
3098 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3099 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3100 (pass_store_merging::process_store): Adjust
3101 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3102 call terminate_all_aliasing_chains newly when adding a store into
3103 a chain with non-NULL chain_info.
3105 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3107 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3109 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3111 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3113 (aarch64_layout_frame): Initialise emit_frame_chain.
3114 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3115 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3117 2017-11-08 Martin Liska <mliska@suse.cz>
3119 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3120 of gimple_call_internal_p.
3122 2017-11-07 Tom de Vries <tom@codesourcery.com>
3124 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3126 2017-11-07 Tom de Vries <tom@codesourcery.com>
3128 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3131 2017-11-08 Martin Liska <mliska@suse.cz>
3134 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3136 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3138 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3140 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3143 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3144 (*cmp<mode>_ccz_1): New insn with $k alternative.
3147 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3148 RTX_COMM_COMPARE as commutative as well.
3149 (ix86_binary_operator_ok): Formatting fix.
3150 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3151 *<code><mode>3<mask_name><round_saeonly_name>,
3152 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3153 *<s>mul<mode>3_highpart<mask_name>,
3154 *vec_widen_umult_even_v16si<mask_name>,
3155 *vec_widen_umult_even_v8si<mask_name>,
3156 *vec_widen_umult_even_v4si<mask_name>,
3157 *vec_widen_smult_even_v16si<mask_name>,
3158 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3159 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3160 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3161 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3162 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3163 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3164 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3165 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3166 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3167 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3168 ix86_binary_operator_ok. Formatting fixes.
3169 (*<plusminus_insn><mode>3<mask_name><round_name>,
3170 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3173 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3175 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3176 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3177 eq and ne if TARGET_ISEL.
3178 (cmp): New code_iterator.
3179 (UNS, UNSU_, UNSIK): New code_attrs.
3180 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3181 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3183 ("eq<mode>3"): ... this.
3184 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3186 ("ne<mode>3"): ... this.
3188 2017-11-07 Julia Koval <julia.koval@intel.com>
3191 * common/config/i386/i386-common.c
3192 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3193 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3194 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3195 * config/i386/i386.opt: Ditto.
3196 * config/i386/i386.c (ix86_target_string): Ditto.
3197 (ix86_option_override_internal): Ditto.
3198 (ix86_init_mpx_builtins): Move MPX to args2.
3199 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3200 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3201 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3202 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3203 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3204 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3205 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3207 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3210 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3211 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3212 (zero-extendsidi peephole2): Remove peephole.
3214 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3217 * stor-layout.c: Include attribs.h.
3218 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3219 explicit lookup of "aligned" attribute.
3221 2017-11-07 Andrew Waterman <andrew@sifive.com>
3223 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3224 (riscv_expand_block_move): Likewise.
3225 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3227 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3228 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3229 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3230 (riscv_adjust_block_mem): Likewise.
3231 (riscv_block_move_loop): Likewise.
3232 (riscv_expand_block_move): Likewise.
3233 gcc/config/riscv/riscv.md (movmemsi): New pattern.
3235 2017-11-07 Michael Clark <michaeljclark@mac.com>
3237 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3238 (MUSL_DYNAMIC_LINKER): Likewise.
3240 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3242 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3243 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3246 2017-11-07 Richard Biener <rguenther@suse.de>
3248 * match.pd: Fix build.
3250 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3251 Jackson Woodruff <jackson.woodruff@arm.com>
3253 PR tree-optimization/71026
3254 * match.pd: Canonicalize negate in division.
3256 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3259 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3261 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3263 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3264 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3266 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3268 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3269 non-scalar integral types.
3270 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3271 (-(A-B), -(~A)): New transformations.
3273 2017-11-07 Tom de Vries <tom@codesourcery.com>
3275 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3276 semicolon after "do {} while (0)".
3277 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3278 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3279 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3280 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3281 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3283 2017-11-07 Tom de Vries <tom@codesourcery.com>
3285 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3286 after "do {} while (0)".
3287 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3288 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3289 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3290 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3291 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3293 2017-11-07 Tom de Vries <tom@codesourcery.com>
3296 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3298 (arm_rtx_costs_internal): Add missing semicolon after
3299 HANDLE_NARROW_SHIFT_ARITH call.
3301 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3304 disable isel if it was not set explicitly.
3306 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3308 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3309 (add_type_attribute) likewise.
3311 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3313 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3314 of struct ix86_frame.
3315 (ix86_initial_elimination_offset): Likewise.
3316 (ix86_expand_split_stack_prologue): Likewise.
3318 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3320 * tree-vrp.h (enum value_range_type): Update stale comment.
3322 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3323 Alan Hayward <alan.hayward@arm.com>
3324 David Sherwood <david.sherwood@arm.com>
3326 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3327 (aarch64_expand_vec_perm_const): Take the number of units too.
3328 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3329 (aarch64_expand_vec_perm_const): Likewise.
3330 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3331 (vec_perm<mode>): Update accordingly.
3333 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3334 Alan Hayward <alan.hayward@arm.com>
3335 David Sherwood <david.sherwood@arm.com>
3337 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3338 Take the number of units too.
3339 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3340 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3341 but check for a vector mode before rather than after the call.
3342 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3343 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3344 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3345 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3346 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3347 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3348 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3349 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3350 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3351 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3352 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3353 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3354 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3355 (aarch64_sqdmull2_n<mode>): Update accordingly.
3357 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3358 Alan Hayward <alan.hayward@arm.com>
3359 David Sherwood <david.sherwood@arm.com>
3361 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3362 the number of units too.
3363 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3364 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3365 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3366 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3367 (vec_store_lanesxi<mode>): Update accordingly.
3369 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3370 Alan Hayward <alan.hayward@arm.com>
3371 David Sherwood <david.sherwood@arm.com>
3373 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3374 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3375 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3376 of units rather than the mode.
3377 * config/aarch64/iterators.md (nunits): New mode attribute.
3378 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3379 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3380 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3381 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3382 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3383 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3384 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3385 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3386 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3387 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3388 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3389 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3390 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3391 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3392 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3393 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3394 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3395 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3396 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3397 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3398 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3399 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3400 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3401 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3402 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3403 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3404 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3405 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3406 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3407 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3408 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3409 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3410 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3411 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3412 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3413 (aarch64_simd_vec_setv2di): Likewise.
3415 2017-11-06 Carl Love <cel@us.ibm.com>
3417 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3419 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3420 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3422 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3424 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3425 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3426 Add power 8 macro expansions.
3427 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3428 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3429 power 8 instructions. (VSX_XXBR): Add iterator.
3431 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3433 * config/arm/arm.md (predicable_short_it): Change default to "no",
3434 improve documentation, remove uses that are identical to the default.
3435 (enabled_for_depr_it): Rename to enabled_for_short_it.
3436 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3437 Remove default uses.
3438 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3439 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3440 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3441 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3443 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3446 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3447 float128 helper macros, which are no longer used after deleting
3448 the old 'q' built-in functions, and moving the round to odd
3449 built-in functions to being special built-in functions.
3450 (BU_FLOAT128_2): Likewise.
3451 (BU_FLOAT128_1_HW): Likewise.
3452 (BU_FLOAT128_2_HW): Likewise.
3453 (BU_FLOAT128_3_HW): Likewise.
3454 (FABSQ): Delete old 'q' built-in functions.
3455 (COPYSIGNQ): Likewise.
3456 (SQRTF128_ODD): Move round to odd built-in functions to be
3457 special built-in functions, so that we can handle
3458 -mabi=ieeelongdouble.
3459 (TRUNCF128_ODD): Likewise.
3460 (ADDF128_ODD): Likewise.
3461 (SUBF128_ODD): Likewise.
3462 (MULF128_ODD): Likewise.
3463 (DIVF128_ODD): Likewise.
3464 (FMAF128_ODD): Likewise.
3465 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3466 built-in names to 'f128'.
3467 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3468 old 'q' built-in functions, as the machine independent code for
3469 'f128' built-in functions handles this.
3470 (rs6000_expand_builtin): Add expansion for float128 round to odd
3471 functions, keying off on -mabi=ieeelongdouble of whether to use
3472 the KFmode or TFmode variant.
3473 (rs6000_init_builtins): Initialize the _Float128 round to odd
3475 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3476 _Float128 'q' built-in functions are now mapped into the new
3477 'f128' built-in functions.
3479 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3481 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3482 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3484 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3487 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3488 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3489 * toplev.c: Include "ipa-fnsummary.h".
3490 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3492 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3494 PR tree-optimization/82838
3495 * gimple-ssa-store-merging.c
3496 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3497 on a separate gimple_seq which is then appended to seq.
3499 2017-11-06 Jeff Law <law@redhat.com>
3502 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3503 (get_probe_interval): New functions.
3504 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3505 (ix86_adjust_stack_and_probe): Likewise.
3506 (output_adjust_stack_and_probe): Likewise.
3507 (ix86_emit_probe_stack_range): Likewise.
3508 (ix86_expand_prologue): Likewise.
3510 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3512 PR tree-optimization/82816
3513 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3514 if the modes of the two types are the same.
3515 (convert_plusminus_to_widen): Likewise.
3517 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3519 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3521 (usadv16qi): New define_expand.
3522 (usadv8hi): New define_expand.
3524 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3527 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3529 (inline_transform): Likewise.
3530 * predict.c (counts_to_freqs): Remove useless conditional.
3531 * profile-count.h (profile_count::apply_scale): Move sanity check.
3532 * tree-inline.c (copy_bb): Always scale.
3533 (copy_cfg_body): Likewise.
3535 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3538 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3540 (*cmp_ite0): Add enabled_for_depr_it attribute.
3541 (*cmp_ite1): Likewise.
3543 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3545 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3548 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3550 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3551 return the value by pointer.
3552 (extract_range_from_multiplicative_op_1): Update accordingly.
3553 Return as soon as an operation fails.
3555 2017-11-05 Tom de Vries <tom@codesourcery.com>
3558 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3559 (DEF_SANITIZER_BUILTIN): ... here.
3560 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3561 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3563 2017-11-05 Tom de Vries <tom@codesourcery.com>
3566 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3568 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3569 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3570 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3572 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3573 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3574 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3576 2017-11-05 Tom de Vries <tom@codesourcery.com>
3579 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3582 2017-11-04 Michael Clark <michaeljclark@mac.com>
3584 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3585 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3587 (*addsi3_extended): Likewise.
3588 (*addsi3_extended2): Likewise.
3589 (<optab>si3): Likewise.
3590 (<optab>di3): Likewise.
3591 (<optab><mode>3): Likewise.
3592 (<*optabe>si3_internal): Likewise.
3593 (zero_extendqi<SUPERQI:mode>2): Likewise.
3594 (*add<mode>hi3): Likewise.
3595 (*xor<mode>hi3): Likewise.
3596 (<optab>di3): Likewise.
3597 (*<optab>si3_extend): Likewise.
3598 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3599 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3600 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3602 2017-11-04 Andrew Waterman <andrew@sifive.com>
3604 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3605 TARGET_STRICT_ALIGN based upon -mtune argument.
3607 2017-11-04 Andrew Waterman <andrew@sifive.com>
3609 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3611 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3613 config/i386/i386.c (choose_basereg): Use optional scratch
3614 register and add assertion.
3615 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3616 needed, and don't allocate stack.
3617 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3618 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3619 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3621 2017-11-03 Jeff Law <law@redhat.com>
3623 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3624 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3625 to probe at the start of a noreturn function.
3627 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3629 PR tree-optimization/78821
3630 * gimple-ssa-store-merging.c: Update the file comment.
3631 (MAX_STORE_ALIAS_CHECKS): Define.
3632 (struct store_operand_info): New type.
3633 (store_operand_info::store_operand_info): New constructor.
3634 (struct store_immediate_info): Add rhs_code and ops data members.
3635 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3636 and op1r arguments to the ctor, initialize corresponding data members.
3637 (struct merged_store_group): Add load_align_base and load_align
3639 (merged_store_group::merged_store_group): Initialize them.
3640 (merged_store_group::do_merge): Update them.
3641 (merged_store_group::apply_stores): Pick the constant for
3642 encode_tree_to_bitpos from one of the two operands, or skip
3643 encode_tree_to_bitpos if neither operand is a constant.
3644 (class pass_store_merging): Add process_store method decl. Remove
3645 bool argument from terminate_all_aliasing_chains method decl.
3646 (pass_store_merging::terminate_all_aliasing_chains): Remove
3647 var_offset_p argument and corresponding handling.
3648 (stmts_may_clobber_ref_p): New function.
3649 (compatible_load_p): New function.
3650 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3651 if there is overlap and rhs_code is not INTEGER_CST. For
3652 non-overlapping stores terminate group if rhs is not mergeable.
3653 (get_alias_type_for_stmts): Change first argument from
3654 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3655 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3656 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3658 (get_location_for_stmts): Change first argument from
3659 auto_vec<gimple *> & to vec<gimple *> &.
3660 (struct split_store): Remove orig_stmts data member, add orig_stores.
3661 (split_store::split_store): Create orig_stores rather than orig_stmts.
3662 (find_constituent_stmts): Renamed to ...
3663 (find_constituent_stores): ... this. Change second argument from
3664 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3665 to info structures rather than the statements.
3666 (split_group): Rename ALLOW_UNALIGNED argument to
3667 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3668 it. Adjust find_constituent_stores caller.
3669 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3670 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3671 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3672 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3673 (mem_valid_for_store_merging): New function.
3674 (handled_load): New function.
3675 (pass_store_merging::process_store): New method.
3676 (pass_store_merging::execute): Use process_store method. Adjust
3677 terminate_all_aliasing_chains caller.
3679 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3681 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3682 Return true for more constants, symbols and label references.
3683 (aarch64_valid_floating_const): Remove unused function.
3685 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3688 * config/aarch64/aarch64.c (aarch64_layout_frame):
3689 Undo forcing of LR at bottom of frame.
3691 2017-11-03 Jeff Law <law@redhat.com>
3694 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3695 for int_registers_saved.
3697 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3698 extracted from tree-ssa-dom.c.
3699 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3700 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3701 (record_equivalences_from_incoming_edge): Add additional argument
3702 to single_pred_edge_ignoring_loop_edges call.
3703 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3704 (uncprop_dom_walker::before_dom_children): Add additional argument
3705 to single_pred_edge_ignoring_loop_edges call.
3706 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3707 single_pred_edge_ignoring_loop_edges rather than open coding.
3708 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3710 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3712 * match.pd (-(-A)): Rewrite.
3714 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3716 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3717 (rs6000_emit_int_cmove): New declaration.
3718 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3719 (rs6000_emit_sISEL): Delete.
3720 (rs6000_emit_int_cmove): Make non-static.
3721 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3722 instead of rs6000_emit_sISEL.
3724 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3726 * asan.c (create_cond_insert_point): Maintain profile.
3727 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3729 * basic-block.h (struct basic_block_def): Remove frequency.
3730 (EDGE_FREQUENCY): Use to_frequency
3731 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3733 (find_traces): Update to use to_frequency.
3734 (find_traces_1_round): Likewise; use only IPA counts.
3735 (bb_to_key): Likewise.
3736 (connect_traces): Use IPA counts only.
3737 (copy_bb_p): Update to use to_frequency.
3738 (fix_up_crossing_landing_pad): Likewise.
3739 (sanitize_hot_paths): Likewise.
3740 * bt-load.c (basic_block_freq): Likewise.
3741 * cfg.c (init_flow): Set count_max to uninitialized.
3742 (check_bb_profile): Remove frequencies; check counts.
3743 (dump_bb_info): Do not dump frequencies.
3744 (update_bb_profile_for_threading): Update counts only.
3745 (scale_bbs_frequencies_int): Likewise.
3746 (MAX_SAFE_MULTIPLIER): Remove.
3747 (scale_bbs_frequencies_gcov_type): Update counts only.
3748 (scale_bbs_frequencies_profile_count): Update counts only.
3749 (scale_bbs_frequencies): Update counts only.
3750 * cfg.h (struct control_flow_graph): Add count-max.
3751 (update_bb_profile_for_threading): Update prototype.
3752 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3753 (find_many_sub_basic_blocks): Likewise.
3754 * cfgcleanup.c (try_forward_edges): Likewise.
3755 (try_crossjump_to_edge): Likewise.
3756 * cfgexpand.c (expand_gimple_cond): Likewise.
3757 (expand_gimple_tailcall): Likewise.
3758 (construct_init_block): Likewise.
3759 (construct_exit_block): Likewise.
3760 * cfghooks.c (verify_flow_info): Check consistency of counts.
3761 (dump_bb_for_graph): Do not dump frequencies.
3762 (split_block_1): Do not update frequencies.
3763 (split_edge): Do not update frequencies.
3764 (make_forwarder_block): Do not update frequencies.
3765 (duplicate_block): Do not update frequencies.
3766 (account_profile_record): Do not update frequencies.
3767 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3768 for global heuristics.
3769 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3770 (expected_loop_iterations_unbounded): Use counts only.
3771 * cfgloopmanip.c (scale_loop_profile): Simplify.
3772 (create_empty_loop_on_edge): Simplify
3774 (duplicate_loop_to_header_edge): Simplify
3775 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3776 (update_br_prob_note): Take care of removing note when profile
3778 (relink_block_chain): Do not dump frequency.
3779 (rtl_account_profile_record): Use to_frequency.
3780 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3781 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3782 (cgraph_update_edges_for_call_stmt_node): Likewise.
3783 (cgraph_edge::verify_count_and_frequency): Update.
3784 (cgraph_node::verify_node): Temporarily disable frequency verification.
3785 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3786 to_cgraph_frequency.
3787 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3788 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3790 (cgraph_node::expand_thunk): Update profile.
3791 * except.c (dw2_build_landing_pads): Do not update frequency.
3792 * final.c (compute_alignments): Use to_frequency.
3793 (dump_basic_block_info): Do not dump frequency.
3794 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3795 (dump_gimple_bb_header): Do not dump frequency.
3796 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3798 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3799 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3800 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3801 (init_before_recovery): Do not update frequency.
3802 (sched_create_recovery_edges): Do not update frequency.
3803 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3804 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3805 (ipa_cp_c_finalize): Set max_count to uninitialized.
3806 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3807 (param_change_prob): Use counts.
3808 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3810 * ipa-split.c (consider_split): Use to_frequency.
3811 (split_function): Use to_frequency.
3812 * ira-build.c (loop_compare_func): Likewise.
3813 (mark_loops_for_removal): Likewise.
3814 (mark_all_loops_for_removal): Likewise.
3815 * loop-doloop.c (doloop_modify): Do not update frequency.
3816 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3818 * lto-streamer-in.c (input_function): Update count_max.
3819 * omp-expand.c (expand_omp_taskreg): Update count_max.
3820 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3821 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3822 (maybe_hot_count_p): Use ipa counts only.
3823 (maybe_hot_bb_p): Simplify.
3824 (maybe_hot_edge_p): Simplify.
3825 (probably_never_executed): Do not take frequency argument.
3826 (probably_never_executed_bb_p): Do not pass frequency.
3827 (probably_never_executed_edge_p): Likewise.
3828 (combine_predictions_for_bb): Check that profile is nonzero.
3829 (propagate_freq): Do not set frequency.
3830 (drop_profile): Simplify.
3831 (counts_to_freqs): Simplify.
3832 (expensive_function_p): Use to_frequency.
3833 (propagate_unlikely_bbs_forward): Simplify.
3834 (determine_unlikely_bbs): Simplify.
3835 (estimate_bb_frequencies): Add hack to silence graphite issues.
3836 (compute_function_frequency): Use ipa counts.
3837 (pass_profile::execute): Update.
3838 (rebuild_frequencies): Use counts only.
3839 (force_edge_cold): Use counts only.
3840 * profile-count.c (profile_count::dump): Dump new count types.
3841 (profile_count::differs_from_p): Check compatiblity.
3842 (profile_count::to_frequency): New function.
3843 (profile_count::to_cgraph_frequency): New function.
3844 * profile-count.h (struct function): Declare.
3845 (enum profile_quality): Add profile_guessed_local and
3846 profile_guessed_global0.
3847 (class profile_proability): Decrease number of bits to 29;
3848 update from_reg_br_prob_note and to_reg_br_prob_note.
3849 (class profile_count: Update comment; decrease number of bits
3850 to 61. Check compatibility.
3851 (profile_count::compatible_p): New private member function.
3852 (profile_count::ipa_p): New member function.
3853 (profile_count::operator<): Handle global zero correctly.
3854 (profile_count::operator>): Handle global zero correctly.
3855 (profile_count::operator<=): Handle global zero correctly.
3856 (profile_count::operator>=): Handle global zero correctly.
3857 (profile_count::nonzero_p): New member function.
3858 (profile_count::force_nonzero): New member function.
3859 (profile_count::max): New member function.
3860 (profile_count::apply_scale): Handle IPA scalling.
3861 (profile_count::guessed_local): New member function.
3862 (profile_count::global0): New member function.
3863 (profile_count::ipa): New member function.
3864 (profile_count::to_frequency): Declare.
3865 (profile_count::to_cgraph_frequency): Declare.
3866 * profile.c (OVERLAP_BASE): Delete.
3867 (compute_frequency_overlap): Delete.
3868 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3869 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3870 * sched-ebb.c (rank): Use counts only.
3871 * shrink-wrap.c (handle_simple_exit): Use counts only.
3872 (try_shrink_wrapping): Use counts only.
3873 (place_prologue_for_one_component): Use counts only.
3874 * tracer.c (find_best_predecessor): Use to_frequency.
3875 (find_trace): Use to_frequency.
3876 (tail_duplicate): Use to_frequency.
3877 * trans-mem.c (expand_transaction): Do not update frequency.
3878 * tree-call-cdce.c: Do not update frequency.
3879 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3880 (gimple_merge_blocks): Likewise.
3881 (gimple_split_edge): Likewise.
3882 (gimple_duplicate_sese_region): Likewise.
3883 (gimple_duplicate_sese_tail): Likewise.
3884 (move_sese_region_to_fn): Likewise.
3885 (gimple_account_profile_record): Likewise.
3886 (insert_cond_bb): Likewise.
3887 * tree-complex.c (expand_complex_div_wide): Likewise.
3888 * tree-eh.c (lower_resx): Update profile.
3889 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3891 (initialize_cfun): Do not initialize frequencies
3892 (freqs_to_counts): Delete.
3893 (copy_cfg_body): Ignore count parameter.
3894 (copy_body): Update.
3895 (expand_call_inline): Update count_max.
3896 (optimize_inline_calls): Update count_max.
3897 (tree_function_versioning): Update count_max.
3898 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3899 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3901 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3902 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3903 (try_peel_loop): Likewise.
3904 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3906 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3907 (tree_transform_and_unroll_loop): Do not use frequencies
3908 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3909 Use reliable prediction only.
3910 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3911 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3912 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3913 probability scaling.
3914 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3915 not update frequency
3916 (any_remaining_duplicated_blocks): Likewise.
3917 (update_profile): Likewise.
3918 (estimated_freqs_path): Delete.
3919 (freqs_to_counts_path): Delete.
3920 (clear_counts_path): Delete.
3921 (ssa_fix_duplicate_block_edges): Likewise.
3922 (duplicate_thread_path): Likewise.
3923 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3924 * tree-tailcall.c (decrease_profile): Do not update frequency.
3925 (eliminate_tail_call): Likewise.
3926 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3927 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3928 (optimize_mask_stores): Likewise.
3929 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3930 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3931 (ubsan_expand_ptr_ifn): Update profile.
3932 * value-prof.c (gimple_ic): Simplify.
3933 * value-prof.h (gimple_ic): Update prototype.
3934 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3935 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3937 (want_inline_self_recursive_call_p): Likewise.
3938 (resolve_noninline_speculation): Only cummulate defined counts.
3939 (inline_small_functions): Use nonzero_p.
3940 (ipa_inline): Do not access freed node.
3942 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3944 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3945 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3947 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3949 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3950 non-legitimate address.
3952 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3954 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3955 (*lt0_<mode>di, *lt0_<mode>si): New.
3957 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3959 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3960 TARGET_PAIRED_FLOAT.
3962 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3963 Jim Wilson <jim.wilson@linaro.org>
3965 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3966 * config/aarch64/aarch64-tune.md: Regenerated.
3967 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3968 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3970 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3972 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3973 (arc_expand_prologue): Restore blink for millicode.
3974 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3976 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3979 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3980 gen_vec_duplicate after forcing the scalar into a register.
3982 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3984 * combine (try_combine): Print the insns input to try_combine to the
3987 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3990 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3991 Remove second argument from aarch64_process_target_attr call.
3992 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3994 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3996 (aarch64_handle_attr_arch): Remove second argument.
3997 (aarch64_handle_attr_cpu): Ditto.
3998 (aarch64_handle_attr_tune): Ditto.
3999 (aarch64_handle_attr_isa_flags): Ditto.
4000 (aarch64_process_one_target_attr): Ditto.
4001 (aarch64_process_target_attr): Ditto.
4002 (aarch64_option_valid_attribute_p): Remove second argument.
4003 on aarch64_process_target_attr call.
4005 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4007 * diagnostic.c: Include "selftest-diagnostic.h".
4008 (selftest::assert_location_text): New function.
4009 (selftest::test_diagnostic_get_location_text): New function.
4010 (selftest::diagnostic_c_tests): Call it.
4012 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4014 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4015 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4016 (class selftest::test_diagnostic_context): Move to...
4017 * selftest-diagnostic.c: New file.
4018 * selftest-diagnostic.h: New file.
4020 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4022 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4023 offset range for FT32B.
4024 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4025 * config/ft32/ft32.md: Add TARGET_NOPM.
4026 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4027 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4029 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4031 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4033 2017-11-02 Jeff Law <law@redhat.com>
4035 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4036 virtual keyword on FINAL OVERRIDE members.
4038 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4039 virtuals together. Add virtual destructor.
4040 (substitute_and_fold_engine): Similarly.
4042 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4044 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4046 2017-11-02 Richard Biener <rguenther@suse.de>
4048 PR tree-optimization/82795
4049 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4051 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4053 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4055 (gcc_GAS_CHECK_FEATURE): Remove.
4056 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4057 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4058 * configure: Regenerate.
4060 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4062 * config/arc/arc.c (hwloop_optimize): Account for empty
4065 2017-11-02 Richard Biener <rguenther@suse.de>
4068 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4069 Truncate ARRAY_REF index and element size.
4071 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4073 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4075 2017-11-01 Jeff Law <law@redhat.com>
4077 * tree-ssa-ccp.c (ccp_folder): New class derived from
4078 substitute_and_fold_engine.
4079 (ccp_folder::get_value): New member function.
4080 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4081 (ccp_fold_stmt): Remove prototype.
4082 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4083 * tree-ssa-copy.c (copy_folder): New class derived from
4084 substitute_and_fold_engine.
4085 (copy_folder::get_value): Renamed from get_value.
4086 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4087 * tree-vrp.c (vrp_folder): New class derived from
4088 substitute_and_fold_engine.
4089 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4090 (vrp_folder::get_value): New member function.
4091 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4092 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4093 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4094 provide a class interface to folder/substitute routines.
4095 (ssa_prop_fold_stmt_fn): Remove typedef.
4096 (ssa_prop_get_value_fn): Likewise.
4097 (subsitute_and_fold): Remove prototype.
4098 (replace_uses_in): Likewise.
4099 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4100 Renamed from replace_uses_in. Call the virtual member function
4101 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4102 (substitute_and_fold_dom_walker): Remove initialization of
4103 data member entries for calbacks. Add substitute_and_fold_engine
4104 member and initialize it.
4105 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4106 member functions for get_value, replace_phi_args_in c
4107 replace_uses_in, and fold_stmt calls.
4108 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4109 substitute_and_fold. Remove assert. Update ctor call.
4111 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4112 (ssa_prop_visit_phi_fn): Likewise.
4113 (class ssa_propagation_engine): New class to provide an interface
4115 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4117 (ssa_prop_visit_phi): Likewise.
4118 (ssa_propagation_engine::simulate_stmt): Moved into class.
4119 Call visit_phi/visit_stmt from the class rather than via
4120 file scoped static variables.
4121 (ssa_propagation_engine::simulate_block): Moved into class.
4122 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4123 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4124 set file scoped statics for the visit_stmt/visit_phi callbacks.
4125 * tree-complex.c (complex_propagate): New class derived from
4126 ssa_propagation_engine.
4127 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4128 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4129 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4131 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4132 ssa_propagation_engine.
4133 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4134 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4135 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4136 * tree-ssa-copy.c (copy_prop): New class derived from
4137 ssa_propagation_engine.
4138 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4139 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4140 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4141 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4142 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4143 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4144 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4146 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4148 PR rtl-optimization/82778
4149 PR rtl-optimization/82597
4150 * compare-elim.c (struct comparison): Add in_a_setter field.
4151 (find_comparison_dom_walker::before_dom_children): Remove killed
4152 bitmap and df_simulate_find_defs call, instead walk the defs.
4153 Compute last_setter and initialize in_a_setter. Merge definitions
4154 with first initialization for a few variables.
4155 (try_validate_parallel): Use insn_invalid_p instead of
4156 recog_memoized. Return insn rather than just the pattern.
4157 (try_merge_compare): Fix up comment. Don't uselessly test if
4158 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4160 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4163 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4164 Alan Hayward <alan.hayward@arm.com>
4165 David Sherwood <david.sherwood@arm.com>
4167 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4168 aarch64_hard_regno_nregs to get the number of registers
4171 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4172 Alan Hayward <alan.hayward@arm.com>
4173 David Sherwood <david.sherwood@arm.com>
4175 * config/aarch64/constraints.md (Upl): Rename to...
4177 * config/aarch64/aarch64.md
4178 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4181 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4182 Alan Hayward <alan.hayward@arm.com>
4183 David Sherwood <david.sherwood@arm.com>
4185 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4186 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4189 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4190 Alan Hayward <alan.hayward@arm.com>
4191 David Sherwood <david.sherwood@arm.com>
4193 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4194 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4195 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4197 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4198 of a single element.
4199 * config/aarch64/iterators.md: Add a comment above the permute
4200 unspecs to say that they are generated directly by
4201 aarch64_expand_vec_perm_const.
4202 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4204 2017-11-01 Nathan Sidwell <nathan@acm.org>
4206 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4208 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4210 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4211 and medany code models, and describe what they do.
4213 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4215 Revert accidental duplicate:
4217 * combine.c (can_change_dest_mode): Reject changes in
4218 REGMODE_NATURAL_SIZE.
4220 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4222 PR rtl-optimization/64682
4223 PR rtl-optimization/69567
4224 PR rtl-optimization/69737
4225 PR rtl-optimization/82683
4226 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4227 register mentioned in the note, drop the note, unless it came from I3,
4228 in which case it should go to I3 again.
4230 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4232 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4233 and return false if not.
4234 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4236 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4238 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4239 range is known to be empty.
4241 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4242 Alan Hayward <alan.hayward@arm.com>
4243 David Sherwood <david.sherwood@arm.com>
4245 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4246 and CONST_VECTOR_NUNITS instead of computing the number of units from
4247 the byte sizes of the vector and element.
4248 (simplify_binary_operation_1): Likewise.
4249 (simplify_const_binary_operation): Likewise.
4250 (simplify_ternary_operation): Likewise.
4252 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4257 (int_mem_offset): ...this new function.
4258 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4259 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4260 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4263 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4264 Alan Hayward <alan.hayward@arm.com>
4265 David Sherwood <david.sherwood@arm.com>
4267 * lower-subreg.c (interesting_mode_p): New function.
4268 (compute_costs, find_decomposable_subregs, decompose_register)
4269 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4270 (resolve_clobber, dump_choices): Use it.
4272 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4273 Alan Hayward <alan.hayward@arm.com>
4274 David Sherwood <david.sherwood@arm.com>
4276 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
4283 truncations. Check the precision rather than the size.
4285 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4286 Alan Hayward <alan.hayward@arm.com>
4287 David Sherwood <david.sherwood@arm.com>
4289 * machmode.h (is_narrower_int_mode): New function
4290 * optabs.c (expand_float, expand_fix): Use it.
4291 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4293 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4294 Alan Hayward <alan.hayward@arm.com>
4295 David Sherwood <david.sherwood@arm.com>
4297 * rtl.h (narrower_subreg_mode): New function.
4298 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
4305 (convert_optab_handler): ...here.
4306 (widening_optab_handler): Delete.
4307 (find_widening_optab_handler): Remove permit_non_widening parameter.
4308 (find_widening_optab_handler_and_mode): Likewise. Provide an
4309 override that operates on mode class wrappers.
4310 * optabs-query.c (widening_optab_handler): Delete.
4311 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4312 parameter. Assert that the two modes are the same class and that
4313 the "from" mode is narrower than the "to" mode. Use
4314 convert_optab_handler instead of widening_optab_handler.
4315 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4316 instead of widening_optab_handler.
4317 * expr.c (expand_expr_real_2): Update calls to
4318 find_widening_optab_handler.
4319 * optabs.c (expand_widen_pattern_expr): Likewise.
4320 (expand_binop_directly): Take the insn_code as a parameter.
4321 (expand_binop): Only call find_widening_optab_handler for
4322 conversion optabs; use optab_handler otherwise. Update calls
4323 to find_widening_optab_handler and expand_binop_directly.
4324 Use convert_optab_handler instead of widening_optab_handler.
4325 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4326 find_widening_optab_handler and use scalar_mode rather than
4328 (convert_plusminus_to_widen): 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 * machmode.h (fixed_size_mode): New class.
4335 * rtl.h (get_pool_mode): Return fixed_size_mode.
4336 * gengtype.c (main): Add fixed_size_mode.
4337 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4338 (get_raw_arg_mode): Likewise.
4339 * doc/tm.texi: Regenerate.
4340 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4341 * targhooks.c (default_get_reg_raw_mode): Likewise.
4342 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4343 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4344 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4345 (msp430_get_raw_result_mode): Likewise.
4346 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4347 * dbxout.c (dbxout_parms): Require fixed-size modes.
4348 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4349 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4350 * omp-low.c (lower_oacc_reductions): Likewise.
4351 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4352 (simplify_subreg): Update accordingly.
4353 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4354 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4355 that aren't fixed-size.
4356 (get_pool_mode): Return a fixed_size_mode.
4357 (output_constant_pool_2): Take a fixed_size_mode.
4359 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4360 Alan Hayward <alan.hayward@arm.com>
4361 David Sherwood <david.sherwood@arm.com>
4363 * doc/rtl.texi (vec_series): Document.
4364 (const): Say that the operand can be a vec_series.
4365 * rtl.def (VEC_SERIES): New rtx code.
4366 * rtl.h (const_vec_series_p_1): Declare.
4367 (const_vec_series_p): New function.
4368 * emit-rtl.h (gen_const_vec_series): Declare.
4369 (gen_vec_series): Likewise.
4370 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4371 (gen_vec_series): Likewise.
4372 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4373 * simplify-rtx.c (simplify_unary_operation): Handle negations
4375 (simplify_binary_operation_series): New function.
4376 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4377 (test_vector_ops_series): New function.
4378 (test_vector_ops): Call it.
4379 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4380 gen_const_vec_series.
4381 (altivec_lvsr): Likewise.
4382 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4384 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4385 Alan Hayward <alan.hayward@arm.com>
4386 David Sherwood <david.sherwood@arm.com>
4388 * doc/rtl.texi (const): Update description of address constants.
4389 Say that vector constants are allowed too.
4390 * common.md (E, F): Use CONSTANT_P instead of checking for
4392 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4393 checking for CONST_VECTOR.
4394 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4396 * expr.c (expand_expr_real_2): Check for vector modes instead
4397 of checking for CONST_VECTOR.
4398 * rtl.h (const_vec_p): New function.
4399 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4400 (unwrap_const_vec_duplicate): Handle them here too.
4402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4403 David Malcolm <dmalcolm@redhat.com>
4404 Alan Hayward <alan.hayward@arm.com>
4405 David Sherwood <david.sherwood@arm.com>
4407 * rtl.h (vec_duplicate_p): New function.
4408 * selftest-rtl.c (assert_rtx_eq_at): New function.
4409 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4410 (assert_rtx_eq_at): Declare.
4411 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4412 * selftest-run-tests.c (selftest::run_tests): Call it.
4413 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4414 (simplify_unary_operation_1): Recursively handle vector duplicates.
4415 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4417 (simplify_subreg): Handle subregs of vector duplicates.
4418 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4419 (selftest::simplify_rtx_c_tests): New functions.
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 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4426 (gen_vec_duplicate): Likewise.
4427 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4429 (gen_const_vector): ...here.
4430 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4431 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4432 whose elements are all equal.
4433 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4434 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4435 (simplify_relational_operation): Likewise.
4436 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4438 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4439 (aarch64_expand_vector_init): Likewise.
4440 * config/arm/arm.c (neon_vdup_constant): Likewise.
4441 (neon_expand_vector_init): Likewise.
4442 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4443 (arm_block_set_unaligned_vect): Likewise.
4444 (arm_block_set_aligned_vect): Likewise.
4445 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4446 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4447 (expand_vec_perm_even_odd_pack): Likewise.
4448 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4449 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4450 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4451 gen_const_vec_duplicate.
4452 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4453 * config/mips/mips.c (mips_gen_const_int_vector): Use
4454 gen_const_vec_duplicate.
4455 (mips_expand_vector_init): Use CONST0_RTX.
4456 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4457 (define_split): Use gen_const_vec_duplicate.
4458 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4459 (define_split): Use gen_const_vec_duplicate.
4460 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4461 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4462 * config/spu/spu.c (spu_const): Likewise.
4464 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4465 Alan Hayward <alan.hayward@arm.com>
4466 David Sherwood <david.sherwood@arm.com>
4468 * combine.c (can_change_dest_mode): Reject changes in
4469 REGMODE_NATURAL_SIZE.
4471 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4473 * configure.ac (--enable-libssp): New.
4474 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4475 trying to determine target-specific default. Adjust indentation.
4476 * configure: Regenerated.
4477 * doc/install.texi (Configuration): Expand --disable-libssp
4480 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4482 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4485 2017-10-31 Martin Jambor <mjambor@suse.cz>
4488 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4490 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4492 * auto-profile.c (autofdo_source_profile::read): Use
4493 UNKNOWN_LOCATION rather than 0.
4494 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4495 (warning_at): ...this overload.
4496 (warning_at_rich_loc_n): Rename to...
4497 (warning_n): ...this overload.
4498 (error_at_rich_loc): Rename to...
4499 (error_at): ...this overload.
4500 (pedwarn_at_rich_loc): Rename to...
4501 (pedwarn): ...this overload.
4502 (permerror_at_rich_loc): Rename to...
4503 (permerror): ...this overload.
4504 (inform_at_rich_loc): Rename to...
4505 (inform): ...this overload.
4506 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4507 (diagnostic_n_impl_richloc): Rename to...
4508 (diagnostic_n_impl): ...this rich_location *-based decl.
4509 (inform_at_rich_loc): Rename to...
4510 (inform): ...this, and add an assertion.
4511 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4512 (warning_at_rich_loc): Rename to...
4513 (warning_at): ...this, and add an assertion.
4514 (warning_at_rich_loc_n): Rename to...
4515 (warning_n): ...this, and add an assertion.
4516 (warning_n): Update location_t-based implementation for removal of
4517 location_t-based diagnostic_n_impl.
4518 (pedwarn_at_rich_loc): Rename to...
4519 (pedwarn): ...this, and add an assertion.
4520 (permerror_at_rich_loc): Rename to...
4521 (permerror): ...this, and add an assertion.
4522 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4523 (error_at_rich_loc): Rename to...
4524 (error_at): ...this, and add an assertion.
4525 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4526 (driver::do_spec_on_infiles): Likewise.
4527 * substring-locations.c (format_warning_va): Update for renaming
4528 of inform_at_rich_loc.
4530 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4532 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4533 _Float<N>X built-in functions so that the variant without the
4534 "__builtin_" prefix is only enabled for the GNU C and Objective C
4535 languages when they are in non-strict ANSI/ISO mode.
4536 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4537 * target.def (floatn_builtin_p): Add a target hook to control
4538 whether _Float<N> and _Float<N>X built-in functions without the
4539 "__builtin_" prefix are enabled, and return true for C and
4540 Objective C in the default hook. Include langhooks.h in
4542 * targhooks.h (default_floatn_builtin_p): Likewise.
4543 * targhooks.c (default_floatn_builtin_p): Likewise.
4544 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4545 floatn_builtin_p target hook.
4546 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4548 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4549 Eric Botcazou <ebotcazou@adacore.com>
4551 PR rtl-optimization/81803
4552 * lra-constraints.c (curr_insn_transform): Also reload the whole
4553 register for a strict subreg no wider than a word if this is for
4554 a WORD_REGISTER_OPERATIONS target.
4556 2017-10-31 Jason Merrill <jason@redhat.com>
4558 * gdbinit.in: Skip over inlines from timevar.h.
4560 2017-10-31 Martin Liska <mliska@suse.cz>
4562 * doc/gcov.texi: Document new option.
4563 * gcov.c (print_usage): Likewise print it.
4564 (process_args): Support the argument.
4565 (format_count): New function.
4566 (format_gcov): Use the function.
4568 2017-10-31 Martin Liska <mliska@suse.cz>
4570 * gcov.c (struct name_map): do not use typedef.
4571 Define operator== and operator<.
4572 (name_search): Remove.
4573 (name_sort): Remove.
4574 (main): Do not allocate names.
4575 (process_file): Add vertical space.
4576 (generate_results): Use std::find.
4577 (release_structures): Do not release memory.
4578 (find_source): Use std::find.
4580 2017-10-31 Martin Liska <mliska@suse.cz>
4582 * gcov.c (struct line_info): Remove it's typedef.
4583 (line_info::line_info): Add proper ctor.
4584 (line_info::has_block): Do not use a typedef.
4585 (struct source_info): Do not use typedef.
4586 (circuit): Likewise.
4587 (get_cycles_count): Likewise.
4588 (output_intermediate_file): Iterate via vector iterator.
4589 (add_line_counts): Use std::vector methods.
4590 (accumulate_line_counts): Likewise.
4591 (output_lines): Likewise.
4593 2017-10-31 Martin Liska <mliska@suse.cz>
4595 * gcov.c (struct source_info): Remove typedef.
4596 (source_info::source_info): Add proper ctor.
4597 (accumulate_line_counts): Use struct, not it's typedef.
4598 (output_gcov_file): Likewise.
4599 (output_lines): Likewise.
4600 (main): Do not allocate an array.
4601 (output_intermediate_file): Use size of vector container.
4602 (process_file): Resize the vector.
4603 (generate_results): Do not preallocate, use newly added vector
4605 (release_structures): Do not release sources.
4606 (find_source): Use vector methods.
4607 (add_line_counts): Do not use typedef.
4609 2017-10-31 Martin Liska <mliska@suse.cz>
4611 * doc/gcov.texi: Document that.
4612 * gcov.c (add_line_counts): Mark lines with a non-executed
4614 (output_line_beginning): Handle such lines.
4615 (output_lines): Pass new argument.
4616 (output_intermediate_file): Print it in intermediate format.
4618 2017-10-31 Martin Liska <mliska@suse.cz>
4620 * color-macros.h: New file.
4621 * diagnostic-color.c: Factor out color related to macros to
4623 * doc/gcov.texi: Document -k option.
4624 * gcov.c (INCLUDE_STRING): Include string.h.
4625 (print_usage): Add -k option.
4626 (process_args): Parse it.
4627 (pad_count_string): New function.
4628 (output_line_beginning): Likewise.
4629 (DEFAULT_LINE_START): New macro.
4630 (output_lines): Support color output.
4632 2017-10-31 Martin Liska <mliska@suse.cz>
4634 PR gcov-profile/82633
4635 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4636 their interaction with GCOV infrastructure.
4637 * configure.ac: Add -fkeep-{inline,static}-functions to
4639 * configure: Regenerate.
4641 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4644 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4646 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4649 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4650 into a register if it does not fit into an immediate offset field.
4652 2017-10-31 Olivier Hainque <hainque@adacore.com>
4654 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4656 2017-10-31 Julia Koval <julia.koval@intel.com>
4658 * config.gcc: Add gfniintrin.h.
4659 * config/i386/gfniintrin.h: New.
4660 * config/i386/i386-builtin-types.def
4661 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4662 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4663 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4664 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4665 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4666 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4667 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4668 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4669 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4670 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4671 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4672 * config/i386/immintrin.h: Include gfniintrin.h.
4673 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4675 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4677 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4679 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4681 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4682 (arm_ashldi3_1bit): Remove pattern.
4683 (ashrdi3): Remove shift by 1 expansion.
4684 (arm_ashrdi3_1bit): Remove pattern.
4685 (lshrdi3): Remove shift by 1 expansion.
4686 (arm_lshrdi3_1bit): Remove pattern.
4687 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4688 cost of ashldi3 by 1.
4689 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4690 (<shift>di3_neon): Likewise.
4692 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4694 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4695 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4696 and (*aarch64_simd_mov<VQ:mode>).
4697 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4698 pattern alternative.
4699 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4700 attributes to match pattern alternative.
4702 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4704 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4705 * config/rs6000/emmintrin.h: New file.
4706 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4708 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4710 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4711 * (movdi_vfp_cortexa8): Remove pattern.
4713 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4715 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4717 (Specific, alpha*-dec-osf5.1): Remove.
4718 (Specific, mips-sgi-irix5): Remove.
4719 (Specific, mips-sgi-irix6): Remove.
4721 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4724 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4725 arguments to clear_bit_region_be.
4727 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4729 * gimplify.c: Include memmodel.h.
4731 2017-10-30 Martin Jambor <mjambor@suse.cz>
4733 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4734 condition whether loop should be executed at all.
4736 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4738 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4739 gimple folding of vec_madd() intrinsics.
4740 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4741 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4742 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4744 2017-10-30 Richard Biener <rguenther@suse.de>
4746 PR tree-optimization/82762
4748 2017-10-23 Richard Biener <rguenther@suse.de>
4750 PR tree-optimization/82129
4752 2017-08-01 Richard Biener <rguenther@suse.de>
4754 PR tree-optimization/81181
4755 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4756 (compute_antic): ... end of iteration here.
4758 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4760 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4761 -std=iso9899:2017 and -std=gnu17.
4762 * doc/standards.texi (C Language): Document C17 support.
4763 * doc/cpp.texi (Overview): Mention -std=c17.
4764 (Standard Predefined Macros): Document C11 and C17 values of
4765 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4766 * doc/extend.texi (Inline): Do not list individual options for
4767 standards newer than C99.
4768 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4770 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4773 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4775 * asan.c (asan_finish_file): Align asan globals array by shadow
4778 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4781 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4782 (struct store_immediate_info): Add bitregion_start and bitregion_end
4784 (store_immediate_info::store_immediate_info): Add brs and bre
4785 arguments and initialize bitregion_{start,end} from those.
4786 (struct merged_store_group): Add bitregion_start, bitregion_end,
4787 align_base and mask fields. Drop unnecessary struct keyword from
4788 struct store_immediate_info. Add do_merge method.
4789 (clear_bit_region_be): Use memset instead of loop storing zeros.
4790 (merged_store_group::do_merge): New method.
4791 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4792 stores as long as the surrounding bitregions have no gaps.
4793 (merged_store_group::merge_overlapping): Use do_merge.
4794 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4795 is byte aligned, rather than requiring that start and width are
4796 byte aligned. Drop unnecessary struct keyword from
4797 struct store_immediate_info. Allocate and populate also mask array.
4798 Make start of the arrays relative to bitregion_start rather than
4799 start and size them according to bitregion_{end,start} difference.
4800 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4801 struct store_immediate_info.
4802 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4803 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4804 struct keyword from struct store_immediate_info.
4805 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4806 between stores as long as the surrounding bitregions have no gaps.
4808 (struct split_store): Add orig non-static data member.
4809 (split_store::split_store): Initialize orig to false.
4810 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4811 if there is exactly a single original stmt. Change stmts argument
4812 to pointer from reference, if NULL, don't push anything to it. Add
4813 first argument, use it to optimize skipping over orig stmts that
4814 are known to be before bitpos already. Simplify.
4815 (split_group): Return unsigned int count how many stores are or
4816 would be needed rather than a bool. Add allow_unaligned argument.
4817 Change split_stores argument from reference to pointer, if NULL,
4818 only do a dry run computing how many stores would be produced.
4819 Rewritten algorithm to use both alignment and misalign if
4820 !allow_unaligned and handle bitfield stores with gaps.
4821 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4822 from bitregion_start instead of start. Compute allow_unaligned
4823 here, if true, do 2 split_group dry runs to compute which one
4824 produces fewer stores and prefer aligned if equal. Punt if
4825 new count is bigger or equal than original before emitting any
4826 statements, rather than during that. Remove no longer needed
4827 new_ssa_names tracking. Replace num_stmts with
4828 split_stores.length (). Use 32-bit stack allocated entries
4829 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4830 if possible. Handle bitfields with gaps.
4831 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4832 Compute bitregion_{start,end} for the stores and construct
4833 store_immediate_info with that. Formatting fixes.
4835 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4838 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4839 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4841 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4843 * gimplify.c: Include tm_p.h.
4845 * common.opt (gcoff): Re-add as ignored option.
4846 (gcoff1, gcoff2, gcoff3): Likewise.
4848 * Makefile.in (OBJS): Delete sdbout.o.
4849 (GTFILES): Delete $(srcdir)/sdbout.c.
4850 * debug.h: Delete sdb_debug_hooks.
4851 * final.c: Delete sdbout.h include.
4852 (final_scan_insn): Delete SDB_DEBUG check.
4853 (rest_of_clean_state): Likewise.
4854 * output.h: Delete sdb_begin_function_line.
4857 * toplev.c: Delete sdbout.h include.
4858 (process_options): Delete SDB_DEBUG check.
4859 * tree-core.h (tree_type_common): Delete pointer field of
4861 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4862 TYPE_SYMTAB_POINTER.
4863 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4864 (TYPE_SYMTAB_IS_POINTER): Delete.
4865 (TYPE_SYMTAB_IS_DIE): Renumber.
4866 * xcoffout.c: Refer to former sdbout.c file.
4867 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4869 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4870 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4871 (Debugging Options): Delete -gcoff.
4872 (-gstabs): Delete SDB reference.
4874 (-gcoff@var{level}): Delete.
4875 * doc/passes.texi (Debugging information output): Delete SDB and
4876 sdbout.c references.
4877 * doc/tm.texi: Regenerate.
4878 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4879 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4881 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4882 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4883 (SDB_DEBUGGING_INFO): Delete.
4884 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4885 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4886 * target.def (output_source_filename): Delete COFF reference.
4888 * common.opt (gcoff): Delete.
4889 (gxcoff+): Update Negative chain.
4890 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4892 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4893 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4894 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4896 (expand_function_start): Change sdb reference to past tense.
4897 (expand_function_end): Change sdb reference to past tense.
4898 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4899 * opts.c (debug_type_names): Delete coff entry.
4900 (common_handle_option): Delete OPT_gcoff case.
4901 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4903 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4904 * config/cris/cris.h: Delete SDB reference in comment.
4905 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4906 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4907 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4908 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4910 (ix86_expand_prologue): Likewise.
4911 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4912 * config/ia64/ia64.h: Likewise.
4913 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4914 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4916 * config/mmix/mmix.h: Likewise.
4917 * config/nds32/nds32.c: Likewise.
4918 * config/stormy/storym16.h: Likewise.
4919 * config/visium/visium.h: Likewise.
4920 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4922 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4924 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4925 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4926 FRAME_POINTER_REGNUM point at high end of local var area.
4928 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4930 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4931 Move comment around. Do not reset best_edge for a copiable
4932 destination if the copy would cause a partition change.
4933 (better_edge_p): Remove redundant check.
4935 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4937 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4939 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4941 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4942 for math functions that have _Float<N> and _Float<N>X variants.
4943 (mathfn_built_in_2): Add support for math functions that have
4944 _Float<N> and _Float<N>X variants.
4945 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4946 (expand_builtin_mathfn_ternary): Add support for fma with
4947 _Float<N> and _Float<N>X variants.
4948 (expand_builtin): Likewise.
4949 (fold_builtin_3): Likewise.
4950 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4951 create math function _Float<N> and _Float<N>X variants as external
4953 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4954 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4955 the __builtin_ prefix and if not strict ansi, without the prefix.
4956 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4957 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4958 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4959 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4960 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4961 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4962 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4963 function signatures for fma _Float<N> and _Float<N>X variants.
4964 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4965 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4966 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4967 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4968 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4969 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4970 * gencfn-macros.c (print_case_cfn): Add support for math functions
4971 that have _Float<N> and _Float<N>X variants.
4972 (print_define_operator_list): Likewise.
4973 (fltfn_suffixes): Likewise.
4975 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4976 for math functions that have _Float<N> and _Float<N>X variants.
4977 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4978 and _Float<N>X variants.
4979 (COPYSIGN): Likewise.
4982 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4983 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4985 (integer_valued_read_call_p): Likewise.
4986 * fold-const-call.c (fold_const_call_ss): Likewise.
4987 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4988 _Float<N> and _Float<N>X variants.
4989 (fold_const_call_ssss): Add support for fma _Float<N> and
4990 _Float<N>X variants.
4991 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4992 support for copysign and fma _Float<N> and _Float<N>X variants.
4993 (backprop::process_builtin_call_use): Likewise.
4994 * tree-call-cdce.c (can_test_argument_range); Add support for
4995 sqrt _Float<N> and _Float<N>X variants.
4996 (edom_only_function): Likewise.
4997 (get_no_error_domain): Likewise.
4998 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4999 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5000 copysign _Float<N> and _Float<N>X variants.
5001 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5002 handled by machine independent code.
5003 (FMAF128): Likewise.
5004 * doc/cpp.texi (Common Predefined Macros): Document defining
5005 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5006 fma _Float<N> and _Float<N>X variants.
5008 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5011 * config/i386/i386-modes.def (CCFPU): Remove definition.
5012 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5013 (ix86_cc_modes_compatible): Ditto.
5014 (ix86_expand_carry_flag_compare): Ditto.
5015 (ix86_expand_int_movcc): Ditto.
5016 (ix86_expand_int_addcc): Ditto.
5017 (ix86_reverse_condition): Ditto.
5018 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5019 Return true/false for unordered/ordered fp comparisons.
5020 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5021 (ix86_prepare_fp_compare_args): Update for rename.
5022 (ix86_expand_fp_compare): Update for rename. Generate unordered
5023 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5024 (ix86_expand_sse_compare_and_jump): Ditto.
5025 * config/i386/predicates.md (fcmov_comparison_operator):
5026 Remove CCFPU mode handling.
5027 (ix86_comparison_operator): Ditto.
5028 (ix86_carry_flag_operator): Ditto.
5029 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5030 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5031 (*cmpu<mode>_cc_i387): Ditto.
5032 (FPCMP): Remove mode iterator.
5033 (unord): Remove mode attribute.
5034 (unord_subst): New define_subst transformation
5035 (unord): New define_subst attribute.
5037 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5038 (*cmpi<unord>xf_i387): Ditto.
5039 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5040 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5041 using unord_subst transformation.
5042 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5043 (round_saeonly): Also handle CCFP mode.
5044 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5045 Remove UNSPEC_SAHF unspec handling.
5047 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5049 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5051 2017-10-27 Jeff Law <law@redhat.com>
5053 * gimple-ssa-sprintf.c: Include domwalk.h.
5054 (class sprintf_dom_walker): New class, derived from dom_walker.
5055 (sprintf_dom_walker::before_dom_children): New function.
5056 (struct call_info): Moved into sprintf_dom_walker class
5057 (compute_formath_length, handle_gimple_call): Likewise.
5058 (sprintf_length::execute): Call the dominator walker rather
5059 than walking the statements.
5061 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5062 gimple statement locations.
5063 (check_array_bounds): Corresponding changes. Get the statement's
5064 location directly from wi->stmt.
5066 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5069 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5071 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5073 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5074 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5076 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5079 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5080 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5081 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5082 maybe_get_pool_constant.
5083 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5086 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5088 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5089 2.26 caveat. Update gas and gld versions.
5090 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5093 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5095 * cgraph.h (set_malloc_flag): Declare.
5096 * cgraph.c (set_malloc_flag_1): New function.
5097 (set_malloc_flag): Likewise.
5098 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5099 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5101 (read_ipa_call_summary): Add support for reading is_return_callee.
5102 (write_ipa_call_summary): Stream is_return_callee.
5103 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5104 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5105 ipa-prop.h, ipa-fnsummary.h.
5106 (pure_const_names): Change to static.
5107 (malloc_state_e): Define.
5108 (malloc_state_names): Define.
5109 (funct_state_d): Add field malloc_state.
5110 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5111 (check_retval_uses): New function.
5112 (malloc_candidate_p): Likewise.
5113 (analyze_function): Add support for malloc attribute.
5114 (pure_const_write_summary): Stream malloc_state.
5115 (pure_const_read_summary): Add support for reading malloc_state.
5116 (dump_malloc_lattice): New function.
5117 (propagate_malloc): New function.
5118 (warn_function_malloc): New function.
5119 (ipa_pure_const::execute): Call propagate_malloc and
5120 ipa_free_fn_summary.
5121 (pass_local_pure_const::execute): Add support for malloc attribute.
5122 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5123 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5125 2017-10-27 Martin Liska <mliska@suse.cz>
5127 PR gcov-profile/82457
5128 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5129 for fork-like functions to be properly instrumented.
5131 2017-10-27 Richard Biener <rguenther@suse.de>
5134 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5135 info when we redirected EH.
5137 2017-10-26 Michael Collison <michael.collison@arm.com>
5139 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5141 (<optab>_trunchf<GPI:mode>2: New pattern.
5142 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5143 * config/aarch64/iterators.md (wv): New mode attribute.
5144 (vf, VF): New mode attributes.
5145 (vgp, VGP): New mode attributes.
5146 (s): Update attribute with SImode and DImode prefixes.
5148 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5150 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5151 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5152 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5153 (nios2_option_overide): Initialize it. Don't allow R0-relative
5154 addressing with PIC.
5155 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5156 (nios2_symbolic_constant_p): Likewise.
5157 (nios2_legitimate_address_p): Likewise.
5158 (nios2_r0rel_section_name_p): New.
5159 (nios2_symbol_ref_in_r0rel_data_p): New.
5160 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5161 (r0rel_constant_p): New.
5162 (nios2_print_operand_address): Handle r0rel_constant_p.
5163 (nios2_cdx_narrow_form_p): Likewise.
5164 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5165 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5166 (Nios II Options): Document -mr0rel-sec.
5168 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5170 * config/nios2/nios2.c: Include xregex.h.
5171 (nios2_gprel_sec_regex): New.
5172 (nios2_option_overide): Initialize it. Don't allow GP-relative
5173 addressing with PIC.
5174 (nios2_small_section_name_p): Check for regex match.
5175 * config/nios2/nios2.opt (mgprel-sec=): New option.
5176 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5177 (Nios II Options): Document -mgprel-sec.
5179 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5181 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5183 2017-10-26 Tom de Vries <tom@codesourcery.com>
5185 PR tree-optimization/82707
5186 * gimple.c (gimple_copy): Fix unsharing of
5187 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5189 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5191 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5192 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5193 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5194 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5195 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5196 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5197 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5198 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5199 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5200 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5201 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5202 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5203 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5204 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5205 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5206 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5207 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5209 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5211 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5213 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5214 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5215 warning to rs6000.c. Remove the Undocumented flag, since it has
5217 (-mabi=ibmlongdouble): Likewise.
5218 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5219 already set, set the default format for long double.
5220 (rs6000_debug_reg_global): Print whether long double is IBM or
5222 (rs6000_option_override_internal): Rework setting long double
5223 format. Only warn if the user is changing the long double default
5224 and they did not use -Wno-psabi.
5225 * doc/invoke.texi (PowerPC options): Update the documentation for
5226 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5228 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5229 Alan Hayward <alan.hayward@arm.com>
5230 David Sherwood <david.sherwood@arm.com>
5232 * rtl.h (wider_subreg_mode): New function.
5233 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5234 rather than an unsigned int *.
5235 * ira-color.c (regno_max_ref_width): Replace with...
5236 (regno_max_ref_mode): ...this new variable.
5237 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5238 Use wider_subreg_mode.
5239 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5240 rather than an unsigned int *.
5241 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5242 (process_alt_operands): Likewise.
5243 (invariant_p): Likewise.
5244 * lra-spills.c (assign_mem_slot): Likewise.
5245 (add_pseudo_to_slot): Likewise.
5246 * lra.c (collect_non_operand_hard_regs): Likewise.
5247 (add_regs_to_insn_regno_info): Likewise.
5248 * reload1.c (regno_max_ref_width): Replace with...
5249 (regno_max_ref_mode): ...this new variable.
5250 (reload): Update accordingly. Update call to
5251 ira_sort_regnos_for_alter_reg.
5252 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5253 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5254 (scan_paradoxical_subregs): Likewise.
5256 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5258 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5259 (aarch64_frame): Add emit_frame_chain boolean.
5260 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5261 Move eh_return case to aarch64_layout_frame.
5262 (aarch64_layout_frame): Initialize emit_frame_chain.
5263 (aarch64_expand_prologue): Use emit_frame_chain.
5265 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5267 * config/aarch64/aarch64.c (aarch64_layout_frame):
5268 Ensure LR is always stored at the bottom of the callee-saves.
5269 Remove rarely used frame layout which saves callee-saves at top of
5270 frame, so the store of LR can be used as a valid probe in all cases.
5272 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5274 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5275 Improve unaligned TImode/TFmode base/offset split.
5277 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5278 Alan Hayward <alan.hayward@arm.com>
5279 David Sherwood <david.sherwood@arm.com>
5281 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5282 * combine.c (find_single_use_1): Likewise.
5283 (expand_field_assignment): Likewise.
5284 (move_deaths): Likewise.
5285 * lra-constraints.c (simplify_operand_subreg): Likewise.
5286 (curr_insn_transform): Likewise.
5287 * lra.c (collect_non_operand_hard_regs): Likewise.
5288 (add_regs_to_insn_regno_info): Likewise.
5289 * rtlanal.c (reg_referenced_p): Likewise.
5290 (covers_regno_no_parallel_p): Likewise.
5292 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5294 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5295 Don't print any bits outside the precision of the value.
5296 * wide-int.cc (test_printing): Add some new tests.
5298 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5300 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5301 supports -xbrace_comment option.
5302 * configure: Regenerate.
5303 * config.in: Regenerate.
5304 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5305 (ASM_CPU_SPEC): Use it.
5307 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5309 * target.def (static_rtx_alignment): New hook.
5310 * targhooks.h (default_static_rtx_alignment): Declare.
5311 * targhooks.c (default_static_rtx_alignment): New function.
5312 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5313 * doc/tm.texi: Regenerate.
5314 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5315 instead of targetm.constant_alignment. Remove call to
5317 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5318 (cris_preferred_mininum_alignment): New function, split out from...
5319 (cris_constant_alignment): ...here.
5320 (cris_static_rtx_alignment): New function.
5321 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5323 (ix86_constant_alignment): ...here.
5324 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5325 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5326 (mmix_static_rtx_alignment): New function.
5327 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5328 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5330 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5333 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5334 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5336 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5338 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5339 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5340 mark as a sign-extending load.
5341 (local_pic_load_u): Define.
5343 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5346 * fold-const.c (operand_equal_for_comparison_p): Also return true
5347 if ARG0 is a simple variant of ARG1 with narrower precision.
5348 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5350 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5352 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5354 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5355 scatter_static, scatter_per_elt.
5356 * x86-tune-costs.h: Add new cost entries.
5358 2017-10-25 Richard Biener <rguenther@suse.de>
5360 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5361 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5362 class pass_fre): Move to ...
5363 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5364 class pass_fre): ... here and adjust for statistics.
5366 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5369 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5370 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5371 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5372 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5375 2017-10-25 Richard Biener <rguenther@suse.de>
5377 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5378 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5379 eliminate_push_avail, eliminate_insert): Move inside...
5380 (class eliminate_dom_walker): ... this class in preparation
5382 (fini_eliminate): Remove by merging with ...
5383 (eliminate): ... this function. Adjust for class changes.
5384 (pass_pre::execute): Remove fini_eliminate call.
5385 (pass_fre::execute): Likewise.
5387 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5390 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5391 (VPERMI2, VPERMI2I): New mode iterators.
5392 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5393 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5395 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5396 mode iterator. Remove 3 old define_insn patterns.
5397 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5398 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5399 VPERMI2 mode iterator, remove the other two expanders.
5400 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5401 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5402 instructions, remove the other two patterns.
5403 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5404 mode iterator, remove the other two patterns.
5405 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5406 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5407 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5408 and adjust argument order accordingly.
5409 (ix86_expand_vec_perm): Adjust caller.
5410 (expand_vec_perm_1): Likewise.
5411 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5412 (expand_vec_perm_vpermt2_vpshub2): ... this.
5413 (ix86_expand_vec_perm_const_1): Adjust caller.
5414 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5417 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5418 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5419 (vec_shl_<mode>): Remove unused expander.
5420 (avx512bw_<shift_insn><mode>3): New define_insn.
5421 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5422 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5424 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5427 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5430 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5432 PR rtl-optimization/82396
5433 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5434 (autopref_multipass_init): Simplify initialization.
5435 (autopref_rank_data): Simplify sort order.
5436 * gcc/sched-int.h (autopref_multipass_data_): Remove
5437 multi_mem_insn_p, min_offset and max_offset.
5439 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5442 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5443 Check special value of flag_omit_frame_pointer.
5444 (aarch64_can_eliminate): Likewise.
5445 (aarch64_override_options_after_change_1): Simplify handling of
5446 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5448 2017-10-24 Richard Biener <rguenther@suse.de>
5450 PR tree-optimization/82697
5451 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5452 zero for conditional load and unconditional store.
5454 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5456 * doc/install.texi: Document bootstrap-cet.
5458 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5461 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5462 ENDBR instruction at function entrance if function is only
5465 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5468 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5469 patterns to better describe from which operation the CF is computed.
5470 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5471 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5472 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5473 is 0, use _0 suffixed expanders instead of emitting a comparison
5476 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5478 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5479 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5481 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5484 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5485 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5486 * loop-iv.c (iv_get_reaching_def): Likewise.
5487 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5488 variable is promoted and the partition contains undefined values.
5490 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5492 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5494 (nios2_address_cost): Define.
5495 (nios2_legitimize_address): Recognize (exp + constant) directly.
5496 (TARGET_ADDRESS_COST): Define.
5498 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5500 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5501 (nios2_symbolic_memory_operand_p): Declare.
5502 (nios2_split_large_constant): Declare.
5503 (nios2_split_symbolic_memory_operand): Declare.
5504 * config/nios2/nios2.c: Adjust includes.
5505 (nios2_symbolic_constant_allowed): New.
5506 (nios2_symbolic_constant_p): New.
5507 (nios2_plus_symbolic_constant_p): New.
5508 (nios2_valid_addr_expr_p): Recognize addresses involving
5510 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5511 (nios2_symbolic_memory_operand_p): New.
5512 (nios2_large_constant_p): New.
5513 (nios2_split_large_constant): New.
5514 (nios2_split_plus_large_constant): New.
5515 (nios2_split_symbolic_memory_operand): New.
5516 (nios2_legitimize_address): Code refactoring. Handle addresses
5517 involving symbolic constants.
5518 (nios2_emit_move_sequence): Likewise.
5519 (nios2_print_operand): Improve error output.
5520 (nios2_print_operand_address): Handle LO_SUM.
5521 (nios2_cdx_narrow_form_p): Likewise.
5522 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5523 operands involving symbolic constants.
5524 (movhi_internal, movsi_internal): Likewise.
5525 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5526 (extendhisi2, extendqi<mode>2): Likewise.
5528 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5530 * tree-pass.h (PROP_rtl_split_insns): Define.
5531 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5533 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5535 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5537 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5540 * target.def (const_not_ok_for_debug_p): Default to
5541 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5542 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5543 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5544 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5545 which targetm.const_not_ok_for_debug_p returned true.
5546 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5548 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5550 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5551 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5552 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5554 (ix86_const_not_ok_for_debug_p): New function.
5555 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5556 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5558 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5561 * system.h: Conditionally include "unique-ptr.h" if
5562 INCLUDE_UNIQUE_PTR is defined.
5563 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5564 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5566 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5568 * config/rl78/rl78.md: New define_expand "subdi3".
5570 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5573 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5574 DF_REF_INSN if DF_REF_INSN_INFO is false.
5576 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5578 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5579 xmm_move instead of sse_move.
5580 (sse_store_index): New function.
5581 (ix86_register_move_cost): Be more sensible about mismatch stall;
5582 model AVX moves correctly; make difference between sse->integer and
5584 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5585 moves; make difference between SSE and AVX.
5586 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5587 and zmm_move. Increase size of sse load and store tables;
5588 add unaligned load and store tables; add ssemmx_to_integer.
5589 * x86-tune-costs.h: Update all entries according to real
5590 move latencies from Agner Fog's manual and chip documentation.
5592 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5595 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5596 * config/i386/constraints.md (Wf): New constraint.
5597 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5598 (cmp<dwi>_doubleword): Removed.
5599 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5600 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5601 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5602 expand with cmp<dwi>_doubleword. For LTU and GEU use
5603 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5605 * common.opt (gcolumn-info): Enable by default.
5606 * doc/invoke.texi (gcolumn-info): Document new default.
5608 2017-10-23 Richard Biener <rguenther@suse.de>
5610 PR tree-optimization/82672
5611 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5612 Fold the stmt if we propagated into it.
5614 2017-10-23 Richard Biener <rguenther@suse.de>
5616 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5617 (bitmap_remove_expr_from_set): ... this. All callers call this
5618 for non-constant values.
5619 (bitmap_set_subtract): Rename to...
5620 (bitmap_set_subtract_expressions): ... this. Adjust and
5622 (bitmap_set_contains_value): Remove superfluous check.
5623 (bitmap_set_replace_value): Inline into single caller ...
5624 (bitmap_value_replace_in_set): ... here and simplify.
5625 (dependent_clean): Merge into ...
5626 (clean): ... this using an overload. Adjust.
5627 (prune_clobbered_mems): Adjust.
5628 (compute_antic_aux): Likewise.
5629 (compute_partial_antic_aux): Likewise.
5631 2017-10-23 Richard Biener <rguenther@suse.de>
5633 PR tree-optimization/82129
5635 2017-08-01 Richard Biener <rguenther@suse.de>
5637 PR tree-optimization/81181
5638 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5639 (compute_antic): ... end of iteration here.
5641 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5643 * target.def (starting_frame_offset): New hook.
5644 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5645 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5646 * doc/tm.texi.in: Regenerate.
5647 * hooks.h (hook_hwi_void_0): Declare.
5648 * hooks.c (hook_hwi_void_0): New function.
5649 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5650 STARTING_FRAME_OFFSET.
5651 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5652 * reload1.c (reload): Likewise.
5653 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5654 instead of STARTING_FRAME_OFFSET.
5655 * function.c (try_fit_stack_local): Likewise.
5656 (assign_stack_local_1): Likewise
5657 (instantiate_virtual_regs): Likewise.
5658 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5659 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5660 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5661 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5662 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5663 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5664 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5665 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5666 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5667 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5668 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5669 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5670 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5671 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5672 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5673 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5674 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5675 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5676 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5677 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5678 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5679 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5680 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5681 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5682 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5683 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5684 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5685 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5686 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5687 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5688 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5689 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5690 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5691 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5692 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5693 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5694 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5695 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5696 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5697 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5698 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5699 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5700 return a HOST_WIDE_INT.
5701 (avr_builtin_setjmp_frame_value): Use it instead of
5702 STARTING_FRAME_OFFSET.
5703 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5704 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5705 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5707 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5708 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5709 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5710 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5711 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5712 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5713 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5714 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5715 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5716 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5717 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5718 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5720 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5721 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5722 * config/mips/mips.c (mips_compute_frame_info): Refer to
5723 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5724 (mips_starting_frame_offset): New function.
5725 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5726 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5727 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5728 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5729 and return a HOST_WIDE_INT.
5730 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5731 (mmix_initial_elimination_offset): Refer to
5732 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5733 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5734 * config/pa/pa.c (pa_starting_frame_offset): New function.
5735 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5736 (pa_expand_prologue): Likewise.
5737 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5738 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5739 !FRAME_GROWS_DOWNWARD handling to...
5740 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5741 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5742 !FRAME_GROWS_DOWNWARD handling to...
5743 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5744 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5745 !FRAME_GROWS_DOWNWARD handling to...
5746 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5747 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5749 (rs6000_starting_frame_offset): New function.
5750 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5751 !FRAME_GROWS_DOWNWARD handling to...
5752 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5753 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5754 !FRAME_GROWS_DOWNWARD handling to...
5755 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5756 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5757 !FRAME_GROWS_DOWNWARD handling to...
5758 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5759 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5760 (rs6000_starting_frame_offset): New function.
5761 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5762 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5763 * config/vax/vax.c (vax_starting_frame_offset): New function.
5764 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5765 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5766 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5767 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5768 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5769 * system.h (STARTING_FRAME_OFFSET): Poison.
5771 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5773 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5774 SCALAR_TYPE_MODE instead of TYPE_MODE.
5776 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5777 Alan Hayward <alan.hayward@arm.com>
5778 David Sherwood <david.sherwood@arm.com>
5780 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5782 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5783 Alan Hayward <alan.hayward@arm.com>
5784 David Sherwood <david.sherwood@arm.com>
5786 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5788 2017-10-23 Richard Biener <rguenther@suse.de>
5790 PR tree-optimization/82129
5791 * tree-ssa-pre.c (bitmap_set_and): Remove.
5792 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5793 canonicalizing expressions in the set to those with lowest
5794 ID rather than taking that from the first edge.
5796 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5798 * combine.c (rtx_equal_for_field_assignment_p): Use
5799 byte_lowpart_offset.
5801 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5802 Alan Hayward <alan.hayward@arm.com>
5803 David Sherwood <david.sherwood@arm.com>
5805 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5806 to a SUBREG_PROMOTED_VAR.
5808 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5809 Alan Hayward <alan.hayward@arm.com>
5810 David Sherwood <david.sherwood@arm.com>
5812 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5813 (expand_debug_source_expr): Likewise.
5814 * combine.c (combine_simplify_rtx): Likewise.
5815 * cse.c (fold_rtx): Likewise.
5816 * optabs.c (expand_float): Likewise.
5817 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5818 (simplify_binary_operation_1): Likewise.
5820 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5821 Alan Hayward <alan.hayward@arm.com>
5822 David Sherwood <david.sherwood@arm.com>
5824 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5825 (record_promoted_value): Likewise.
5826 * expr.c (expand_expr_real_2): Likewise.
5827 * ree.c (update_reg_equal_equiv_notes): Likewise.
5828 (combine_set_extension): Likewise.
5829 * rtlanal.c (low_bitmask_len): Likewise.
5830 * simplify-rtx.c (neg_const_int): Likewise.
5831 (simplify_binary_operation_1): Likewise.
5833 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5834 Alan Hayward <alan.hayward@arm.com>
5835 David Sherwood <david.sherwood@arm.com>
5837 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5838 * regcprop.c (maybe_mode_change): Likewise.
5839 * reload1.c (alter_reg): Likewise.
5841 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5843 * inchash.h (inchash::hash::add_wide_int): New function.
5844 * lto-streamer-out.c (hash_tree): Use it.
5846 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5848 * inchash.h (inchash::hash::add_wide_int): Rename to...
5849 (inchash::hash::add_hwi): ...this.
5850 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5851 (polymorphic_call_target_hasher::hash): Likewise.
5852 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5853 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5854 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5855 * lto-streamer-out.c (hash_tree): Likewise.
5856 * optc-save-gen.awk: Likewise.
5857 * tree.c (add_expr): Likewise.
5859 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5862 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5863 for ordered inequality comparisons even with TARGET_IEEE_FP.
5865 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5868 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5869 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5870 Expand with cmp<dwi>_doubleword.
5872 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5874 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5875 List CET intrinsics.
5876 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5877 specific to -fcf-protection option.
5879 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5881 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5882 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5883 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5884 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5885 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5886 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5887 (extra_objs): Add cet.o for Linux/x86 targets.
5888 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5889 * config/i386/cet.c: New file.
5890 * config/i386/cetintrin.h: Likewise.
5891 * config/i386/t-cet: Likewise.
5892 * config/i386/cpuid.h (bit_SHSTK): New.
5893 (bit_IBT): Likewise.
5894 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5895 pass IBT and SHSTK bits.
5896 * config/i386/i386-builtin-types.def
5897 (VOID_FTYPE_UNSIGNED_PVOID): New.
5898 (VOID_FTYPE_UINT64_PVOID): Likewise.
5899 * config/i386/i386-builtin.def: Add CET intrinsics.
5900 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5901 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5902 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5903 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5905 * config/i386/i386.c (rest_of_insert_endbranch): New.
5906 (pass_data_insert_endbranch): Likewise.
5907 (pass_insert_endbranch): Likewise.
5908 (make_pass_insert_endbranch): Likewise.
5909 (ix86_notrack_prefixed_insn_p): Likewise.
5910 (ix86_target_string): Add -mibt, -mshstk flags.
5911 (ix86_option_override_internal): Add flag_cf_protection
5913 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5914 (ix86_print_operand): Add 'notrack' prefix output.
5915 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5916 (ix86_expand_builtin): Expand CET intrinsics.
5917 (x86_output_mi_thunk): Add 'endbranch' instruction.
5918 * config/i386/i386.h (TARGET_IBT): New.
5919 (TARGET_IBT_P): Likewise.
5920 (TARGET_SHSTK): Likewise.
5921 (TARGET_SHSTK_P): Likewise.
5922 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5923 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5924 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5925 (builtin_setjmp_setup): New pattern.
5926 (builtin_longjmp): Likewise.
5927 (rdssp<mode>): Likewise.
5928 (incssp<mode>): Likewise.
5929 (saveprevssp): Likewise.
5930 (rstorssp): Likewise.
5931 (wrss<mode>): Likewise.
5932 (wruss<mode>): Likewise.
5933 (setssbsy): Likewise.
5934 (clrssbsy): Likewise.
5935 (nop_endbr): Likewise.
5936 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5938 * config/i386/immintrin.h: Include <cetintrin.h>.
5939 * config/i386/linux-common.h
5940 (file_end_indicate_exec_stack_and_cet): New prototype.
5941 (TARGET_ASM_FILE_END): New.
5943 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5945 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5946 latencies instead of having separate table; make difference between
5947 integer and float costs.
5948 * i386.h (processor_costs): Remove scalar_stmt_cost,
5949 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5950 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5952 * x86-tune-costs.h: Remove entries which has been removed in
5953 procesor_costs from all tables; make cond_taken_branch_cost
5954 and cond_not_taken_branch_cost COST_N_INSNS based.
5956 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5958 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5960 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5962 * config/i386/i386.md (isa): Remove fma_avx512f.
5963 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5964 <avx512>_fmadd_<mode>_mask3<round_name>,
5965 <avx512>_fmsub_<mode>_mask<round_name>,
5966 <avx512>_fmsub_<mode>_mask3<round_name>,
5967 <avx512>_fnmadd_<mode>_mask<round_name>,
5968 <avx512>_fnmadd_<mode>_mask3<round_name>,
5969 <avx512>_fnmsub_<mode>_mask<round_name>,
5970 <avx512>_fnmsub_<mode>_mask3<round_name>,
5971 <avx512>_fmaddsub_<mode>_mask<round_name>,
5972 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5973 <avx512>_fmsubadd_<mode>_mask<round_name>,
5974 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5975 (*vec_widen_umult_even_v16si<mask_name>,
5976 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5977 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5979 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5981 * extend.texi: Add 'nocf_check' documentation.
5982 * gimple.texi: Add second parameter to
5983 gimple_build_call_from_tree.
5984 * invoke.texi: Add -fcf-protection documentation.
5985 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5987 2017-10-20 Richard Biener <rguenther@suse.de>
5989 PR tree-optimization/82473
5990 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5991 the largest input type.
5993 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5995 * c-attribs.c (handle_nocf_check_attribute): New function.
5996 (c_common_attribute_table): Add 'nocf_check' handling.
5997 * gimple-parser.c: Add second argument NULL to
5998 gimple_build_call_from_tree.
5999 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6000 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6002 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6003 * common.opt: Add fcf-protection flag.
6004 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6005 * flag-types.h: Add enum cf_protection_level.
6006 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6007 Add 'nocf_check' attribute propagation to gimple call.
6008 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6009 (gimple_build_call_from_tree): Update prototype.
6010 (gimple_call_nocf_check_p): New function.
6011 (gimple_call_set_nocf_check): Likewise.
6012 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6013 * ipa-icf.c: Add nocf_check attribute in statement hash.
6014 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6015 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6016 * toplev.c (process_options): Add flag_cf_protection handling.
6018 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6020 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6022 2017-10-20 Richard Biener <rguenther@suse.de>
6024 PR tree-optimization/82603
6025 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6026 remove false predicated stores.
6028 2017-10-20 Richard Biener <rguenther@suse.de>
6030 * graphite-isl-ast-to-gimple.c
6031 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6032 Remove return value and simplify, dump copied stmt after lhs
6034 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6035 Reduce dump verbosity.
6036 (gsi_insert_earliest): Likewise.
6037 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6038 * graphite.c (print_global_statistics): Adjust dumping.
6039 (print_graphite_scop_statistics): Likewise.
6040 (print_graphite_statistics): Do not dump loops here.
6041 (graphite_transform_loops): But here.
6043 2017-10-20 Nicolas Roche <roche@adacore.com>
6045 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6046 * configure: Regenerate.
6048 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6051 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6052 functions when optimizing replace GIMPLE_RETURN stmts with
6053 calls to __builtin_unreachable ().
6056 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6057 for -fsanitize=thread link of executables.
6058 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6059 link of executables.
6062 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6064 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6065 define_insns for logical vector shifts to use VI248_AVX512BW
6066 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6067 condition, useless isa and prefix attributes. Change the first
6068 2 of these define_insns to ...
6069 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6070 define_insn for avx512vl.
6071 (<shift_insn><mode>3): ... and this, new define_insn without
6072 masking for non-avx512vl.
6075 * config/i386/sse.md (*andnot<mode>3,
6076 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6077 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6078 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6079 not applied use empty suffix even for TARGET_AVX512VL.
6080 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6081 is applied, supply evex,evex or evex,evex,evex instead of just
6084 2017-10-20 Julia Koval <julia.koval@intel.com>
6086 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6087 (OPTION_MASK_ISA_GFNI_UNSET): New.
6088 (ix86_handle_option): Handle OPT_mgfni.
6089 * config/i386/cpuid.h (bit_GFNI): New.
6090 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6091 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6092 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6093 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6094 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6095 * config/i386/i386.opt: Add mgfni.
6097 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6099 * config/msp430/msp430.c (msp430_option_override): Disable
6100 -fdelete-null-pointer-checks.
6101 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6103 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6105 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6106 of x87 and SSE instructions.
6108 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6110 * asan.c (create_cond_insert_point): Do not update edge count.
6111 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6112 (afdo_propagate_circuit): Likewise.
6113 (afdo_calculate_branch_prob): Likewise.
6114 (afdo_annotate_cfg): Likewise.
6115 * basic-block.h (struct edge_def): Remove count.
6116 (edge_def::count): New accessor.
6117 * bb-reorder.c (rotate_loop): Update.
6118 (find_traces_1_round): Update.
6119 (connect_traces): Update.
6120 (sanitize_hot_paths): Update.
6121 * cfg.c (unchecked_make_edge): Update.
6122 (make_single_succ_edge): Update.
6123 (check_bb_profile): Update.
6124 (dump_edge_info): Update.
6125 (update_bb_profile_for_threading): Update.
6126 (scale_bbs_frequencies_int): Update.
6127 (scale_bbs_frequencies_gcov_type): Update.
6128 (scale_bbs_frequencies_profile_count): Update.
6129 (scale_bbs_frequencies): Update.
6130 * cfganal.c (connect_infinite_loops_to_exit): Update.
6131 * cfgbuild.c (compute_outgoing_frequencies): Update.
6132 (find_many_sub_basic_blocks): Update.
6133 * cfgcleanup.c (try_forward_edges): Update.
6134 (try_crossjump_to_edge): Update
6135 * cfgexpand.c (expand_gimple_cond): Update
6136 (expand_gimple_tailcall): Update
6137 (construct_exit_block): Update
6138 * cfghooks.c (verify_flow_info): Update
6139 (redirect_edge_succ_nodup): Update
6140 (split_edge): Update
6141 (make_forwarder_block): Update
6142 (duplicate_block): Update
6143 (account_profile_record): Update
6144 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6145 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6146 * cfgloopmanip.c (scale_loop_profile): Update.
6148 (lv_adjust_loop_entry_edge): Update.
6149 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6150 (force_nonfallthru_and_redirect): Update.
6151 (purge_dead_edges): Update.
6152 (rtl_flow_call_edges_add): Update.
6153 * cgraphunit.c (init_lowered_empty_function): Update.
6154 (cgraph_node::expand_thunk): Update.
6155 * gimple-pretty-print.c (dump_probability): Update.
6156 (dump_edge_probability): Update.
6157 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6158 * haifa-sched.c (sched_create_recovery_edges): Update.
6159 * hsa-gen.c (convert_switch_statements): Update.
6160 * ifcvt.c (dead_or_predicable): Update.
6161 * ipa-inline-transform.c (inline_transform): Update.
6162 * ipa-split.c (split_function): Update.
6163 * ipa-utils.c (ipa_merge_profiles): Update.
6164 * loop-doloop.c (add_test): Update.
6165 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6166 * lto-streamer-in.c (input_cfg): Update.
6167 (input_function): Update.
6168 * lto-streamer-out.c (output_cfg): Update.
6169 * modulo-sched.c (sms_schedule): Update.
6170 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6171 * predict.c (maybe_hot_edge_p): Update.
6172 (unlikely_executed_edge_p): Update.
6173 (probably_never_executed_edge_p): Update.
6174 (dump_prediction): Update.
6175 (drop_profile): Update.
6176 (propagate_unlikely_bbs_forward): Update.
6177 (determine_unlikely_bbs): Update.
6178 (force_edge_cold): Update.
6179 * profile.c (compute_branch_probabilities): Update.
6180 * reg-stack.c (better_edge): Update.
6181 * shrink-wrap.c (handle_simple_exit): Update.
6182 * tracer.c (better_p): Update.
6183 * trans-mem.c (expand_transaction): Update.
6184 (split_bb_make_tm_edge): Update.
6185 * tree-call-cdce.c: Update.
6186 * tree-cfg.c (gimple_find_sub_bbs): Update.
6187 (gimple_split_edge): Update.
6188 (gimple_duplicate_sese_region): Update.
6189 (gimple_duplicate_sese_tail): Update.
6190 (gimple_flow_call_edges_add): Update.
6191 (insert_cond_bb): Update.
6192 (execute_fixup_cfg): Update.
6193 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6194 * tree-complex.c (expand_complex_div_wide): Update.
6195 * tree-eh.c (lower_resx): Update.
6196 (unsplit_eh): Update.
6197 (cleanup_empty_eh_move_lp): Update.
6198 * tree-inline.c (copy_edges_for_bb): Update.
6199 (freqs_to_counts): Update.
6200 (copy_cfg_body): Update.
6201 * tree-ssa-dce.c (remove_dead_stmt): Update.
6202 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6203 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6204 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6205 (unloop_loops): Update.
6206 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6207 * tree-ssa-loop-split.c (connect_loops): Update.
6208 (split_loop): Update.
6209 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6210 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6211 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6212 * tree-ssa-reassoc.c (branch_fixup): Update.
6213 * tree-ssa-tail-merge.c (replace_block_by): Update.
6214 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6215 (compute_path_counts): Update.
6216 (update_profile): Update.
6217 (recompute_probabilities): Update.
6218 (update_joiner_offpath_counts): Update.
6219 (estimated_freqs_path): Update.
6220 (freqs_to_counts_path): Update.
6221 (clear_counts_path): Update.
6222 (ssa_fix_duplicate_block_edges): Update.
6223 (duplicate_thread_path): Update.
6224 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6225 (case_bit_test_cmp): Update.
6226 (collect_switch_conv_info): Update.
6227 (gen_inbound_check): Update.
6228 (do_jump_if_equal): Update.
6229 (emit_cmp_and_jump_insns): Update.
6230 * tree-tailcall.c (decrease_profile): Update.
6231 (eliminate_tail_call): Update.
6232 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6233 (vect_do_peeling): Update.
6234 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6235 * ubsan.c (ubsan_expand_null_ifn): Update.
6236 (ubsan_expand_ptr_ifn): Update.
6237 * value-prof.c (gimple_divmod_fixed_value): Update.
6238 (gimple_mod_pow2): Update.
6239 (gimple_mod_subtract): Update.
6240 (gimple_ic): Update.
6241 (gimple_stringop_fixed_value): Update.
6243 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6246 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6248 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6250 PR rtl-optimization/82395
6251 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6252 based on non_spilled_static_chain_regno_p.
6254 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6256 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6257 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6258 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6260 2017-10-19 Martin Sebor <msebor@redhat.com>
6262 PR tree-optimization/82596
6263 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6265 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6267 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6268 (get_mem_refs_of_builtin_call): Likewise.
6269 * builtins.c (expand_builtin_apply): Adjust call to
6270 allocate_dynamic_stack_space.
6271 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6272 the third argument to allocate_dynamic_stack_space, otherwise -1.
6273 (expand_builtin): Deal with all alloca variants.
6274 (is_inexpensive_builtin): Likewise.
6275 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6276 * calls.c (special_function_p): Deal with all alloca variants.
6277 (initialize_argument_information): Adjust call to
6278 allocate_dynamic_stack_space.
6279 (expand_call): Likewise.
6280 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6281 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6282 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6283 use it for the stack usage computation.
6284 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6285 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6286 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6287 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6288 (in_loop_p): Remove first argument and useless check.
6289 (pass_walloca::execute): Remove useless test and adjust call to above.
6290 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6291 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6292 (gimplify_call_expr): Deal with all alloca variants.
6293 * hsa-gen.c (gen_hsa_alloca): Likewise.
6294 (gen_hsa_insns_for_call): Likewise.
6295 * ipa-pure-const.c (special_builtin_state): Likewise.
6296 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6297 * tree-object-size.c (alloc_object_size): Likewise.
6298 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6299 (call_may_clobber_ref_p_1): Likewise.
6300 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6301 (ccp_fold_stmt): Likewise.
6302 (optimize_stack_restore): Likewise.
6303 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6304 (mark_all_reaching_defs_necessary_1): Likewise.
6305 (propagate_necessity): Likewise.
6306 (eliminate_unnecessary_stmts): Likewise.
6307 * tree.c (build_common_builtin_nodes): Build
6308 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6309 (build_alloca_call_expr): New function.
6310 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6311 (CASE_BUILT_IN_ALLOCA): Likewise.
6312 (build_alloca_call_expr): Declare.
6313 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6315 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6318 * dwarf2out.c (new_die_raw): New static inline function.
6319 (new_die): Use it to create the DIE.
6320 (add_AT_external_die_ref): Likewise.
6321 (clone_die): Likewise.
6322 (clone_as_declaration): Likewise.
6323 (dwarf2out_vms_debug_main_pointer): Likewise.
6324 (base_type_die): Likewise. Remove early return for corner cases.
6325 Do not call add_pubtype on the DIE here.
6326 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6327 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6328 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6329 native order exists for base types, attach the DIE manually and call
6330 add_pubtype on it. Do not equate a reverse order DIE to the type.
6332 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6334 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6335 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6337 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6339 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6341 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6342 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6343 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6344 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6345 * opts.c (sanitizer_opts): Add builtin.
6346 * ubsan.c (instrument_builtin): New function.
6347 (pass_ubsan::execute): Call it.
6348 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6349 * doc/invoke.texi: Document -fsanitize=builtin.
6351 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6352 builtins, store max (log2 (align), 0) into uchar field instead of
6353 align into uptr field.
6354 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6355 store uchar 0 field instead of uptr 0 field.
6356 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6357 instead of passing one address of struct with 2 locations pass
6358 two addresses of structs with 1 location each.
6359 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6360 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6361 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6362 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6363 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6364 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6365 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6366 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6368 2017-10-19 Martin Liska <mliska@suse.cz>
6371 * file-find.c (remove_prefix): Remove.
6372 * file-find.h (remove_prefix): Likewise.
6373 * gcc-ar.c: Remove smartness of lookup.
6375 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6377 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6378 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6379 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6381 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6384 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6385 (setcc + and to xor + setcc): New peephole2.
6387 2017-10-19 Tom de Vries <tom@codesourcery.com>
6389 * doc/sourcebuild.texi (Test Directives, Variants of
6390 dg-require-support): Add dg-require-stack-size.
6392 2017-10-19 Martin Liska <mliska@suse.cz>
6395 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6396 that have a large alignment.
6397 (gimplify_target_expr): Likewise.
6399 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6401 PR rtl-optimization/82602
6402 * ira.c (rtx_moveable_p): Return false for volatile asm.
6404 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6407 * config/i386/i386-modes.def (CCGZ): New CC mode.
6408 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6409 * config/i386/predicates.md (ix86_comparison_operator):
6411 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6412 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6413 with double-word subtraction.
6414 (put_condition_code): Handle CCGZmode.
6416 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6418 * wide-int.cc (debug (const wide_int &)): New.
6419 (debug (const wide_int *)): New.
6420 (debug (const widest_int &)): New.
6421 (debug (const widest_int *)): New.
6423 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6426 * lra-constraints.c (curr_insn_transform): Use non-input operand
6427 instead of output one for matched reload.
6429 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6431 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6432 (tree-ssa-loop-ivopts.h): New header file.
6433 (struct builtin_info): New fields.
6434 (classify_builtin_1): Compute and record base and offset parts for
6435 memset builtin partition by calling strip_offset.
6436 (offset_cmp, fuse_memset_builtins): New functions.
6437 (finalize_partitions): Fuse adjacent memset partitions by calling
6439 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6440 Expose the interface.
6441 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6443 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6445 PR tree-optimization/82574
6446 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6447 that data reference must be executed exactly once per iteration
6448 against the outermost loop in nest.
6449 (classify_partition): Update call to above function.
6451 2017-10-18 Richard Biener <rguenther@suse.de>
6453 PR tree-optimization/82591
6454 * graphite.c (graphite_transform_loops): Move code gen message
6456 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6457 Here. Handle scop_to_isl_ast failing.
6458 (scop_to_isl_ast): Limit the number of ISL operations.
6460 2017-10-18 Richard Biener <rguenther@suse.de>
6462 * graphite-isl-ast-to-gimple.c
6463 (translate_isl_ast_to_gimple::set_rename): Simplify.
6464 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6465 (graphite_copy_stmts_from_block): ... here.
6466 (copy_bb_and_scalar_dependences): Simplify.
6467 (add_parameters_to_ivs_params): Canonicalize.
6468 (generate_entry_out_of_ssa_copies): Simplify.
6469 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6470 by passing in ISL dimension.
6471 (parameter_index_in_region_1): Rename to ...
6472 (parameter_index_in_region): ... this.
6473 (extract_affine): Adjust assert, pass down parameter index.
6474 (add_param_constraints): Use range-info when available.
6475 (build_scop_context): Adjust.
6476 * sese.c (new_sese_info): Adjust.
6477 (free_sese_info): Likewise.
6478 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6479 Remove unused typedefs.
6480 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6482 2017-10-18 Martin Liska <mliska@suse.cz>
6484 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6486 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6488 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6489 (s390_sched_init): Do not reset s390_sched_state if we entered the
6490 current basic block via a fallthru edge and all others are unlikely.
6492 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6494 * config/s390/s390.c (NUM_SIDES): New variable.
6495 (LONGRUNNING_THRESHOLD): New variable.
6496 (LATENCY_FACTOR): New variable.
6497 (s390_sched_score): Decrease score for long-running instructions on
6499 (s390_sched_variable_issue): Perform bookkeeping for long-running
6502 2017-10-18 Richard Biener <rguenther@suse.de>
6504 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6505 Simplify with removal of the parameter rename map.
6506 (set_rename): Likewise.
6507 (should_copy_to_new_region): Likewise.
6508 (graphite_copy_stmts_from_block): Likewise.
6509 (copy_bb_and_scalar_dependences): Remove initialization of
6510 unused copied_bb_map.
6512 (copy_internal_parameters): Likewise.
6513 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6514 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6515 Use INTEGRAL_TYPE_P.
6516 (parameter_index_in_region_1): Rename to ...
6517 (assign_parameter_index_in_region): ... this. Assert we have
6518 a parameter we handle.
6519 (scan_tree_for_params): Adjust.
6520 * sese.h (parameter_rename_map_t): Remove.
6521 (struct sese_info_t): Remove unused parameter_rename_map and
6522 copied_bb_map members.
6523 * sese.c (new_sese_info): Adjust.
6524 (free_sese_info): Likewise.
6526 2017-10-18 Martin Liska <mliska@suse.cz>
6529 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6530 on an abnormal edge.
6532 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6534 * doc/invoke.texi (ffunction-sections and fdata-sections):
6537 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6539 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6540 the use statement can throw internally.
6542 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6544 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6545 any RTX present on the RHS of a SET.
6546 * compare-elim.c (try_eliminate_compare): Restore comment.
6548 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6550 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6551 may be also called on tcc_type nodes.
6552 * langhooks.c (lhd_tree_size): Likewise.
6554 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6556 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6557 format_warning_at_substring.
6558 (maybe_warn): Convert source_range * param to a location_t. Pass
6559 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6560 (format_directive): Remove code to extract source_ranges and
6561 source_range * in favor of just a location_t.
6562 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6564 * substring-locations.c (format_warning_va): Convert
6565 source_range * param to a location_t.
6566 (format_warning_at_substring): Likewise.
6567 * substring-locations.h (format_warning_va): Likewise.
6568 (format_warning_at_substring): Likewise.
6570 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6572 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6574 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6575 and scatter/gather ops.
6577 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6578 vec_gather_load and vec_scatter_store.
6579 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6580 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6581 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6582 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6583 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6584 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6586 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6588 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6589 Detect FTST instruction and handle its register pops. Only pop
6590 second operand if can_pop_second_op is true.
6591 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6592 set can_pop_second_op to false in the compare_for_stack_reg call.
6594 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6595 output_fp_compare for stack register operands.
6596 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6597 instructions here. Do not emit stack register pops here. Assert
6598 that FCOMPP pops next to top stack register. Rewrite function.
6600 2017-10-17 Nathan Sidwell <nathan@acm.org>
6603 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6604 use DECL_ASSEMBLER_NAME_RAW.
6607 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6610 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6611 Wilco Dijkstra <wilco.dijkstra@arm.com>
6613 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6614 converstion to Pmode from the buf_addr.
6616 2017-10-17 Richard Biener <rguenther@suse.de>
6618 * graphite-dependences.c (scop_get_reads_and_writes): Change
6619 output parameters to references.
6621 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6623 PR 71026/tree-optimization
6624 * fold-const.c (distribute_real_division): Removed.
6625 (fold_binary_loc): Remove calls to distribute_real_divison.
6627 2017-10-17 Richard Biener <rguenther@suse.de>
6629 * graphite-scop-detection.c
6630 (scop_detection::stmt_has_simple_data_refs_p): Always use
6631 the full nest as region.
6632 (try_generate_gimple_bb): Likewise.
6633 * sese.c (scalar_evolution_in_region): Simplify now that
6634 SCEV can handle instantiation in regions.
6635 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6636 in the non-loop part of a function if requested.
6638 2017-10-17 Richard Biener <rguenther@suse.de>
6640 PR tree-optimization/82563
6641 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6643 (graphite_regenerate_ast_isl): Call it.
6644 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6646 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6648 PR tree-optimization/82549
6649 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6650 Formatting fixes. Instead of calling make_bit_field_ref with negative
6653 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6655 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6656 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6657 _mm_maskz_reduce_ss): New.
6658 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6659 __builtin_ia32_reducess_mask): Ditto..
6660 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6661 * config/i386/sse.md (reduces<mode>): Renamed to ...
6662 (reduces<mode><mask_scalar_name>): ... this.
6663 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6665 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6666 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6668 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6670 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6671 * selftest-run-tests.c (selftest::run_tests): Call
6672 selftest::unique_ptr_tests_cc_tests.
6673 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6674 * unique-ptr-tests.cc: New file.
6676 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6679 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6681 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6682 (make_hard_regno_born, make_hard_regno_dead): Update
6683 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6685 2017-10-16 Jeff Law <law@redhat.com>
6687 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6689 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6691 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6693 2017-10-16 Olivier Hainque <hainque@adacore.com>
6695 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6696 with_cpu if we were configured for an e500v2 target cpu name.
6698 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6700 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6701 * doc/invoke.texi: Document +nodsp as a valid extension for
6704 2017-10-16 Martin Liska <mliska@suse.cz>
6706 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6707 (test_set_range): Likewise.
6708 (test_range_functions): Rename to ...
6709 (test_bit_in_range): ... this.
6710 (sbitmap_c_tests): Add new test.
6712 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6714 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6716 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6717 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6719 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6721 * config/aarch64/aarch64-builtins.c
6722 (aarch64_types_quadopu_lane_qualifiers): New.
6723 (TYPES_QUADOPU_LANE): New.
6724 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6725 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6726 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6727 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6728 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6729 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6730 (Vdottype, DOTPROD): New.
6731 (sur): Add SDOT and UDOT.
6733 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6735 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6736 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6737 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6739 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6740 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6741 Enable TARGET_DOTPROD.
6742 (cortex-a75.cortex-a55): Likewise.
6743 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6745 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6747 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6748 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6750 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6752 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6753 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6754 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6755 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6756 * config/arm/types.md (neon_dot, neon_dot_q): New.
6757 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6759 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6761 * config/arm/arm.h (TARGET_DOTPROD): New.
6762 * config/arm/arm.c (arm_arch_dotprod): New.
6763 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6764 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6765 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6766 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6767 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6768 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6769 * doc/invoke.texi (armv8.2-a): Document dotprod
6771 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6773 * i386.c (ix86_vec_cost): New function.
6774 (ix86_rtx_costs): Handle vector operations better.
6775 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6776 * x86-tune-costs.h: Add new costs to all tables.
6778 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6780 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6782 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6783 divsd, sqrtss and sqrtsd
6784 * x86-tune-costs.h: Add new entries to all costs.
6785 (znver1_cost): Fix to match real instruction latencies.
6787 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6788 Michael Collison <michael.collison@arm.com>
6790 * compare-elim.c: Include emit-rtl.h.
6791 (can_merge_compare_into_arith): New function.
6792 (try_validate_parallel): Likewise.
6793 (try_merge_compare): Likewise.
6794 (try_eliminate_compare): Call the above when no previous clobber
6796 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6799 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6803 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6804 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6808 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6809 to be any operand_equal_p operands. For & (B - 1) require
6810 B to be power of 2. Recognize
6811 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6813 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6816 * optabs.c (expand_memory_blockage): Fix call of
6817 targetm.have_memory_blockage.
6819 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6822 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6823 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6824 objects to extra_objs instead of overwriting it.
6826 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6828 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6829 Use any_fp_register_operand as operand[3] predicate. Simplify
6830 equality test for operands[2] and operands[4] memory location.
6831 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6832 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6833 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6834 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6835 any_fp_register_operand as operand[1] predicate. Simplify
6836 equality test for operands[0] and operands[3] memory location.
6837 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6838 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6839 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6841 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6843 * target-insns.def: Add memory_blockage.
6844 * optabs.c (expand_memory_blockage): New function.
6845 (expand_asm_memory_barrier): Rename ...
6846 (expand_asm_memory_blockage): ... to this.
6847 (expand_mem_thread_fence): Call expand_memory_blockage
6848 instead of expand_asm_memory_barrier.
6849 (expand_mem_singnal_fence): Ditto.
6850 (expand_atomic_load): Ditto.
6851 (expand_atomic_store): Ditto.
6852 * doc/md.texi (Standard Pattern Names For Generation):
6853 Document memory_blockage instruction pattern.
6855 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6857 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6858 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6859 * config/rl78/rl78.md: New define_expand "adddi3".
6861 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6863 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6866 2017-10-13 Jeff Law <law@redhat.com>
6868 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6870 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6873 * internal-fn.c (expand_mul_overflow): If both operands have
6874 the same highpart of -1 or 0 and the topmost bit of lowpart
6875 is different, overflow is if res <= 0 rather than res < 0.
6877 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6879 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6880 TARGET_P9_VECTOR code for unaligned_load case.
6882 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6884 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6886 2017-10-13 Nathan Sidwell <nathan@acm.org>
6888 * tree-core.h (tree_contains_struct): Make bool.
6889 * tree.c (tree_contains_struct): Likewise.
6890 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6891 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6892 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6893 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6894 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6896 2017-10-13 Richard Biener <rguenther@suse.de>
6898 * graphite-isl-ast-to-gimple.c
6899 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6900 parameters and dominance check.
6901 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6902 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6903 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6904 Do not update SSA form here or do intermediate IL verification.
6905 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6906 (graphite_initialize): Remove check on the number of loops in
6907 the function and inline into graphite_transform_loops.
6908 (graphite_finalize): Inline into graphite_transform_loops.
6909 (graphite_transform_loops): Perform SSA update and IL verification
6911 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6913 2017-10-13 Richard Biener <rguenther@suse.de>
6915 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6916 graphite_expression_type_precision): Avoid global constructor
6918 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6919 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6920 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6921 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6922 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6923 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6925 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6928 * config/i386/i386.h (ix86_red_zone_size): New.
6929 * config/i386/i386.md (push peephole2s): Replace
6930 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6932 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6933 Alan Hayward <alan.hayward@arm.com>
6934 David Sherwood <david.sherwood@arm.com>
6936 * combine.c (can_change_dest_mode): Reject changes in
6937 REGMODE_NATURAL_SIZE.
6939 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6940 Alan Hayward <alan.hayward@arm.com>
6941 David Sherwood <david.sherwood@arm.com>
6943 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6944 (expand_debug_source_expr): Likewise.
6945 * combine.c (combine_simplify_rtx): Likewise.
6946 * cse.c (fold_rtx): Likewise.
6947 * fwprop.c (canonicalize_address): Likewise.
6948 * targhooks.c (default_shift_truncation_mask): Likewise.
6950 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6951 Alan Hayward <alan.hayward@arm.com>
6952 David Sherwood <david.sherwood@arm.com>
6954 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6955 (widened_mode): Likewise.
6956 (expand_unop): Likewise.
6957 * ree.c (transform_ifelse): Likewise.
6958 (merge_def_and_ext): Likewise.
6959 (combine_reaching_defs): Likewise.
6960 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6962 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6963 Alan Hayward <alan.hayward@arm.com>
6964 David Sherwood <david.sherwood@arm.com>
6966 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6967 * combine.c (gen_lowpart_for_combine): Likewise.
6968 * dwarf2out.c (rtl_for_decl_location): Likewise.
6969 * final.c (alter_subreg): Likewise.
6970 * rtlhooks.c (gen_lowpart_general): Likewise.
6971 (gen_lowpart_if_possible): Likewise.
6973 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6974 Alan Hayward <alan.hayward@arm.com>
6975 David Sherwood <david.sherwood@arm.com>
6977 * calls.c (expand_call): Use subreg_lowpart_offset.
6978 * cse.c (cse_insn): Likewise.
6979 * regcprop.c (copy_value): Likewise.
6980 (copyprop_hardreg_forward_1): Likewise.
6982 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6985 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6986 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6987 =Q constraints to +Q and into insn condition add check
6988 that operands[0] and operands[1] are equal.
6989 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6990 =Q constraints to +Q and into insn condition add check
6991 that operands[0] is equal to either operands[1] or operands[2].
6994 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6995 instead of handling MINUS_EXPR twice (once for each argument),
6996 canonicalize operand order and handle just once, use rtype where
6997 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7000 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7001 any values of __C while still being pattern recognizable as a simple
7004 2017-10-13 Richard Biener <rguenther@suse.de>
7006 PR tree-optimization/82451
7008 2017-10-02 Richard Biener <rguenther@suse.de>
7010 PR tree-optimization/82355
7011 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7012 a mapping for the enclosing loop but avoid generating one for
7014 (copy_bb_and_scalar_dependences): Remove premature codegen
7015 error on PHIs in blocks duplicated into multiple places.
7016 * graphite-scop-detection.c
7017 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7018 in the region use it as loop and nest to analyze the DR in.
7019 (try_generate_gimple_bb): Likewise.
7020 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7021 (add_loop_constraints): For blocks in a loop not in the region
7022 create a dimension with a single iteration.
7023 * sese.h (gbb_loop_at_index): Remove assert.
7025 * cfgloop.c (loop_preheader_edge): For the loop tree root
7026 return the single successor of the entry block.
7027 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7028 Reset the SCEV hashtable and niters.
7029 * graphite-scop-detection.c
7030 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7031 assert that we only have POLYNOMIAL_CHREC that vary in loops
7032 contained in the region.
7033 (scop_detection::graphite_can_represent_expr): Adjust.
7034 (scop_detection::stmt_has_simple_data_refs_p): For loops
7035 not in the region set loop to NULL. The nest is now the
7036 entry edge to the region.
7037 (try_generate_gimple_bb): Likewise.
7038 * sese.c (scalar_evolution_in_region): Adjust for
7039 instantiate_scev change.
7040 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7041 Make nest parameter the edge into the region.
7042 (create_data_ref): Likewise.
7043 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7044 entry edge into a region and adjust instantiate_scev calls.
7045 (create_data_ref): Likewise.
7046 (graphite_find_data_references_in_stmt): Likewise.
7047 (find_data_references_in_stmt): Pass the loop preheader edge
7048 from the nest argument.
7049 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7050 parameter the edge into the region.
7051 (instantiate_parameters): Use the loop preheader edge as entry.
7052 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7054 (get_instantiated_value_entry): Make instantiate_below parameter
7055 the edge into the region.
7056 (instantiate_scev_name): Likewise. Adjust dominance checks,
7057 when we cannot use loop-based instantiation instantiate by
7058 walking use-def chains.
7059 (instantiate_scev_poly): Adjust.
7060 (instantiate_scev_binary): Likewise.
7061 (instantiate_scev_convert): Likewise.
7062 (instantiate_scev_not): Likewise.
7063 (instantiate_array_ref): Remove.
7064 (instantiate_scev_3): Likewise.
7065 (instantiate_scev_2): Likewise.
7066 (instantiate_scev_1): Likewise.
7067 (instantiate_scev_r): Do not blindly handle N-operand trees.
7068 Do not instantiate array-refs. Handle all constants and invariants.
7069 (instantiate_scev): Make instantiate_below parameter
7070 the edge into the region.
7071 (resolve_mixers): Use the loop preheader edge for the region
7072 parameter to instantiate_scev_r.
7073 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7075 2017-10-13 Richard Biener <rguenther@suse.de>
7077 PR tree-optimization/82525
7078 * graphite-isl-ast-to-gimple.c
7079 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7081 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7082 Fail code generation when we cannot represent the isl integer.
7083 (binary_op_to_tree): Elide modulo operations that are no-ops
7084 in the type we code generate. Remove now superfluous code
7087 2017-10-13 Richard Biener <rguenther@suse.de>
7089 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7090 (scop_detection::harmful_loop_in_region): Remove premature
7091 IV type restriction.
7092 (scop_detection::graphite_can_represent_scev): We can handle
7093 pointer IVs just fine.
7095 2017-10-13 Alan Modra <amodra@gmail.com>
7097 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7098 "Clobbers and Scratch Registers". Add paragraph on
7099 alternative to clobbers for scratch registers and OpenBLAS
7102 2017-10-13 Alan Modra <amodra@gmail.com>
7104 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7105 example of a memory input for a string of known length. Move
7106 commentary out of table. Add a number of new examples
7107 covering array memory inputs.
7109 2017-10-12 Martin Liska <mliska@suse.cz>
7111 PR tree-optimization/82493
7112 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7113 (test_range_functions): New function.
7114 (sbitmap_c_tests): Likewise.
7115 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7116 * selftest.h (sbitmap_c_tests): New function.
7118 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7120 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7122 * config/rs6000/amo.h: Fix spacing issue.
7124 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7127 * config/i386/i386.md (*ashl<mode>3_mask_1,
7128 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7129 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7132 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7134 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7135 (profile_probability): Set max_probability
7136 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7138 * profile-count.c (profile_probability::differs_from_p): Do not
7139 rely on max_probaiblity == 10000
7141 2017-10-12 Jeff Law <law@redhat.com>
7143 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7146 2017-10-12 Martin Sebor <msebor@redhat.com>
7150 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7151 (handle_alias_pairs): Call it.
7152 * common.opt (-Wattribute-alias): New option.
7153 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7154 * doc/invoke.texi (-Wattribute-alias): Document.
7156 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7159 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7161 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7163 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7164 (make_hard_regno_born, make_hard_regno_dead): Update
7165 bb_killed_pseudos and bb_gen_pseudos.
7167 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7169 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7171 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7173 * config/alpha/alpha.c (alpha_split_conditional_move):
7174 Use std::swap instead of manually swapping.
7175 (alpha_stdarg_optimize_hook): Ditto.
7176 (alpha_canonicalize_comparison): Ditto.
7178 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7180 * tree-loop-distribution.c (struct builtin_info): New struct.
7181 (struct partition): Refactor fields into struct builtin_info.
7182 (partition_free): Free struct builtin_info.
7183 (build_size_arg_loc, build_addr_arg_loc): Delete.
7184 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7185 information from struct builtin_info.
7186 (find_single_drs): New function refactored from classify_partition.
7187 Also moved builtin validity checks to this function.
7188 (compute_access_range, alloc_builtin): New functions.
7189 (classify_builtin_st, classify_builtin_ldst): New functions.
7190 (classify_partition): Refactor code into functions find_single_drs,
7191 classify_builtin_st and classify_builtin_ldst.
7192 (distribute_loop): Don't do runtime alias check when distributing
7194 (find_seed_stmts_for_distribution): New function.
7195 (pass_loop_distribution::execute): Refactor code finding seed
7196 stmts into above function. Support distribution for the innermost
7197 two-level loop nest. Adjust dump information.
7199 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7201 * tree-loop-distribution.c: Adjust the general comment.
7202 (NUM_PARTITION_THRESHOLD): New macro.
7203 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7204 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7205 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7206 in call to build_partition_graph.
7207 (finalize_partitions): New parameter. Make loop distribution more
7208 conservative by fusing more partitions.
7209 (distribute_loop): Don't do runtime alias check in case of loop nest
7211 (find_seed_stmts_for_distribution): New function.
7212 (prepare_perfect_loop_nest): New function.
7213 (pass_loop_distribution::execute): Refactor code finding seed stmts
7214 and loop nest into above functions. Support loop nest distribution.
7215 Adjust dump information accordingly.
7217 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7219 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7220 and set PTYPE_SEQUENTIAL for merged partition.
7222 2017-10-12 Richard Biener <rguenther@suse.de>
7224 PR tree-optimization/69728
7226 2017-09-19 Richard Biener <rguenther@suse.de>
7228 PR tree-optimization/69728
7229 * graphite-sese-to-poly.c (schedule_error): New global.
7230 (add_loop_schedule): Handle empty domain by failing the
7232 (build_original_schedule): Handle schedule_error.
7234 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7235 domain by returning an unchanged schedule.
7237 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7239 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7240 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7242 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7244 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7247 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7250 * expr.c (store_field): Don't optimize away bitsize == 0 store
7251 from CALL_EXPR with addressable return type.
7253 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7255 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7256 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7257 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7258 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7259 TARGET_ISEL instead of TARGET_ISEL<sel>.
7261 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7263 * config/rs6000/rs6000.c
7264 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7266 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7268 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7270 (reg_or_cint_operand): Fix comment.
7271 (reg_or_zero_operand): New predicate.
7272 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7273 * config/rs6000/rs6000.c (output_isel): Delete.
7274 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7275 instead of reg_or_cint_operand. Output instruction directly (not via
7277 (isel_unsigned_<mode>): Ditto.
7278 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7279 gpc_reg_operand. Add an instruction alternative for this. Output
7280 instruction directly.
7281 (*isel_reversed_unsigned_<mode>): Ditto.
7283 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7285 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7286 (TARGET_CANONICALIZE_COMPARISON): Define.
7288 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7291 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7292 Check whether the dest is REG before adding REG_EQUIV note.
7294 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7297 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7299 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7300 (make_hard_regno_born, make_hard_regno_dead): Update
7301 bb_killed_pseudos and bb_gen_pseudos.
7303 2017-10-11 Nathan Sidwell <nathan@acm.org>
7305 * incpath.h (enum incpath_kind): Name enum, prefix values.
7306 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7307 * incpath.c (heads, tails): Use INC_MAX.
7308 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7309 (merge_include_chains, split_quote_chain,
7310 register_include_chains): Update incpath_kind names.
7311 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7312 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7314 (add_framework_path, darwin_register_objc_includes): Likewise.
7315 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7317 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7319 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7320 Do not use float_operator operator predicate.
7321 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7322 * config/i386/predicates.md (float_operator): Remove predicate.
7324 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7326 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7327 (*jccxf_i387): Ditto.
7328 (*jcc<mode>_i387): Ditto.
7329 (*jccu<mode>_i387): Ditto.
7330 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7331 (*jcc_*_i387 splitters): Remove.
7332 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7333 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7334 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7337 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7339 * profile-count.h (slow_safe_scale_64bit): New function.
7340 (safe_scale_64bit): New inline.
7341 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7342 * profile-count.c: Include wide-int.h
7343 (slow_safe_scale_64bit): New.
7345 2017-10-11 Nathan Sidwell <nathan@acm.org>
7347 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7348 HAS_DECL_ASSEMBLER_NAME_P.
7349 * gimple-expr.c (gimple_decl_printable_name: Check
7350 HAS_DECL_ASSEMBLER_NAME_P too.
7351 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7352 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7353 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7354 * passes.c (rest_of_decl_compilation): Check
7355 HAS_DECL_ASSEMBLER_NAME_P too.
7356 * recog.c (verify_changes): Likewise.
7357 * tree-pretty-print.c (dump_decl_name): Likewise.
7358 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7360 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7361 (SET_DECL_ASSEMBLER_NAME): Use it.
7362 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7363 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7364 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7366 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7368 * config.gcc (i386, x86_64): Add extra objects.
7369 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7370 (ix86_min_insn_size): Declare.
7371 (ix86_issue_rate): Declare.
7372 (ix86_adjust_cost): Declare.
7373 (ia32_multipass_dfa_lookahead): Declare.
7374 (ix86_macro_fusion_p): Declare.
7375 (ix86_macro_fusion_pair_p): Declare.
7376 (ix86_bd_has_dispatch): Declare.
7377 (ix86_bd_do_dispatch): Declare.
7378 (ix86_core2i7_init_hooks): Declare.
7379 (ix86_atom_sched_reorder): Declare.
7380 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7381 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7382 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7383 (rip_relative_addr_p): Rename to ...
7384 (ix86_rip_relative_addr_p): ... this one; export.
7385 (memory_address_length): Update.
7386 (ix86_issue_rate): Move to x86-tune-sched.c.
7387 (ix86_flags_dependent): Move to x86-tune-sched.c.
7388 (ix86_agi_dependent): Move to x86-tune-sched.c.
7389 (exact_dependency_1): Move to x86-tune-sched.c.
7390 (exact_store_load_dependency): Move to x86-tune-sched.c.
7391 (ix86_adjust_cost): Move to x86-tune-sched.c.
7392 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7393 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7394 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7395 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7396 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7397 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7398 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7399 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7400 (min_insn_size): Rename to ...
7401 (ix86_min_insn_size): ... this one; export.
7402 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7403 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7404 (core2i7_first_cycle_multipass_backtrack): Move to
7405 x86-tune-sched-core.c.
7406 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7407 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7408 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7409 (ix86_avoid_jump_mispredicts): Update.
7410 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7411 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7412 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7413 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7414 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7415 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7416 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7417 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7418 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7419 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7420 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7421 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7422 (BIG): Move to ix86-tune-sched-bd.c.
7423 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7424 (enum insn_path): Move to ix86-tune-sched-bd.c.
7425 (get_mem_group): Move to ix86-tune-sched-bd.c.
7426 (is_cmp): Move to ix86-tune-sched-bd.c.
7427 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7428 (is_branch): Move to ix86-tune-sched-bd.c.
7429 (is_prefetch): Move to ix86-tune-sched-bd.c.
7430 (init_window): Move to ix86-tune-sched-bd.c.
7431 (allocate_window): Move to ix86-tune-sched-bd.c.
7432 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7433 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7434 (process_end_window): Move to ix86-tune-sched-bd.c.
7435 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7436 (find_constant): Move to ix86-tune-sched-bd.c.
7437 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7438 (has_immediate): Move to ix86-tune-sched-bd.c.
7439 (get_insn_path): Move to ix86-tune-sched-bd.c.
7440 (get_insn_group): Move to ix86-tune-sched-bd.c.
7441 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7442 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7443 (add_insn_window): Move to ix86-tune-sched-bd.c.
7444 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7445 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7446 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7447 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7448 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7449 (do_dispatch): Move to ix86-tune-sched-bd.c.
7450 (has_dispatch): Move to ix86-tune-sched-bd.c.
7451 * config/i386/t-i386: Add new object files.
7452 * config/i386/x86-tune-costs.h: New file.
7453 * config/i386/x86-tune-sched-atom.c: New file.
7454 * config/i386/x86-tune-sched-bd.c: New file.
7455 * config/i386/x86-tune-sched-core.c: New file.
7456 * config/i386/x86-tune-sched.c: New file.
7458 2017-10-11 Liu Hao <lh_mouse@126.com>
7460 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7461 the generic version below instead.
7462 (should_colorize): Recognize Windows consoles as terminals
7464 * pretty-print.c [__MINGW32__] (write_all): New function.
7465 [__MINGW32__] (find_esc_head): Likewise.
7466 [__MINGW32__] (find_esc_terminator): Likewise.
7467 [__MINGW32__] (eat_esc_sequence): Likewise.
7468 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7470 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7473 2017-10-11 Richard Biener <rguenther@suse.de>
7475 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7476 Properly call analyze_scalar_evolution with the loop of the stmt.
7478 2017-10-11 Richard Biener <rguenther@suse.de>
7480 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7481 * tree-core.h (tree_base): Add chrec_var union member.
7482 * tree.h (CHREC_VAR): Remove.
7483 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7484 * tree-chrec.h (build_polynomial_chrec): Adjust.
7485 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7486 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7488 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7490 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7491 * match.pd: ... here.
7492 ((T) X == (T) Y): Relax condition.
7494 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7496 PR tree-optimization/82472
7497 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7499 (break_alias_scc_partitions): Update postorder number.
7501 2017-10-11 Martin Liska <mliska@suse.cz>
7504 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7506 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7508 2017-10-11 Martin Liska <mliska@suse.cz>
7510 * print-rtl.c (print_insn): Move declaration of idbuf
7511 to same scope as name.
7513 2017-10-11 Martin Liska <mliska@suse.cz>
7518 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7521 2017-10-11 Martin Liska <mliska@suse.cz>
7524 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7527 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7529 * config.gcc (default_gnu_indirect_function): Default to yes for
7530 arm*-*-linux* with glibc.
7532 2017-10-11 Richard Biener <rguenther@suse.de>
7534 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7535 default-defs and types we do not want to analyze.
7536 (interpret_loop_phi): Replace unreachable code with an assert.
7537 (compute_scalar_evolution_in_loop): Remove and inline ...
7538 (analyze_scalar_evolution_1): ... here, replacing condition with
7539 what makes the intent clearer. Remove handling of cases
7540 get_scalar_evolution now handles.
7542 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7544 PR rtl-optimization/81434
7545 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7546 comment for main loop. In sched_group_found if, also add checks for
7547 pass and min_cost_group.
7549 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7551 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7552 (rs6000_insn_cost): New function.
7553 * config/rs6000/rs6000.md (cost): New attribute.
7555 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7556 H.J. Lu <hongjiu.lu@intel.com>
7560 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7561 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7562 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7563 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7564 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7565 OPTION_MASK_ISA_MMX and some other bit set require both
7566 mmx and the ISAs without the mmx bit.
7567 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7568 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7569 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7570 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7571 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7572 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7573 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7574 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7575 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7576 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7577 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7578 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7579 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7580 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7581 Add OPTION_MASK_ISA_MMX.
7583 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7585 * config.gcc (armv7*-*-freebsd*): New target.
7586 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7588 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7590 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7591 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7594 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7596 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7597 a template parameter.
7598 (WIDE_INT_REF_FOR): Update accordingly.
7599 * tree.h (wi::int_traits <const_tree>): Delete.
7600 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7601 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7602 (wi::tree_to_wide_ref): New typedef.
7603 (wi::to_wide): New function.
7604 * calls.c (get_size_range): Use wi::to_wide when operating on
7606 * cgraph.c (cgraph_node::create_thunk): Likewise.
7607 * config/i386/i386.c (ix86_data_alignment): Likewise.
7608 (ix86_local_alignment): Likewise.
7609 * dbxout.c (stabstr_O): Likewise.
7610 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7611 * expr.c (const_vector_from_tree): Likewise.
7612 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7613 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7614 (fold_negate_expr_1, int_const_binop_1, const_binop)
7615 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7616 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7617 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7618 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7619 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7620 (fold_not_const, round_up_loc): Likewise.
7621 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7622 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7623 (alloca_call_type): Likewise.
7624 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7625 * godump.c (go_output_typedef): Likewise.
7626 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7627 * internal-fn.c (get_min_precision): Likewise.
7628 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7629 * ipa-polymorphic-call.c
7630 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7631 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7632 (ipa_modify_call_arguments): Likewise.
7633 * match.pd: Likewise.
7634 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7635 * print-tree.c (print_node_brief, print_node): Likewise.
7636 * stmt.c (expand_case): Likewise.
7637 * stor-layout.c (layout_type): Likewise.
7638 * tree-affine.c (tree_to_aff_combination): Likewise.
7639 * tree-cfg.c (group_case_labels_stmt): Likewise.
7640 * tree-data-ref.c (dr_analyze_indices): Likewise.
7641 (prune_runtime_alias_test_list): Likewise.
7642 * tree-dump.c (dequeue_and_dump): Likewise.
7643 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7644 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7645 * tree-pretty-print.c (dump_generic_node): Likewise.
7646 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7647 (simple_iv_with_niters): Likewise.
7648 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7649 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7650 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7651 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7652 (refine_value_range_using_guard, number_of_iterations_ne_max)
7653 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7654 (get_cst_init_from_scev, record_nonwrapping_iv)
7655 (scev_var_range_cant_overflow): Likewise.
7656 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7657 * tree-ssa-pre.c (compute_avail): Likewise.
7658 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7659 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7660 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7661 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7662 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7663 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7664 (dump_case_nodes, try_switch_expansion): Likewise.
7665 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7666 (vect_do_peeling): Likewise.
7667 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7668 * tree-vect-stmts.c (vectorizable_load): Likewise.
7669 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7670 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7671 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7672 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7673 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7674 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7675 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7676 (evrp_dom_walker::before_dom_children): Likewise.
7677 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7678 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7679 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7680 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7681 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7682 (get_range_pos_neg): Likewise.
7683 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7684 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7685 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7686 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7687 * config/avr/avr.c (avr_fold_builtin): Likewise.
7688 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7689 * config/msp430/msp430.c (msp430_attr): Likewise.
7690 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7691 * config/powerpcspe/powerpcspe-c.c
7692 (altivec_resolve_overloaded_builtin): Likewise.
7693 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7694 (rs6000_expand_ternop_builtin): Likewise.
7695 * config/rs6000/rs6000-c.c
7696 (altivec_resolve_overloaded_builtin): Likewise.
7697 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7698 (rs6000_expand_ternop_builtin): Likewise.
7699 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7701 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7703 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7704 when copying loop nest with only one inner loop.
7706 2017-10-10 Richard Biener <rguenther@suse.de>
7708 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7709 blocks if SCEV is active.
7710 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7712 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7713 (scev_initialize): Assert we are not yet initialized.
7715 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7717 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7718 inner loop's exit stmt by making it always exit the loop, otherwise
7719 we would generate an infinite empty loop.
7721 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7723 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7724 renaming variables in new preheader if it's deleted.
7726 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7728 * tree-loop-distribution.c (struct partition): Remove unused field
7729 loops of the structure.
7730 (partition_alloc, partition_free): Ditto.
7731 (build_rdg_partition_for_vertex): Ditto.
7733 2017-10-09 Jeff Law <law@redhat.com>
7735 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7736 return type to match prototype and documentation.
7738 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7740 * config/rs6000/rs6000.c (processor_costs): Move to ...
7741 * config/rs6000/rs6000.h: ... here.
7742 (rs6000_cost): Declare.
7744 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7746 * except.c (setjmp_fn): New global variable.
7747 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7748 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7749 if DONT_USE_BUILTIN_SETJMP is defined.
7751 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7753 * target.def (insn_cost): New hook.
7754 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7755 * doc/tm.texi: Regenerate.
7756 * rtlanal.c (insn_cost): Use the new hook.
7758 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7760 * combine.c (combine_validate_cost): Compute the new insn_cost,
7761 not just pattern_cost.
7762 (try_combine): Adjust comment.
7764 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7766 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7768 * combine.c (uid_insn_cost): Adjust comment.
7769 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7771 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7772 * dse.c (find_shift_sequence): Ditto.
7773 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7774 (bb_valid_for_noce_process_p): Use pattern_cost.
7775 * rtl.h (insn_rtx_cost): Delete.
7776 (pattern_cost): New prototype.
7777 (insn_cost): New prototype.
7778 * rtlanal.c (insn_rtx_cost): Rename to...
7779 (pattern_cost): ... this.
7782 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7784 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7785 (*jcc<mode>_0_r_i387): Ditto.
7786 (*jccxf_r_i387): Ditto.
7787 (*jcc<mode>_r_i387): Ditto.
7788 (*jccu<mode>_r_i387): Ditto.
7789 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7790 (*jcc): Rename from *jcc_1.
7792 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7794 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7795 deferred rescans after the lvx/stvx recombination pre-pass.
7797 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7799 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7800 memory operation instruction support.
7801 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7802 (rs6000-ibm-aix[789]*): Likewise.
7803 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7804 Document new functions.
7806 2017-10-09 Richard Biener <rguenther@suse.de>
7808 PR tree-optimization/82397
7809 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7810 equality only for semantically equal trees.
7812 2017-10-09 Richard Biener <rguenther@suse.de>
7814 PR tree-optimization/82449
7815 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7816 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7817 allow constant addresses.
7818 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7821 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7823 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7826 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7829 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7832 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7835 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7837 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7840 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7841 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7843 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7845 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7846 (WI_BINARY_PREDICATE_RESULT): Likewise.
7847 (wi::binary_traits::operator_result): New type.
7848 (wi::binary_traits::predicate_result): Likewise.
7849 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7850 (generic_wide_int::operator==, generic_wide_int::operator!=)
7851 (generic_wide_int::operator&, generic_wide_int::and_not)
7852 (generic_wide_int::operator|, generic_wide_int::or_not)
7853 (generic_wide_int::operator^, generic_wide_int::operator+
7854 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7856 (operator~, unary operator-, operator==, operator!=, operator&)
7857 (operator|, operator^, operator+, binary operator-, operator*): New
7859 * expr.c (get_inner_reference): Use wi::bit_and_not.
7860 * fold-const.c (fold_binary_loc): Likewise.
7861 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7862 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7863 (bit_value_binop): Likewise.
7864 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7865 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7866 (extract_range_from_binary_expr_1): Likewise.
7867 (masked_increment): Likewise.
7868 (simplify_bit_ops_using_ranges): Likewise.
7870 2017-10-09 Martin Jambor <mjambor@suse.cz>
7873 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7874 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7875 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7877 (hsa_fixup_mov_insn_type): New function.
7878 (hsa_op_with_type::get_in_type): Use it.
7879 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7880 immediates in an assert.
7881 (hsa_op_with_type::extend_int_to_32bit): New method.
7882 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7883 types. Convert to dest type if necessary.
7884 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7885 (reg_for_gimple_ssa): Pass false as min32int to
7886 hsa_type_for_scalar_tree_type.
7887 (gen_hsa_addr): Fixup type when creating addresable temporary.
7888 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7889 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7890 necessary. Call hsa_fixup_mov_insn_type.
7891 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7892 extend operands and convert to dest type if necessary.
7893 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7894 to dest type if necessary.
7895 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7896 if conversion nt necessary and size matches.
7897 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7898 to dest type if necessary.
7899 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7900 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7902 (gen_hsa_clrsb): Likewise.
7903 (gen_hsa_ffs): Likewise.
7904 (gen_hsa_divmod): Extend operands and convert to dest type if
7906 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7908 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7910 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7911 Remove empty default arguments. Use a brace block as output
7913 (conditional return): Ditto.
7915 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7916 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7918 (group_ending_nop): Ditto.
7919 (doloop_end): Ditto.
7920 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7921 (splitters for those): Ditto.
7923 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7925 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7926 a conditional jump (and the compare for it) so that pc_rtx is the
7928 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7929 for the deleted and renamed ctr<mode>_internal[234] patterns.
7930 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7931 Delete second conditional return pattern.
7932 (ctr<mode>_internal2): Delete this second bdnz pattern.
7933 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7934 (ctr<mode>_internal4): Delete this second bdz pattern.
7936 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7938 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7939 (always_initialized_rtx_for_ssa_name_p): New predicate.
7940 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7941 (finish_out_of_ssa): Free new field of SA.
7942 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7943 * tree-ssa-coalesce.c: Include tree-ssa.h.
7944 (get_parm_default_def_partitions): Remove extern keyword.
7945 (get_undefined_value_partitions): New function.
7946 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7947 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7949 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7951 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7953 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7955 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7957 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7958 for targets that preffer 128bit.
7960 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7962 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7964 2017-10-08 Olivier Hainque <hainque@adacore.com>
7966 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7967 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7969 (thumb_set_return_address): Likewise.
7971 2017-10-08 Olivier Hainque <hainque@adacore.com>
7973 * common/config/arm/arm-common.c (arm_except_unwind_info):
7974 Handle DWARF2_UNWIND_INFO.
7976 2017-10-07 Michael Collison <michael.collison@arm.com>
7978 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7981 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7983 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7984 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7985 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7986 defined, force the creation of a new block for a dispatch label.
7988 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7990 * invoke.texi (Wsuggest-attribute=cold): Document.
7991 * common.opt (Wsuggest-attribute=cold): New
7992 * ipa-pure-const.c (warn_function_cold): New function.
7993 * predict.c (compute_function_frequency): Use it.
7994 * predict.h (warn_function_cold): Declare.
7996 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7998 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8001 2017-10-06 Martin Liska <mliska@suse.cz>
8003 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8004 keyword for member functions.
8005 (struct sanopt_tree_couple): New struct.
8006 (struct sanopt_tree_couple_hash): New function.
8007 (struct sanopt_ctx): Add new hash_map.
8008 (has_dominating_ubsan_ptr_check): New function.
8009 (record_ubsan_ptr_check_stmt): Likewise.
8010 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8011 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8012 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8014 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8017 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8018 aarch64_simd_valid_immediate on CONST_VECTORs.
8019 (aarch64_reg_or_bic_imm): Likewise.
8021 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8023 PR rtl-optimization/82396
8024 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8026 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8028 * graphite-dependences.c (scop_get_reads): Move code to...
8029 (scop_get_must_writes): Move code to...
8030 (scop_get_may_writes): Move code to...
8031 (scop_get_reads_and_writes): ... here.
8032 (scop_get_dependences): Call scop_get_reads_and_writes.
8034 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8036 PR tree-optimization/82434
8037 * fold-const.h (can_native_encode_type_p,
8038 can_native_encode_string_p): Remove.
8039 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8040 don't encode anything, just return what would be otherwise returned.
8041 (native_encode_fixed, native_encode_complex, native_encode_vector):
8043 (native_encode_string): Likewise. Inline by hand
8044 can_native_encode_string_p.
8045 (can_native_encode_type_p): Remove.
8046 (can_native_encode_string_p): Remove.
8047 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8048 STRING_CSTs using can_native_encode_string_p, test all
8049 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8050 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8051 argument from native_encode_expr.
8052 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8053 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8056 2017-10-06 Richard Biener <rguenther@suse.de>
8058 PR tree-optimization/82397
8059 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8060 operand_equal_p but rely on data_ref_compare_tree for detecting
8062 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8063 to match up with dr_group_sort_cmp.
8065 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8068 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8070 * config/s390/s390-builtin-types.def: Regenerate.
8072 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8075 * config/s390/s390-builtin-types.def: Regenerate.
8076 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8077 Change flag from B_VXE to B_VX.
8078 (s390_vec_min_dbl): Remove B_VXE flag.
8080 2017-10-06 Richard Biener <rguenther@suse.de>
8082 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8083 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8084 translate_isl_ast_to_gimple::is_valid_rename,
8085 translate_isl_ast_to_gimple::get_rename,
8086 translate_isl_ast_to_gimple::get_def_bb_for_const,
8087 translate_isl_ast_to_gimple::get_new_name,
8088 translate_isl_ast_to_gimple::collect_all_ssa_names,
8089 translate_isl_ast_to_gimple::copy_loop_phi_args,
8090 translate_isl_ast_to_gimple::collect_all_ssa_names,
8091 translate_isl_ast_to_gimple::copy_loop_phi_args,
8092 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8093 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8094 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8095 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8096 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8097 translate_isl_ast_to_gimple::copy_cond_phi_args,
8098 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8099 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8100 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8101 translate_isl_ast_to_gimple::rename_uses,
8102 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8103 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8104 (set_rename_for_each_def): Likewise.
8105 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8106 here. Handle rewriting SCEV analyzable uses here.
8107 (copy_bb_and_scalar_dependences): Generate code for PHI
8109 (graphite_regenerate_ast_isl): Adjust.
8110 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8111 (add_write, add_read): New functions.
8112 (build_cross_bb_scalars_def): Use it and simplify.
8113 (build_cross_bb_scalars_use): Likewise.
8114 (graphite_find_cross_bb_scalar_vars): Inline into...
8115 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8116 simulating out-of-SSA. Compute liveout and add dependencies.
8117 (build_scops): Force an empty entry block.
8118 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8120 (sese_build_liveouts): Declare.
8121 (sese_trivially_empty_bb_p): Likewise.
8122 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8123 compute liveout and debug_liveout.
8124 (sese_bad_liveouts_use): Remove.
8125 (sese_reset_debug_liveouts_bb): Likewise.
8126 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8127 (sese_build_liveouts): Build liveout and debug_liveout and store
8129 (new_sese_info): Adjust.
8130 (free_sese_info): Likewise.
8131 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8132 do not build liveout here.
8133 (move_sese_in_condition): Adjust region entry.
8134 (scev_analyzable_p): Match up with chrec_apply requirements.
8135 (sese_trivially_empty_bb_p): New.
8136 * tree-into-ssa.c (get_reaching_def): Properly support generating
8137 default-defs for incremental rewrite of anonymous names.
8139 2017-10-06 Richard Biener <rguenther@suse.de>
8141 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8142 precision do not perform modulo reduction.
8144 2017-10-06 Richard Biener <rguenther@suse.de>
8146 PR tree-optimization/82436
8147 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8148 conservatively choose the vectorization factor when checking
8149 whether we can perform the required load permutation.
8150 (vect_transform_slp_perm_load): Assert when we may not fail.
8152 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8154 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8155 message for incompatible -msdata=* and -mcall-* options.
8157 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8159 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8160 rate for post-reload scheduling.
8162 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8164 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8166 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8168 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8169 to improve monte carlo in scimark.
8171 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8173 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8174 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8175 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8176 pentium4_cost, nocona_cost): Set reassociation width to 1.
8177 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8178 width to 2 for fp operations and 1 otherwise.
8179 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8181 (atom_cost): Set reassociation width to 2.
8182 (slm_cost, generic_cost): Set fp reassociation width
8183 to 2 and 1 otherwise.
8184 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8185 (core_cost): Set fp reassociation width to 4 and vector to 2.
8186 (ix86_reassociation_width): Rewrite using cost table; special case
8187 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8188 and TARGET_AVX128_OPTIMAL.
8189 * config/i386/i386.h (processor_costs): Add
8190 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8191 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8192 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8193 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8194 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8195 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8197 2017-10-05 Nathan Sidwell <nathan@acm.org>
8199 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8201 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8203 * config/arm/arm.c (arm_test_fpu_data): New.
8204 (arm_run_selftests): Call arm_test_fpu_data.
8206 2017-10-04 Nathan Sidwell <nathan@acm.org>
8208 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8210 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8212 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8214 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8215 check type for aarch64_simd_valid_immediate.
8216 (aarch64_output_simd_mov_immediate): Update prototype.
8217 (aarch64_simd_valid_immediate): Update prototype.
8218 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8219 support for ORR-immediate.
8220 (and<mode>3): modified pattern to add support for BIC-immediate.
8221 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8222 now checks for valid immediate for BIC and ORR based on new enum
8224 (aarch64_output_simd_mov_immediate): Function now used to output
8225 BIC/ORR imm as well based on new enum argument.
8226 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8228 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8229 (aarch64_reg_or_bic_imm): Likewise.
8231 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8233 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8234 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8236 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8240 PR rtl-optimization/82396
8241 * haifa-sched.c (autopref_multipass_init): Simplify
8243 (autopref_rank_data): Simplify sort order.
8244 * sched-int.h (autopref_multipass_data_): Remove
8245 multi_mem_insn_p, min_offset and max_offset.
8247 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8249 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8251 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8253 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8256 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8258 * doc/sourcebuild.texi: Document vect_long_mult.
8260 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8262 PR tree-optimization/82413
8263 * fold-const.c (build_range_check): Use widest_int when comparing
8264 the maximum ETYPE value with HIGH.
8266 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8268 PR rtl-optimization/82396
8269 * haifa-sched.c (autopref_multipass_init): Simplify
8271 (autopref_rank_data): Simplify sort order.
8272 * sched-int.h (autopref_multipass_data_): Remove
8273 multi_mem_insn_p, min_offset and max_offset.
8275 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8277 PR tree-optimization/82381
8278 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8279 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8282 PR tree-optimization/82374
8283 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8284 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8285 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8286 current_function_decl to the new decl.
8288 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8290 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8291 helper macro for IEEE float128 hardware built-in functions.
8292 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8294 (TRUNCF128_ODD): Likewise.
8295 (ADDF128_ODD): Likewise.
8296 (SUBF128_ODD): Likewise.
8297 (MULF128_ODD): Likewise.
8298 (DIVF128_ODD): Likewise.
8299 (FMAF128_ODD): Likewise.
8300 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8301 UNSPEC_TRUNC_ROUND_TO_ODD.
8302 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8303 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8304 floating point round to odd instructions.
8305 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8306 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8307 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8308 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8309 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8310 (trunc<mode>sf2_hw): Change the truncate with round to odd
8311 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8312 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8313 to odd hardware instructions.
8314 (sub<mode>3_odd): Likewise.
8315 (mul<mode>3_odd): Likewise.
8316 (div<mode>3_odd): Likewise.
8317 (sqrt<mode>2_odd): Likewise.
8318 (fma<mode>4_odd): Likewise.
8319 (fms<mode>4_odd): Likewise.
8320 (nfma<mode>4_odd): Likewise.
8321 (nfms<mode>4_odd): Likewise.
8322 (trunc<mode>df2_odd): Change the truncate with round to odd
8323 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8325 * doc/extend.texi (PowerPC built-in functions): Update documentation
8326 for existing IEEE float128-bit built-in functions. Add built-in
8327 functions that generate the IEEE 128-bit floating point round to
8330 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8332 PR rtl-optimization/77729
8333 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8334 to (X&(C1&~C2))|C2 transformations.
8336 2017-10-03 Martin Jambor <mjambor@suse.cz>
8338 PR tree-optimization/82363
8339 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8340 mismatch, mark lacc written regardless of racc.
8342 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8344 PR tree-optimization/82381
8345 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8346 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8347 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8348 fallthrough into reassoc_stmt_dominates_stmt_p.
8351 * combine.c (combine_instructions): Don't combine in unreachable
8354 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8358 function to not use the have_cpu variable. Do not set cpu_index,
8359 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8361 (rs6000_valid_attribute_p): Remove duplicate initializations of
8362 old_optimize and func_optimize.
8363 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8364 (rs6000_activate_target_options): Make global.
8365 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8368 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8370 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8371 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8372 Set *poffset to 0 and *psize and *pmax_size to -1 if
8373 *poffset + *psize overflows in HOST_WIDE_INT.
8375 PR tree-optimization/82387
8376 PR tree-optimization/82388
8377 PR tree-optimization/82389
8378 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8379 instead of live_bytes non-NULL.
8381 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8384 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8387 2017-10-02 Richard Biener <rguenther@suse.de>
8389 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8390 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8391 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8393 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8395 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8396 requested precision matches the type's.
8397 * calls.c (alloc_max_size): Calculate the new candidate size as
8398 a widest_int and use wi::to_widest when comparing it with the
8399 current candidate size.
8400 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8401 zero rather than integer_zero_node.
8402 * match.pd: Check for a no-op conversion before using wi::add
8403 rather than after. Use tree_to_uhwi when summing small shift
8404 counts into an unsigned int.
8406 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8407 Alan Hayward <alan.hayward@arm.com>
8408 David Sherwood <david.sherwood@arm.com>
8411 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8412 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8413 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8414 POINTER_AND_FP_REGS.
8416 2017-10-02 Richard Biener <rguenther@suse.de>
8418 PR tree-optimization/82355
8419 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8420 a mapping for the enclosing loop but avoid generating one for
8422 (copy_bb_and_scalar_dependences): Remove premature codegen
8423 error on PHIs in blocks duplicated into multiple places.
8424 * graphite-scop-detection.c
8425 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8426 in the region use it as loop and nest to analyze the DR in.
8427 (try_generate_gimple_bb): Likewise.
8428 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8429 (add_loop_constraints): For blocks in a loop not in the region
8430 create a dimension with a single iteration.
8431 * sese.h (gbb_loop_at_index): Remove assert.
8433 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8435 * omp-expand.c (adjust_context_scope): New function.
8436 (expand_parallel_call): Call adjust_context_scope.
8438 2017-10-01 Jeff Law <law@redhat.com>
8440 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8441 dom_opt_dom_walker class with direct access to private members.
8442 Add comments. Call test_for_singularity.
8443 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8444 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8445 m_dummy_cond anymore.
8446 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8448 (pass_dominator:execute): Build the dummy_cond here and pass it
8449 to the dom_opt_dom_walker ctor.
8450 (test_for_singularity): New function.
8452 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8453 Maya Rashish <coypu@sdf.org>
8455 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8456 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8457 (alpha*-*-netbsd*) Use nbsd_tm_file.
8458 (arm*-*-netbsdelf*) Likewise.
8459 (i[34567]86-*-netbsdelf*) Likewise.
8460 (x86_64-*-netbsd*) Likewise.
8461 (mips*-*-netbsd*) Likewise.
8462 (powerpc-*-netbsd*) Likewise.
8463 (sh*-*-netbsd*) Likewise.
8464 (sparc-*-netbsdelf*) Likewise.
8465 (sparc64-*-netbsd*) Likewise.
8466 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8468 (vax-*-netbsdelf*) Likewise.
8469 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8470 (UINT_FAST8_TYPE) Likewise.
8471 (INT_FAST16_TYPE) Check CHAR_FAST16.
8472 (UINT_FAST16_TYPE) Likewise.
8474 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8477 * config/i386/i386.md
8478 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8479 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8480 *divmodsi4_zext_2): New define_insn_and_split.
8481 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8482 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8483 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8484 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8485 New define_insn_and_split.
8486 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8487 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8488 operands[1] having DImode when mode is SImode.
8490 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8491 always SImode for DIV and MOD in REG_EQUAL notes.
8493 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8496 * match.pd: Fix handling of NaNs in pattern.
8498 2017-09-29 Jeff Law <law@redhat.com>
8500 * sbitmap.c (bitmap_bit_in_range_p): New function.
8501 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8502 * tree-ssa-dse.c (live_bytes_read): New function.
8503 (dse_classify_store): Ignore reads of dead bytes.
8505 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8506 typos and whitespace errors.
8507 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8508 * config/s390/s390.c (s390_emit_prologue): Likewise.
8510 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8513 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8514 with a symbol for LRA.
8516 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8518 PR rtl-optimization/82338
8519 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8521 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8523 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8524 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8525 (qsort_chk): Declare.
8526 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8527 (qsort_chk): New function.
8529 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8531 PR tree-optimization/82337
8532 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8533 phi definition if the PHI result appears in an abnormal PHI.
8534 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8535 basis appears in an abnormal PHI.
8537 2017-09-29 Richard Biener <rguenther@suse.de>
8539 * graphite-isl-ast-to-gimple.c
8540 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8541 (binary_op_to_tree): Use it.
8542 (get_rename_from_scev): Likewise.
8543 (copy_loop_phi_nodes): Likewise.
8544 (copy_bb_and_scalar_dependences): Likewise.
8545 (translate_pending_phi_nodes): Likewise.
8547 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8550 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8551 for movabsq $(i32 << shift), r64.
8553 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8555 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8556 index when encoding %esp as %rsp to avoid 0x67 prefix.
8558 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8560 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8561 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8563 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8565 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8566 Extensions with more than 16 double VFP registers.
8567 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8568 to_clear_mask and all code related to it. Replace the remaining
8569 entry by a sbitmap and adapt code accordingly.
8571 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8573 * brig-builtins.def: Change pure attributes to const.
8575 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8577 * config.gcc (default_gnu_indirect_function): Default to yes for
8578 sparc*-*-linux* with glibc.
8580 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8582 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8583 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8584 when creating .init_array and .fini_array sections with priority
8587 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8590 * config/aarch64/aarch64.c
8591 (aarch64_builtin_support_vector_misalignment): Always return false
8592 when misalignment is unknown.
8594 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8596 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8597 this function to return false if the definition used by the swap
8598 instruction is artificial, or if the memory address from which the
8599 constant value is loaded is not represented by a base address held
8600 in a register or if the base address register is a frame or stack
8601 pointer. Additionally, return false if the base address of the
8602 loaded constant is a SYMBOL_REF but is not considered to be a
8604 (replace_swapped_load_constant): New function.
8605 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8606 loaded constant vector with a load of a swapped constant vector.
8608 2017-09-27 Carl Love <cel@us.ibm.com>
8610 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8611 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8612 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8615 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8617 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8618 first, always call autopref_rank_data otherwise.
8620 2017-09-27 Richard Biener <rguenther@suse.de>
8622 * graphite-scop-detection.c (find_scop_parameters): Move
8623 loop bound handling ...
8624 (gather_bbs::before_dom_children): ... here, avoiding the need
8625 to build scop_info->loop_nest.
8626 (record_loop_in_sese): Remove.
8627 * sese.h (sese_info_t::loop_nest): Remove.
8628 * sese.c (new_sese_info): Do not allocate loop_nest.
8629 (free_sese_info): Do not free loop_nest.
8631 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8634 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8635 lhs from calls if the lhs has addressable type.
8637 2017-09-27 Richard Biener <rguenther@suse.de>
8639 * graphite.h (scop::max_alias_set): New member.
8640 * graphite-scop-detection.c: Remove references to non-existing
8641 --param in comments.
8642 (build_alias_sets): Record the maximum alias set used for drs.
8643 (build_scops): Support zero as unlimited for
8644 --param graphite-max-arrays-per-scop.
8645 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8647 (build_poly_sr_1): ... here. Compute alias set based on the
8648 maximum alias set used for drs rather than
8649 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8651 2017-09-27 Richard Biener <rguenther@suse.de>
8653 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8654 --param loop-block-tile-size=0 to disable tiling.
8656 2017-09-27 Richard Biener <rguenther@suse.de>
8658 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8659 (graphite-max-nb-scop-params): Document special value zero.
8660 * domwalk.h (dom_walker::STOP): New symbolical constant.
8661 (dom_walker::dom_walker): Add optional parameter for bb to
8663 (dom_walker::~dom_walker): Declare.
8664 (dom_walker::before_dom_children): Document STOP return value.
8665 (dom_walker::m_user_bb_to_rpo): New member.
8666 (dom_walker::m_bb_to_rpo): Likewise.
8667 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8668 mapping here if not provided by the user.
8669 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8670 provided by the user.
8671 (dom_walker::STOP): Define.
8672 (dom_walker::walk): Do not compute bb to RPO mapping here.
8673 Support STOP return value from before_dom_children to stop
8675 * graphite-optimize-isl.c (optimize_isl): If the schedule
8676 is the same still generate code if -fgraphite-identity
8677 or -floop-parallelize-all are given.
8678 * graphite-scop-detection.c: Include cfganal.h.
8679 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8681 (gather_bbs::before_dom_children): Return STOP for BBs
8683 (build_scops): Compute bb to RPO mapping and pass it to
8684 the domwalk. Treat --param graphite-max-nb-scop-params=0
8685 as not limiting the number of params.
8686 * graphite.c (graphite_initialize): Remove limit on the
8687 number of basic-blocks in a function.
8688 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8689 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8690 default value of 10.
8692 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8694 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8695 Adjust code to eliminate needing to do the shift right 32-bits
8696 operation after XSCVDPSPN.
8698 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8700 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8701 ((X / Y) != 0 -> X >= Y): Likewise.
8703 2017-09-26 Carl Love <cel@us.ibm.com>
8705 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8706 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8707 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8708 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8709 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8710 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8711 definitions and overloading.
8712 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8713 statement for P9V_BUILTIN_XST_LEN_R.
8714 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8715 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8716 define_expand and define_insn for the instructions and builtins.
8717 * doc/extend.texi: Update the built-in documentation file for the new
8719 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8720 define_insn for the instructions
8722 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8725 * gcc/config/netbsd-protos.h: New file.
8726 * gcc/config/netbsd.c: New file.
8727 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8728 * gcc/config/t-netbsd: New file.
8729 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8730 (tmake_file) Add t-netbsd.
8731 (extra_objs) Add netbsd.o.
8733 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8737 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8739 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8741 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8742 sign extension from a vector register to a GPR by doing a 32-bit
8743 direct move and then an EXTSW.
8744 (extendsi<mode>2 splitter): Likewise.
8745 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8746 right or vector extract after doing XSCVDPSPN. Use
8747 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8749 (movdi_from_sf_zero_ext): Likewise.
8750 (reload_gpr_from_vsxsf): Likewise.
8751 (p8_mfvsrd_4_disf): Delete, no longer used.
8752 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8753 then needing to move the SFmode to a GPR to use the XSCVDPSP
8754 instruction instead of FRSP and XSCVDPSPN.
8755 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8756 it is adjacent to the other XSCVSPDP insns.
8757 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8758 SFmode to be in traditional Altivec registers.
8759 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8760 (vsx_xscvspdpn): Likewise.
8761 (vsx_xscvspdpn_scalar): Likewise.
8763 2017-09-26 Martin Jambor <mjambor@suse.cz>
8765 * tree-sra.c (compare_access_positions): Put integral types first,
8766 stabilize sorting of integral types, remove conditions putting
8767 non-full-precision integers last.
8768 (sort_and_splice_var_accesses): Disable scalarization if a
8769 non-integert would be represented by a non-full-precision integer.
8771 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8773 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8774 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8775 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8776 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8777 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8778 conditionals inside the function instead of around it. Call
8779 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8780 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8782 2017-09-26 Richard Biener <rguenther@suse.de>
8784 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8786 (scop_detection::build_scop_breadth): ... this. Removed.
8787 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8788 (scop_detection::harmful_stmt_in_bb): Likewise.
8789 (scop_detection::graphite_can_represent_stmt): Likewise.
8790 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8791 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8792 (scop_detection::can_represent_loop_1): ... this. Removed.
8793 (scop_detection::harmful_loop_in_region): Simplify after inlining
8794 the above and remove more quadraticness.
8795 (build_scops): Adjust.
8796 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8799 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8802 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8803 REGNO (base) == SP_REG if base is a REG.
8806 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8807 if it is different SSA_NAME.
8808 (optimize_range_tests_cmp_bitwise): New function.
8809 (optimize_range_tests): Call it.
8811 2017-09-26 Richard Biener <rguenther@suse.de>
8813 PR tree-optimization/82321
8814 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8815 for the def being inside the loop.
8817 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8819 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8821 * config/s390/s390-builtins.def: Fix constraint on op4.
8823 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8825 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8826 independent expanders.
8827 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8828 ("vec_ordered", "vec_unordered"): New expanders.
8830 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8832 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8835 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8837 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8838 vec_unpacks_lo_v16qi.
8839 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8841 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8843 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8844 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8845 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8847 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8849 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8851 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8853 ("*vec_slb<mode>"): New insn pattern.
8854 ("vec_shr_<mode>"): New expander.
8855 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8856 and force the shift count operand to V16QImode.
8857 ("vec_srb<mode>"): Set shift count mode to V16QI.
8859 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8861 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8862 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8863 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8865 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8868 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8870 2017-09-26 Richard Biener <rguenther@suse.de>
8872 PR tree-optimization/82320
8873 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8876 2017-09-25 Jeff Law <law@redhat.com>
8878 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8879 prototype for new argument.
8880 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8881 mostly extracted from rs6000_emit_allocate_stack.
8882 (rs6000_emit_probe_stack_range_stack_clash): New function.
8883 (rs6000_emit_allocate_stack): Call
8884 rs6000_emit_probe_stack_range_stack_clash as needed.
8885 (rs6000_emit_probe_stack_range): Add additional argument
8886 to call to gen_probe_stack_range{si,di}.
8887 (output_probe_stack_range): New.
8888 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8889 (output_probe_stack_range_stack_clash): New.
8890 (rs6000_emit_prologue): Emit notes into dump file as requested.
8891 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8892 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8893 Add additional operand and pass it to output_probe_stack_range.
8895 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8897 PR tree-optimization/82163
8898 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8899 (checking_verify_loop_closed_ssa): New parameter.
8900 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8901 (check_loop_closed_ssa_stmt): Delete.
8902 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8903 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8904 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8907 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8909 * brig-builtins.def: Treat HSAIL barrier builtins as
8910 setjmp/longjump style functions.
8912 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8914 * target.def (constant_alignment): New hook.
8915 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8916 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8917 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8918 * doc/tm.texi: Regenerate.
8919 * targhooks.h (default_constant_alignment): Declare.
8920 (constant_alignment_word_strings): Likewise.
8921 * targhooks.c (default_constant_alignment): New function.
8922 (constant_alignment_word_strings): Likewise.
8923 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8924 instead of CONSTANT_ALIGNMENT.
8925 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8926 (force_const_mem): Likewise.
8927 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8928 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8929 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8930 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8931 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8933 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8934 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8935 constant_alignment_word_strings.
8936 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8937 (CONSTANT_ALIGNMENT): Likewise.
8938 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8939 (arm_constant_alignment): New function.
8940 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8941 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8942 constant_alignment_word_strings.
8943 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8944 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8945 constant_alignment_word_strings.
8946 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8947 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8948 (cris_constant_alignment): New function.
8949 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8950 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8951 (epiphany_constant_alignment): New function.
8952 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8953 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8954 constant_alignment_word_strings.
8955 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8956 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8957 constant_alignment_word_strings.
8958 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8959 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8960 constant_alignment_word_strings.
8961 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8962 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8963 * config/i386/i386.c (ix86_constant_alignment): Make static.
8964 Use the same interface as the target hook.
8965 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8966 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8967 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8968 constant_alignment_word_strings.
8969 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8970 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8971 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8972 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8973 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8974 constant_alignment_word_strings.
8975 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8976 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8977 constant_alignment_word_strings.
8978 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8979 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8980 constant_alignment_word_strings.
8981 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8982 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8984 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8985 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8986 * config/mips/mips.c (mips_constant_alignment): New function.
8987 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8988 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8989 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8990 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8991 (mmix_constant_alignment): Make static. Use the same interface
8993 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8994 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8995 constant_alignment_word_strings.
8996 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8997 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8998 constant_alignment_word_strings.
8999 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9000 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9001 constant_alignment_word_strings.
9002 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9003 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9004 (rs6000_constant_alignment): New function.
9005 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9006 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9007 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9008 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9009 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9010 (rs6000_constant_alignment): New function.
9011 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9012 * config/s390/s390.c (s390_constant_alignment): New function.
9013 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9014 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9015 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9016 constant_alignment_word_strings.
9017 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9018 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9019 (sparc_constant_alignment): New function.
9020 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9021 * config/spu/spu.c (spu_constant_alignment): New function.
9022 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9023 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9024 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9025 constant_alignment_word_strings.
9026 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9027 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9028 constant_alignment_word_strings.
9029 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9030 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9031 constant_alignment_word_strings.
9032 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9033 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9034 (visium_constant_alignment): New function.
9035 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9036 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9037 (xtensa_constant_alignment): New function.
9038 * system.h (CONSTANT_ALIGNMENT): Poison.
9040 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9042 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9043 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9044 (rs6000_builtin_valid_without_lhs): New helper function.
9045 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9046 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9048 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9050 * target.h (vec_perm_indices): Use unsigned short rather than
9052 (auto_vec_perm_indices): Likewise.
9053 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9054 Use unsigned int rather than unsigned char.
9055 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9057 2017-09-25 Richard Biener <rguenther@suse.de>
9059 * cfgloop.h (sort_sibling_loops): Declare.
9060 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9061 (sort_sibling_loops): New function sorting the sibling loop list
9063 * graphite.c (graphite_transform_loops): Sort sibling loops.
9065 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9067 * target.def (vec_perm_const_ok): Change sel parameter to
9069 * optabs-query.c (can_vec_perm_p): Update accordingly.
9070 * doc/tm.texi: Regenerate.
9071 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9072 auto_vec_perm_indices and remove separate nelt field.
9073 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9074 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9075 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9076 (aarch64_expand_vec_perm_const): Update accordingly.
9077 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9078 to vec_perm_indices.
9079 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9080 auto_vec_perm_indices and remove separate nelt field.
9081 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9082 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9083 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9085 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9086 to vec_perm_indices.
9087 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9088 sel to vec_perm_indices.
9089 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9090 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9091 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9093 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9096 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9099 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9100 on the FUNCTION_DECL function context if it has a DIE that is a
9103 2017-09-25 Richard Biener <rguenther@suse.de>
9105 PR tree-optimization/82285
9106 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9109 2017-09-25 Tom de Vries <tom@codesourcery.com>
9113 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9116 2017-09-25 Richard Biener <rguenther@suse.de>
9118 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9119 ISL errors other than isl_error_quota happen. Dump if the
9120 schedule is the same.
9121 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9122 errors instead of aborting inside ISL.
9124 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9127 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9128 of libgcc_eh for m64.
9129 * config/i386/darwin64.h: Likewise.
9131 2017-09-25 Richard Biener <rguenther@suse.de>
9134 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9135 attribute for incomplete types nor twice for complete ones.
9137 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9140 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9141 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9144 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9147 * config/i386/i386.opt (mprefer-avx256): Use
9148 ix86_target_flags variable.
9149 * config/i386/i386.c (ix86_target_string): Move
9150 -mprefer-avx256 to flag2_opts.
9152 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9155 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9156 and x != -1 | y != -1 into (x & y) != -1.
9158 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9160 * config.gcc: Add new case statement to set
9161 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9162 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9163 s390x-*-linux* case statements. Added aarch64 to the list of
9164 supported architectures.
9166 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9168 PR tree-optimization/82289
9169 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9170 STMT_VINFO_RELEVANT_P.
9172 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9173 Alan Hayward <alan.hayward@arm.com>
9174 David Sherwood <david.sherwood@arm.com>
9176 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9177 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9178 (extract_range_from_binary_expr_1): Don't call
9179 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9181 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9182 Alan Hayward <alan.hayward@arm.com>
9183 David Sherwood <david.sherwood@arm.com>
9185 * target.def (preferred_vector_alignment): New hook.
9186 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9188 * doc/tm.texi: Regenerate.
9189 * targhooks.h (default_preferred_vector_alignment): Declare.
9190 * targhooks.c (default_preferred_vector_alignment): New function.
9191 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9193 (DR_TARGET_ALIGNMENT): New macro.
9194 (aligned_access_p): Update commentary.
9195 (vect_known_alignment_in_bytes): New function.
9196 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9198 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9199 Calculate the misalignment based on the target alignment rather than
9201 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9202 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9203 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9204 the target alignment, rather than masking the element misalignment
9205 with the number of elements in a vector. Also use the target
9206 alignment when calculating the maximum number of peels.
9207 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9208 instead of TYPE_ALIGN_UNIT.
9209 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9210 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9211 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9212 (vect_create_data_ref_ptr): Likewise.
9213 (vect_setup_realignment): Realign by ANDing with
9214 -DR_TARGET_MISALIGNMENT.
9215 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9216 the number of peels based on DR_TARGET_ALIGNMENT.
9217 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9218 with the guaranteed alignment boundary when deciding whether
9220 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9221 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9222 (ensure_base_align): Remove stmt_info parameter. Get the
9223 target base alignment from DR_TARGET_ALIGNMENT.
9224 (vectorizable_store): Update call accordingly. Interpret
9225 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9227 (vectorizable_load): Likewise.
9229 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9230 Alan Hayward <alan.hayward@arm.com>
9231 David Sherwood <david.sherwood@arm.com>
9233 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9234 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9235 (vect_enhance_data_refs_alignment): Likewise.
9237 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9239 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9240 error. Only quit immediately if parsing is complete.
9241 (BEGIN): Initialize fatal_err and parse_done.
9242 (begin fpu, end fpu): Check number of arguments.
9243 (begin arch, end arch): Likewise.
9244 (begin cpu, end cpu): Likewise.
9245 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9246 (optalias): Likewise.
9248 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9250 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9251 * config/arm/arm-isa.h: Delete. Move definitions to ...
9252 * arm-cpus.in: ... here. Use new feature and fgroup values.
9253 * config/arm/arm.c (arm_option_override): Use lower case for feature
9255 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9256 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9257 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9259 (print_isa_bits_for): New function.
9260 (gen_isa): New function.
9261 (gen_comm_data): Use print_isa_bits_for.
9262 (define feature): New keyword.
9263 (define fgroup): New keyword.
9264 * config/arm/t-arm (TM_H): Remove.
9265 (GTM_H): Add arm-isa.h.
9266 (arm-isa.h): Add rule to generate file.
9267 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9268 case for feature bit names.
9270 2017-09-22 Richard Biener <rguenther@suse.de>
9272 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9274 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9275 print of no dependency loops ...
9276 * graphite.c (graphite_transform_loops): ... here.
9277 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9279 (same_close_phi_node, remove_duplicate_close_phi,
9280 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9281 (canonicalize_loop_closed_ssa): ... here and simplify.
9282 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9283 (optimize_isl): Use dump_printf_loc to tell when we stopped
9284 optimizing because of an ISL timeout.
9286 2017-09-22 Richard Biener <rguenther@suse.de>
9288 PR tree-optimization/82291
9289 * tree-if-conv.c (predicate_mem_writes): Make sure to
9290 remove writes in blocks predicated with false.
9292 2017-09-22 Richard Biener <rguenther@suse.de>
9294 * sese.c: Include cfganal.h.
9295 (if_region_set_false_region): Remove.
9296 (create_if_region_on_edge): Likewise.
9297 (move_sese_in_condition): Re-implement without destroying
9300 2017-09-22 Richard Biener <rguenther@suse.de>
9302 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9303 Verify both BBs contain loop PHI nodes before dispatching to
9305 (graphite_regenerate_ast_isl): Do not recompute dominators,
9306 do not verify three times. Restructure for clarity.
9307 * graphite-scop-detection.c (same_close_phi_node,
9308 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9309 defined_in_loop_p, canonicalize_loop_closed_ssa,
9310 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9311 checking and SSA rewrite, move to ...
9312 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9313 (graphite_initialize): Do not pass in ctx, do not reset the
9314 SCEV cache, compute only dominators.
9315 (graphite_transform_loops): Allocate ISL ctx after
9316 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9317 Maintain post-dominators only around build_scops.
9318 * sese.c (if_region_set_false_region): Make static. Free
9319 and recompute dominators.
9320 (move_sese_in_condition): Assert we don't get called with
9321 post-dominators computed.
9322 * sese.h (if_region_set_false_region): Remove.
9324 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9326 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9327 mode attribute for TARGET_AVX512VL.
9329 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9331 * config/i386/i386.opt (mprefer-avx256): New option.
9332 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9334 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9335 for TARGET_PREFER_AVX256.
9336 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9338 2017-09-21 Jeff Law <law@redhat.com>
9340 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9341 Fix dump output if the only stack space is for pushed registers.
9343 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9345 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9348 2017-09-21 Martin Sebor <msebor@redhat.com>
9351 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9352 code (in C++) or code that triggers warnings.
9354 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9356 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9358 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9360 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9361 * sched-rgn.c: Ditto.
9362 * sel-sched-ir.c: Ditto.
9364 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9366 * toplev.h (set_random_seed): Adjust return type.
9367 * toplev.c (init_local_tick): Move eager initialization of random_seed
9368 to get_random_seed. Adjust comment.
9369 (init_random_seed): Inline to get_random_seed, delete.
9370 (get_random_seed): Initialize random_seed lazily.
9371 (set_random_seed): Do not return previous value.
9372 (print_switch_value): Do not call get_random_seed.
9374 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9376 * cgraph.c (delete_function_version): New, broken out from...
9377 (cgraph_node::delete_function_version): ...here. Rename to
9378 cgraph_node::delete_function_version_by_decl. Update all uses.
9379 (cgraph_node::remove): Call delete_function_version.
9381 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9384 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9385 VAR_DECLs to which addressable non-volatile parameters are mapped
9386 and for id->retvar after the return value assignment. Clear
9387 id->retval and id->retbnd after inlining.
9389 2017-09-21 Richard Biener <rguenther@suse.de>
9391 PR tree-optimization/82276
9392 PR tree-optimization/82244
9393 * tree-vrp.c (build_assert_expr_for): Set
9394 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9396 (remove_range_assertions): Revert earlier change.
9398 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9401 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9403 2017-09-21 Richard Biener <rguenther@suse.de>
9405 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9406 Restore valid IL after code generation errors.
9407 * graphite.c (graphite_transform_loops): Diagnose code
9408 generation issues as MSG_MISSED_OPTIMIZATION and continue
9409 with processing SCOPs.
9411 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9412 Alan Hayward <alan.hayward@arm.com>
9413 David Sherwood <david.sherwood@arm.com>
9415 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9416 rather than choosing between plus_constant and gen_rtx_<CODE>.
9417 * expr.c (emit_push_insn): Likewise.
9418 (expand_expr_real_2): Likewise.
9420 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9421 Alan Hayward <alan.hayward@arm.com>
9422 David Sherwood <david.sherwood@arm.com>
9424 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9426 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9427 Alan Hayward <alan.hayward@arm.com>
9428 David Sherwood <david.sherwood@arm.com>
9430 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9431 calling tree_to_uhwi.
9433 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9434 Alan Hayward <alan.hayward@arm.com>
9435 David Sherwood <david.sherwood@arm.com>
9437 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9438 INTEGER_CST rather than a negative test for ADDR_EXPR.
9440 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9441 Alan Hayward <alan.hayward@arm.com>
9442 David Sherwood <david.sherwood@arm.com>
9444 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9445 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9447 2017-09-21 Richard Biener <rguenther@suse.de>
9449 PR tree-optimization/71351
9450 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9451 graphite_create_new_loop_guard): Remove, fold remaining parts
9453 (translate_isl_ast_node_for): ... here and simplify.
9455 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9458 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9459 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9460 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9461 alternative always use QI mode, for -Os imov (=R,R) alternative
9462 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9465 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9466 Jeff Law <law@redhat.com>
9468 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9469 (allocate_stack_space): New function, partially extracted from
9471 (s390_emit_prologue): Track offset to most recent stack probe.
9472 Code to allocate space moved into allocate_stack_space.
9473 Dump actions when no stack is allocated.
9474 (s390_prologue_plus_offset): New function.
9475 (s390_emit_stack_probe): Likewise.
9477 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9479 * common.opt (Wa, Wl, Wp, g, gz=): Add
9481 (gno-column-info): Remove.
9482 (gcolumn-info): Drop RejectNegative.
9484 (gno-record-gcc-switches): Remove.
9485 (grecord-gcc-switches): Drop RejectNegative.
9486 (gno-split-dwarf): Remove.
9487 (gsplit-dwarf): Drop RejectNegative.
9488 (gno-strict-dwarf): Remove.
9489 (gstrict-dwarf): Drop RejectNegative.
9490 * config/darwin.opt (gfull, gused): Add RejectNegative.
9491 * dwarf2out.c (gen_producer_string): Drop
9492 gno-record-gcc-switches handler.
9493 * optc-gen.awk: Add g to prefixes with negative forms.
9494 * opts-common.c (remapping_prefix_p): New.
9495 (find_opt): Check it.
9496 (generate_canonical_option): Test g prefix.
9497 (option_map): Add -gno- mapping.
9498 (add_misspelling_candidates): Check remapping_prefix_p.
9500 2017-09-20 Jeff Law <law@redhat.com>
9502 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9503 thinko in stack clash protection support.
9505 * explow.c (compute_stack_clash_protection_loop_data): Use
9506 CONST_INT_P instead of explicit test. Verify object is a
9507 CONST_INT_P before looking at INTVAL.
9508 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9509 instead of explicit test.
9511 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9514 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9515 address instead of to r1 and r11.
9517 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9519 * config.gcc: Support "knm".
9520 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9521 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9523 * config/i386/i386.c (m_KNM): Define.
9524 (processor_target_table): Add "knm".
9526 (ix86_option_override_internal): Add "knm".
9527 (ix86_issue_rate): Add PROCESSOR_KNM.
9528 (ix86_adjust_cost): Ditto.
9529 (ia32_multipass_dfa_lookahead): Ditto.
9530 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9531 (fold_builtin_cpu): Add M_INTEL_KNM.
9532 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9533 (processor_type): Add PROCESSOR_KNM.
9534 * config/i386/x86-tune.def: Add m_KNM.
9535 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9537 2017-09-20 Richard Biener <rguenther@suse.de>
9539 PR tree-optimization/80213
9540 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9541 are allowed in empty BBs as well.
9542 (canonicalize_loop_closed_ssa): Also look for other complex
9544 (scop_detection::get_sese): Include the loop-closed PHI block
9546 (scop_detection::merge_sese): Remove code adding extra blocks.
9547 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9548 (build_scops): Assert the final returned scop is invalid.
9550 2017-09-20 Richard Biener <rguenther@suse.de>
9552 PR tree-optimization/82264
9553 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9555 (vn_phi_lookup): Likewise.
9556 (vn_phi_insert): Likewise.
9558 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9560 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9561 that fits into uhwi or shwi, add DW_AT_const_value regardless
9562 of early_dwarf without going through RTL, using add_AT_unsigned
9565 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9566 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9567 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9568 (DEBUG_MACRO_SECTION): Likewise.
9569 (DEBUG_LTO_MACRO_SECTION): Likewise.
9570 (DEBUG_STR_DWO_SECTION): Likewise.
9571 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9572 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9573 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9574 (DEBUG_LTO_LINE_STR_SECTION): Define.
9575 (init_sections_and_labels): Initialize debug_line_str_section
9576 variable. Initialize debug_loc_section for -gdwarf-5 to
9577 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9579 2017-09-20 Richard Biener <rguenther@suse.de>
9581 * graphite-sese-to-poly.c (extract_affine): Properly handle
9582 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9584 2017-09-20 Richard Biener <rguenther@suse.de>
9586 PR tree-optimization/81373
9587 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9588 Force SESE live-out defs to be handled even if they are
9591 2017-09-19 Jeff Law <law@redhat.com>
9593 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9594 nothing for stack adjustments with REG_STACK_CHECK.
9595 * sched-deps.c (parse_add_or_inc): Reject insns with
9596 REG_STACK_CHECK from dependency breaking.
9597 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9598 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9599 * reg-notes.def (STACK_CHECK): New note.
9601 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9602 (ix86_expand_prologue): Dump stack clash info as needed.
9603 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9605 * function.c (dump_stack_clash_frame_info): New function.
9606 * function.h (dump_stack_clash_frame_info): Prototype.
9607 (enum stack_clash_probes): New enum.
9609 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9610 flag_stack_clash_protection.
9611 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9612 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9613 (arm_frame_pointer_required): Likewise.
9614 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9615 (ia64_expand_prologue): Likewise.
9616 * config/mips/mips.c (mips_expand_prologue): Likewise.
9617 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9618 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9619 (sparc_flat_expand_prologue): Likewise.
9620 * config/spu/spu.c (spu_expand_prologue): Likewise.
9622 * explow.c: Include "params.h".
9623 (anti_adjust_stack_and_probe_stack_clash): New function.
9624 (get_stack_check_protect): Likewise.
9625 (compute_stack_clash_protection_loop_data): Likewise.
9626 (emit_stack_clash_protection_loop_start): Likewise.
9627 (emit_stack_clash_protection_loop_end): Likewise.
9628 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9629 Use anti_adjust_stack_and_probe_stack_clash.
9630 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9631 (emit_stack_clash_protection_loop_start): Likewise.
9632 (emit_stack_clash_protection_loop_end): Likewise.
9633 * rtl.h (get_stack_check_protect): Prototype.
9634 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9635 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9636 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9638 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9640 * doc/tm.texi: Rebuilt.
9641 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9642 get_stack_check_protect.
9643 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9644 * config/arm/arm.c (arm_expand_prologue): Likewise.
9645 (arm_frame_pointer_required): Likewise.
9646 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9647 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9648 * config/mips/mips.c (mips_expand_prologue): Likewise.
9649 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9650 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9651 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9652 (sparc_flat_expand_prologue): Likewise.
9654 * common.opt (-fstack-clash-protection): New option.
9655 * flag-types.h (enum stack_check_type): Note difference between
9656 -fstack-check= and -fstack-clash-protection.
9657 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9658 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9659 * toplev.c (process_options): Issue warnings/errors for cases
9660 not handled with -fstack-clash-protection.
9661 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9662 (-fstack-check): Note additional problem with -fstack-check=generic.
9663 Note that -fstack-check is primarily for Ada and refer users
9664 to -fstack-clash-protection for stack-clash-protection.
9665 Document new params for stack clash protection.
9667 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9669 * config/i386/i386.c (ix86_split_long_move): Do not handle
9670 address used for LEA in a special way.
9672 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9674 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9676 2017-09-19 Martin Sebor <msebor@redhat.com>
9679 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9680 of incompatible types.
9682 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9684 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9685 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9686 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9687 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9689 2017-09-19 Richard Biener <rguenther@suse.de>
9691 PR tree-optimization/82244
9692 * tree-vrp.c (remove_range_assertions): Do not propagate
9693 a constant to abnormals but replace the assert with a copy.
9695 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9697 PR rtl-optimization/57878
9698 PR rtl-optimization/68988
9699 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9700 avoidance test involving non_reload_pseudos. Move frequency test
9701 below the general fragmentation avoidance test.
9703 2017-09-19 Richard Biener <rguenther@suse.de>
9705 PR tree-optimization/69728
9706 * graphite-sese-to-poly.c (schedule_error): New global.
9707 (add_loop_schedule): Handle empty domain by failing the
9709 (build_original_schedule): Handle schedule_error.
9711 2017-09-19 Richard Biener <rguenther@suse.de>
9713 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9714 Do not iterate to sibling loops but only to siblings of inner
9717 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9720 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9721 that the registers are different.
9723 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9725 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9726 to processor_model and "amdfam17h" to arch_names_table.
9727 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9729 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9732 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9734 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9735 Alan Hayward <alan.hayward@arm.com>
9736 David Sherwood <david.sherwood@arm.com>
9738 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9740 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9742 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9743 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9744 vect_schedule_slp_instance.
9745 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9746 Update call to vect_slp_analyze_node_operations. Simplify return
9748 (vect_slp_analyze_bb_1): Update call accordingly.
9749 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9750 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9751 (vect_schedule_slp): Update call accordingly.
9753 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9754 Alan Hayward <alan.hayward@arm.com>
9755 David Sherwood <david.sherwood@arm.com>
9757 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9758 with types that aren't in fact scalar.
9760 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9762 * tree-vect-slp.c (vect_record_max_nunits): New function,
9764 (vect_build_slp_tree_1): ...here.
9765 (vect_build_slp_tree_2): Call it for phis too.
9767 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9769 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9770 to vect_get_vec_def_for_operand when getting the mask operand.
9772 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9773 Alan Hayward <alan.hayward@arm.com>
9774 David Sherwood <david.sherwood@arm.com>
9776 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9779 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9780 Alan Hayward <alan.hayward@arm.com>
9781 David Sherwood <david.sherwood@arm.com>
9783 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9784 calculation for vector booleans.
9786 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9787 Alan Hayward <alan.hayward@arm.com>
9788 David Sherwood <david.sherwood@arm.com>
9790 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9792 (vect_transform_stmt): ...here.
9793 (vect_analyze_stmt): Use it instead of calling
9794 vectorizable_live_operation directly.
9796 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9798 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9799 non-SIMT targets in acc vector loops.
9801 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9803 * configure.ac: Add arc and check if assembler supports gdwarf2.
9804 * configure: Regenerate.
9806 2017-09-18 Richard Biener <rguenther@suse.de>
9808 PR tree-optimization/82220
9809 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9810 epilogue niters from the min_profitable_iters compute.
9812 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9815 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9816 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9817 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9819 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9822 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9823 switching to a new text section.
9825 2017-09-18 Richard Biener <rguenther@suse.de>
9827 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9829 (build_alias_set): Reject aliases with no access function.
9831 2017-09-18 Richard Biener <rguenther@suse.de>
9833 PR tree-optimization/79622
9834 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9836 (build_cross_bb_scalars_use): Likewise.
9838 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9840 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9842 2017-09-18 Alan Modra <amodra@gmail.com>
9845 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9846 stack_pointer_rtx for count 0. Update comments. Break up
9847 large rtl expression.
9849 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9851 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9852 Increase to 20 bytes.
9853 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9854 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9855 or avx version of the stub.
9857 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9860 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9861 compute the minimum stack alignment. Also update preferred stack
9862 boundary for leaf functions.
9864 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9866 PR tree-optimization/82228
9867 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9870 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9872 * common/config/nds32/nds32-common.c
9873 (nds32_option_optimization_table): Refine formatting.
9874 (nds32_option_optimization_table): Use -fsched-pressure and
9875 -fomit-frame-pointer for specific optimization level.
9877 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9879 * config/nds32/nds32.c: Refine formatting and comments.
9880 * config/nds32/nds32.h: Likewise.
9881 * config/nds32/nds32.md: Likewise.
9882 * config/nds32/nds32-cost.c: Likewise.
9883 * config/nds32/nds32-isr.c: Likewise.
9884 * config/nds32/nds32-md-auxiliary.c: Likewise.
9885 * config/nds32/nds32-multiple.md: Likewise.
9886 * config/nds32/nds32-predicates.c: Likewise.
9888 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9889 Jakub Jelinek <jakub@redhat.com>
9891 Add support for -std=c++2a.
9892 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9894 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9896 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9899 * doc/extend.texi (Common Function Attributes): Add
9900 references to ARM, AArch64, and S/390 specific attributes.
9901 (Function Specific Option Pragmas): Add AArch64 and S/390
9902 to list of back ends that support the target pragma.
9904 2017-09-15 Nathan Sidwell <nathan@acm.org>
9906 * doc/standards.texi: Fix C++17 description. Update URLs for
9909 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9911 * common.opt (Wcast-align=strict): New warning option.
9912 * doc/invoke.texi: Document -Wcast-align=strict.
9914 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9916 * cgraph.h (cgraph_thunk_info): Add comments.
9917 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9918 assert for VIRTUAL_* arguments stricter.
9920 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9922 PR tree-optimization/71026
9923 * match.pd: Move RDIV patterns from fold-const.c
9924 * fold-const.c (distribute_real_division): Removed.
9925 (fold_binary_loc): Remove calls to distribute_real_divison.
9927 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9929 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9930 c++1z and gnu++1z as deprecated. Change other references to
9931 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9932 Change -Wc++1z-compat to -Wc++17-compat.
9933 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9934 * dwarf2out.c (highest_c_language): Handle C++17.
9935 (gen_compile_unit_die): Likewise.
9937 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9939 PR rtl-optimization/82192
9940 * combine.c (make_extraction): Don't look through non-paradoxical
9941 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9944 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9945 Alan Hayward <alan.hayward@arm.com>
9946 David Sherwood <david.sherwood@arm.com>
9948 * target.def (function_arg_offset): New hook.
9949 * targhooks.h (default_function_arg_offset): Declare.
9950 * targhooks.c (default_function_arg_offset): New function.
9951 * function.c (locate_and_pad_parm): Use
9952 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9953 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9954 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9955 * doc/tm.texi: Regenerate.
9956 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9957 * config/spu/spu.c (spu_function_arg_offset): New function.
9958 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9959 * system.h (FUNCTION_ARG_OFFSET): Poison.
9961 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9962 Alan Hayard <alan.hayward@arm.com>
9963 David Sherwood <david.sherwood@arm.com>
9965 * target.def (truly_noop_truncation): New hook.
9966 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9967 than TRULY_NOOP_TRUNCATION.
9968 * hooks.h (hook_bool_uint_uint_true): Declare.
9969 * hooks.c (hook_bool_uint_uint_true): New function.
9970 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9971 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9972 * doc/tm.texi: Regenerate.
9973 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9974 rather than TRULY_NOOP_TRUNCATION in comments.
9975 (simplify_comparison): Likewise.
9976 (record_truncated_value): Likewise.
9977 * expmed.c (extract_bit_field_1): Likewise.
9978 (extract_split_bit_field): Likewise.
9979 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9980 instead of TRULY_NOOP_TRUNCATION.
9981 * function.c (assign_parm_setup_block): Likewise.
9982 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9983 * rtlhooks.c: Include target.h.
9984 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9985 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9986 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9987 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9988 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9989 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9990 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9991 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9992 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9993 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9994 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9995 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9996 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9997 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9998 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9999 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10000 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10001 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10002 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10003 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10004 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10005 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10006 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10007 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10008 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10009 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10010 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10011 rather than TRULY_NOOP_TRUNCATION in comments.
10012 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10013 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10014 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10015 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10016 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10017 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10018 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10019 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10020 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10021 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10022 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10023 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10024 rather than TRULY_NOOP_TRUNCATION in comments.
10025 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10026 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10027 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10028 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10029 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10030 TRULY_NOOP_TRUNCATION condition.
10031 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10032 (TRULY_NOOP_TRUNCATION): Delete.
10033 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10034 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10035 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10036 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10037 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10038 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10039 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10040 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10041 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10042 rather than TRULY_NOOP_TRUNCATION in comments.
10043 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10044 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10045 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10046 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10047 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10048 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10050 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10053 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10054 (*cmp_ior): Likewise.
10055 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10056 (*ior_scc_scc_cmp): Likewise.
10057 (*and_scc_scc): Likewise.
10058 (*and_scc_scc_cmp): Likewise.
10060 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10061 Alan Hayard <alan.hayward@arm.com>
10062 David Sherwood <david.sherwood@arm.com>
10064 * target.def (can_change_mode_class): New hook.
10065 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10066 (hard_regno_nregs): Likewise.
10067 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10068 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10069 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10070 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10071 (LOAD_EXTEND_OP): Update accordingly.
10072 * doc/tm.texi: Regenerate.
10073 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10074 CANNOT_CHANGE_MODE_CLASS.
10075 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10076 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10077 * combine.c (simplify_set): Update accordingly.
10078 * emit-rtl.c (validate_subreg): Likewise.
10079 * recog.c (general_operand): Likewise.
10080 * regcprop.c (mode_change_ok): Likewise.
10081 * reload1.c (choose_reload_regs): Likewise.
10082 (inherit_piecemeal_p): Likewise.
10083 * rtlanal.c (simplify_subreg_regno): Likewise.
10084 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10085 instead of CANNOT_CHANGE_MODE_CLASS.
10086 (reload_cse_simplify_operands): Likewise.
10087 * reload.c (push_reload): Use targetm.can_change_mode_class
10088 instead of CANNOT_CHANGE_MODE_CLASS.
10089 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10090 REG_CANNOT_CHANGE_MODE_P.
10091 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10092 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10093 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10094 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10095 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10096 (arm_can_change_mode_class): New function.
10097 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10098 than CANNOT_CHANGE_MODE_CLASS in comments.
10099 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10100 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10101 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10102 (ix86_can_change_mode_class): ...this new function, inverting the
10103 sense of the return value.
10104 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10105 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10106 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10107 (ia64_can_change_mode_class): New function.
10108 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10109 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10110 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10111 (m32c_can_change_mode_class): ...this new function, inverting the
10112 sense of the return value.
10113 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10114 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10115 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10116 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10117 (mips_can_change_mode_class): ...this new function, inverting the
10118 sense of the return value.
10119 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10120 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10121 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10122 (msp430_can_change_mode_class): New function.
10123 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10124 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10125 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10126 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10127 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10128 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10129 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10130 (pa_cannot_change_mode_class): Replace with...
10131 (pa_can_change_mode_class): ...this new function, inverting the
10132 sense of the return value.
10133 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10134 than CANNOT_CHANGE_MODE_CLASS in comments.
10135 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10136 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10137 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10138 (pdp11_cannot_change_mode_class): Replace with...
10139 (pdp11_can_change_mode_class): ...this new function, inverting the
10140 sense of the return value.
10141 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10142 * config/powerpcspe/powerpcspe-protos.h
10143 (rs6000_cannot_change_mode_class_ptr): Delete.
10144 * config/powerpcspe/powerpcspe.c
10145 (rs6000_cannot_change_mode_class_ptr): Delete.
10146 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10147 (rs6000_option_override_internal): Assign to
10148 targetm.can_change_mode_class instead of
10149 rs6000_cannot_change_mode_class_ptr.
10150 (rs6000_cannot_change_mode_class): Replace with...
10151 (rs6000_can_change_mode_class): ...this new function, inverting the
10152 sense of the return value.
10153 (rs6000_debug_cannot_change_mode_class): Replace with...
10154 (rs6000_debug_can_change_mode_class): ...this new function.
10155 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10156 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10157 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10158 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10159 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10161 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10162 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10163 (rs6000_option_override_internal): Assign to
10164 targetm.can_change_mode_class instead of
10165 rs6000_cannot_change_mode_class_ptr.
10166 (rs6000_cannot_change_mode_class): Replace with...
10167 (rs6000_can_change_mode_class): ...this new function, inverting the
10168 sense of the return value.
10169 (rs6000_debug_cannot_change_mode_class): Replace with...
10170 (rs6000_debug_can_change_mode_class): ...this new function.
10171 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10172 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10173 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10174 (s390_can_change_mode_class): ...this new function, inverting the
10175 sense of the return value.
10176 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10177 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10178 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10179 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10180 (sh_cannot_change_mode_class): Replace with...
10181 (sh_can_change_mode_class): ...this new function, inverting the
10182 sense of the return value.
10183 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10184 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10185 (sparc_can_change_mode_class): New function.
10186 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10187 * config/spu/spu.c (spu_can_change_mode_class): New function.
10188 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10189 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10190 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10191 (visium_can_change_mode_class): New function.
10192 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10194 2017-09-15 Richard Biener <rguenther@suse.de>
10196 PR tree-optimization/82217
10197 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10198 but not undefined case.
10200 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10203 * postreload.c (reload_cse_simplify_operands): Skip
10204 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10206 2017-09-15 Richard Biener <rguenther@suse.de>
10208 PR tree-optimization/68823
10209 * graphite-scop-detection.c (build_alias_set): If we have a
10210 possible dependence check whether we can handle them by just
10211 looking at the DRs DR_ACCESS_FNs.
10212 (build_scops): If build_alias_set fails, fail the SCOP.
10214 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10216 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10217 to support float128 built-in functions that require the ISA 3.0
10219 (BU_FLOAT128_3_HW): Likewise.
10220 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10221 built-in functions.
10222 (FMAF128): Likewise.
10224 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10225 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10226 floating point instructions.
10227 (rs6000_invalid_builtin): Likewise.
10228 (rs6000_builtin_mask_names): Likewise.
10229 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10230 (RS6000_BTM_FLOAT128_HW): Likewise.
10231 (RS6000_BTM_COMMON): Likewise.
10232 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10234 * doc/extend.texi (RS/6000 built-in functions): Document the
10235 IEEE 128-bit floating point square root and fused multiply-add
10236 built-in functions.
10238 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10240 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10241 reg (r2) isn't in the set of registers defined in the prologue.
10243 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10244 Alan Hayward <alan.hayward@arm.com>
10245 David Sherwood <david.sherwood@arm.com>
10247 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10248 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10249 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10250 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10251 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10253 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10254 max_vectorization_factor.
10255 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10257 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10258 Alan Hayward <alan.hayward@arm.com>
10259 David Sherwood <david.sherwood@arm.com>
10261 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10262 (vect_worthwhile_without_simd_p): Declare.
10263 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10264 (vectorizable_reduction): Use it.
10265 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10266 (vectorizable_operation): Likewise.
10268 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10269 Alan Hayward <alan.hayward@arm.com>
10270 David Sherwood <david.sherwood@arm.com>
10272 * tree-vectorizer.h (vect_get_num_copies): New function.
10273 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10274 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10275 (vectorizable_induction): Likewise.
10276 (vectorizable_live_operation): Likewise.
10277 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10278 (vectorizable_bswap): Likewise.
10279 (vectorizable_call): Likewise.
10280 (vectorizable_conversion): Likewise.
10281 (vectorizable_assignment): Likewise.
10282 (vectorizable_shift): Likewise.
10283 (vectorizable_operation): Likewise.
10284 (vectorizable_store): Likewise.
10285 (vectorizable_load): Likewise.
10286 (vectorizable_condition): Likewise.
10287 (vectorizable_comparison): Likewise.
10288 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10290 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10291 Alan Hayward <alan.hayward@arm.com>
10292 David Sherwood <david.sherwood@arm.com>
10294 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10295 of vect_init_vector.
10297 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10298 Alan Hayward <alan.hayward@arm.com>
10299 David Sherwood <david.sherwood@arm.com>
10301 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10302 an inline wrapper that provides a location.
10303 (gimple_build_vector): Likewise.
10304 * gimple-fold.c (gimple_build_vector_from_val): New function.
10305 (gimple_build_vector): Likewise.
10306 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10307 functions to build the initial value. Always return a gimple value.
10308 (get_initial_defs_for_reduction): Likewise. Only compute
10310 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10311 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10312 (vectorizable_induction): Use gimple_build_vector rather than
10315 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10316 Alan Hayward <alan.hayward@arm.com>
10317 David Sherwood <david.sherwood@arm.com>
10319 * target.h (vec_perm_indices): New typedef.
10320 (auto_vec_perm_indices): Likewise.
10321 * optabs-query.h: Include target.h
10322 (can_vec_perm_p): Take a vec_perm_indices *.
10323 * optabs-query.c (can_vec_perm_p): Likewise.
10324 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10325 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10326 * tree-vect-generic.c (lower_vec_perm): Likewise.
10327 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10328 (vect_grouped_load_supported): Likewise.
10329 (vect_shift_permute_load_chain): Likewise.
10330 (vect_permute_store_chain): Use auto_vec_perm_indices.
10331 (vect_permute_load_chain): Likewise.
10332 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10333 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10334 Update uses of can_vec_perm_p.
10335 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10336 mode with a number of elements. Take a vec_perm_indices *.
10337 (vect_create_epilog_for_reduction): Update accordingly.
10338 Use auto_vec_perm_indices.
10339 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10340 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10341 (vect_transform_slp_perm_load): Likewise.
10342 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10343 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10344 (vect_gen_perm_mask_checked): Likewise.
10345 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10346 (vect_gen_perm_mask_checked): Likewise.
10347 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10348 (vectorizable_store): Likewise.
10349 (vectorizable_load): Likewise.
10350 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10351 (vectorizable_bswap): Likewise.
10353 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10354 Alan Hayward <alan.hayward@arm.com>
10355 David Sherwood <david.sherwood@arm.com>
10357 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10358 * tree.c (build_vector): Likewise.
10359 (build_vector_from_ctor): Update accordingly.
10360 (build_vector_from_val): Likewise.
10361 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10362 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10363 * tree-vect-generic.c (add_rshift): Likewise.
10364 (expand_vector_divmod): Likewise.
10365 (optimize_vector_constructor): Likewise.
10366 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10367 (vect_transform_slp_perm_load): Likewise.
10368 (vect_schedule_slp_instance): Likewise.
10369 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10370 (vectorizable_call): Likewise.
10371 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10372 * expmed.c (make_tree): Likewise.
10373 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10374 a vector passed to build_vector.
10375 (fold_convert_const): Likewise.
10376 (exact_inverse): Likewise.
10377 (fold_ternary_loc): Likewise.
10378 (fold_relational_const): Likewise.
10379 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10380 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10381 (const_unop): Likewise. Store the reduction accumulator in a
10382 variable rather than an array.
10383 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10384 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10385 the new vector, rather than constructing it after the input arrays.
10386 (native_interpret_vector): Use auto_vec<tree> when building
10387 a vector passed to build_vector. Add elements in order.
10388 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10389 auto_vec<tree> when building a vector passed to build_vector.
10390 (vect_create_epilog_for_reduction): Likewise.
10391 (vectorizable_induction): Likewise.
10392 (get_initial_def_for_reduction): Likewise. Fix indentation of
10394 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10396 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10397 passed to build_vector.
10399 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10400 Alan Hayward <alan.hayward@arm.com>
10401 David Sherwood <david.sherwood@arm.com>
10403 * tree-core.h (tree_base::u): Add an "nelts" field.
10404 (tree_vector): Use VECTOR_CST_NELTS as the length.
10405 * tree.c (tree_size): Likewise.
10406 (make_vector): Initialize VECTOR_CST_NELTS.
10407 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10408 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10409 TYPE_VECTOR_SUBPARTS.
10410 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10411 as the number of units, setting it from VECTOR_CST_NELTS.
10412 (const_vector_from_tree): Likewise.
10413 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10414 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10415 (fold_negate_expr_1): Likewise.
10416 (fold_convert_const): Likewise.
10417 (const_binop): Likewise. Differentiate the number of output and
10419 (const_unop): Likewise.
10420 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10421 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10422 in cases that did the opposite.
10424 2017-09-14 Richard Biener <rguenther@suse.de>
10426 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10429 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10431 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10433 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10436 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10437 if and where to split a bb, except for splitting before debug insn
10438 sequences followed by non-label real insn. Delete debug insns
10439 in between basic blocks.
10441 * combine.c (make_compound_operation_int): Formatting fixes.
10443 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10444 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10445 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10446 * config/sol2.h (LINK_EH_SPEC): Likewise.
10447 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10448 * config/s390/linux.h (LINK_SPEC): Likewise.
10449 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10450 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10451 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10452 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10453 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10454 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10455 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10456 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10458 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10460 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10462 (ENDFILE_LINUX_SPEC): Likewise.
10463 (LINK_EH_SPEC): Likewise.
10464 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10465 (LINK_OS_LINUX_SPEC32): Likewise.
10466 (LINK_OS_LINUX_SPEC64): Likewise.
10467 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10468 (LINK_OS_LINUX_SPEC): Likewise.
10470 2017-09-13 Martin Liska <mliska@suse.cz>
10472 PR middle-end/82154
10473 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10474 CASE_HIGH is NULL_TREE.
10476 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10477 Alan Hayward <alan.hayward@arm.com>
10478 David Sherwood <david.sherwood@arm.com>
10480 * target.def (secondary_memory_needed): New hook.
10481 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10482 instead of SECONDARY_MEMORY_NEEDED.
10483 (secondary_memory_needed_mode): Likewise.
10484 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10485 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10486 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10487 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10488 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10489 * doc/tm.texi: Regenerate.
10490 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10491 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10492 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10493 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10494 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10495 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10496 mode argument first and change the reg_class arguments to reg_class_t.
10497 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10498 Make static. Update the call to inline_secondary_memory_needed.
10499 (ix86_register_move_cost): Update the call to
10500 inline_secondary_memory_needed.
10501 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10502 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10504 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10505 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10507 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10508 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10509 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10510 and match hook interface. Add comment from mips.h.
10511 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10512 * config/mmix/mmix.md (truncdfsf2): Refer to
10513 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10515 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10516 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10517 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10518 (pa_secondary_memory_needed): New function.
10519 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10520 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10521 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10522 (pdp11_secondary_memory_needed): Make static and match hook interface.
10523 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10524 * config/powerpcspe/powerpcspe-protos.h
10525 (rs6000_secondary_memory_needed_ptr): Delete.
10526 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10528 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10529 (rs6000_option_override_internal): Assign to
10530 targetm.secondary_memory_needed rather than
10531 rs6000_secondary_memory_needed_ptr.
10532 (rs6000_secondary_memory_needed): Match hook interface.
10533 (rs6000_debug_secondary_memory_needed): Likewise.
10534 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10535 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10536 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10537 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10538 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10539 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10541 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10542 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10543 (rs6000_option_override_internal): Assign to
10544 targetm.secondary_memory_needed rather than
10545 rs6000_secondary_memory_needed_ptr.
10546 (rs6000_secondary_memory_needed): Match hook interface.
10547 (rs6000_debug_secondary_memory_needed): Likewise.
10548 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10549 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10550 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10551 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10552 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10553 (sparc_secondary_memory_needed): New function.
10554 * lra-constraints.c (check_and_process_move): Refer to
10555 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10557 (curr_insn_transform): Likewise.
10558 (process_alt_operands): Use targetm.secondary_memory_needed
10559 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10560 (check_secondary_memory_needed_p): Likewise.
10561 (choose_split_class): Likewise.
10562 * reload.c: Unconditionally include code that was previously
10563 conditional on SECONDARY_MEMORY_NEEDED.
10564 (push_secondary_reload): Use targetm.secondary_memory_needed
10565 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10566 (push_reload): Likewise.
10567 * reload1.c: Unconditionally include code that was previously
10568 conditional on SECONDARY_MEMORY_NEEDED.
10569 (choose_reload_regs): Use targetm.secondary_memory_needed
10570 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10571 (gen_reload): Likewise.
10572 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10574 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10575 Alan Hayward <alan.hayward@arm.com>
10576 David Sherwood <david.sherwood@arm.com>
10578 * target.def (secondary_memory_needed_mode): New hook:
10579 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10580 * targhooks.h (default_secondary_memory_needed_mode): New function.
10581 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10582 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10583 * doc/tm.texi: Regenerate.
10584 * lra-constraints.c (check_and_process_move): Use
10585 targetm.secondary_memory_needed_mode instead of
10586 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10587 (curr_insn_transform): Likewise.
10588 * reload.c (get_secondary_mem): Likewise.
10589 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10590 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10592 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10593 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10594 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10595 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10596 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10598 * config/powerpcspe/powerpcspe-protos.h
10599 (rs6000_secondary_memory_needed_mode): Delete.
10600 * config/powerpcspe/powerpcspe.c
10601 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10602 (rs6000_secondary_memory_needed_mode): Make static.
10603 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10604 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10606 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10608 (rs6000_secondary_memory_needed_mode): Make static.
10609 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10610 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10611 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10612 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10613 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10615 (sparc_secondary_memory_needed_mode): New function.
10616 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10618 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10620 * config/aarch64/constraints.md (Umq): New constraint.
10621 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10623 (mov<mode>): Update condition.
10625 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10627 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10628 when bitposition is the same.
10630 2017-09-13 Richard Biener <rguenther@suse.de>
10632 * dwarf2out.c (output_die_symbol): Remove.
10633 (output_die): Do not output a DIEs symbol.
10635 2017-09-13 Richard Biener <rguenther@suse.de>
10637 PR middle-end/82128
10638 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10639 default-def to avoid breaking iterator update with the weird
10640 interaction with cgraph_update_edges_for_call_stmt_node.
10642 2017-09-13 Richard Biener <rguenther@suse.de>
10644 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10645 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10646 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10647 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10649 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10651 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10652 Disable pc relative literal load irrespective of
10653 TARGET_FIX_ERR_A53_84341 for default.
10655 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10657 * config/sparc/sparc.c (output_return): Output the source location of
10658 the insn in the delay slot, if any.
10659 (output_sibcall): Likewise.
10661 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10664 * common.opt (-static-pie): New alias.
10665 (shared): Negate static-pie.
10666 (-no-pie): Update help text.
10668 (static-pie): New option.
10669 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10670 -static-pie support.
10671 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10672 (LINK_EH_SPEC): Likewise.
10673 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10674 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10675 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10676 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10677 (init_gcc_specs): Likewise.
10678 (init_spec): Likewise.
10679 (display_help): Update help message for -pie.
10680 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10683 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10685 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10686 (movdi_aarch64): Likewise.
10687 (movti_aarch64): Likewise.
10689 2017-09-12 Simon Wright <simon@pushface.org>
10692 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10693 calculation of the minor version, always output as 0.
10695 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10698 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10699 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10700 on it early, rather than manual conversion late. For
10701 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10702 instead of performing manual conversion.
10704 2017-09-12 Carl Love <cel@us.ibm.com>
10706 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10707 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10709 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10710 VMULOSW): Add definitions.
10711 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10712 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10713 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10714 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10715 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10717 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10719 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10721 (movti_aarch64): Likewise.
10722 (movdf_aarch64): Likewise.
10723 (movtf_aarch64): Likewise.
10724 (load_pairdi): Likewise.
10725 (store_pairdi): Likewise.
10726 (load_pairdf): Likewise.
10727 (store_pairdf): Likewise.
10728 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10729 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10730 (ldr_got_small_<mode>): Likewise.
10731 (ldr_got_small_28k_<mode>): Likewise.
10732 (ldr_got_tiny): Likewise.
10733 * config/aarch64/iterators.md (ldst_sz): New.
10734 (ldpstp_sz): Likewise.
10735 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10737 (thunderx_load): Split load_8 to load_16.
10738 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10740 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10741 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10742 (xgene1_store_pair): Split store_8 to store_16.
10743 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10744 (falkor_st_0_st_sd): Split store_8 to store_16.
10746 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10748 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10749 and store1/2/3/4 to store_4/8/12/16.
10750 * config/aarch64/aarch64.md: Update for rename.
10751 * config/arm/arm.md: Likewise.: Likewise.
10752 * config/arm/arm.c: Likewise.
10753 * config/arm/thumb1.md: Likewise.
10754 * config/arm/thumb2.md: Likewise.
10755 * config/arm/vfp.md: Likewise.
10756 * config/arm/arm-generic.md: Likewise.
10757 * config/arm/arm1020e.md: Likewise.
10758 * config/arm/arm1026ejs.md: Likewise.
10759 * config/arm/arm1136jfs.md: Likewise.
10760 * config/arm/arm926ejs.md: Likewise.
10761 * config/arm/cortex-a15.md: Likewise.
10762 * config/arm/cortex-a17.md: Likewise.
10763 * config/arm/cortex-a5.md: Likewise.
10764 * config/arm/cortex-a53.md: Likewise.
10765 * config/arm/cortex-a57.md: Likewise.
10766 * config/arm/cortex-a7.md: Likewise.
10767 * config/arm/cortex-a8.md: Likewise.
10768 * config/arm/cortex-a9.md: Likewise.
10769 * config/arm/cortex-m4.md: Likewise.
10770 * config/arm/cortex-m7.md: Likewise.
10771 * config/arm/cortex-r4.md: Likewise.
10772 * config/arm/exynos-m1.md: Likewise.
10773 * config/arm/fa526.md: Likewise.
10774 * config/arm/fa606te.md: Likewise.
10775 * config/arm/fa626te.md: Likewise.
10776 * config/arm/fa726te.md: Likewise.
10777 * config/arm/fmp626.md: Likewise.
10778 * config/arm/iwmmxt.md: Likewise.
10779 * config/arm/ldmstm.md: Likewise.
10780 * config/arm/marvell-pj4.md: Likewise.
10781 * config/arm/xgene1.md: Likewise.
10782 * config/aarch64/thunderx.md: Likewise.
10783 * config/aarch64/thunderx2t99.md: Likewise.
10784 * config/aarch64/falkor.md: Likewise.
10786 2017-09-12 Martin Liska <mliska@suse.cz>
10788 * attribs.c (private_lookup_attribute): New function.
10789 * attribs.h (private_lookup_attribute): Declared here.
10790 (lookup_attribute): Called from this place.
10792 2017-09-12 Richard Biener <rguenther@suse.de>
10794 PR tree-optimization/82157
10795 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10796 stmts with side-effects.
10798 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10799 Alan Hayward <alan.hayward@arm.com>
10800 David Sherwood <david.sherwood@arm.com>
10802 * target.def (hard_regno_nregs): New hook.
10803 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10804 * targhooks.h (default_hard_regno_nregs): Declare.
10805 * targhooks.c (default_hard_regno_nregs): New function.
10806 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10807 (TARGET_HARD_REGNO_NREGS): ...this hook.
10808 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10809 (CLASS_MAX_NREGS): Likewise.
10810 * doc/tm.texi: Regenerate.
10811 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10812 instead of HARD_REGNO_NREGS.
10813 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10814 HARD_REGNO_NREGS in the comment.
10815 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10816 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10817 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10818 Return an unsigned int.
10819 (TARGET_HARD_REGNO_NREGS): Redefine.
10820 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10821 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10822 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10823 (arc_hard_regno_nregs): New function.
10824 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10825 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10826 (arm_hard_regno_nregs): New function.
10827 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10828 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10829 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10830 (TARGET_HARD_REGNO_NREGS): Redefine.
10831 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10832 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10833 (HARD_REGNO_NREGS): Delete.
10834 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10835 (cr16_hard_regno_nregs): New function.
10836 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10837 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10838 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10839 (cris_hard_regno_nregs): New function.
10840 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10841 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10842 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10843 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10844 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10845 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10846 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10847 (frv_hard_regno_nregs): Make static. Take and return an
10849 (frv_class_max_nregs): Remove outdated copy of documentation.
10850 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10851 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10852 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10853 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10854 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10855 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10856 (TARGET_HARD_REGNO_NREGS): Redefine.
10857 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10858 (CLASS_MAX_NREGS): Update comment.
10859 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10860 (ia64_hard_regno_nregs): New function.
10861 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10862 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10863 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10864 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10865 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10867 (m32c_hard_regno_nregs): Likewise. Make static.
10868 (TARGET_HARD_REGNO_NREGS): Redefine.
10869 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10870 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10871 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10872 (m68k_hard_regno_nregs): New function.
10873 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10874 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10875 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10876 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10877 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10878 Take and return an unsigned int.
10879 (TARGET_HARD_REGNO_NREGS): Redefine.
10880 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10881 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10882 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10883 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10884 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10885 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10886 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10887 (msp430_hard_regno_nregs): Make static. Take and return an
10889 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10890 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10891 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10892 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10893 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10894 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10895 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10896 (TARGET_HARD_REGNO_NREGS): Redefine.
10897 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10898 (PA_HARD_REGNO_NREGS): ...this.
10899 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10900 (PA_HARD_REGNO_NREGS): ...this.
10901 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10902 (pa_hard_regno_nregs): New function.
10903 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10904 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10905 (pdp11_hard_regno_nregs): New function.
10906 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10907 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10908 (rs6000_hard_regno_nregs_hook): New function.
10909 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10910 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10911 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10912 Take and return an unsigned int. Move earlier in file.
10913 (TARGET_HARD_REGNO_NREGS): Redefine.
10914 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10915 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10916 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10917 (rl78_hard_regno_nregs): Make static. Take and return an
10919 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10920 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10921 (rs6000_hard_regno_nregs_hook): New function.
10922 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10923 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10924 (TARGET_HARD_REGNO_NREGS): Redefine.
10925 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10926 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10927 instead of HARD_REGNO_NREGS.
10928 (s390_hard_regno_nregs): New function.
10929 (s390_hard_regno_mode_ok): Add comment from s390.h.
10930 (TARGET_HARD_REGNO_NREGS): Redefine.
10931 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10932 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10933 (sh_hard_regno_nregs): New function.
10934 (sh_pass_in_reg_p): Use it.
10935 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10936 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10937 (sparc_hard_regno_nregs): New function.
10938 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10939 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10940 (spu_function_arg_advance): Use it, supplying a valid register number.
10941 (TARGET_HARD_REGNO_NREGS): Redefine.
10942 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10943 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10944 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10945 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10946 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10947 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10948 (CLASS_MAX_NREGS): Remove copy of old documentation.
10949 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10950 (visium_hard_regno_nregs): New function.
10951 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10952 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10953 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10954 xtensa_hard_regno_nregs): New function.
10955 * system.h (HARD_REGNO_NREGS): Poison.
10957 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10959 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10960 hard_regno_nregs instead of HARD_REGNO_NREGS.
10961 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10962 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10963 (c6x_expand_epilogue): Likewise.
10964 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10965 (frv_read_iacc_argument): Likewise.
10966 * config/sh/sh.c: Include regs.h.
10967 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10968 (regs_used): Likewise.
10969 (output_stack_adjust): Likewise.
10970 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10971 * expmed.c: Include regs.h.
10972 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10973 * ree.c: Include regs.h.
10974 (combine_reaching_defs): Use hard_regno_nregs instead of
10976 (add_removable_extension): Likewise.
10978 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10980 * regs.h (hard_regno_nregs): Turn into a function.
10981 (end_hard_regno): Update accordingly.
10982 * caller-save.c (setup_save_areas): Likewise.
10983 (save_call_clobbered_regs): Likewise.
10984 (replace_reg_with_saved_mem): Likewise.
10985 (insert_restore): Likewise.
10986 (insert_save): Likewise.
10987 * combine.c (can_change_dest_mode): Likewise.
10988 (move_deaths): Likewise.
10989 (distribute_notes): Likewise.
10990 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10991 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10992 (rs6000_split_multireg_move): Likewise.
10993 (rs6000_register_move_cost): Likewise.
10994 (rs6000_memory_move_cost): Likewise.
10995 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10996 (rs6000_split_multireg_move): Likewise.
10997 (rs6000_register_move_cost): Likewise.
10998 (rs6000_memory_move_cost): Likewise.
10999 * cselib.c (cselib_reset_table): Likewise.
11000 (cselib_lookup_1): Likewise.
11001 * emit-rtl.c (set_mode_and_regno): Likewise.
11002 * function.c (aggregate_value_p): Likewise.
11003 * ira-color.c (setup_profitable_hard_regs): Likewise.
11004 (check_hard_reg_p): Likewise.
11005 (calculate_saved_nregs): Likewise.
11006 (assign_hard_reg): Likewise.
11007 (improve_allocation): Likewise.
11008 (calculate_spill_cost): Likewise.
11009 * ira-emit.c (modify_move_list): Likewise.
11010 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11011 (ira_hard_reg_in_set_p): Likewise.
11012 * ira.c (setup_reg_mode_hard_regset): Likewise.
11013 (clarify_prohibited_class_mode_regs): Likewise.
11014 (check_allocation): Likewise.
11015 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11016 (lra_setup_reg_renumber): Likewise.
11017 (setup_try_hard_regno_pseudos): Likewise.
11018 (spill_for): Likewise.
11019 (assign_hard_regno): Likewise.
11020 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11021 * lra-constraints.c (in_class_p): Likewise.
11022 (lra_constraint_offset): Likewise.
11023 (simplify_operand_subreg): Likewise.
11024 (lra_constraints): Likewise.
11025 (split_reg): Likewise.
11026 (split_if_necessary): Likewise.
11027 (invariant_p): Likewise.
11028 (inherit_in_ebb): Likewise.
11029 * lra-lives.c (process_bb_lives): Likewise.
11030 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11031 (get_hard_regs): Likewise.
11032 (do_remat): Likewise.
11033 * lra-spills.c (assign_spill_hard_regs): Likewise.
11034 * mode-switching.c (create_pre_exit): Likewise.
11035 * postreload.c (reload_combine_recognize_pattern): Likewise.
11036 * recog.c (peep2_find_free_register): Likewise.
11037 * regcprop.c (kill_value_regno): Likewise.
11038 (set_value_regno): Likewise.
11039 (copy_value): Likewise.
11040 (maybe_mode_change): Likewise.
11041 (find_oldest_value_reg): Likewise.
11042 (copyprop_hardreg_forward_1): Likewise.
11043 * regrename.c (check_new_reg_p): Likewise.
11044 (regrename_do_replace): Likewise.
11045 * reload.c (push_reload): Likewise.
11046 (combine_reloads): Likewise.
11047 (find_dummy_reload): Likewise.
11048 (operands_match_p): Likewise.
11049 (find_reloads): Likewise.
11050 (find_equiv_reg): Likewise.
11051 (reload_adjust_reg_for_mode): Likewise.
11052 * reload1.c (count_pseudo): Likewise.
11053 (count_spilled_pseudo): Likewise.
11054 (find_reg): Likewise.
11055 (clear_reload_reg_in_use): Likewise.
11056 (free_for_value_p): Likewise.
11057 (allocate_reload_reg): Likewise.
11058 (choose_reload_regs): Likewise.
11059 (reload_adjust_reg_for_temp): Likewise.
11060 (emit_reload_insns): Likewise.
11061 (delete_output_reload): Likewise.
11062 * rtlanal.c (subreg_get_info): Likewise.
11063 * sched-deps.c (sched_analyze_reg): Likewise.
11064 * sel-sched.c (init_regs_for_mode): Likewise.
11065 (mark_unavailable_hard_regs): Likewise.
11066 (choose_best_reg_1): Likewise.
11067 (verify_target_availability): Likewise.
11068 * valtrack.c (dead_debug_insert_temp): Likewise.
11069 * var-tracking.c (track_loc_p): Likewise.
11070 (emit_note_insn_var_location): Likewise.
11071 * varasm.c (make_decl_rtl): Likewise.
11072 * reginfo.c (choose_hard_reg_mode): Likewise.
11073 (init_reg_modes_target): Refer directly to
11074 this_target_regs->x_hard_regno_nregs.
11076 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11078 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11079 instead of hard_regno_nregs.
11081 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11083 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11084 end_hard_regno instead of hard_regno_nregs.
11085 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11086 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11087 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11088 * ira-color.c (improve_allocation): Likewise.
11089 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11090 * lra-lives.c (mark_regno_live): Likewise.
11091 (mark_regno_dead): Likewise.
11092 * lra-remat.c (operand_to_remat): Likewise.
11093 * lra.c (collect_non_operand_hard_regs): Likewise.
11094 * postreload.c (reload_combine_note_store): Likewise.
11095 (move2add_valid_value_p): Likewise.
11096 * reload.c (regno_clobbered_p): Likewise.
11098 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11100 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11102 * config/v850/v850.c (v850_reorg): Likewise.
11103 * reload.c (refers_to_regno_for_reload_p): Likewise.
11104 (find_equiv_reg): Likewise.
11105 * reload1.c (reload_reg_reaches_end_p): Likewise.
11107 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11109 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11111 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11112 * config/arm/arm.c (output_move_neon): Likewise.
11113 (arm_attr_length_move_neon): Likewise.
11114 (neon_split_vcombine): Likewise.
11115 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11116 (c6x_mark_reg_written): Likewise.
11117 (c6x_dwarf_register_span): Likewise.
11118 * config/i386/i386.c (ix86_save_reg): Likewise.
11119 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11120 (rws_access_reg): Likewise.
11121 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11122 * mode-switching.c (create_pre_exit): Likewise.
11123 * ree.c (combine_reaching_defs): Likewise.
11124 (add_removable_extension): Likewise.
11125 * regcprop.c (find_oldest_value_reg): Likewise.
11126 (copyprop_hardreg_forward_1): Likewise.
11127 * reload.c (reload_inner_reg_of_subreg): Likewise.
11128 (push_reload): Likewise.
11129 (combine_reloads): Likewise.
11130 (find_dummy_reload): Likewise.
11131 (reload_adjust_reg_for_mode): Likewise.
11132 * reload1.c (find_reload_regs): Likewise.
11133 (forget_old_reloads_1): Likewise.
11134 (reload_reg_free_for_value_p): Likewise.
11135 (reload_adjust_reg_for_temp): Likewise.
11136 (emit_reload_insns): Likewise.
11137 (delete_output_reload): Likewise.
11138 * sel-sched.c (choose_best_reg_1): Likewise.
11139 (choose_best_pseudo_reg): Likewise.
11141 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11142 Alan Hayward <alan.hayward@arm.com>
11143 David Sherwood <david.sherwood@arm.com>
11145 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11146 * target.def (slow_unaligned_access): New hook.
11147 * targhooks.h (default_slow_unaligned_access): Declare.
11148 * targhooks.c (default_slow_unaligned_access): New function.
11149 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11150 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11151 * doc/tm.texi: Regenerate.
11152 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11153 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11154 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11156 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11157 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11159 (rs6000_slow_unaligned_access): New function.
11160 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11161 (expand_block_compare): Likewise.
11162 (expand_strn_compare): Likewise.
11163 (rs6000_rtx_costs): Likewise.
11164 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11165 (riscv_slow_unaligned_access): Likewise.
11166 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11167 (riscv_slow_unaligned_access_p): ...this and make static.
11168 (riscv_option_override): Update accordingly.
11169 (riscv_slow_unaligned_access): New function.
11170 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11171 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11172 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11173 (rs6000_slow_unaligned_access): New function.
11174 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11175 (rs6000_rtx_costs): Likewise.
11176 * config/rs6000/rs6000-string.c (expand_block_compare)
11177 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11178 of SLOW_UNALIGNED_ACCESS.
11179 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11180 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11181 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11182 of SLOW_UNALIGNED_ACCESS.
11183 * expmed.c (simple_mem_bitfield_p): Likewise.
11184 * expr.c (alignment_for_piecewise_move): Likewise.
11185 (emit_group_load_1): Likewise.
11186 (emit_group_store): Likewise.
11187 (copy_blkmode_from_reg): Likewise.
11188 (emit_push_insn): Likewise.
11189 (expand_assignment): Likewise.
11190 (store_field): Likewise.
11191 (expand_expr_real_1): Likewise.
11192 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11193 * lra-constraints.c (simplify_operand_subreg): Likewise.
11194 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11195 * gimple-ssa-store-merging.c: Likewise in block comment at start
11197 * tree-ssa-strlen.c: Include target.h.
11198 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11199 of SLOW_UNALIGNED_ACCESS.
11200 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11202 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11204 PR rtl-optimization/82185
11205 * expmed.c (emit_store_flag_int): Only test tem if it has been
11208 2017-09-12 Richard Biener <rguenther@suse.de>
11210 PR middle-end/82149
11211 * match.pd ((FTYPE) N CMP CST): Fix typo.
11213 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11215 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11217 (mips_near_type_p): Add 'short_call' attribute as a synonym
11219 * doc/extend.texi (short_call): Document new function attribute.
11221 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11224 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11225 assertion check that in the condition.
11226 (get_atomic_generic_size): Likewise. Before testing if parameter
11227 has pointer type, if it has array type, call for C++
11228 default_conversion to perform array-to-pointer conversion.
11230 2017-09-12 Richard Biener <rguenther@suse.de>
11232 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11233 for operations we cannot scalarize.
11235 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11237 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11238 vectors heap vectors. Clean up comments.
11239 Make visited_bbs a reference.
11240 (profitable_jump_thread_path): Make GC
11241 vectors heap vectors. Clean up comments.
11243 (convert_and_register_jump_thread_path): Make GC vectors heap
11245 (check_subpath_and_update_thread_path): Same. Clean up comments.
11246 Make visited_bbs a reference.
11247 (handle_phi): Abstract common code to to
11248 register_jump_thread_path_if_profitable.
11249 Rename VAR_BB to DEF_BB.
11251 Make GC vectors heap vectors.
11252 Make visited_bbs a reference.
11253 (handle_assignment): Same.
11254 (register_jump_thread_path_if_profitable): New.
11255 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11257 Make GC vectors heap vectors. Clean up comments.
11258 Make visited_bbs a reference.
11259 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11260 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11263 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11266 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11267 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11269 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11271 Revert r251800 and r251799.
11273 2017-09-11 Martin Jambor <mjambor@suse.cz>
11276 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11277 arguments in advance.
11278 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11279 use it to find predecessor edges.
11280 (naive_outof_ssa): Collect vector of predecessors.
11282 2017-09-08 Jason Merrill <jason@redhat.com>
11284 PR c++/70029 - ICE with ref-qualifier and -flto
11285 * langhooks.h (struct lang_hooks_for_types): Add
11286 copy_lang_qualifiers.
11287 * attribs.c (build_type_attribute_qual_variant): Use it.
11288 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11290 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11291 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11293 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11296 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11297 (*mulsi3_sp64): New instruction.
11298 (mulsi3): New expander.
11300 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11302 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11304 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11306 * sancov.c: Include memmodel.h.
11308 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11311 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11314 2017-09-07 Carl Love <cel@us.ibm.com>
11316 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11317 the sldi instruction.
11319 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11321 * sancov.c: Include tm_p.h.
11323 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11326 * output.h (switch_to_other_text_partition): New declaration.
11327 * varasm.c (switch_to_other_text_partition): New function.
11328 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11329 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11330 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11331 to the other text partition before emitting LCL label and switch back
11332 after emitting the word after it.
11334 2017-09-07 Richard Biener <rguenther@suse.de>
11336 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11337 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11338 critical edges here, after loop init.
11339 (pass_data_pre): Remove PROP_no_crit_edges flags.
11340 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11341 for valueization of call args to avoid leaking VN_TOP.
11342 (visit_use): Assert we do not visit default defs.
11343 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11344 Use error_mark_node to more easily detect leaking VN_TOP.
11345 All default-defs are varying, not VN_TOP. Mark them visited.
11346 (run_scc_vn): Make code match comment.
11348 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11350 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11351 OPTION_MASK_FLOAT128_KEYWORD.
11352 (POWERPC_MASKS): Likewise.
11353 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11354 support for the -mfloat128-type option, and make -mfloat128
11355 default on PowerPC Linux systems. Define or undefine
11356 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11357 Define __float128 to be __ieee128 if IEEE 128-bit support is
11358 enabled, or undefine it.
11359 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11360 Delete defining __FLOAT128_TYPE__.
11361 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11362 -mfloat128-type option and make -mfloat128 default on PowerPC
11364 (TARGET_FLOAT128_TYPE): Likewise.
11365 (-mfloat128-type): Likewise.
11366 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11367 Delete the -mfloat128-type option and make -mfloat128 default on
11368 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11369 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11370 128-bit floating point is enabled. Change tests from using
11371 -mfloat128-type to -mfloat128.
11372 (rs6000_mangle_type): Use the correct mangling for the __float128
11373 type even if normal long double is restricted to 64-bits.
11374 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11376 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11377 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11378 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11380 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11381 documentation for -mfloat128.
11383 2017-09-06 Olivier Hainque <hainque@adacore.com>
11385 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11387 2017-09-06 Wish Wu <wishwu007@gmail.com>
11388 Jakub Jelinek <jakub@redhat.com>
11390 * asan.c (initialize_sanitizer_builtins): Add
11391 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11392 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11393 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11394 BT_FN_VOID_UINT64_PTR variables.
11395 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11396 (BT_FN_VOID_UINT16_UINT16): Likewise.
11397 (BT_FN_VOID_UINT32_UINT32): Likewise.
11398 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11399 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11400 (BT_FN_VOID_UINT64_PTR): Likewise.
11401 * common.opt (flag_sanitize_coverage): New variable.
11402 (fsanitize-coverage=trace-pc): Remove.
11403 (fsanitize-coverage=): Add.
11404 * flag-types.h (enum sanitize_coverage_code): New enum.
11405 * fold-const.c (fold_range_test): Disable non-short-circuit
11406 optimization if flag_sanitize_coverage.
11407 (fold_truth_andor): Likewise.
11408 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11409 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11410 (coverage_sanitizer_opts): New array.
11411 (get_closest_sanitizer_option): Add OPTS argument, handle also
11412 OPT_fsanitize_coverage_.
11413 (parse_sanitizer_options): Adjusted to also handle
11414 OPT_fsanitize_coverage_.
11415 (common_handle_option): Add OPT_fsanitize_coverage_.
11416 * sancov.c (instrument_comparison, instrument_switch): New function.
11417 (sancov_pass): Add trace-cmp support.
11418 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11419 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11420 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11421 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11422 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11423 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11424 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11425 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11426 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11427 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11429 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11431 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11432 error. Only quit immediately if parsing is complete.
11433 (BEGIN): Initialize fatal_err and parse_done.
11434 (begin fpu, end fpu): Check number of arguments.
11435 (begin arch, end arch): Likewise.
11436 (begin cpu, end cpu): Likewise.
11437 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11438 (optalias): Likewise.
11440 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11442 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11443 * config/arm/arm-isa.h: Delete. Move definitions to ...
11444 * arm-cpus.in: ... here. Use new feature and fgroup values.
11445 * config/arm/arm.c (arm_option_override): Use lower case for feature
11447 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11448 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11449 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11451 (print_isa_bits_for): New function.
11452 (gen_isa): New function.
11453 (gen_comm_data): Use print_isa_bits_for.
11454 (define feature): New keyword.
11455 (define fgroup): New keyword.
11456 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11457 (arm-isa.h): Add rule to generate file.
11458 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11459 case for feature bit names.
11461 2017-09-06 Richard Biener <rguenther@suse.de>
11463 * tree-ssa-pre.c (NECESSARY): Remove.
11464 (create_expression_by_pieces): Do not touch pass-local flags.
11465 (insert_into_preds_of_block): Likewise.
11466 (do_pre_regular_insertion): Likewise.
11467 (eliminate_insert): Likewise.
11468 (eliminate_dom_walker::before_dom_children): Likewise.
11469 (fini_eliminate): Do not look at inserted_exprs.
11470 (mark_operand_necessary): Remove.
11471 (remove_dead_inserted_code): Replace with simple work-list
11472 algorithm based on inserted_exprs and SSA uses.
11473 (pass_pre::execute): Re-order fini_eliminate and
11474 remove_dead_inserted_code.
11476 2017-09-06 Olivier Hainque <hainque@adacore.com>
11478 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11479 for VxWorks 7. Adjust surrounding comments.
11481 2017-09-06 Richard Biener <rguenther@suse.de>
11483 * gimple-ssa-strength-reduction.c
11484 (find_candidates_dom_walker::before_dom_children): Also allow
11487 2017-09-06 Richard Biener <rguenther@suse.de>
11489 PR tree-optimization/82108
11490 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11491 for gap in the non-permutation SLP case.
11493 2017-09-06 Martin Jambor <mjambor@suse.cz>
11495 PR tree-optimization/82078
11496 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11497 add_access_to_work_queue...
11498 (build_accesses_from_assign): ...here.
11499 (propagate_all_subaccesses): Make sure racc is the group
11500 representative, if there is one.
11502 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11504 PR middle-end/82095
11505 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11508 2017-09-06 Richard Biener <rguenther@suse.de>
11510 * gimple-ssa-strength-reduction.c
11511 (find_candidates_dom_walker::before_doom_children): Use a
11512 type and not a mode check.
11514 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11517 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11519 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11521 2017-09-05 Jeff Law <law@redhat.com>
11523 PR tree-optimization/64910
11524 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11525 cases where we have 3 or more operands.
11527 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11529 PR middle-end/81768
11530 * omp-low.c (lower_omp_for): Recompute tree invariant if
11531 gimple_omp_for_initial/final is ADDR_EXPR.
11533 PR middle-end/81768
11534 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11535 into gimple val before gimplification fo the COND_EXPR.
11537 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11539 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11540 REGION_COPY argument.
11541 (thread_through_all_blocks): Remove unused argument to
11542 duplicate_thread_path.
11544 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11545 Alan Hayward <alan.hayward@arm.com>
11546 David Sherwood <david.sherwood@arm.com>
11548 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11549 Take a scalar_mode rather than a machine_mode.
11550 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11551 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11552 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11553 (aarch64_gen_adjusted_ldpstp): Likewise.
11554 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11556 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11557 Alan Hayward <alan.hayward@arm.com>
11558 David Sherwood <david.sherwood@arm.com>
11560 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11561 Take a scalar_int_mode instead of a machine_mode.
11562 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11563 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11564 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11565 (aarch64_simd_attr_length_rglist): Delete.
11566 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11567 a scalar_int_mode instead of a machine_mode.
11568 (aarch64_add_offset): Likewise.
11569 (aarch64_internal_mov_immediate): Likewise
11570 (aarch64_add_constant_internal): Likewise.
11571 (aarch64_add_constant): Likewise.
11572 (aarch64_movw_imm): Likewise.
11573 (aarch64_rtx_arith_op_extract_p): Likewise.
11574 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11575 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11576 Remove assert that the mode isn't a vector.
11577 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11578 (aarch64_expand_mov_immediate): Update calls after above changes.
11579 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11580 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11581 (aarch64_move_imm): Likewise.
11582 (aarch64_can_const_movi_rtx_p): Likewise.
11583 (aarch64_strip_extend): Likewise.
11584 (aarch64_extr_rtx_p): Likewise.
11585 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11586 a CONST_INT when the mode parameter is VOIDmode.
11587 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11589 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11591 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11592 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11593 (bitwise_type_for_mode): Update accordingly.
11595 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11597 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11598 * stor-layout.c (mode_for_size_tree): Likewise.
11599 (mode_for_array): Update accordingly.
11600 (layout_decl): Likewise.
11601 (compute_record_mode): Likewise. Only set the mode once.
11603 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11605 * target.def (get_mask_mode): Change return type to opt_mode.
11607 * doc/tm.texi: Regenerate.
11608 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11609 * targhooks.c (default_get_mask_mode): Likewise.
11610 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11611 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11612 targetm.get_mask_mode.
11613 * tree.c (build_truth_vector_type): Likewise.
11615 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11617 * machmode.h (mode_for_vector): Return an opt_mode.
11618 * stor-layout.c (mode_for_vector): Likewise.
11619 (mode_for_int_vector): Update accordingly.
11620 (layout_type): Likewise.
11621 * config/i386/i386.c (emit_memmov): Likewise.
11622 (ix86_expand_set_or_movmem): Likewise.
11623 (ix86_expand_vector_init): Likewise.
11624 (ix86_get_mask_mode): Likewise.
11625 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11627 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11628 * expmed.c (extract_bit_field_1): Likewise.
11629 * expr.c (expand_expr_real_2): Likewise.
11630 * optabs-query.c (can_vec_perm_p): Likewise.
11631 (can_vec_mask_load_store_p): Likewise.
11632 * optabs.c (expand_vec_perm): Likewise.
11633 * targhooks.c (default_get_mask_mode): Likewise.
11634 * tree-vect-stmts.c (vectorizable_store): Likewise.
11635 (vectorizable_load): Likewise.
11636 (get_vectype_for_scalar_type_and_size): Likewise.
11638 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11640 * machmode.h (mode_for_int_vector): New function.
11641 * stor-layout.c (mode_for_int_vector): Likewise.
11642 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11643 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11644 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11645 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11646 (s390_expand_vcond): Likewise.
11648 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11650 * machmode.h (opt_machine_mode): New type.
11651 (opt_mode<T>): Allow construction from anything that can be
11653 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11654 (mode_for_size): Return an opt_machine_mode.
11655 * stor-layout.c (mode_for_size): Likewise.
11656 (mode_for_size_tree): Update call accordingly.
11657 (bitwise_mode_for_mode): Likewise.
11658 (make_fract_type): Likewise.
11659 (make_accum_type): Likewise.
11660 * caller-save.c (replace_reg_with_saved_mem): Update call
11662 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11663 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11664 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11665 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11666 * expmed.c (extract_bit_field_1): Likewise.
11667 * reload.c (get_secondary_mem): Likewise.
11668 * varasm.c (assemble_integer): Likewise.
11669 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11672 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11674 * machmode.h (decimal_float_mode_for_size): New function.
11675 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11676 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11677 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11678 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11679 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11680 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11682 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11684 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11685 (get_builtin_sync_mode): Likewise.
11686 (expand_ifn_atomic_compare_exchange): Likewise.
11687 (expand_builtin_atomic_clear): Likewise.
11688 (expand_builtin_atomic_test_and_set): Likewise.
11689 (fold_builtin_atomic_always_lock_free): Likewise.
11690 * calls.c (compute_argument_addresses): Likewise.
11691 (emit_library_call_value_1): Likewise.
11692 (store_one_arg): Likewise.
11693 * combine.c (combine_instructions): Likewise.
11694 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11695 * config/arm/arm.c (arm_function_value): Likewise.
11696 (aapcs_allocate_return_reg): Likewise.
11697 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11698 * config/i386/i386.c (construct_container): Likewise.
11699 (ix86_gimplify_va_arg): Likewise.
11700 (ix86_expand_sse_cmp): Likewise.
11701 (emit_memmov): Likewise.
11702 (emit_memset): Likewise.
11703 (expand_small_movmem_or_setmem): Likewise.
11704 (ix86_expand_pextr): Likewise.
11705 (ix86_expand_pinsr): Likewise.
11706 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11707 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11709 * config/mips/mips.c (mips_function_value_1) Likewise.
11710 (mips_block_move_straight): Likewise.
11711 (mips_expand_ins_as_unaligned_store): Likewise.
11712 * config/powerpcspe/powerpcspe.c
11713 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11714 (rs6000_darwin64_record_arg_flush): Likewise.
11715 * config/rs6000/rs6000.c
11716 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11717 (rs6000_darwin64_record_arg_flush): Likewise.
11718 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11719 (sparc_function_value_1): Likewise.
11720 * config/spu/spu.c (adjust_operand): Likewise.
11721 (spu_emit_branch_or_set): Likewise.
11722 (arith_immediate_p): Likewise.
11723 * emit-rtl.c (gen_lowpart_common): Likewise.
11724 * expr.c (expand_expr_real_1): Likewise.
11725 * function.c (assign_parm_setup_block): Likewise.
11726 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11727 * reload1.c (alter_reg): Likewise.
11728 * stor-layout.c (mode_for_vector): Likewise.
11729 (layout_type): Likewise.
11731 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11733 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11734 (spu_convert_move): Likewise.
11735 * lower-subreg.c (resolve_simple_move): Likewise.
11737 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11740 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11741 define_insn to a define_expand.
11742 (altivec_vsum2sws_direct): New define_insn.
11743 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11745 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11747 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11750 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11755 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11756 realign stack if stack alignment needed is less than incoming
11759 2017-09-05 Marek Polacek <polacek@redhat.com>
11762 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11765 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11767 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11769 2017-09-05 Richard Biener <rguenther@suse.de>
11771 PR tree-optimization/82084
11772 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11774 2017-09-05 Richard Biener <rguenther@suse.de>
11776 PR tree-optimization/82102
11777 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11779 2017-09-05 Martin Liska <mliska@suse.cz>
11781 PR tree-optimization/82032
11782 * tree-cfg.c (generate_range_test): New function.
11783 * tree-cfg.h (generate_range_test): Declared here.
11784 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11785 (cleanup_control_expr_graph): Use it.
11786 * tree-switch-conversion.c (try_switch_expansion): Remove
11788 (emit_case_nodes): Use generate_range_test.
11790 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11793 * config/i386/i386.md (*<btsc><mode>_mask): Add
11794 TARGET_USE_BT to insn constraint.
11795 (*btr<mode>_mask): Ditto.
11797 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11799 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11800 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11802 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11805 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11806 TARGET_NEON and TARGET_IWMMXT.
11807 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11808 TARGET_NEON and TARGET_IWMMXT.
11809 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11811 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11813 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11814 (ix86_rewrite_tls_address): Ditto.
11815 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11816 (ix86_rewrite_tls_address_1): Ditto.
11817 (ix86_rewrite_tls_address): Ditto.
11818 * config/i386/predicates.md (tls_address_pattern): New predicate.
11819 * config/i386/i386.md (TLS address splitter): New splitter.
11821 2017-09-04 Richard Biener <rguenther@suse.de>
11823 PR tree-optimization/82084
11824 * fold-const.h (can_native_encode_string_p): Declare.
11825 * fold-const.c (can_native_encode_string_p): Factor out from ...
11826 (native_encode_string): ... here.
11827 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11828 vectorizing stores from constants we later cannot handle.
11830 2017-09-04 Marek Polacek <polacek@redhat.com>
11833 * doc/invoke.texi: Update -Wtautological-compare documentation.
11835 2017-09-04 Jeff Law <law@redhat.com>
11837 PR tree-optimization/64910
11838 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11839 swap the first and last operand if the last is a constant.
11841 2017-09-04 Marek Polacek <polacek@redhat.com>
11844 * convert.c (do_narrow): When sanitizing signed integer overflows,
11845 bail out for signed types.
11846 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11848 2017-09-04 Richard Biener <rguenther@suse.de>
11850 PR tree-optimization/82060
11851 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11852 Move devirtualization after stmt folding and before EH/AB/noreturn
11853 cleanup to get the stmt refs canonicalized. Use a bool instead
11854 of gimple_modified_p since that doesn't work for NOPs. Schedule
11855 NOPs generated by folding for removal.
11857 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11858 Alan Hayward <alan.hayward@arm.com>
11859 David Sherwood <david.sherwood@arm.com>
11861 * coretypes.h (pad_direction): New enum.
11862 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11863 (FUNCTION_ARG_PADDING): Likewise.
11864 * target.def (function_arg_padding): New hook.
11865 * targhooks.h (default_function_arg_padding): Declare.
11866 * targhooks.c (default_function_arg_padding): New function.
11867 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11868 (TARGET_FUNCTION_ARG_PADDING): ...this.
11869 * doc/tm.texi: Regenerate.
11870 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11871 instead of direction.
11872 (compute_argument_addresses): Likewise.
11873 (load_register_parameters): Likewise.
11874 (emit_library_call_value_1): Likewise.
11875 (store_one_arg): Use targetm.calls.function_arg_padding instead
11876 of FUNCTION_ARG_PADDING.
11877 (must_pass_in_stack_var_size_or_pad): Likewise.
11878 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11879 (emit_group_store): Likewise.
11880 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11881 instead of FUNCTION_ARG_PADDING.
11882 (emit_push_insn): Likewise, and propagate enum change throughout
11884 * function.h (direction): Delete.
11885 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11887 * function.c (assign_parm_find_stack_rtl): Likewise.
11888 (assign_parm_setup_block_p): Likewise.
11889 (assign_parm_setup_block): Likewise.
11890 (gimplify_parameters): Likewise.
11891 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11892 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11894 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11895 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11896 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11897 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11898 (aarch64_function_arg_padding): ...this new function.
11899 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11900 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11901 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11902 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11903 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11904 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11905 (arm_pad_arg_upward): Replace with...
11906 (arm_function_arg_padding): ...this new function.
11907 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11909 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11910 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11911 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11912 (ia64_hpux_function_arg_padding): Replace with...
11913 (ia64_function_arg_padding): ...this new function. Use pad_direction
11914 instead of direction. Check for TARGET_HPUX.
11915 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11916 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11917 (iq2000_function_arg_padding): New function.
11918 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11919 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11920 (mips_function_arg_padding): ...this new function.
11921 (mips_pad_reg_upward): Update accordingly.
11922 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11923 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11924 targetm.calls.function_arg_padding.
11925 (FUNCTION_ARG_PADDING): Delete.
11926 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11927 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11928 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11929 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11930 (nios2_block_reg_padding): Return pad_direction instead of direction.
11931 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11932 instead of direction.
11933 (nios2_function_arg_padding): Likewise. Make static.
11934 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11935 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11936 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11937 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11938 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11939 (pa_function_arg_padding): Make static. Return pad_direction instead
11941 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11942 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11943 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11944 instead of direction. Use targetm.calls.function_arg_padding.
11945 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11946 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11947 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11948 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11949 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11951 (function_arg_padding): Rename to...
11952 (rs6000_function_arg_padding): ...this. Make static. Return
11953 pad_direction instead of direction.
11954 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11955 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11956 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11957 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11958 instead of direction. Use targetm.calls.function_arg_padding.
11959 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11960 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11961 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11962 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11963 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11964 (function_arg_padding): Rename to...
11965 (rs6000_function_arg_padding): ...this. Make static. Return
11966 pad_direction instead of direction.
11967 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11968 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11969 * config/s390/s390.c (s390_function_arg_padding): New function.
11970 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11971 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11972 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11973 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11974 (function_arg_padding): Rename to...
11975 (sparc_function_arg_padding): ...this. Make static. Return
11976 pad_direction instead of direction.
11977 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11978 * config/spu/spu.c (spu_function_arg_padding): New function.
11979 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11980 * system.h (FUNCTION_ARG_PADDING): Poison.
11982 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11983 Alan Hayward <alan.hayward@arm.com>
11984 David Sherwood <david.sherwood@arm.com>
11986 * target.def (modes_tieable_p): New hook.
11987 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11988 (TARGET_MODES_TIEABLE_P): ...this.
11989 * doc/tm.texi.in: Regenerate.
11990 * hooks.h (hook_bool_mode_mode_true): Declare.
11991 * hooks.c (hook_bool_mode_mode_true): New function.
11992 * combine.c (subst): Use targetm.modes_tieable_p instead of
11994 * dse.c (find_shift_sequence): Likewise.
11995 * expmed.c (extract_low_bits): Likewise.
11996 * lower-subreg.c: Include target.h.
11997 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11999 * rtlanal.c (rtx_cost): Likewise.
12000 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12001 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12002 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12003 (TARGET_MODES_TIEABLE_P): Redefine.
12004 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12005 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12006 (TARGET_MODES_TIEABLE_P): Redefine.
12007 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12008 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12009 (arc_modes_tieable_p): New function.
12010 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12011 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12012 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12013 (arm_modes_tieable_p): Make static.
12014 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12015 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12016 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12017 (TARGET_MODES_TIEABLE_P): Redefine.
12018 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12019 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12020 (TARGET_MODES_TIEABLE_P): Redefine.
12021 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12022 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12023 (cr16_modes_tieable_p): New function.
12024 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12025 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12026 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12027 (TRULY_NOOP_TRUNCATION): Update comment.
12028 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12029 (TRULY_NOOP_TRUNCATION): Update comment.
12030 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12031 (frv_modes_tieable_p): New function.
12032 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12033 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12034 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12035 (TARGET_MODES_TIEABLE_P): Redefine.
12036 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12037 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12038 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12039 (TARGET_MODES_TIEABLE_P): Redefine.
12040 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12041 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12042 (ia64_modes_tieable_p): New function.
12043 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12044 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12045 (iq2000_modes_tieable_p): New function.
12046 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12047 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12048 (lm32_modes_tieable_p): New function.
12049 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12050 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12051 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12052 (TARGET_MODES_TIEABLE_P): Redefine.
12053 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12054 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12055 (m32r_modes_tieable_p): New function.
12056 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12057 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12058 (m68k_modes_tieable_p): New function.
12059 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12060 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12061 (mcore_modes_tieable_p): New function.
12062 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12063 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12065 (TARGET_MODES_TIEABLE_P): Redefine.
12066 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12067 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12068 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12069 (TARGET_MODES_TIEABLE_P): Redefine.
12070 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12071 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12072 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12073 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12074 (mn10300_modes_tieable_p): ...this and make static.
12075 (TARGET_MODES_TIEABLE_P): Redefine.
12076 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12077 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12078 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12079 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12080 (msp430_modes_tieable_p): Make static.
12081 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12082 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12083 (TARGET_MODES_TIEABLE_P): Redefine.
12084 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12085 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12086 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12087 (TARGET_MODES_TIEABLE_P): Redefine.
12088 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12089 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12090 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12091 (TARGET_MODES_TIEABLE_P): Redefine.
12092 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12093 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12094 (pdp11_modes_tieable_p): New function.
12095 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12096 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12097 (rs6000_modes_tieable_p): New function.
12098 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12099 * config/powerpcspe/powerpcspe.md: Update comment.
12100 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12101 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12102 (TARGET_MODES_TIEABLE_P): Redefine.
12103 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12104 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12105 (rl78_modes_tieable_p): New function.
12106 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12107 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12108 (rs6000_modes_tieable_p): New function.
12109 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12110 * config/rs6000/rs6000.md: Update comment.
12111 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12112 * config/rx/rx.c (rx_modes_tieable_p): New function.
12113 (TARGET_MODES_TIEABLE_P): Redefine.
12114 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12115 * config/s390/s390.c (s390_modes_tieable_p): New function.
12116 (TARGET_MODES_TIEABLE_P): Redefine.
12117 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12118 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12119 (sh_modes_tieable_p): New function.
12120 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12121 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12122 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12123 (sparc_modes_tieable_p): Make static.
12124 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12125 * config/spu/spu.c (spu_modes_tieable_p): New function.
12126 (TARGET_MODES_TIEABLE_P): Redefine.
12127 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12128 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12129 (TARGET_MODES_TIEABLE_P): Redefine.
12130 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12131 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12132 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12133 * config/v850/v850.c (v850_modes_tieable_p): New function.
12134 (TARGET_MODES_TIEABLE_P): Redefine.
12135 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12136 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12137 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12138 (visium_modes_tieable_p): New function.
12139 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12140 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12141 (xtensa_modes_tieable_p): New function.
12142 * system.h (MODES_TIEABLE_P): Poison.
12144 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12145 Alan Hayward <alan.hayward@arm.com>
12146 David Sherwood <david.sherwood@arm.com>
12148 * target.def (hard_regno_mode_ok): New hook.
12149 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12150 (TARGET_HARD_REGNO_MODE_OK): ...this.
12151 * doc/tm.texi.in: Regenerate.
12152 * hooks.h (hook_bool_uint_mode_true): Declare.
12153 * hooks.c (hook_bool_uint_mode_true): New function.
12154 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12155 HARD_REGNO_MODE_OK.
12156 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12157 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12158 instead of HARD_REGNO_MODE_OK.
12159 * caller-save.c: Include target.h.
12160 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12161 HARD_REGNO_MODE_OK.
12162 * combine.c (can_combine_p): Likewise.
12163 (combinable_i3pat): Likewise.
12164 (can_change_dest_mode): Likewise.
12165 * expr.c (init_expr_target): Likewise.
12166 (convert_move): Likewise.
12167 (convert_modes): Likewise.
12168 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12169 (setup_prohibited_mode_move_regs): Likewise.
12170 * ira.h (target_ira): Likewise.
12171 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12172 * lra-constraints.c (process_alt_operands): Likewise.
12173 (split_reg): Likewise.
12174 * recog.c (peep2_find_free_register): Likewise.
12175 * ree.c (combine_reaching_defs): Likewise.
12176 * regcprop.c (maybe_mode_change): Likewise.
12177 * reginfo.c (init_reg_sets_1): Likewise.
12178 (choose_hard_reg_mode): Likewise.
12179 (simplifiable_subregs): Likewise.
12180 * regrename.c (check_new_reg_p): Likewise.
12181 * reload.c (find_valid_class): Likewise.
12182 (find_valid_class_1): Likewise.
12183 (reload_inner_reg_of_subreg): Likewise.
12184 (push_reload): Likewise.
12185 (combine_reloads): Likewise.
12186 (find_dummy_reload): Likewise.
12187 (find_reloads): Likewise.
12188 * reload1.c (find_reg): Likewise.
12189 (set_reload_reg): Likewise.
12190 (allocate_reload_reg): Likewise.
12191 (choose_reload_regs): Likewise.
12192 (reload_adjust_reg_for_temp): Likewise.
12193 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12194 (simplify_subreg_regno): Likewise.
12195 * sel-sched.c (init_regs_for_mode): Likewise.
12196 * varasm.c (make_decl_rtl): Likewise.
12197 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12198 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12199 HARD_REGNO_MODE_OK.
12200 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12201 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12203 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12204 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12205 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12206 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12207 (arc_mode_class): Delete.
12208 (HARD_REGNO_MODE_OK): Delete.
12209 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12210 (arc_hard_regno_mode_ok): Rename old array to...
12211 (arc_hard_regno_mode_ok_modes): ...this.
12212 (arc_conditional_register_usage): Update accordingly.
12213 (arc_mode_class): Make static.
12214 (arc_hard_regno_mode_ok): New function.
12215 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12216 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12217 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12218 (arm_hard_regno_mode_ok): Make static.
12219 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12220 HARD_REGNO_MODE_OK.
12221 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12222 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12223 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12225 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12226 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12227 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12228 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12229 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12230 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12231 * config/bfin/predicates.md (valid_reg_operand): Use
12232 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12233 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12234 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12235 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12236 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12237 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12238 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12239 (cr16_hard_regno_mode_ok): Make static and return a bool.
12240 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12241 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12242 (cris_hard_regno_mode_ok): New function.
12243 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12244 (epiphany_mode_class): Delete.
12245 (HARD_REGNO_MODE_OK): Delete.
12246 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12247 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12248 (hard_regno_mode_ok): Rename to...
12249 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12250 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12251 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12252 HARD_REGNO_MODE_OK.
12253 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12254 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12255 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12256 (frv_hard_regno_mode_ok): Make static and return a bool.
12257 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12258 HARD_REGNO_MODE_OK.
12259 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12260 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12261 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12262 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12265 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12266 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12267 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12269 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12270 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12271 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12272 (ia64_hard_regno_mode_ok): New function.
12273 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12274 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12275 (iq2000_hard_regno_mode_ok): New function.
12276 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12277 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12278 (lm32_hard_regno_mode_ok): New function.
12279 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12280 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12281 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12282 instead of HARD_REGNO_MODE_OK.
12283 (m32c_hard_regno_ok): Rename to...
12284 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12285 (m32c_cannot_change_mode_class): Update accordingly.
12286 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12287 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12288 (m32r_mode_class): Delete.
12289 (HARD_REGNO_MODE_OK): Delete.
12290 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12291 (m32r_hard_regno_mode_ok): Rename to...
12292 (m32r_hard_regno_modes): ...this.
12293 (m32r_mode_class): Make static.
12294 (m32r_hard_regno_mode_ok): New function.
12295 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12296 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12297 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12298 (m68k_hard_regno_mode_ok): Make static.
12299 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12300 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12301 (mcore_hard_regno_mode_ok): New function.
12302 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12303 (HARD_REGNO_MODE_OK): Delete.
12304 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12306 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12307 (microblaze_hard_regno_mode_ok): New function.
12308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12309 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12310 (mips_hard_regno_mode_ok): Delete.
12311 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12312 (mips_hard_regno_mode_ok_p): ...this and make static.
12313 (mips_hard_regno_mode_ok_p): Rename to...
12314 (mips_hard_regno_mode_ok_uncached): ...this.
12315 (mips_hard_regno_mode_ok): New function.
12316 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12317 of HARD_REGNO_MODE_OK.
12318 (mips_option_override): Update after above name changes.
12319 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12320 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12321 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12322 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12323 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12325 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12326 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12327 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12328 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12329 (msp430_hard_regno_mode_ok): Make static and return a bool.
12330 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12331 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12332 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12334 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12335 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12336 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12337 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12338 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12339 (PA_HARD_REGNO_MODE_OK): ...this
12340 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12341 (PA_HARD_REGNO_MODE_OK): ...this.
12342 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12343 (pa_hard_regno_mode_ok): New function.
12344 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12345 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12346 (pdp11_hard_regno_mode_ok): New function.
12347 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12348 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12350 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12352 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12353 (rs6000_hard_regno_mode_ok): Rename to...
12354 (rs6000_hard_regno_mode_ok_uncached): ...this.
12355 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12356 (rs6000_hard_regno_mode_ok): New function.
12357 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12358 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12359 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12360 (riscv_hard_regno_mode_ok): ...this and make static.
12361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12362 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12363 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12364 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12365 (rl78_hard_regno_mode_ok): Make static and return bool.
12366 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12367 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12369 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12370 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12371 (rs6000_hard_regno_mode_ok): Rename to...
12372 (rs6000_hard_regno_mode_ok_uncached): ...this.
12373 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12374 (rs6000_hard_regno_mode_ok): New function.
12375 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12376 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12377 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12378 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12379 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12380 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12381 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12382 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12383 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12384 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12385 (sh_hard_regno_mode_ok): Make static.
12386 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12387 instead of HARD_REGNO_MODE_OK.
12388 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12389 (sparc_mode_class): Delete.
12390 (HARD_REGNO_MODE_OK): Delete.
12391 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12392 (hard_regno_mode_classes): Make static.
12393 (sparc_mode_class): Likewise.
12394 (sparc_hard_regno_mode_ok): New function.
12395 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12396 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12397 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12399 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12400 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12401 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12402 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12403 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12405 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12406 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12407 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12408 (visium_hard_regno_mode_ok): New function.
12409 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12410 instead of HARD_REGNO_MODE_OK.
12411 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12412 (HARD_REGNO_MODE_OK): Delete.
12413 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12414 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12415 (xtensa_option_override): Update accordingly.
12416 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12417 (xtensa_hard_regno_mode_ok): New function.
12418 * system.h (HARD_REGNO_MODE_OK): Poison.
12420 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12421 Alan Hayward <alan.hayward@arm.com>
12422 David Sherwood <david.sherwood@arm.com>
12424 * target.def (hard_regno_call_part_clobbered): New hook.
12425 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12426 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12427 * doc/tm.texi: Regenerate.
12428 * hooks.h (hook_bool_uint_mode_false): Declare.
12429 * hooks.c (hook_bool_uint_mode_false): New function.
12430 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12431 * cselib.c (cselib_process_insn): Use
12432 targetm.hard_regno_call_part_clobbered instead of
12433 HARD_REGNO_CALL_PART_CLOBBERED.
12434 * ira-conflicts.c (ira_build_conflicts): Likewise.
12435 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12436 * lra-constraints.c (need_for_call_save_p): Likewise.
12437 * lra-lives.c: Include target.h.
12438 (check_pseudos_live_through_calls): Use
12439 targetm.hard_regno_call_part_clobbered instead of
12440 HARD_REGNO_CALL_PART_CLOBBERED.
12441 * regcprop.c: Include target.h.
12442 (copyprop_hardreg_forward_1): Use
12443 targetm.hard_regno_call_part_clobbered instead of
12444 HARD_REGNO_CALL_PART_CLOBBERED.
12445 * reginfo.c (choose_hard_reg_mode): Likewise.
12446 * regrename.c (check_new_reg_p): Likewise.
12447 * reload.c (find_equiv_reg): Likewise.
12448 * reload1.c (emit_reload_insns): Likewise.
12449 * sched-deps.c (deps_analyze_insn): Likewise.
12450 * sel-sched.c (init_regs_for_mode): Likewise.
12451 (mark_unavailable_hard_regs): Likewise.
12452 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12453 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12454 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12456 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12457 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12458 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12460 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12462 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12463 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12464 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12466 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12467 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12468 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12470 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12471 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12473 * config/powerpcspe/powerpcspe.c
12474 (rs6000_hard_regno_call_part_clobbered): New function.
12475 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12476 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12477 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12479 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12480 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12481 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12483 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12484 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12485 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12487 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * rtl.h (subreg_memory_offset): Declare.
12492 * emit-rtl.c (subreg_memory_offset): New function.
12493 * expmed.c (store_bit_field_1): Use it.
12494 * expr.c (undefined_operand_subword_p): Likewise.
12495 * simplify-rtx.c (simplify_subreg): Likewise.
12497 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12499 PR rtl-optimization/57448
12502 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12503 using atomic_load pattern.
12504 (expand_atomic_store): Likewise.
12506 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12509 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12510 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12511 handling. Use replace_call_with_value with NULL instead of
12512 gsi_replace, unlink_stmt_vdef and release_defs.
12514 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12517 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12519 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12522 * rtl.h (emit_library_call_value_1): Declare.
12523 (emit_library_call): Replace declaration with a series of overloads.
12524 Remove the parameter count argument.
12525 (emit_library_call_value): Likewise.
12526 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12527 with an "rtx_mode_t *".
12528 (emit_library_call_value): Delete.
12529 (emit_library_call): Likewise.
12530 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12531 (asan_emit_allocas_unpoison): Likewise.
12532 * builtins.c (expand_builtin_powi): Likewise.
12533 (expand_asan_emit_allocas_unpoison): Likewise.
12534 * cfgexpand.c (expand_main_function): Likewise.
12535 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12536 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12537 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12538 * config/arm/arm.c (arm_trampoline_init): Likewise.
12539 (arm_call_tls_get_addr): Likewise.
12540 (arm_expand_divmod_libfunc): Likewise.
12541 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12542 (smulsi3_highpart): Likewise.
12543 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12544 (c6x_expand_compare): Likewise.
12545 (c6x_expand_movmem): Likewise.
12546 * config/frv/frv.c (frv_trampoline_init): Likewise.
12547 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12548 (ix86_expand_divmod_libfunc): Likewise.
12549 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12550 (ia64_expand_compare): Likewise.
12551 (ia64_profile_hook): Likewise.
12552 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12553 (nonlocal_goto): Likewise.
12554 (restore_stack_nonlocal): Likewise.
12555 * config/m32r/m32r.c (block_move_call): Likewise.
12556 (m32r_trampoline_init): Likewise.
12557 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12558 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12559 (m68k_call_m68k_read_tp): Likewise.
12560 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12561 (microblaze_expand_divide): Likewise.
12562 * config/mips/mips.h (mips_args): Likewise.
12563 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12564 (MIPS_ICACHE_SYNC): Likewise.
12565 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12566 (nios2_trampoline_init): Likewise.
12567 * config/pa/pa.c (hppa_tls_call): Likewise.
12568 (pa_trampoline_init): Likewise.
12569 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12570 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12571 (expand_strn_compare): Likewise.
12572 (rs6000_generate_compare): Likewise.
12573 (rs6000_expand_float128_convert): Likewise.
12574 (output_profile_hook): Likewise.
12575 (rs6000_trampoline_init): Likewise.
12576 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12577 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12578 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12579 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12580 (rs6000_generate_compare): Likewise.
12581 (rs6000_expand_float128_convert): Likewise.
12582 (output_profile_hook): Likewise.
12583 (rs6000_trampoline_init): Likewise.
12584 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12585 * config/sh/sh.c (sh_trampoline_init): Likewise.
12586 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12587 (sparc_emit_float_lib_cmp): Likewise.
12588 (sparc32_initialize_trampoline): Likewise.
12589 (sparc64_initialize_trampoline): Likewise.
12590 (sparc_profile_hook): Likewise.
12591 * config/spu/spu.c (ea_load_store): Likewise.
12592 * config/spu/spu.md (floatunssidf2): Likewise.
12593 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12594 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12595 * config/visium/visium.c (expand_block_move_4): Likewise.
12596 (expand_block_move_2): Likewise.
12597 (expand_block_move_1): Likewise.
12598 (expand_block_set_4): Likewise.
12599 (expand_block_set_2): Likewise.
12600 (expand_block_set_1): Likewise.
12601 (visium_trampoline_init): Likewise.
12602 (visium_profile_hook): Likewise.
12603 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12604 (xtensa_setup_frame_addresses): Likewise.
12605 (xtensa_trampoline_init): Likewise.
12606 * except.c (sjlj_emit_function_enter): Likewise.
12607 (sjlj_emit_function_exit): Likewise.
12608 * explow.c (allocate_dynamic_stack_space): Likewise.
12609 (probe_stack_range): Likewise.
12610 * expr.c (convert_mode_scalar): Likewise.
12611 * optabs.c (expand_binop): Likewise.
12612 (expand_twoval_binop_libfunc): Likewise.
12613 (expand_unop): Likewise.
12614 (prepare_cmp_insn): Likewise.
12615 (prepare_float_lib_cmp): Likewise.
12616 (expand_float): Likewise.
12617 (expand_fix): Likewise.
12618 (expand_fixed_convert): Likewise.
12619 (maybe_emit_sync_lock_test_and_set): Likewise.
12620 (expand_atomic_compare_and_swap): Likewise.
12621 (expand_mem_thread_fence): Likewise.
12622 (expand_atomic_fetch_op): Likewise.
12624 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12626 * doc/generic.texi (OpenACC): Adjust URL.
12627 * doc/invoke.texi (C Dialect Options): Ditto.
12629 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12631 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12632 predicate for operand 1. Add (m,<S>) constraint.
12633 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12634 Prevent memory operand 1 with register operand 2.
12636 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12638 PR rtl-optimization/82024
12639 * combine.c (try_combine): If the combination result is a PARALLEL,
12640 and we only need to retain the SET in there that would be placed
12641 at I2, check that we can place that at I3 instead, before doing so.
12643 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12646 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12648 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12649 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12652 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12653 Jeff Law <law@redhat.com>
12655 * varasm.c (bss_initializer_p): Do not put constants into .bss
12656 (categorize_decl_for_section): Handle bss_initializer_p returning
12657 false when DECL_INITIAL is NULL.
12659 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12662 * config/s390/s390.c (s390_can_inline_p): New function.
12664 2017-09-01 Jeff Law <law@redhat.com>
12666 PR tree-optimization/82052
12667 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12668 Always initialize the returned slot after a hash table miss
12669 when INSERT is true.
12671 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12673 * config/s390/s390.md (mem_signal_fence): Remove.
12674 * doc/md.texi (mem_signal_fence): Remove.
12675 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12677 * target-insns.def (mem_signal_fence): Remove.
12679 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12682 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12685 * asan.c (create_odr_indicator): Strip name encoding from assembler
12686 name before appending it after __odr_asan_.
12688 2017-09-01 Martin Liska <mliska@suse.cz>
12690 PR tree-optimization/82059
12691 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12692 frequency only when an edge is redirected.
12694 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12696 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12697 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12698 (arc_conditional_register_usage): Remove ARC600 lp_count
12700 (arc_file_start): Emit Tag_ARC_CPU_variation.
12701 (arc_can_use_doloop_p): New conditions to use ZOLs.
12702 (hwloop_fail): New function.
12703 (hwloop_optimize): Likewise.
12704 (hwloop_pattern_reg): Likewise.
12705 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12706 (arc_reorg_loops): New function, calls reorg_loops.
12707 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12708 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12710 (arc_loop_hazard): Remove function, functionality moved into
12712 (arc_hazard): Remove arc_loop_hazard call.
12713 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12714 into hwloop_optimize.
12715 (arc_label_align): Remove ZOL handling.
12716 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12717 * config/arc/arc.md (doloop_begin): Remove pattern.
12718 (doloop_begin_i): Likewise.
12719 (doloop_end_i): Likewise.
12720 (doloop_fallback): Likewise.
12721 (doloop_fallback_m): Likewise.
12722 (doloop_end): Reimplement expand.
12723 (arc_lp): New pattern for LP instruction.
12724 (loop_end): New pattern.
12725 (loop_fail): Likewise.
12726 (decrement_and_branch_until_zero): Likewise.
12727 * config/arc/arc.opt (mlpc-width): New option.
12728 * doc/invoke.texi (mlpc-width): Document option.
12730 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12732 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12733 (arc_ccfsm_advance): Fix checking for delay slots.
12734 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12736 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12738 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12740 (movhi_insn): Update store instruction constraint which are saving
12741 6-bit short immediates.
12742 (movsi_insn): Consider also short scaled load operations.
12743 (zero_extendhisi2_i): Use Usd constraint instead of T.
12744 (extendhisi2_i): Add q constraint.
12745 (arc_clzsi2): Add type and length attributes.
12746 (arc_ctzsi2): Likewise.
12747 * config/arc/constraints.md (Usc): Update constraint, the
12748 assembler can parse two relocations for a single instruction.
12750 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12752 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12753 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12755 2017-08-31 Olivier Hainque <hainque@adacore.com>
12757 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12758 match as powerpc-wrs-vxworks*.
12760 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12762 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12763 register constraint for by-element operand.
12764 (aarch64_mls_elt_merge<mode>): Likewise.
12766 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12768 * config/arc/arc.c (arc_can_follow_jump): Check for short
12771 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12773 * config.gcc: Use g.opt for arc.
12774 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12775 functionality moved to ...
12776 (legitimate_scaled_address_p): New function, ...here.
12777 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12778 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12779 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12781 (arc_override_options): Handle G option.
12782 (arc_output_pic_addr_const): Correct function definition.
12783 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12784 (arc_decl_anon_ns_mem_p): Delete.
12785 (arc_in_small_data_p): Overhaul this function to take into
12786 consideration the value given via G option.
12787 (arc_rewrite_small_data_1): Renamed and corrected old
12788 arc_rewrite_small_data function.
12789 (arc_rewrite_small_data): New function.
12790 (small_data_pattern): Don't use pic_offset_table_rtx.
12791 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12792 * config/arc/simdext.md (movmisalignv2hi): Use
12793 prepare_move_operands function.
12795 (movmisalign*): Likewise.
12796 * doc/invoke.texi (ARC options): Document -G option.
12798 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12800 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12802 * config/arc/arc.c (arc_print_operand): Output scalled address for
12803 sdata whenever is possible.
12804 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12805 load/stores are available.
12806 (compact_sda_memory_operand): Check for the alignment required by
12807 code density instructions.
12808 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12810 * config/arc/constraints.md (Usd): Update constraint.
12811 (Us0): New constraint.
12812 (Usc): Update constraint.
12814 2017-08-31 Richard Biener <rguenther@suse.de>
12816 PR middle-end/82054
12817 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12818 function only once.
12820 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12822 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12823 Resize type_signature.
12825 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12826 Alan Hayward <alan.hayward@arm.com>
12827 David Sherwood <david.sherwood@arm.com>
12829 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12830 subregs whose inner modes can be stored in GPRs.
12831 (aarch64_classify_index): Likewise.
12833 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12834 Alan Hayward <alan.hayward@arm.com>
12835 David Sherwood <david.sherwood@arm.com>
12837 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12838 (V_INT_EQUIV): ...this.
12839 (v_cmp_result): Rename to...
12840 (v_int_equiv): ...this.
12841 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12842 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12843 (copysign<mode>3): Likewise.
12844 (aarch64_simd_bsl<mode>_internal): Likewise.
12845 (aarch64_simd_bsl<mode>): Likewise.
12846 (vec_cmp<mode><mode>): Likewise.
12847 (vcond<mode><mode>): Likewise.
12848 (vcond<v_cmp_mixed><mode>): Likewise.
12849 (vcondu<mode><v_cmp_mixed>): Likewise.
12850 (aarch64_cm<optab><mode>): Likewise.
12851 (aarch64_cmtst<mode>): Likewise.
12852 (aarch64_fac<optab><mode>): Likewise.
12853 (vec_perm_const<mode>): Likewise.
12854 (vcond_mask_<mode><v_cmp_result>): Rename to...
12855 (vcond_mask_<mode><v_int_equiv>): ...this.
12856 (vec_cmp<mode><v_cmp_result>): Rename to...
12857 (vec_cmp<mode><v_int_equiv>): ...this.
12859 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12860 Alan Hayward <alan.hayward@arm.com>
12861 David Sherwood <david.sherwood@arm.com>
12863 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12865 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12866 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12867 (UNSPEC_LD4_DREG): New unspecs.
12868 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12869 (aarch64_ld2<mode>_dreg_be): Replace with...
12870 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12872 (aarch64_ld3<mode>_dreg_le)
12873 (aarch64_ld3<mode>_dreg_be): Replace with...
12874 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12876 (aarch64_ld4<mode>_dreg_le)
12877 (aarch64_ld4<mode>_dreg_be): Replace with...
12878 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12881 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12883 PR tree-optimization/81987
12884 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12885 insert an initializer in a location not dominated by the stride
12888 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12890 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12891 on the entire header of the finally block in the fallthru case.
12893 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12895 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12897 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12899 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12900 rs6000_emit_move_from_cr and call renamed function.
12901 (rs6000_emit_prologue): Call renamed functions.
12902 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12903 movesi_from_cr, remove volatile CRs.
12905 2017-08-30 Jon Beniston <jon@beniston.com>
12906 Richard Biener <rguenther@suse.de>
12908 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12909 of VECTOR_MODE_P check.
12910 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12911 element vector types.
12913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12915 * df.h (df_read_modify_subreg_p): Remove in favor of...
12916 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12917 const_rtx instead of an rtx.
12918 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12919 * df-problems.c (df_word_lr_mark_ref): Likewise.
12920 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12921 (mark_pseudo_reg_dead): Likewise.
12922 (mark_ref_dead): Likewise.
12923 * reginfo.c (init_subregs_of_mode): Likewise.
12924 * sched-deps.c (sched_analyze_1): Likewise.
12925 * df-scan.c (df_def_record_1): Likewise.
12926 (df_uses_record): Likewise.
12927 (df_read_modify_subreg_p): Remove in favor of...
12928 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12929 const_rtx instead of an rtx.
12931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12932 Alan Hayward <alan.hayward@arm.com>
12933 David Sherwood <david.sherwood@arm.com>
12935 * rtl.h (partial_subreg_p): New function.
12936 * caller-save.c (save_call_clobbered_regs): Use it.
12937 * calls.c (expand_call): Likewise.
12938 * combine.c (combinable_i3pat): Likewise.
12939 (simplify_set): Likewise.
12940 (make_extraction): Likewise.
12941 (make_compound_operation_int): Likewise.
12942 (gen_lowpart_or_truncate): Likewise.
12943 (force_to_mode): Likewise.
12944 (make_field_assignment): Likewise.
12945 (reg_truncated_to_mode): Likewise.
12946 (record_truncated_value): Likewise.
12947 (move_deaths): Likewise.
12948 * cse.c (record_jump_cond): Likewise.
12949 (cse_insn): Likewise.
12950 * cselib.c (cselib_lookup_1): Likewise.
12951 * expmed.c (extract_bit_field_using_extv): Likewise.
12952 * function.c (assign_parm_setup_reg): Likewise.
12953 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12954 * ira-build.c (create_insn_allocnos): Likewise.
12955 * lra-coalesce.c (merge_pseudos): Likewise.
12956 * lra-constraints.c (match_reload): Likewise.
12957 (simplify_operand_subreg): Likewise.
12958 (curr_insn_transform): Likewise.
12959 * lra-lives.c (process_bb_lives): Likewise.
12960 * lra.c (new_insn_reg): Likewise.
12961 (lra_substitute_pseudo): Likewise.
12962 * regcprop.c (mode_change_ok): Likewise.
12963 (maybe_mode_change): Likewise.
12964 (copyprop_hardreg_forward_1): Likewise.
12965 * reload.c (push_reload): Likewise.
12966 (find_reloads): Likewise.
12967 (find_reloads_subreg_address): Likewise.
12968 * reload1.c (alter_reg): Likewise.
12969 (eliminate_regs_1): Likewise.
12970 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12972 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12974 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12975 back to if statements, including unpack.
12977 2017-08-30 Martin Liska <mliska@suse.cz>
12979 PR inline-asm/82001
12980 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12982 (func_checker::compare_asm_inputs_outputs): ... this function.
12983 (func_checker::compare_gimple_asm): Use the function to compare
12984 also ASM constrains.
12985 * ipa-icf-gimple.h: Rename the function.
12987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12988 Alan Hayward <alan.hayward@arm.com>
12989 David Sherwood <david.sherwood@arm.com>
12991 * coretypes.h (complex_mode): New type.
12992 * gdbhooks.py (build_pretty_printer): Handle it.
12993 * machmode.h (complex_mode): New class.
12994 (complex_mode::includes_p): New function.
12995 (is_complex_int_mode): Likewise.
12996 (is_complex_float_mode): Likewise.
12997 * genmodes.c (get_mode_class): Handle complex mode classes.
12998 * function.c (expand_function_end): Use is_complex_int_mode.
13000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13001 Alan Hayward <alan.hayward@arm.com>
13002 David Sherwood <david.sherwood@arm.com>
13004 * coretypes.h (scalar_mode_pod): New typedef.
13005 * gdbhooks.py (build_pretty_printer): Handle it.
13006 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13007 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13008 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13009 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13010 as_a <scalar_mode>.
13012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13013 Alan Hayward <alan.hayward@arm.com>
13014 David Sherwood <david.sherwood@arm.com>
13016 * machmode.h (mode_for_vector): Take a scalar_mode instead
13018 * stor-layout.c (mode_for_vector): Likewise.
13019 * explow.c (promote_mode): Use as_a <scalar_mode>.
13020 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13023 Alan Hayward <alan.hayward@arm.com>
13024 David Sherwood <david.sherwood@arm.com>
13026 * target.def (preferred_simd_mode): Take a scalar_mode
13027 instead of a machine_mode.
13028 * targhooks.h (default_preferred_simd_mode): Likewise.
13029 * targhooks.c (default_preferred_simd_mode): Likewise.
13030 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13031 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13032 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13033 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13034 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13035 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13036 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13037 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13039 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13040 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13041 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13042 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13043 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13044 * doc/tm.texi: Regenerate.
13045 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13049 Alan Hayward <alan.hayward@arm.com>
13050 David Sherwood <david.sherwood@arm.com>
13052 * target.def (scalar_mode_supported_p): Take a scalar_mode
13053 instead of a machine_mode.
13054 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13055 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13056 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13057 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13058 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13059 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13060 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13061 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13062 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13063 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13064 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13065 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13066 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13067 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13069 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13070 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13071 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13072 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13073 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13075 * doc/tm.texi: Regenerate.
13077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13078 Alan Hayward <alan.hayward@arm.com>
13079 David Sherwood <david.sherwood@arm.com>
13081 * coretypes.h (opt_scalar_mode): New typedef.
13082 * gdbhooks.py (build_pretty_printers): Handle it.
13083 * machmode.h (mode_iterator::get_2xwider): Add overload for
13085 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13087 * expr.c (convert_mode_scalar): Likewise.
13088 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13089 * optabs.c (expand_float): Likewise.
13090 (expand_fix): Likewise.
13091 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13094 Alan Hayward <alan.hayward@arm.com>
13095 David Sherwood <david.sherwood@arm.com>
13097 * optabs.c (expand_float): Explicitly check for scalars before
13098 using a branching expansion.
13099 (expand_fix): Likewise.
13101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13102 Alan Hayward <alan.hayward@arm.com>
13103 David Sherwood <david.sherwood@arm.com>
13105 * expr.c (convert_mode): Split scalar handling out into...
13106 (convert_mode_scalar): ...this new function. Treat the modes
13109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13110 Alan Hayward <alan.hayward@arm.com>
13111 David Sherwood <david.sherwood@arm.com>
13113 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13115 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13118 Alan Hayward <alan.hayward@arm.com>
13119 David Sherwood <david.sherwood@arm.com>
13121 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13122 rather than a machine_mode.
13123 (fixed_from_string): Likewise.
13124 (fixed_convert): Likewise.
13125 (fixed_convert_from_int): Likewise.
13126 (fixed_convert_from_real): Likewise.
13127 (real_convert_from_fixed): Likewise.
13128 * fixed-value.c (fixed_from_double_int): Likewise.
13129 (fixed_from_string): Likewise.
13130 (fixed_convert): Likewise.
13131 (fixed_convert_from_int): Likewise.
13132 (fixed_convert_from_real): Likewise.
13133 (real_convert_from_fixed): Likewise.
13134 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13137 Alan Hayward <alan.hayward@arm.com>
13138 David Sherwood <david.sherwood@arm.com>
13140 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13141 of separate mode class checks. Do not allow vector modes here.
13142 (immed_wide_int_const): Use as_a <scalar_mode>.
13143 * explow.c (trunc_int_for_mode): Likewise.
13144 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13145 (wi::shwi): Likewise.
13146 (wi::min_value): Likewise.
13147 (wi::max_value): Likewise.
13148 * dwarf2out.c (loc_descriptor): Likewise.
13149 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13150 for CONST_WIDE_INT.
13152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13153 Alan Hayward <alan.hayward@arm.com>
13154 David Sherwood <david.sherwood@arm.com>
13156 * tree.h (SCALAR_TYPE_MODE): New macro.
13157 * expr.c (expand_expr_addr_expr_1): Use it.
13158 (expand_expr_real_2): Likewise.
13159 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13160 (fold_convert_const_fixed_from_int): Likewise.
13161 (fold_convert_const_fixed_from_real): Likewise.
13162 (native_encode_fixed): Likewise
13163 (native_encode_complex): Likewise
13164 (native_encode_vector): Likewise.
13165 (native_interpret_fixed): Likewise.
13166 (native_interpret_real): Likewise.
13167 (native_interpret_complex): Likewise.
13168 (native_interpret_vector): Likewise.
13169 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13170 (simd_clone_adjust_argument_types): Likewise.
13171 (simd_clone_init_simd_arrays): Likewise.
13172 (simd_clone_adjust): Likewise.
13173 * stor-layout.c (layout_type): Likewise.
13174 * tree.c (build_minus_one_cst): Likewise.
13175 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13176 * tree-inline.c (estimate_move_cost): Likewise.
13177 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13178 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13179 (vectorizable_reduction): Likewise.
13180 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13181 (vect_recog_mixed_size_cond_pattern): Likewise.
13182 (check_bool_pattern): Likewise.
13183 (adjust_bool_pattern): Likewise.
13184 (search_type_for_mask_1): Likewise.
13185 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13186 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13187 (vectorizable_load): Likewise.
13188 (vectorizable_store): Likewise.
13189 * ubsan.c (ubsan_encode_value): Likewise.
13190 * varasm.c (output_constant): Likewise.
13192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13193 Alan Hayward <alan.hayward@arm.com>
13194 David Sherwood <david.sherwood@arm.com>
13196 * coretypes.h (scalar_mode): New class.
13197 * machmode.h (scalar_mode): Likewise.
13198 (scalar_mode::includes_p): New function.
13199 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13200 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13201 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13202 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13203 * expmed.c (store_bit_field_1): Likewise.
13204 (extract_bit_field_1): Likewise.
13205 * expr.c (write_complex_part): Likewise.
13206 (read_complex_part): Likewise.
13207 (emit_move_complex_push): Likewise.
13208 (expand_expr_real_2): Likewise.
13209 * function.c (assign_parm_setup_reg): Likewise.
13210 (assign_parms_unsplit_complex): Likewise.
13211 * optabs.c (expand_binop): Likewise.
13212 * rtlanal.c (subreg_get_info): Likewise.
13213 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13214 * varasm.c (output_constant_pool_2): Likewise.
13216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13217 Alan Hayward <alan.hayward@arm.com>
13218 David Sherwood <david.sherwood@arm.com>
13220 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13222 (expmed_mult_highpart_optab): Likewise.
13223 (expmed_mult_highpart): Likewise.
13225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13226 Alan Hayward <alan.hayward@arm.com>
13227 David Sherwood <david.sherwood@arm.com>
13229 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13230 instead of a machine_mode.
13231 (builtin_memset_read_str): Likewise.
13232 * builtins.c (c_readstr): Likewise.
13233 (builtin_memcpy_read_str): Likewise.
13234 (builtin_strncpy_read_str): Likewise.
13235 (builtin_memset_read_str): Likewise.
13236 (builtin_memset_gen_str): Likewise.
13237 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13238 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13239 instead of a machine_mode.
13240 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13242 (make_extraction): Likewise.
13243 (try_widen_shift_mode): Take and return scalar_int_modes instead
13245 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13246 a scalar_int_mode instead of a machine_mode.
13247 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13248 (avr_addr_space_pointer_mode): Likewise.
13249 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13250 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13251 (msp430_unwind_word_mode): Likewise.
13252 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13253 (spu_addr_space_pointer_mode): Likewise.
13254 (spu_addr_space_address_mode): Likewise.
13255 (spu_libgcc_cmp_return_mode): Likewise.
13256 (spu_libgcc_shift_count_mode): Likewise.
13257 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13258 (rl78_addr_space_pointer_mode): Likewise.
13259 (fl78_unwind_word_mode): Likewise.
13260 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13262 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13263 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13264 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13265 (mips_valid_pointer_mode): Likewise.
13266 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13267 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13268 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13270 (ft32_addr_space_address_mode): Likewise.
13271 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13272 scalar_int_mode instead of a machine_mode.
13273 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13275 (m32c_addr_space_address_mode): Likewise.
13276 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13277 (rs6000_eh_return_filter_mode): Likewise.
13278 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13279 (rs6000_eh_return_filter_mode): Likewise.
13280 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13281 (s390_libgcc_shift_count_mode): Likewise.
13282 (s390_unwind_word_mode): Likewise.
13283 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13285 * target.def (mode_rep_extended): Likewise.
13286 (valid_pointer_mode): Likewise.
13287 (addr_space.valid_pointer_mode): Likewise.
13288 (eh_return_filter_mode): Return a scalar_int_mode rather than
13290 (libgcc_cmp_return_mode): Likewise.
13291 (libgcc_shift_count_mode): Likewise.
13292 (unwind_word_mode): Likewise.
13293 (addr_space.pointer_mode): Likewise.
13294 (addr_space.address_mode): Likewise.
13295 * doc/tm.texi: Regenerate.
13296 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13298 (do_jump): Use scalar_int_mode for local variables.
13299 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13300 rather than a machine_mode.
13301 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13302 (scompare_loc_descriptor_wide): Likewise.
13303 (scompare_loc_descriptor_narrow): Likewise.
13304 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13306 * except.c (sjlj_emit_dispatch_table): Likewise.
13307 (expand_builtin_eh_copy_values): Likewise.
13308 * explow.c (convert_memory_address_addr_space_1): Likewise.
13309 Take a scalar_int_mode rather than a machine_mode.
13310 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13311 than a machine_mode.
13312 (memory_address_addr_space): Use scalar_int_mode for local variables.
13313 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13314 rather than a machine_mode.
13315 * expmed.c (mask_rtx): Likewise.
13316 (init_expmed_one_conv): Likewise.
13317 (expand_mult_highpart_adjust): Likewise.
13318 (extract_high_half): Likewise.
13319 (expmed_mult_highpart_optab): Likewise.
13320 (expmed_mult_highpart): Likewise.
13321 (expand_smod_pow2): Likewise.
13322 (expand_sdiv_pow2): Likewise.
13323 (emit_store_flag_int): Likewise.
13324 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13326 (extract_low_bits): Likewise.
13327 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13329 * expr.c (pieces_addr::adjust): Likewise.
13330 (can_store_by_pieces): Likewise.
13331 (store_by_pieces): Likewise.
13332 (clear_by_pieces_1): Likewise.
13333 (expand_expr_addr_expr_1): Likewise.
13334 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13335 (expand_expr_real_1): Likewise.
13336 (try_casesi): Likewise.
13337 * final.c (shorten_branches): Likewise.
13338 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13339 type of "mode" to machine_mode.
13340 * internal-fn.c (expand_arith_overflow_result_store): Take a
13341 scalar_int_mode rather than a machine_mode.
13342 (expand_mul_overflow): Use scalar_int_mode for local variables.
13343 * loop-doloop.c (doloop_modify): Likewise.
13344 (doloop_optimize): Likewise.
13345 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13346 than a machine_mode.
13347 (expand_doubleword_shift_condmove): Likewise.
13348 (expand_doubleword_shift): Likewise.
13349 (expand_doubleword_clz): Likewise.
13350 (expand_doubleword_popcount): Likewise.
13351 (expand_doubleword_parity): Likewise.
13352 (expand_absneg_bit): Use scalar_int_mode for local variables.
13353 (prepare_float_lib_cmp): Likewise.
13354 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13355 rather than a machine_mode.
13356 (convert_memory_address_addr_space): Likewise.
13357 (get_mode_bounds): Likewise.
13358 (get_address_mode): Return a scalar_int_mode rather than a
13360 * rtlanal.c (get_address_mode): Likewise.
13361 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13362 than a machine_mode.
13363 * targhooks.c (default_mode_rep_extended): Likewise.
13364 (default_valid_pointer_mode): Likewise.
13365 (default_addr_space_valid_pointer_mode): Likewise.
13366 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13367 than a machine_mode.
13368 (default_libgcc_cmp_return_mode): Likewise.
13369 (default_libgcc_shift_count_mode): Likewise.
13370 (default_unwind_word_mode): Likewise.
13371 (default_addr_space_pointer_mode): Likewise.
13372 (default_addr_space_address_mode): Likewise.
13373 * targhooks.h (default_eh_return_filter_mode): Likewise.
13374 (default_libgcc_cmp_return_mode): Likewise.
13375 (default_libgcc_shift_count_mode): Likewise.
13376 (default_unwind_word_mode): Likewise.
13377 (default_addr_space_pointer_mode): Likewise.
13378 (default_addr_space_address_mode): Likewise.
13379 (default_mode_rep_extended): Take a scalar_int_mode rather than
13381 (default_valid_pointer_mode): Likewise.
13382 (default_addr_space_valid_pointer_mode): Likewise.
13383 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13385 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13386 rather than a machine_mode.
13387 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13388 for local variables.
13389 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13390 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13391 than a machine_mode.
13393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13394 Alan Hayward <alan.hayward@arm.com>
13395 David Sherwood <david.sherwood@arm.com>
13397 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13398 the mode argument to scalar_int_mode.
13399 (do_jump_by_parts_zero_rtx): Likewise.
13400 (do_jump_by_parts_equality_rtx): Likewise.
13401 (do_jump_by_parts_greater): Take a mode argument.
13402 (do_jump_by_parts_equality): Likewise.
13403 (do_jump_1): Update calls accordingly.
13405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13406 Alan Hayward <alan.hayward@arm.com>
13407 David Sherwood <david.sherwood@arm.com>
13409 * is-a.h (safe_dyn_cast): New function.
13410 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13411 (jump_table_for_label): Likewise.
13412 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13413 instead of an rtx_insn *.
13414 (shorten_branches): Use dyn_cast instead of LABEL_P and
13415 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13416 rtx_jump_table_data::get_data_mode.
13417 (final_scan_insn): Likewise.
13419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13420 Alan Hayward <alan.hayward@arm.com>
13421 David Sherwood <david.sherwood@arm.com>
13423 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13424 trying to combine a full-register integer set with a subreg
13427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13428 Alan Hayward <alan.hayward@arm.com>
13429 David Sherwood <david.sherwood@arm.com>
13431 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13432 that is always either address_mode or pointer_mode.
13434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13435 Alan Hayward <alan.hayward@arm.com>
13436 David Sherwood <david.sherwood@arm.com>
13438 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13439 when the two are known to be equal.
13441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13442 Alan Hayward <alan.hayward@arm.com>
13443 David Sherwood <david.sherwood@arm.com>
13445 * simplify-rtx.c (simplify_const_unary_operation): Use
13446 is_a <scalar_int_mode> instead of checking for a nonzero
13447 precision. Forcibly convert op_mode to a scalar_int_mode
13448 in that case. More clearly differentiate the operand and
13449 result modes and use the former when deciding what the value
13450 of a count-bits operation should be. Use is_int_mode instead
13451 of checking for a MODE_INT. Remove redundant check for whether
13452 this mode has a zero precision.
13454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13455 Alan Hayward <alan.hayward@arm.com>
13456 David Sherwood <david.sherwood@arm.com>
13458 * optabs.c (widen_leading): Change the type of the mode argument
13459 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13460 (widen_bswap): Likewise.
13461 (expand_parity): Likewise.
13462 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13463 (expand_ffs): Likewise.
13464 (epand_unop): Check for scalar integer modes before calling the
13467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13468 Alan Hayward <alan.hayward@arm.com>
13469 David Sherwood <david.sherwood@arm.com>
13471 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13473 (expand_expr_real_1): Update call accordingly.
13475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13476 Alan Hayward <alan.hayward@arm.com>
13477 David Sherwood <david.sherwood@arm.com>
13479 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13480 value_mode arguments. Use scalar_int_mode internally.
13481 (store_bit_field_1): Rename the new integer mode from imode
13482 to op0_mode and use it instead of GET_MODE (op0). Update calls
13483 to store_split_bit_field, store_bit_field_using_insv and
13484 store_fixed_bit_field.
13485 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13486 Use scalar_int_mode internally. Use a bit count rather than a mode
13487 when calculating the largest bit size for get_best_mode.
13488 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13489 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13490 Remove assertion that OP0 has a scalar integer mode.
13491 (store_split_bit_field): Add op0_mode and value_mode arguments.
13492 Update calls to extract_fixed_bit_field.
13493 (extract_bit_field_using_extv): Add an op0_mode argument.
13494 Use scalar_int_mode internally.
13495 (extract_bit_field_1): Rename the new integer mode from imode to
13496 op0_mode and use it instead of GET_MODE (op0). Update calls to
13497 extract_split_bit_field, extract_bit_field_using_extv and
13498 extract_fixed_bit_field.
13499 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13500 to extract_split_bit_field and extract_fixed_bit_field_1.
13501 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13502 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13503 on the target mode.
13504 (extract_split_bit_field): Add an op0_mode argument. Update call
13505 to extract_fixed_bit_field.
13507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13508 Alan Hayward <alan.hayward@arm.com>
13509 David Sherwood <david.sherwood@arm.com>
13511 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13512 * explow.c (hard_function_value): Likewise.
13513 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13514 convert_to_mode call outside the loop.
13515 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13516 for the mode iterator. Require the mode specified by max_pieces
13518 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13520 (copy_blkmode_to_reg): Likewise.
13521 (set_storage_via_setmem): Likewise.
13522 * optabs.c (prepare_cmp_insn): Likewise.
13523 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13524 * stor-layout.c (finish_bitfield_representative): Likewise.
13526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13527 Alan Hayward <alan.hayward@arm.com>
13528 David Sherwood <david.sherwood@arm.com>
13530 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13531 * expr.c (convert_move): Use them.
13532 (convert_modes): Likewise.
13533 (store_expr_with_bounds): Likewise.
13535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13536 Alan Hayward <alan.hayward@arm.com>
13537 David Sherwood <david.sherwood@arm.com>
13539 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13540 parameter for the mode of "x". Remove the "known_x", "known_mode"
13541 and "known_ret" arguments. Change the type of the mode argument
13542 to scalar_int_mode.
13543 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13544 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13545 (reg_num_sign_bit_copies_for_combine): Likewise.
13546 * rtlanal.c (nonzero_bits1): Likewise.
13547 (num_sign_bit_copies1): Likewise.
13548 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13549 (reg_num_sign_bit_copies_general): Likewise.
13550 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13551 (reg_nonzero_bits_general): Likewise.
13553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13554 Alan Hayward <alan.hayward@arm.com>
13555 David Sherwood <david.sherwood@arm.com>
13557 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13558 than in subroutines. Return 1 for non-integer modes.
13559 (cached_num_sign_bit_copies): Change the type of the mode parameter
13560 to scalar_int_mode.
13561 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13562 classes. Handle CONST_INT_P first and then check whether X also
13563 has a scalar integer mode. Check the same thing for inner registers
13564 of a SUBREG and for values that are being extended or truncated.
13566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13567 Alan Hayward <alan.hayward@arm.com>
13568 David Sherwood <david.sherwood@arm.com>
13570 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13571 in subroutines. Return the mode mask for non-integer modes.
13572 (cached_nonzero_bits): Change the type of the mode parameter
13573 to scalar_int_mode.
13574 (nonzero_bits1): Likewise. Remove early exit for other mode
13575 classes. Handle CONST_INT_P first and then check whether X
13576 also has a scalar integer mode.
13578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13579 Alan Hayward <alan.hayward@arm.com>
13580 David Sherwood <david.sherwood@arm.com>
13582 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13583 Return a scalar_int_mode and assert that the size is greater than
13585 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13586 (op_by_pieces_d::op_by_pieces_d): Likewise.
13587 (op_by_pieces_d::run): Likewise.
13588 (can_store_by_pieces): 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 * combine.c (extract_left_shift): Add a mode argument and update
13596 (make_compound_operation_int): Change the type of the mode parameter
13597 to scalar_int_mode and update the call to extract_left_shift.
13599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13600 Alan Hayward <alan.hayward@arm.com>
13601 David Sherwood <david.sherwood@arm.com>
13603 * combine.c (simplify_and_const_int): Change the type of the mode
13604 parameter to scalar_int_mode.
13605 (simplify_and_const_int_1): Likewise. Update recursive call.
13607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13608 Alan Hayward <alan.hayward@arm.com>
13609 David Sherwood <david.sherwood@arm.com>
13611 * combine.c (simplify_compare_const): Check that the mode is a
13612 scalar_int_mode (rather than VOIDmode) before testing its
13614 (simplify_comparison): Move COMPARISON_P handling out of the
13615 loop and restrict the latter part of the loop to scalar_int_modes.
13616 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13617 and when considering SUBREG_REGs. Use is_int_mode instead of
13618 checking GET_MODE_CLASS against MODE_INT.
13620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13621 Alan Hayward <alan.hayward@arm.com>
13622 David Sherwood <david.sherwood@arm.com>
13624 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13625 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13626 shift_unit_mode and for modes involved in scalar shifts.
13628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13629 Alan Hayward <alan.hayward@arm.com>
13630 David Sherwood <david.sherwood@arm.com>
13632 * combine.c (force_int_to_mode): New function, split out from...
13633 (force_to_mode): ...here. Keep xmode up-to-date and use it
13634 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
13641 opt_scalar_int_mode and update comment.
13642 (extraction_insn::field_mode): Change type to scalar_int_mode.
13643 (extraction_insn::pos_mode): Likewise.
13644 * combine.c (make_extraction): Update accordingly.
13645 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13646 (get_optab_extraction_insn): Likewise.
13647 * recog.c (simplify_while_replacing): Likewise.
13648 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13649 parameter to opt_scalar_int_mode.
13651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13652 Alan Hayward <alan.hayward@arm.com>
13653 David Sherwood <david.sherwood@arm.com>
13655 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13656 to a scalar_int_mode instead of a machine_mode.
13657 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13658 (get_best_mode): Return a boolean and use a pointer argument to store
13659 the selected mode. Replace the limit mode parameter with a bit limit.
13660 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13661 for the values returned by bit_field_mode_iterator::next_mode.
13662 (store_bit_field): Update call to get_best_mode.
13663 (store_fixed_bit_field): Likewise.
13664 (extract_fixed_bit_field): Likewise.
13665 * expr.c (optimize_bitfield_assignment_op): Likewise.
13666 * fold-const.c (optimize_bit_field_compare): Likewise.
13667 (fold_truth_andor_1): Likewise.
13668 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13669 Update for new type of m_mode.
13670 (get_best_mode): As above.
13672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13673 Alan Hayward <alan.hayward@arm.com>
13674 David Sherwood <david.sherwood@arm.com>
13676 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13677 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13678 (store_bit_field): Check is_a <scalar_int_mode> before calling
13679 strict_volatile_bitfield_p.
13680 (extract_bit_field): Likewise.
13682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13683 Alan Hayward <alan.hayward@arm.com>
13684 David Sherwood <david.sherwood@arm.com>
13686 * target.def (cstore_mode): Return a scalar_int_mode.
13687 * doc/tm.texi: Regenerate.
13688 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13689 * targhooks.h (default_cstore_mode): Likewise.
13690 * targhooks.c (default_cstore_mode): Likewise, using a forced
13692 * expmed.c (emit_cstore): Expect the target of the cstore to be
13695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13696 Alan Hayward <alan.hayward@arm.com>
13697 David Sherwood <david.sherwood@arm.com>
13699 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13701 (niter_desc): Likewise mode.
13702 (iv_analyze): Add a mode parameter.
13704 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13705 and change its type to scalar_int_mode.
13706 * loop-iv.c: Update commentary at head of file.
13707 (iv_constant): Pass the mode paraeter before the rtx it describes
13708 and change its type to scalar_int_mode. Remove VOIDmode handling.
13709 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13710 (iv_extend): Likewise.
13711 (shorten_into_mode): Likewise.
13712 (iv_add): Use scalar_int_mode.
13713 (iv_mult): Likewise.
13714 (iv_shift): Likewise.
13715 (canonicalize_iv_subregs): Likewise.
13716 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13717 it describes and change its mode to scalar_int_mode. Also change
13718 the type of the returned inner_mode to scalar_int_mode.
13719 (get_biv_step): Likewise, turning outer_mode from a pointer
13720 into a direct parameter. Update call to get_biv_step_1.
13721 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13722 iv_constant and get_biv_step.
13723 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13724 and change its type to scalar_int_mode. Don't initialise iv->mode
13725 to VOIDmode and remove later checks for its still being VOIDmode.
13726 Update calls to iv_analyze_op and iv_analyze_expr. Check
13727 is_a <scalar_int_mode> when changing the mode under consideration.
13728 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13729 Update call to iv_analyze_expr.
13730 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13731 inner register is not also a scalar_int_mode. Update call to
13733 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13734 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13735 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13736 separate mode class checks. Update calls to iv_analyze. Remove
13737 fix-up of VOIDmodes after iv_analyze_biv.
13738 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13739 don't have a scalar_int_mode. Update call to biv_p.
13741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13742 Alan Hayward <alan.hayward@arm.com>
13743 David Sherwood <david.sherwood@arm.com>
13745 * cfgexpand.c (convert_debug_memory_address): Use
13746 as_a <scalar_int_mode>.
13747 * combine.c (expand_compound_operation): Likewise.
13748 (make_extraction): Likewise.
13749 (change_zero_ext): Likewise.
13750 (simplify_comparison): Likewise.
13751 * cse.c (cse_insn): Likewise.
13752 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13753 (mem_loc_descriptor): Likewise.
13754 (loc_descriptor): Likewise.
13755 * expmed.c (init_expmed_one_mode): Likewise.
13756 (synth_mult): Likewise.
13757 (emit_store_flag_1): Likewise.
13758 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13759 of a comparison with size.
13760 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13761 (reduce_to_bit_field_precision): Likewise.
13762 * function.c (expand_function_end): Likewise.
13763 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13764 * loop-doloop.c (doloop_modify): Likewise.
13765 * optabs.c (expand_binop): Likewise.
13766 (expand_unop): Likewise.
13767 (expand_copysign_absneg): Likewise.
13768 (prepare_cmp_insn): Likewise.
13769 (maybe_legitimize_operand): Likewise.
13770 * recog.c (const_scalar_int_operand): Likewise.
13771 * rtlanal.c (get_address_mode): Likewise.
13772 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13773 (simplify_cond_clz_ctz): Likewise.
13774 * tree-nested.c (get_nl_goto_field): Likewise.
13775 * tree.c (build_vector_type_for_mode): Likewise.
13776 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13783 * builtins.c (expand_builtin_signbit): Use it.
13784 * cfgexpand.c (expand_debug_expr): Likewise.
13785 * dojump.c (do_jump): Likewise.
13786 (do_compare_and_jump): Likewise.
13787 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13788 * expmed.c (make_tree): Likewise.
13789 * expr.c (expand_expr_real_2): Likewise.
13790 (expand_expr_real_1): Likewise.
13791 (try_casesi): Likewise.
13792 * fold-const-call.c (fold_const_call_ss): Likewise.
13793 * fold-const.c (unextend): Likewise.
13794 (extract_muldiv_1): Likewise.
13795 (fold_single_bit_test): Likewise.
13796 (native_encode_int): Likewise.
13797 (native_encode_string): Likewise.
13798 (native_interpret_int): Likewise.
13799 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13800 * internal-fn.c (expand_addsub_overflow): Likewise.
13801 (expand_neg_overflow): Likewise.
13802 (expand_mul_overflow): Likewise.
13803 (expand_arith_overflow): Likewise.
13804 * match.pd: Likewise.
13805 * stor-layout.c (layout_type): Likewise.
13806 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13807 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13808 * tree-ssanames.c (get_range_info): Likewise.
13809 * tree-switch-conversion.c (array_value_type) Likewise.
13810 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13811 (vect_recog_divmod_pattern): Likewise.
13812 (vect_recog_mixed_size_cond_pattern): Likewise.
13813 * tree-vrp.c (extract_range_basic): Likewise.
13814 (simplify_float_conversion_using_ranges): Likewise.
13815 * tree.c (int_fits_type_p): Likewise.
13816 * ubsan.c (instrument_bool_enum_load): Likewise.
13817 * varasm.c (mergeable_string_section): Likewise.
13818 (narrowing_initializer_constant_valid_p): Likewise.
13819 (output_constant): Likewise.
13821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13822 Alan Hayward <alan.hayward@arm.com>
13823 David Sherwood <david.sherwood@arm.com>
13825 * machmode.h (NARROWEST_INT_MODE): New macro.
13826 * expr.c (alignment_for_piecewise_move): Use it instead of
13827 GET_CLASS_NARROWEST_MODE (MODE_INT).
13828 (push_block): Likewise.
13829 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13831 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13834 Alan Hayward <alan.hayward@arm.com>
13835 David Sherwood <david.sherwood@arm.com>
13837 * postreload.c (move2add_valid_value_p): Change the type of the
13838 mode parameter to scalar_int_mode.
13839 (move2add_use_add2_insn): Add a mode parameter and use it instead
13841 (move2add_use_add3_insn): Likewise.
13842 (reload_cse_move2add): Update accordingly.
13844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13845 Alan Hayward <alan.hayward@arm.com>
13846 David Sherwood <david.sherwood@arm.com>
13848 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13850 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13851 * optabs.c (expand_unop): Likewise.
13853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13854 Alan Hayward <alan.hayward@arm.com>
13855 David Sherwood <david.sherwood@arm.com>
13857 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13858 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13859 (popcount_loc_descriptor): Likewise.
13860 (bswap_loc_descriptor): Likewise.
13861 (rotate_loc_descriptor): Likewise.
13862 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13863 calling the functions above.
13865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13866 Alan Hayward <alan.hayward@arm.com>
13867 David Sherwood <david.sherwood@arm.com>
13869 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13871 (try_combine): Likewise.
13872 (simplify_if_then_else): Likewise.
13873 * cse.c (cse_insn): Likewise.
13874 * dwarf2out.c (mem_loc_descriptor): Likewise.
13875 * emit-rtl.c (gen_lowpart_common): Likewise.
13876 * simplify-rtx.c (simplify_truncation): Likewise.
13877 (simplify_binary_operation_1): Likewise.
13878 (simplify_const_relational_operation): Likewise.
13879 (simplify_ternary_operation): Likewise.
13880 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13883 Alan Hayward <alan.hayward@arm.com>
13884 David Sherwood <david.sherwood@arm.com>
13886 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13887 * reload.c (push_reload): Likewise.
13888 (find_reloads): 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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13895 (make_compound_operation_int): Likewise.
13896 (change_zero_ext): Likewise.
13897 * expr.c (convert_move): Likewise.
13898 (convert_modes): Likewise.
13899 * fwprop.c (forward_propagate_subreg): Likewise.
13900 * loop-iv.c (get_biv_step_1): Likewise.
13901 * optabs.c (widen_operand): Likewise.
13902 * postreload.c (move2add_valid_value_p): Likewise.
13903 * recog.c (simplify_while_replacing): Likewise.
13904 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13905 (simplify_binary_operation_1): Likewise. Remove redundant
13906 mode equality check.
13908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13909 Alan Hayward <alan.hayward@arm.com>
13910 David Sherwood <david.sherwood@arm.com>
13912 * combine.c (combine_simplify_rtx): Add checks for
13913 is_a <scalar_int_mode>.
13914 (simplify_if_then_else): Likewise.
13915 (make_field_assignment): Likewise.
13916 (simplify_comparison): Likewise.
13917 * ifcvt.c (noce_try_bitop): Likewise.
13918 * loop-invariant.c (canonicalize_address_mult): Likewise.
13919 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13922 Alan Hayward <alan.hayward@arm.com>
13923 David Sherwood <david.sherwood@arm.com>
13925 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13926 is_a <scalar_int_mode> instead of != BLKmode.
13928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13929 Alan Hayward <alan.hayward@arm.com>
13930 David Sherwood <david.sherwood@arm.com>
13932 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13933 instead of != VOIDmode.
13934 * combine.c (if_then_else_cond): Likewise.
13935 (change_zero_ext): Likewise.
13936 * dwarf2out.c (mem_loc_descriptor): Likewise.
13937 (loc_descriptor): Likewise.
13938 * rtlanal.c (canonicalize_condition): Likewise.
13939 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13942 Alan Hayward <alan.hayward@arm.com>
13943 David Sherwood <david.sherwood@arm.com>
13945 * simplify-rtx.c (simplify_binary_operation_1): Use
13946 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13949 Alan Hayward <alan.hayward@arm.com>
13950 David Sherwood <david.sherwood@arm.com>
13952 * wide-int.h (int_traits<unsigned char>) New class.
13953 (int_traits<unsigned short>) Likewise.
13954 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13955 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13957 * combine.c (set_nonzero_bits_and_sign_copies): Use
13958 is_a <scalar_int_mode>.
13959 (find_split_point): Likewise.
13960 (combine_simplify_rtx): Likewise.
13961 (simplify_logical): Likewise.
13962 (expand_compound_operation): Likewise.
13963 (expand_field_assignment): Likewise.
13964 (make_compound_operation): Likewise.
13965 (extended_count): Likewise.
13966 (change_zero_ext): Likewise.
13967 (simplify_comparison): Likewise.
13968 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13969 (ucompare_loc_descriptor): Likewise.
13970 (minmax_loc_descriptor): Likewise.
13971 (mem_loc_descriptor): Likewise.
13972 (loc_descriptor): Likewise.
13973 * expmed.c (init_expmed_one_mode): Likewise.
13974 * lra-constraints.c (lra_constraint_offset): Likewise.
13975 * optabs.c (prepare_libcall_arg): Likewise.
13976 * postreload.c (move2add_note_store): Likewise.
13977 * reload.c (operands_match_p): Likewise.
13978 * rtl.h (load_extend_op): Likewise.
13979 * rtlhooks.c (gen_lowpart_general): Likewise.
13980 * simplify-rtx.c (simplify_truncation): Likewise.
13981 (simplify_unary_operation_1): Likewise.
13982 (simplify_binary_operation_1): Likewise.
13983 (simplify_const_binary_operation): Likewise.
13984 (simplify_const_relational_operation): Likewise.
13985 (simplify_subreg): Likewise.
13986 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13987 * var-tracking.c (adjust_mems): Likewise.
13988 (prepare_call_arguments): Likewise.
13990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13991 Alan Hayward <alan.hayward@arm.com>
13992 David Sherwood <david.sherwood@arm.com>
13994 * machmode.h (is_int_mode): New fuction.
13995 * combine.c (find_split_point): Use it.
13996 (combine_simplify_rtx): Likewise.
13997 (simplify_if_then_else): Likewise.
13998 (simplify_set): Likewise.
13999 (simplify_shift_const_1): Likewise.
14000 (simplify_comparison): Likewise.
14001 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14002 * cse.c (notreg_cost): Likewise.
14003 (cse_insn): Likewise.
14004 * cselib.c (cselib_lookup_1): Likewise.
14005 * dojump.c (do_jump_1): Likewise.
14006 (do_compare_rtx_and_jump): Likewise.
14007 * dse.c (get_call_args): Likewise.
14008 * dwarf2out.c (rtl_for_decl_init): Likewise.
14009 (native_encode_initializer): Likewise.
14010 * expmed.c (emit_store_flag_1): Likewise.
14011 (emit_store_flag): Likewise.
14012 * expr.c (convert_modes): Likewise.
14013 (store_field): Likewise.
14014 (expand_expr_real_1): Likewise.
14015 * fold-const.c (fold_read_from_constant_string): Likewise.
14016 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14017 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14018 * optabs.c (expand_binop): Likewise.
14019 (expand_unop): Likewise.
14020 (expand_abs_nojump): Likewise.
14021 (expand_one_cmpl_abs_nojump): Likewise.
14022 * simplify-rtx.c (mode_signbit_p): Likewise.
14023 (val_signbit_p): Likewise.
14024 (val_signbit_known_set_p): Likewise.
14025 (val_signbit_known_clear_p): Likewise.
14026 (simplify_relational_operation_1): Likewise.
14027 * tree.c (vector_type_mode): Likewise.
14029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * machmode.h (smallest_mode_for_size): Fix formatting.
14034 (smallest_int_mode_for_size): New function.
14035 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14036 instead of smallest_mode_for_size.
14037 * combine.c (make_extraction): Likewise.
14038 * config/arc/arc.c (arc_expand_movmem): Likewise.
14039 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14040 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14041 * config/s390/s390.c (s390_expand_insv): Likewise.
14042 * config/sparc/sparc.c (assign_int_registers): Likewise.
14043 * config/spu/spu.c (spu_function_value): Likewise.
14044 (spu_function_arg): Likewise.
14045 * coverage.c (get_gcov_type): Likewise.
14046 (get_gcov_unsigned_t): Likewise.
14047 * dse.c (find_shift_sequence): Likewise.
14048 * expmed.c (store_bit_field_1): Likewise.
14049 * expr.c (convert_move): Likewise.
14050 (store_field): Likewise.
14051 * internal-fn.c (expand_arith_overflow): Likewise.
14052 * optabs-query.c (get_best_extraction_insn): Likewise.
14053 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14054 * stor-layout.c (layout_type): Likewise.
14055 (initialize_sizetypes): Likewise.
14056 * targhooks.c (default_get_mask_mode): Likewise.
14057 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14060 Alan Hayward <alan.hayward@arm.com>
14061 David Sherwood <david.sherwood@arm.com>
14063 * machmode.h (opt_mode::else_blk): New function.
14064 (int_mode_for_mode): Declare.
14065 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14066 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14068 * cfgexpand.c (expand_debug_expr): Likewise.
14069 * combine.c (gen_lowpart_or_truncate): Likewise.
14070 (gen_lowpart_for_combine): Likewise.
14071 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14072 * config/avr/avr.c (avr_to_int_mode): Likewise.
14073 (avr_out_plus_1): Likewise.
14074 (avr_out_plus): Likewise.
14075 (avr_out_round): Likewise.
14076 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14077 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14078 (s390_expand_vcond): Likewise.
14079 * config/spu/spu.c (spu_split_immediate): Likewise.
14080 (spu_expand_mov): Likewise.
14081 * dse.c (get_stored_val): Likewise.
14082 * expmed.c (store_bit_field_1): Likewise.
14083 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14085 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14086 (extract_low_bits): Likewise.
14087 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14088 handling rather than repeating the check.
14089 (emit_group_store): Likewise.
14090 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14091 * optabs.c (expand_absneg_bit): Likewise.
14092 (expand_copysign_absneg): Likewise.
14093 (expand_copysign_bit): Likewise.
14094 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14095 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14096 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14097 * var-tracking.c (prepare_call_arguments): Likewise.
14098 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14099 int_mode_for_mode instead of mode_for_size.
14100 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14103 Alan Hayward <alan.hayward@arm.com>
14104 David Sherwood <david.sherwood@arm.com>
14106 * machmode.h (int_mode_for_size): New function.
14107 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14108 instead of mode_for_size.
14109 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14111 * combine.c (expand_field_assignment): Use int_mode_for_size
14112 instead of mode_for_size.
14113 (make_extraction): Likewise.
14114 (simplify_shift_const_1): Likewise.
14115 (simplify_comparison): Likewise.
14116 * dojump.c (do_jump): Likewise.
14117 * dwarf2out.c (mem_loc_descriptor): Likewise.
14118 * emit-rtl.c (init_derived_machine_modes): Likewise.
14119 * expmed.c (flip_storage_order): Likewise.
14120 (convert_extracted_bit_field): Likewise.
14121 * expr.c (copy_blkmode_from_reg): Likewise.
14122 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14123 * internal-fn.c (expand_mul_overflow): Likewise.
14124 * lower-subreg.c (simple_move): Likewise.
14125 * optabs-libfuncs.c (init_optabs): Likewise.
14126 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14127 * tree.c (vector_type_mode): Likewise.
14128 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14129 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14130 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14131 * tree-vect-stmts.c (vectorizable_load): Likewise.
14132 (vectorizable_store): Likewise.
14134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14135 Alan Hayward <alan.hayward@arm.com>
14136 David Sherwood <david.sherwood@arm.com>
14138 * coretypes.h (pod_mode): New type.
14139 (scalar_int_mode_pod): New typedef.
14140 * machmode.h (pod_mode): New class.
14141 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14142 * genmodes.c (emit_mode_int_n): Update accordingly.
14143 * lower-subreg.h (target_lower_subreg): Change type to
14144 scalar_int_mode_pod.
14145 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14146 scalar_int_mode_pod.
14148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14149 Alan Hayward <alan.hayward@arm.com>
14150 David Sherwood <david.sherwood@arm.com>
14152 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14153 machine_mode to scalar_int_mode.
14154 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14155 (rs6000_option_override_internal): Remove cast to int.
14156 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14157 machine_mode to scalar_int_mode.
14158 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14159 (rs6000_option_override_internal): Remove cast to int.
14160 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14161 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14163 * config/s390/s390.c (s390_expand_builtin): Likewise.
14164 * coretypes.h (scalar_int_mode): New type.
14165 (opt_scalar_int_mode): New typedef.
14166 * machmode.h (scalar_int_mode): New class.
14167 (scalar_int_mode::includes_p): New function.
14168 (byte_mode): Change type to scalar_int_mode.
14169 (word_mode): Likewise.
14170 (ptr_mode): Likewise.
14171 * emit-rtl.c (byte_mode): Likewise.
14172 (word_mode): Likewise.
14173 (ptr_mode): Likewise.
14174 (init_derived_machine_modes): Update accordingly.
14175 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14176 and MODE_PARTIAL_INT.
14177 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14178 opt_scalar_int_mode.
14180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14181 Alan Hayward <alan.hayward@arm.com>
14182 David Sherwood <david.sherwood@arm.com>
14184 * target.def (libgcc_floating_mode_supported_p): Take a
14186 * doc/tm.texi: Regenerate.
14187 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14189 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14190 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14194 Alan Hayward <alan.hayward@arm.com>
14195 David Sherwood <david.sherwood@arm.com>
14197 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14198 * doc/tm.texi: Regenerate.
14199 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14200 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14201 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14202 * targhooks.h (default_floatn_mode): Likewise.
14203 * targhooks.c (default_floatn_mode): Likewise.
14204 * tree.c (build_common_tree_nodes): Update accordingly.
14206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14207 Alan Hayward <alan.hayward@arm.com>
14208 David Sherwood <david.sherwood@arm.com>
14210 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14211 (mode_iterator::iterate_p): Likewise.
14212 (mode_iterator::get_wider): Likewise.
14213 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14216 Alan Hayward <alan.hayward@arm.com>
14217 David Sherwood <david.sherwood@arm.com>
14219 * coretypes.h (opt_scalar_float_mode): New typedef.
14220 * machmode.h (float_mode_for_size): New function.
14221 * emit-rtl.c (double_mode): Delete.
14222 (init_emit_once): Use float_mode_for_size.
14223 * stor-layout.c (layout_type): Likewise.
14224 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14227 Alan Hayward <alan.hayward@arm.com>
14228 David Sherwood <david.sherwood@arm.com>
14230 * output.h (assemble_real): Take a scalar_float_mode.
14231 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14232 * config/arm/arm.md (consttable_4): Likewise.
14233 (consttable_8): Likewise.
14234 (consttable_16): Likewise.
14235 * config/mips/mips.md (consttable_float): Likewise.
14236 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14237 * varasm.c (assemble_real): Take a scalar_float_mode.
14238 (output_constant_pool_2): Update accordingly.
14239 (output_constant): Likewise.
14241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14242 Alan Hayward <alan.hayward@arm.com>
14243 David Sherwood <david.sherwood@arm.com>
14245 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14246 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14247 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14248 (native_encode_real): Likewise.
14249 (native_interpret_real): Likewise.
14250 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14251 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14254 Alan Hayward <alan.hayward@arm.com>
14255 David Sherwood <david.sherwood@arm.com>
14257 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14258 <scalar_float_mode>. Simplify.
14259 (gen_extend_conv_libfunc): Likewise.
14261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14262 Alan Hayward <alan.hayward@arm.com>
14263 David Sherwood <david.sherwood@arm.com>
14265 * coretypes.h (scalar_float_mode): New type.
14266 * machmode.h (mode_traits::from_int): Use machine_mode if
14267 USE_ENUM_MODES is defined.
14268 (is_a): New function.
14270 (dyn_cast): Likewise.
14271 (scalar_float_mode): New class.
14272 (scalar_float_mode::includes_p): New function.
14273 (is_float_mode): Likewise.
14274 * gdbhooks.py (MachineModePrinter): New class.
14275 (build_pretty_printer): Use it for scalar_float_mode.
14276 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14277 (format_helper::format_helper): Turn into a template.
14278 * genmodes.c (get_mode_class): New function.
14279 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14280 or machine_mode if none.
14281 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14282 as_a <scalar_float_mode>.
14283 * dwarf2out.c (mem_loc_descriptor): Likewise.
14284 (insert_float): Likewise.
14285 (add_const_value_attribute): Likewise.
14286 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14287 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14288 (expand_unop): Update accordingly.
14289 (expand_abs_nojump): Likewise.
14290 (expand_copysign_absneg): Take a scalar_float_mode.
14291 (expand_copysign_bit): Likewise.
14292 (expand_copysign): Update accordingly.
14294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14295 Alan Hayward <alan.hayward@arm.com>
14296 David Sherwood <david.sherwood@arm.com>
14298 * coretypes.h (opt_mode): New class.
14299 * machmode.h (opt_mode): Likewise.
14300 (opt_mode::else_void): New function.
14301 (opt_mode::require): Likewise.
14302 (opt_mode::exists): Likewise.
14303 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14304 (GET_MODE_2XWIDER_MODE): Likewise.
14305 (mode_iterator::get_wider): Update accordingly.
14306 (mode_iterator::get_2xwider): Likewise.
14307 (mode_iterator::get_known_wider): Likewise, turning into a template.
14308 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14309 forcing a wider mode to exist.
14310 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14311 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14312 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14313 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14314 * lower-subreg.c (init_lower_subreg): Likewise.
14315 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14316 on the final iteration.
14317 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14318 a wider mode exists before asking for a move pattern.
14319 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14320 forcing a wider mode to exist.
14321 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14322 returning false if no such mode exists.
14323 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14324 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14325 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14326 Avoid checking for a MODE_INT if we already know the mode is not a
14328 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14329 forcing a wider mode to exist.
14330 (expmed_mult_highpart_optab): Likewise.
14331 (expmed_mult_highpart): Likewise.
14332 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14334 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14335 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14336 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14337 * internal-fn.c (expand_mul_overflow): Update use of
14338 GET_MODE_2XWIDER_MODE.
14339 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14340 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14341 GET_MODE_WIDER_MODE.
14342 (convert_plusminus_to_widen): Likewise.
14343 * tree-switch-conversion.c (array_value_type): Likewise.
14344 * var-tracking.c (emit_note_insn_var_location): Likewise.
14345 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14346 Return false inside rather than outside the loop if no wider mode
14348 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14349 and GET_MODE_2XWIDER_MODE
14350 (can_compare_p): Use else_void.
14351 * gdbhooks.py (OptMachineModePrinter): New class.
14352 (build_pretty_printer): Use it for opt_mode.
14354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14355 Alan Hayward <alan.hayward@arm.com>
14356 David Sherwood <david.sherwood@arm.com>
14358 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14359 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14362 Alan Hayward <alan.hayward@arm.com>
14363 David Sherwood <david.sherwood@arm.com>
14365 * machmode.h (mode_traits): New structure.
14366 (get_narrowest_mode): New function.
14367 (mode_iterator::start): Likewise.
14368 (mode_iterator::iterate_p): Likewise.
14369 (mode_iterator::get_wider): Likewise.
14370 (mode_iterator::get_known_wider): Likewise.
14371 (mode_iterator::get_2xwider): Likewise.
14372 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14373 (FOR_EACH_MODE): Likewise.
14374 (FOR_EACH_MODE_FROM): Likewise.
14375 (FOR_EACH_MODE_UNTIL): Likewise.
14376 (FOR_EACH_WIDER_MODE): Likewise.
14377 (FOR_EACH_2XWIDER_MODE): Likewise.
14378 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14379 * combine.c (simplify_comparison): Likewise
14380 * config/i386/i386.c (type_natural_mode): Likewise.
14381 * cse.c (cse_insn): Likewise.
14382 * dse.c (find_shift_sequence): Likewise.
14383 * emit-rtl.c (init_derived_machine_modes): Likewise.
14384 (init_emit_once): Likewise.
14385 * explow.c (hard_function_value): Likewise.
14386 * expmed.c (extract_fixed_bit_field_1): Likewise.
14387 (extract_bit_field_1): Likewise.
14388 (expand_divmod): Likewise.
14389 (emit_store_flag_1): Likewise.
14390 * expr.c (init_expr_target): Likewise.
14391 (convert_move): Likewise.
14392 (alignment_for_piecewise_move): Likewise.
14393 (widest_int_mode_for_size): Likewise.
14394 (emit_block_move_via_movmem): Likewise.
14395 (copy_blkmode_to_reg): Likewise.
14396 (set_storage_via_setmem): Likewise.
14397 (compress_float_constant): Likewise.
14398 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14399 * optabs-query.c (get_best_extraction_insn): Likewise.
14400 * optabs.c (expand_binop): Likewise.
14401 (expand_twoval_unop): Likewise.
14402 (expand_twoval_binop): Likewise.
14403 (widen_leading): Likewise.
14404 (widen_bswap): Likewise.
14405 (expand_parity): Likewise.
14406 (expand_unop): Likewise.
14407 (prepare_cmp_insn): Likewise.
14408 (prepare_float_lib_cmp): Likewise.
14409 (expand_float): Likewise.
14410 (expand_fix): Likewise.
14411 (expand_sfix_optab): Likewise.
14412 * postreload.c (move2add_use_add2_insn): Likewise.
14413 * reg-stack.c (reg_to_stack): Likewise.
14414 * reginfo.c (choose_hard_reg_mode): Likewise.
14415 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14416 * stor-layout.c (mode_for_size): Likewise.
14417 (smallest_mode_for_size): Likewise.
14418 (mode_for_vector): Likewise.
14419 (finish_bitfield_representative): Likewise.
14420 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14421 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14422 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14423 * var-tracking.c (prepare_call_arguments): Likewise.
14425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14426 Alan Hayward <alan.hayward@arm.com>
14427 David Sherwood <david.sherwood@arm.com>
14429 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14430 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14431 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14432 * machmode.h (mode_size): Move earlier in file.
14433 (mode_precision): Likewise.
14434 (mode_inner): Likewise.
14435 (mode_nunits): Likewise.
14436 (mode_unit_size): Likewise.
14437 (unit_unit_precision): Likewise.
14438 (mode_wider): Likewise.
14439 (mode_2xwider): Likewise.
14440 (machine_mode): New class.
14441 (mode_to_bytes): New function.
14442 (mode_to_bits): Likewise.
14443 (mode_to_precision): Likewise.
14444 (mode_to_inner): Likewise.
14445 (mode_to_unit_size): Likewise.
14446 (mode_to_unit_precision): Likewise.
14447 (mode_to_nunits): Likewise.
14448 (GET_MODE_SIZE): Use mode_to_bytes.
14449 (GET_MODE_BITSIZE): Use mode_to_bits.
14450 (GET_MODE_PRECISION): Use mode_to_precision.
14451 (GET_MODE_INNER): Use mode_to_inner.
14452 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14453 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14454 (GET_MODE_NUNITS): Use mode_to_nunits.
14455 * system.h (ALWAYS_INLINE): New macro.
14456 * config/powerpcspe/powerpcspe-c.c
14457 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14458 int for arg1_mode and arg2_mode.
14460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14461 Alan Hayward <alan.hayward@arm.com>
14462 David Sherwood <david.sherwood@arm.com>
14464 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14465 Prefix mode names with E_ in case statements.
14466 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14467 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14468 (aarch64_split_simd_move): Likewise.
14469 (aarch64_gen_storewb_pair): Likewise.
14470 (aarch64_gen_loadwb_pair): Likewise.
14471 (aarch64_gen_store_pair): Likewise.
14472 (aarch64_gen_load_pair): Likewise.
14473 (aarch64_get_condition_code_1): Likewise.
14474 (aarch64_constant_pool_reload_icode): Likewise.
14475 (get_rsqrte_type): Likewise.
14476 (get_rsqrts_type): Likewise.
14477 (get_recpe_type): Likewise.
14478 (get_recps_type): Likewise.
14479 (aarch64_gimplify_va_arg_expr): Likewise.
14480 (aarch64_simd_container_mode): Likewise.
14481 (aarch64_emit_load_exclusive): Likewise.
14482 (aarch64_emit_store_exclusive): Likewise.
14483 (aarch64_expand_compare_and_swap): Likewise.
14484 (aarch64_gen_atomic_cas): Likewise.
14485 (aarch64_emit_bic): Likewise.
14486 (aarch64_emit_atomic_swap): Likewise.
14487 (aarch64_emit_atomic_load_op): Likewise.
14488 (aarch64_evpc_trn): Likewise.
14489 (aarch64_evpc_uzp): Likewise.
14490 (aarch64_evpc_zip): Likewise.
14491 (aarch64_evpc_ext): Likewise.
14492 (aarch64_evpc_rev): Likewise.
14493 (aarch64_evpc_dup): Likewise.
14494 (aarch64_gen_ccmp_first): Likewise.
14495 (aarch64_gen_ccmp_next): Likewise.
14496 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14497 (alpha_emit_xfloating_libcall): Likewise.
14498 (emit_insxl): Likewise.
14499 (alpha_arg_type): Likewise.
14500 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14501 (arc_preferred_simd_mode): Likewise.
14502 (arc_secondary_reload): Likewise.
14503 (get_arc_condition_code): Likewise.
14504 (arc_print_operand): Likewise.
14505 (arc_legitimate_constant_p): Likewise.
14506 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14507 * config/arc/arc.md (casesi_load): Likewise.
14508 (casesi_compact_jump): Likewise.
14509 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14510 (cc_use_register): Likewise.
14511 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14512 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14513 (arm_init_iwmmxt_builtins): Likewise.
14514 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14515 (neon_expand_vector_init): Likewise.
14516 (arm_attr_length_move_neon): Likewise.
14517 (maybe_get_arm_condition_code): Likewise.
14518 (arm_emit_vector_const): Likewise.
14519 (arm_preferred_simd_mode): Likewise.
14520 (arm_output_iwmmxt_tinsr): Likewise.
14521 (thumb1_output_casesi): Likewise.
14522 (thumb2_output_casesi): Likewise.
14523 (arm_emit_load_exclusive): Likewise.
14524 (arm_emit_store_exclusive): Likewise.
14525 (arm_expand_compare_and_swap): Likewise.
14526 (arm_evpc_neon_vuzp): Likewise.
14527 (arm_evpc_neon_vzip): Likewise.
14528 (arm_evpc_neon_vrev): Likewise.
14529 (arm_evpc_neon_vtrn): Likewise.
14530 (arm_evpc_neon_vext): Likewise.
14531 (arm_validize_comparison): Likewise.
14532 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14533 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14534 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14535 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14536 (c6x_preferred_simd_mode): Likewise.
14537 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14538 (epiphany_rtx_costs): Likewise.
14539 * config/epiphany/predicates.md (proper_comparison_operator):
14541 * config/frv/frv.c (condexec_memory_operand): Likewise.
14542 (frv_emit_move): Likewise.
14543 (output_move_single): Likewise.
14544 (output_condmove_single): Likewise.
14545 (frv_hard_regno_mode_ok): Likewise.
14546 (frv_matching_accg_mode): Likewise.
14547 * config/h8300/h8300.c (split_adds_subs): Likewise.
14548 (h8300_rtx_costs): Likewise.
14549 (h8300_print_operand): Likewise.
14550 (compute_mov_length): Likewise.
14551 (output_logical_op): Likewise.
14552 (compute_logical_op_length): Likewise.
14553 (compute_logical_op_cc): Likewise.
14554 (h8300_shift_needs_scratch_p): Likewise.
14555 (output_a_shift): Likewise.
14556 (compute_a_shift_length): Likewise.
14557 (compute_a_shift_cc): Likewise.
14558 (expand_a_rotate): Likewise.
14559 (output_a_rotate): Likewise.
14560 * config/i386/i386.c (classify_argument): Likewise.
14561 (function_arg_advance_32): Likewise.
14562 (function_arg_32): Likewise.
14563 (function_arg_64): Likewise.
14564 (function_value_64): Likewise.
14565 (ix86_gimplify_va_arg): Likewise.
14566 (ix86_legitimate_constant_p): Likewise.
14567 (put_condition_code): Likewise.
14568 (split_double_mode): Likewise.
14569 (ix86_avx256_split_vector_move_misalign): Likewise.
14570 (ix86_expand_vector_logical_operator): Likewise.
14571 (ix86_split_idivmod): Likewise.
14572 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14573 (ix86_build_const_vector): Likewise.
14574 (ix86_build_signbit_mask): Likewise.
14575 (ix86_match_ccmode): Likewise.
14576 (ix86_cc_modes_compatible): Likewise.
14577 (ix86_expand_branch): Likewise.
14578 (ix86_expand_sse_cmp): Likewise.
14579 (ix86_expand_sse_movcc): Likewise.
14580 (ix86_expand_int_sse_cmp): Likewise.
14581 (ix86_expand_vec_perm_vpermi2): Likewise.
14582 (ix86_expand_vec_perm): Likewise.
14583 (ix86_expand_sse_unpack): Likewise.
14584 (ix86_expand_int_addcc): Likewise.
14585 (ix86_split_to_parts): Likewise.
14586 (ix86_vectorize_builtin_gather): Likewise.
14587 (ix86_vectorize_builtin_scatter): Likewise.
14588 (avx_vpermilp_parallel): Likewise.
14589 (inline_memory_move_cost): Likewise.
14590 (ix86_tieable_integer_mode_p): Likewise.
14591 (x86_maybe_negate_const_int): Likewise.
14592 (ix86_expand_vector_init_duplicate): Likewise.
14593 (ix86_expand_vector_init_one_nonzero): Likewise.
14594 (ix86_expand_vector_init_one_var): Likewise.
14595 (ix86_expand_vector_init_concat): Likewise.
14596 (ix86_expand_vector_init_interleave): Likewise.
14597 (ix86_expand_vector_init_general): Likewise.
14598 (ix86_expand_vector_set): Likewise.
14599 (ix86_expand_vector_extract): Likewise.
14600 (emit_reduc_half): Likewise.
14601 (ix86_emit_i387_round): Likewise.
14602 (ix86_mangle_type): Likewise.
14603 (ix86_expand_round_sse4): Likewise.
14604 (expand_vec_perm_blend): Likewise.
14605 (canonicalize_vector_int_perm): Likewise.
14606 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14607 (expand_vec_perm_1): Likewise.
14608 (expand_vec_perm_interleave3): Likewise.
14609 (expand_vec_perm_even_odd_pack): Likewise.
14610 (expand_vec_perm_even_odd_1): Likewise.
14611 (expand_vec_perm_broadcast_1): Likewise.
14612 (ix86_vectorize_vec_perm_const_ok): Likewise.
14613 (ix86_expand_vecop_qihi): Likewise.
14614 (ix86_expand_mul_widen_hilo): Likewise.
14615 (ix86_expand_sse2_abs): Likewise.
14616 (ix86_expand_pextr): Likewise.
14617 (ix86_expand_pinsr): Likewise.
14618 (ix86_preferred_simd_mode): Likewise.
14619 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14620 * config/i386/sse.md (*andnot<mode>3): Likewise.
14621 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14622 (*<code><mode>3): Likewise.
14623 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14624 (ia64_expand_atomic_op): Likewise.
14625 (ia64_arg_type): Likewise.
14626 (ia64_mode_to_int): Likewise.
14627 (ia64_scalar_mode_supported_p): Likewise.
14628 (ia64_vector_mode_supported_p): Likewise.
14629 (expand_vec_perm_broadcast): Likewise.
14630 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14631 (iq2000_function_arg_advance): Likewise.
14632 (iq2000_function_arg): Likewise.
14633 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14634 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14635 (m68k_libcall_value): Likewise.
14636 (m68k_function_value): Likewise.
14637 (sched_attr_op_type): Likewise.
14638 * config/mcore/mcore.c (mcore_output_move): Likewise.
14639 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14641 (microblaze_function_arg): Likewise.
14642 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14643 (mips_print_operand): Likewise.
14644 (mips_mode_ok_for_mov_fmt_p): Likewise.
14645 (mips_vector_mode_supported_p): Likewise.
14646 (mips_preferred_simd_mode): Likewise.
14647 (mips_expand_vpc_loongson_even_odd): Likewise.
14648 (mips_expand_vec_unpack): Likewise.
14649 (mips_expand_vi_broadcast): Likewise.
14650 (mips_expand_vector_init): Likewise.
14651 (mips_expand_vec_reduc): Likewise.
14652 (mips_expand_msa_cmp): Likewise.
14653 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14654 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14655 (cc_flags_for_mode): Likewise.
14656 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14657 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14658 (nds32_output_casesi_pc_relative): Likewise.
14659 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14660 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14661 (nvptx_gen_unpack): Likewise.
14662 (nvptx_gen_pack): Likewise.
14663 (nvptx_gen_shuffle): Likewise.
14664 (nvptx_gen_wcast): Likewise.
14665 (nvptx_preferred_simd_mode): Likewise.
14666 * config/pa/pa.c (pa_secondary_reload): Likewise.
14667 * config/pa/predicates.md (base14_operand): Likewise.
14668 * config/powerpcspe/powerpcspe-c.c
14669 (altivec_resolve_overloaded_builtin): Likewise.
14670 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14672 (rs6000_preferred_simd_mode): Likewise.
14673 (output_vec_const_move): Likewise.
14674 (rs6000_expand_vector_extract): Likewise.
14675 (rs6000_split_vec_extract_var): Likewise.
14676 (reg_offset_addressing_ok_p): Likewise.
14677 (rs6000_legitimate_offset_address_p): Likewise.
14678 (rs6000_legitimize_address): Likewise.
14679 (rs6000_emit_set_const): Likewise.
14680 (rs6000_const_vec): Likewise.
14681 (rs6000_emit_move): Likewise.
14682 (spe_build_register_parallel): Likewise.
14683 (rs6000_darwin64_record_arg_recurse): Likewise.
14684 (swap_selector_for_mode): Likewise.
14685 (spe_init_builtins): Likewise.
14686 (paired_init_builtins): Likewise.
14687 (altivec_init_builtins): Likewise.
14688 (do_load_for_compare): Likewise.
14689 (rs6000_generate_compare): Likewise.
14690 (rs6000_expand_float128_convert): Likewise.
14691 (emit_load_locked): Likewise.
14692 (emit_store_conditional): Likewise.
14693 (rs6000_output_function_epilogue): Likewise.
14694 (rs6000_handle_altivec_attribute): Likewise.
14695 (rs6000_function_value): Likewise.
14696 (emit_fusion_gpr_load): Likewise.
14697 (emit_fusion_p9_load): Likewise.
14698 (emit_fusion_p9_store): Likewise.
14699 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14700 (fusion_gpr_mem_load): Likewise.
14701 (fusion_addis_mem_combo_load): Likewise.
14702 (fusion_addis_mem_combo_store): Likewise.
14703 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14704 (fusion_gpr_mem_load): Likewise.
14705 (fusion_addis_mem_combo_load): Likewise.
14706 (fusion_addis_mem_combo_store): Likewise.
14707 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14709 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14710 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14711 (rs6000_preferred_simd_mode): Likewise.
14712 (output_vec_const_move): Likewise.
14713 (rs6000_expand_vector_extract): Likewise.
14714 (rs6000_split_vec_extract_var): Likewise.
14715 (reg_offset_addressing_ok_p): Likewise.
14716 (rs6000_legitimate_offset_address_p): Likewise.
14717 (rs6000_legitimize_address): Likewise.
14718 (rs6000_emit_set_const): Likewise.
14719 (rs6000_const_vec): Likewise.
14720 (rs6000_emit_move): Likewise.
14721 (rs6000_darwin64_record_arg_recurse): Likewise.
14722 (swap_selector_for_mode): Likewise.
14723 (paired_init_builtins): Likewise.
14724 (altivec_init_builtins): Likewise.
14725 (rs6000_expand_float128_convert): Likewise.
14726 (emit_load_locked): Likewise.
14727 (emit_store_conditional): Likewise.
14728 (rs6000_output_function_epilogue): Likewise.
14729 (rs6000_handle_altivec_attribute): Likewise.
14730 (rs6000_function_value): Likewise.
14731 (emit_fusion_gpr_load): Likewise.
14732 (emit_fusion_p9_load): Likewise.
14733 (emit_fusion_p9_store): Likewise.
14734 * config/rx/rx.c (rx_gen_move_template): Likewise.
14735 (flags_from_mode): Likewise.
14736 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14737 (s390_slb_comparison): Likewise.
14738 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14739 (s390_vector_mode_supported_p): Likewise.
14740 (s390_cc_modes_compatible): Likewise.
14741 (s390_match_ccmode_set): Likewise.
14742 (s390_canonicalize_comparison): Likewise.
14743 (s390_emit_compare_and_swap): Likewise.
14744 (s390_branch_condition_mask): Likewise.
14745 (s390_rtx_costs): Likewise.
14746 (s390_secondary_reload): Likewise.
14747 (__SECONDARY_RELOAD_CASE): Likewise.
14748 (s390_expand_cs): Likewise.
14749 (s390_preferred_simd_mode): Likewise.
14750 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14751 * config/sh/sh.c (sh_print_operand): Likewise.
14752 (dump_table): Likewise.
14753 (sh_secondary_reload): Likewise.
14754 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14755 * config/sh/sh.md (casesi_worker_1): Likewise.
14756 (casesi_worker_2): Likewise.
14757 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14758 (fcc_comparison_operator): Likewise.
14759 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14760 (emit_soft_tfmode_cvt): Likewise.
14761 (sparc_preferred_simd_mode): Likewise.
14762 (output_cbranch): Likewise.
14763 (sparc_print_operand): Likewise.
14764 (sparc_expand_vec_perm_bmask): Likewise.
14765 (vector_init_bshuffle): Likewise.
14766 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14767 (spu_vector_mode_supported_p): Likewise.
14768 (spu_expand_insv): Likewise.
14769 (spu_emit_branch_or_set): Likewise.
14770 (spu_handle_vector_attribute): Likewise.
14771 (spu_builtin_splats): Likewise.
14772 (spu_builtin_extract): Likewise.
14773 (spu_builtin_promote): Likewise.
14774 (spu_expand_sign_extend): Likewise.
14775 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14776 (tilegx_simd_int): Likewise.
14777 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14778 (tilepro_simd_int): Likewise.
14779 * config/v850/v850.c (const_double_split): Likewise.
14780 (v850_print_operand): Likewise.
14781 (ep_memory_offset): Likewise.
14782 * config/vax/vax.c (vax_rtx_costs): Likewise.
14783 (vax_output_int_move): Likewise.
14784 (vax_output_int_add): Likewise.
14785 (vax_output_int_subtract): Likewise.
14786 * config/visium/predicates.md (visium_branch_operator): Likewise.
14787 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14788 (visium_print_operand_address): Likewise.
14789 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14790 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14791 (xtensa_expand_conditional_branch): Likewise.
14792 (xtensa_copy_incoming_a7): Likewise.
14793 (xtensa_output_literal): Likewise.
14794 * dfp.c (decimal_real_maxval): Likewise.
14795 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14798 Alan Hayward <alan.hayward@arm.com>
14799 David Sherwood <david.sherwood@arm.com>
14801 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14802 (mode_nunits_inline): Likewise.
14803 (mode_inner_inline): Likewise.
14804 (mode_unit_size_inline): Likewise.
14805 (mode_unit_precision_inline): Likewise.
14806 (emit_insn_modes_h): Likewise. Also emit a #define of the
14808 (emit_mode_wider): Add an E_ prefix to mode names.
14809 (emit_mode_complex): Likewise.
14810 (emit_mode_inner): Likewise.
14811 (emit_mode_adjustments): Likewise.
14812 (emit_mode_int_n): Likewise.
14813 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14814 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14815 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14816 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14817 (CRC32_BUILTIN, ENTRY): Likewise.
14818 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14819 (aarch64_pop_regs): Likewise.
14820 (aarch64_process_components): Likewise.
14821 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14822 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14823 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14824 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14825 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14826 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14828 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14829 (pcmpistr): Likewise.
14830 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14831 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14832 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14834 * config/rl78/rl78.c (mduc_regs): Likewise.
14835 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14836 (htm_expand_builtin): Likewise.
14837 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14838 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14839 * config/xtensa/xtensa.c (print_operand): Likewise.
14840 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14841 (NUM_MODE_VECTOR_INT): Likewise.
14842 * genoutput.c (null_operand): Likewise.
14843 (output_operand_data): Likewise.
14844 * genrecog.c (print_parameter_value): Likewise.
14845 * lra.c (debug_operand_data): Likewise.
14847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14849 * dwarf2out.c (scompare_loc_descriptor_wide)
14850 (scompare_loc_descriptor_narrow): New functions, split out from...
14851 (scompare_loc_descriptor): ...here.
14852 * expmed.c (emit_store_flag_int): New function, split out from...
14853 (emit_store_flag): ...here.
14855 2017-08-30 Richard Biener <rguenther@suse.de>
14857 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14858 (dwarf2out_early_finish): Move setting of AT_pubnames from
14859 early debug output to early finish.
14861 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14863 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14864 and -mdata-region to the assembler.
14866 2017-08-30 Richard Biener <rguenther@suse.de>
14868 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14870 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14871 (dwarf2out_early_global_decl): Only generate a DIE for the
14872 abstract origin if it doesn't already exist or is a declaration DIE.
14873 (resolve_addr): Do not add the linkage name twice when
14874 generating a stub DIE for the DW_TAG_GNU_call_site target.
14876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14878 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14879 Use machine_mode rather than int for arg1_mode.
14881 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14884 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14885 that the second argument of the built-in functions to unpack
14886 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14887 switch statement instead a lot of if statements.
14888 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14889 Allow 64-bit values to be in Altivec registers as well as
14890 traditional floating point registers.
14891 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14893 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14895 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14896 MAX_REGS_PER_ADDRESS == 1.
14898 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14900 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14901 * config/i386/i386.c (ix86_option_override_internal): Simplify
14902 setting of opts->x_flag_entry.
14904 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14905 Jakub Jelinek <jakub@redhat.com>
14906 Richard Biener <rguenther@suse.de>
14908 PR tree-optimization/81503
14909 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14910 folded constant fits in the target type; reorder tests for clarity.
14912 2017-08-29 Martin Liska <mliska@suse.cz>
14914 * passes.def: Include pass_lower_switch.
14915 * stmt.c (dump_case_nodes): Remove and move to
14916 tree-switch-conversion.
14917 (case_values_threshold): Likewise.
14918 (expand_switch_as_decision_tree_p): Likewise.
14919 (emit_case_decision_tree): Likewise.
14920 (expand_case): Likewise.
14921 (balance_case_nodes): Likewise.
14922 (node_has_low_bound): Likewise.
14923 (node_has_high_bound): Likewise.
14924 (node_is_bounded): Likewise.
14925 (emit_case_nodes): Likewise.
14926 (struct simple_case_node): New struct.
14927 (add_case_node): Remove.
14928 (emit_case_dispatch_table): Use vector instead of case_list.
14929 (reset_out_edges_aux): Remove.
14930 (compute_cases_per_edge): Likewise.
14931 (expand_case): Build list of simple_case_node.
14932 (expand_sjlj_dispatch_table): Use it.
14933 * tree-switch-conversion.c (struct case_node): Moved from
14934 stmt.c and adjusted.
14935 (emit_case_nodes): Likewise.
14936 (node_has_low_bound): Likewise.
14937 (node_has_high_bound): Likewise.
14938 (node_is_bounded): Likewise.
14939 (case_values_threshold): Likewise.
14940 (reset_out_edges_aux): Likewise.
14941 (compute_cases_per_edge): Likewise.
14942 (add_case_node): Likewise.
14943 (dump_case_nodes): Likewise.
14944 (balance_case_nodes): Likewise.
14945 (expand_switch_as_decision_tree_p): Likewise.
14946 (emit_jump): Likewise.
14947 (emit_case_decision_tree): Likewise.
14948 (try_switch_expansion): Likewise.
14949 (do_jump_if_equal): Likewise.
14950 (emit_cmp_and_jump_insns): Likewise.
14951 (fix_phi_operands_for_edge): New function.
14952 (record_phi_operand_mapping): Likewise.
14953 (class pass_lower_switch): New pass.
14954 (pass_lower_switch::execute): New function.
14955 (make_pass_lower_switch): Likewise.
14956 (conditional_probability):
14957 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14958 * tree-pass.h: Add make_pass_lower_switch.
14960 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14963 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14966 2017-08-29 Richard Biener <rguenther@suse.de>
14968 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14969 we do not add a DW_AT_inline attribute twice.
14970 (gen_subprogram_die): Remove code setting DW_AT_inline on
14971 DECL_ABSTRACT_P nodes.
14973 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14975 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14976 calls to internal functions.
14977 (gimplify_modify_expr): Likewise.
14978 * tree-call-cdce.c (use_internal_fn): Likewise.
14979 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14980 (convert_to_divmod): Set the nothrow flag.
14981 * tree-if-conv.c (predicate_mem_writes): Likewise.
14982 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14983 (vectorizable_call): Likewise.
14984 (vectorizable_store): Likewise.
14985 (vectorizable_load): Likewise.
14986 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14987 (vect_recog_mask_conversion_pattern): Likewise.
14989 2017-08-29 Martin Liska <mliska@suse.cz>
14992 * gcc.c (driver_handle_option): Add new argument.
14993 * opts-common.c (handle_option): Pass
14994 target_option_override_hook.
14995 * opts-global.c (lang_handle_option): Add new option.
14996 (set_default_handlers): Add new argument.
14997 (decode_options): Likewise.
14998 * opts.c (target_handle_option): Likewise.
14999 (common_handle_option): Call target_option_override_hook.
15000 * opts.h (struct cl_option_handler_func): Add hook for
15001 target option override.
15002 (struct cl_option_handlers): Likewise.
15003 (set_default_handlers): Add new argument.
15004 (decode_options): Likewise.
15005 (common_handle_option): Likewise.
15006 (target_handle_option): Likewise.
15007 * toplev.c (toplev::main): Pass targetm.target_option.override
15010 2017-08-29 Richard Biener <rguenther@suse.de>
15011 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15013 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15014 life to the active subtree.
15016 2017-08-28 Jeff Law <law@redhat.com>
15018 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15019 derive_equivalences.
15020 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15021 Code moved into....
15022 (edge_info::derive_equivalences): New private member function
15024 * tree-ssa-dom.c (class edge_info): Changed from a struct
15025 to a class. Add ctor/dtor, methods and data members.
15026 (edge_info::edge_info): Renamed from allocate_edge_info.
15027 Initialize additional members.
15028 (edge_info::~edge_info): New.
15029 (free_dom_edge_info): Delete the edge info.
15030 (record_edge_info): Use new class & associated member functions.
15031 Tighten forms for testing for edge equivalences.
15032 (record_temporary_equivalences): Iterate over the simple
15033 equivalences rather than assuming there's only one per edge.
15034 (cprop_into_successor_phis): Iterate over the simple
15035 equivalences rather than assuming there's only one per edge.
15036 (optimize_stmt): Use operand_equal_p rather than pointer
15037 equality for mini-DSE code.
15039 2017-08-28 Nathan Sidwell <nathan@acm.org>
15041 * gcc.c (execute): Fold SIGPIPE handling into switch
15042 statement. Adjust internal error message.
15044 2017-08-28 Richard Biener <rguenther@suse.de>
15047 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15048 Do nothing for removed DIEs.
15050 2017-08-28 Richard Biener <rguenther@suse.de>
15052 PR tree-optimization/81977
15053 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15056 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15059 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15060 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15061 using targetm.gen_mem_thread_fence.
15063 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15066 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15067 predicate to register_operand. Reorder operands.
15068 (*btr<mode>): Ditto.
15069 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15070 (*btr<mode>_mask): Ditto.
15072 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15074 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15075 * config/rs6000/xmmintrin.h: New file.
15076 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15078 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15081 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15082 parameter and_insn and return it.
15083 (recombine_lvx_pattern): Insert a copy to ensure availability of
15084 the base register of the copied masking operation at the point of
15085 the instruction replacement.
15086 (recombine_stvx_pattern): Likewise.
15088 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15090 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15091 undocumented switches.
15092 (-mpower9-dform-vector): Likewise.
15093 (-mpower9-dform): Likewise.
15094 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15095 comments to delete references to -mpower9-dform* switches.
15096 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15097 Delete reference to -mpower9-dform* switches, test for
15098 -mpower9-vector instead.
15099 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15100 (OTHER_P9_VECTOR_MASKS): Likewise.
15101 (POWERPC_MASKS): Likewise.
15102 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15103 tests against -mpower9-dform* to -mpower9-vector. Delete code
15104 that checked for -mpower9-dform* consistancy with other options.
15105 Add test for -mpower9-misc to enable other power9 switches.
15106 (rs6000_init_hard_regno_mode_ok): Likewise.
15107 (rs6000_option_override_internal): Likewise.
15108 (rs6000_emit_prologue): Likewise.
15109 (rs6000_emit_epilogue): Likewise.
15110 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15111 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15112 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15114 (emit_fusion_p9_store): Likewise.
15115 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15116 resetting these macros if the assembler does not support ISA 3.0
15118 (TARGET_P9_DFORM_VECTOR): Likewise.
15119 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15120 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15122 2017-08-25 Alan Modra <amodra@gmail.com>
15124 PR rtl-optimization/81747
15125 * cse.c (cse_extended_basic_block): Don't attempt to record
15126 equivalences for degenerate conditional branches that branch
15127 to their fall-through.
15129 2017-08-24 Martin Sebor <msebor@redhat.com>
15131 PR middle-end/81908
15132 * gimple-fold.c (size_must_be_zero_p): New function.
15133 (gimple_fold_builtin_memory_op): Call it.
15135 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15137 * config/rs6000/mm_malloc.h: New file.
15139 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15141 PR tree-optimization/81913
15142 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15143 analysis when either IVs in condition can wrap.
15145 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15147 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15148 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15150 2017-08-24 Richard Biener <rguenther@suse.de>
15153 * targhooks.c (default_target_can_inline_p): Properly
15154 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15155 is present and always compare.
15156 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15157 infer -mfpmath=sse from TARGET_SSE_P.
15158 (ix86_can_inline_p): Properly use target_option_default_node when
15159 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15161 2017-08-24 Richard Biener <rguenther@suse.de>
15164 * dwarf2out.c (output_die): Handle flag_generate_offload like
15166 (output_comp_unit): Likewise.
15167 (gen_array_type_die): Likewise.
15168 (dwarf2out_early_finish): Likewise.
15169 (note_variable_value_in_expr): Likewise.
15170 (dwarf2out_finish): Likewise. Adjust assert.
15171 * cgraphunit.c (symbol_table::compile): Move setting of
15172 flag_generate_offload earlier ...
15173 (symbol_table::finalize_compilation_unit): ... here, before
15174 early debug finalization.
15176 2017-08-24 Richard Biener <rguenther@suse.de>
15178 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15179 and ipa-fnsummary.h.
15180 (ix86_can_inline_p): When ix86_fpmath flags do not match
15181 check whether the callee uses FP math at all.
15183 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15185 PR middle-end/81931
15186 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15187 instead of TYPE_PRECISION.
15189 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15190 Alan Hayward <alan.hayward@arm.com>
15191 David Sherwood <david.sherwood@arm.com>
15193 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15195 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15197 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15199 (ix86_frame::stack_realign_allocate): New field.
15200 (struct machine_frame_state): Modify comments.
15201 (machine_frame_state::sp_realigned_fp_end): New field.
15202 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15203 layout calculation.
15204 (sp_valid_at): Add assertion to assure no attempt to access invalid
15205 offset of a realigned stack.
15206 (fp_valid_at): Likewise.
15207 (choose_baseaddr): Modify comments.
15208 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15209 ix86_expand_prologue.
15210 (ix86_expand_prologue): Modify stack realignment and allocation.
15211 (ix86_expand_epilogue): Modify comments.
15212 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15213 avx2_runtime, avx512f, and avx512f_runtime.
15215 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15217 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15218 (mstackrealign): Do not init to -1.
15219 * config/i386/i386.c (ix86_option_override_internal):
15220 Check opts_set, not opts when setting default value of
15221 opts->x_ix86_force_align_arg_pointer.
15223 2017-08-23 Richard Biener <rguenther@suse.de>
15225 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15226 lang_hooks.decl_printable_name.
15227 * print-rtl-function.c (print_rtx_function): Likewise.
15228 * tree-pretty-print.c (dump_function_header): Likewise.
15230 2017-08-23 Richard Biener <rguenther@suse.de>
15233 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15234 -g0 at compile-time.
15236 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15238 PR middle-end/19706
15239 * doc/sourcebuild.texi (Other hardware attributes):
15242 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15244 PR middle-end/19706
15245 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15246 Fix single-use check.
15248 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15250 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15252 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15254 * doc/install.texi: Modify to add more details on running selected
15257 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15259 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15260 is combined with -mabi=ms.
15261 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15264 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15266 PR tree-optimization/81488
15267 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15268 and cached_basis fields.
15269 (MAX_SPREAD): New constant.
15270 (alloc_cand_and_find_basis): Initialize new fields.
15271 (clear_visited): New function.
15272 (create_phi_basis_1): Rename from create_phi_basis, set visited
15273 and cached_basis fields.
15274 (create_phi_basis): New wrapper function.
15275 (phi_add_costs_1): Rename from phi_add_costs, add spread
15276 parameter, set visited field, short-circuit when limits reached.
15277 (phi_add_costs): New wrapper function.
15278 (record_phi_increments_1): Rename from record_phi_increments, set
15280 (record_phi_increments): New wrapper function.
15281 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15282 (phi_incr_cost): New wrapper function.
15283 (all_phi_incrs_profitable_1): Rename from
15284 all_phi_incrs_profitable, set visited field.
15285 (all_phi_incrs_profitable): New wrapper function.
15287 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15288 Alan Hayward <alan.hayward@arm.com>
15289 David Sherwood <david.sherwood@arm.com>
15291 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15292 that takes the outer and inner modes.
15293 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15294 comparison as the canonical test for a paradoxical subreg.
15295 * combine.c (simplify_set): Use paradoxical_subreg_p.
15296 (make_extraction): Likewise.
15297 (force_to_mode): Likewise.
15298 (rtx_equal_for_field_assignment_p): Likewise.
15299 (gen_lowpart_for_combine): Likewise.
15300 (simplify_comparison): Likewise.
15301 * cse.c (equiv_constant): Likewise.
15302 * expmed.c (store_bit_field_1): Likewise.
15303 * final.c (alter_subreg): Likewise.
15304 * fwprop.c (propagate_rtx): Likewise.
15305 (forward_propagate_subreg): Likewise.
15306 * ira-conflicts.c (ira_build_conflicts): Likewise.
15307 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15308 * lra-constraints.c (curr_insn_transform): Likewise.
15309 (split_reg): Likewise.
15310 * lra-eliminations.c (move_plus_up): Likewise.
15311 (lra_eliminate_regs_1): Likewise.
15312 * recog.c (general_operand): Likewise.
15313 * ree.c (combine_reaching_defs): Likewise.
15314 * reload.c (push_reload): Likewise.
15315 (find_reloads): Likewise.
15316 * reload1.c (elimination_effects): Likewise.
15317 (compute_reload_subreg_offset): Likewise.
15318 (choose_reload_regs): Likewise.
15319 * rtlanal.c (subreg_lsb_1): Likewise.
15320 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15321 (simplify_subreg): Likewise.
15322 * var-tracking.c (track_loc_p): Likewise.
15323 * emit-rtl.c (byte_lowpart_offset): Likewise.
15324 (paradoxical_subreg_p): Delete out-of-line definition.
15326 2017-08-22 Jeff Law <law@redhat.com>
15328 PR tree-optimization/81741
15329 PR tree-optimization/71947
15330 * tree-ssa-dom.c: Include tree-inline.h.
15331 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15332 equivalences if one is more expensive to compute than the other.
15333 * tree-ssa-scopedtables.h (class const_or_copies): Make
15334 record_const_or_copy_raw method private.
15335 (class avail_exprs_stack): New method simplify_binary_operation.
15336 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15337 avail_exprs_stack::simplify_binary_operation as needed.
15338 (avail_exprs_stack::simplify_binary_operation): New function.
15340 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15342 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15343 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15344 (DOT_SYMBOLS): Likewise.
15345 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15346 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15347 (RS6000_ABI_NAME): Likewise.
15348 (TARGET_CMODEL): Likewise.
15349 (TOC_SECTION_ASM_OP): Likewise.
15350 (SET_CMODEL): New macro.
15351 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15353 2017-08-22 Richard Biener <rguenther@suse.de>
15355 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15356 to allow for free-lang-data replacements similar to verify_type_variant.
15358 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15359 Alan Hayward <alan.hayward@arm.com>
15360 David Sherwood <david.sherwood@arm.com>
15362 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15363 VOIDmode for the LABEL_REF.
15365 2017-08-22 Richard Biener <rguenther@suse.de>
15367 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15369 2017-08-22 Richard Biener <rguenther@suse.de>
15371 * common.opt (feliminate-dwarf2-dups): Ignore.
15372 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15373 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15374 same_die_p_wrap, compute_section_prefix,
15375 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15376 (comdat_symbol_id, comdat_symbol_number): Likewise.
15377 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15379 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15380 (output_die): Mark unreachable path unreachable.
15381 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15382 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15383 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15384 (dwarf2out_early_finish): Likewise.
15386 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15388 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15390 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15393 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15394 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15395 (avr_attribute_table) <io, io_low, address>: Initialize
15396 .decl_required with true.
15398 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15400 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15401 undocumented debugging options.
15402 (-mvsx-scalar-double): Likewise.
15403 (-mallow-df-permute): Likewise.
15404 (-mvectorize-builtins): Likewise.
15405 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15406 (rs6000_builtin_vectorized_function): Likewise.
15407 (rs6000_builtin_md_vectorized_function): Likewise.
15408 (rs6000_opt_vars): Likewise.
15410 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15413 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15414 (*btrq_imm): Rename from *btrq.
15415 (*btcq_imm): Rename from *btcq.
15416 (btsc): New code attribute.
15417 (*<btsc><mode>): New insn pattern.
15418 (*btr<mode>): Ditto.
15419 (*<btsc><mode>_mask): New insn_and_split pattern.
15420 (*btr<mode>_mask): Ditto.
15422 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15423 Alan Hayward <alan.hayward@arm.com>
15424 David Sherwood <david.sherwood@arm.com>
15426 * function.c (pad_below): Simplify padding calculation.
15428 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15429 Alan Hayward <alan.hayward@arm.com>
15430 David Sherwood <david.sherwood@arm.com>
15432 * target.def (function_prologue): Remove frame size argument.
15433 (function_epilogue): Likewise.
15434 * doc/tm.texi: Regenerate.
15435 * final.c (final_start_function): Update call to function_prologue.
15436 (final_end_function): Update call to function_epilogue.
15437 (default_function_pro_epilogue): Remove frame size argument.
15438 * output.h (default_function_pro_epilogue): Likewise.
15439 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15440 (arm_output_function_prologue): Likewise.
15441 * config/frv/frv.c (frv_function_prologue): Likewise.
15442 (frv_function_epilogue): Likewise.
15443 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15444 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15445 (ia64_output_function_epilogue): Likewise.
15446 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15447 (m32r_output_function_epilogue): Likewise.
15448 * config/microblaze/microblaze.c (microblaze_function_prologue)
15449 (microblaze_function_epilogue): Likewise.
15450 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15451 (mips_output_function_epilogue): Likewise.
15452 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15453 (mmix_target_asm_function_epilogue): Likewise.
15454 * config/msp430/msp430.c (msp430_start_function): Likewise.
15455 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15456 (nds32_asm_function_epilogue): Likewise.
15457 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15458 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15459 (pa_output_function_epilogue): Likewise.
15460 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15461 (rs6000_output_function_epilogue): Likewise.
15462 * config/rl78/rl78.c (rl78_start_function): Likewise.
15463 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15464 (rs6000_output_function_epilogue): Likewise.
15465 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15466 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15467 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15468 (sparc_asm_function_epilogue): Likewise.
15470 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15472 * tree.h (type_has_mode_precision_p): New function.
15473 * convert.c (convert_to_integer_1): Use it.
15474 * expr.c (expand_expr_real_2): Likewise.
15475 (expand_expr_real_1): Likewise.
15476 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15477 * match.pd: Likewise.
15478 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15479 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15480 * tree-tailcall.c (process_assignment): Likewise.
15481 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15482 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15483 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15484 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15485 (vectorizable_assignment): Likewise.
15486 (vectorizable_shift): Likewise.
15487 (vectorizable_operation): Likewise.
15488 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15490 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15492 * match.pd: Add pow (C, x) simplification.
15494 2017-08-21 Richard Biener <rguenther@suse.de>
15496 PR tree-optimization/81900
15497 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15498 for blocks with abnormal predecessors.
15499 (compute_antic): Do not set visited flag prematurely.
15501 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15504 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15506 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15508 * stor-layout.h (vector_type_mode): Move to...
15509 * tree.h (vector_type_mode): ...here.
15510 * stor-layout.c (vector_type_mode): Move to...
15511 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15513 2017-08-21 Richard Biener <rguenther@suse.de>
15515 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15516 register_external_die hooks.
15517 (debug_false_tree_charstarstar_uhwistar): Declare.
15518 (debug_nothing_tree_charstar_uhwi): Likewise.
15519 * debug.c (do_nothing_debug_hooks): Adjust.
15520 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15521 (debug_nothing_tree_charstar_uhwi): Likewise.
15522 * dbxout.c (dbx_debug_hooks): Adjust.
15523 (xcoff_debug_hooks): Likewise.
15524 * sdbout.c (sdb_debug_hooks): Likewise.
15525 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15526 * dwarf2out.c (macinfo_label_base): New global.
15527 (dwarf2out_register_external_die): New function for the
15528 register_external_die hook.
15529 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15530 (dwarf2_debug_hooks): Use them.
15531 (dwarf2_lineno_debug_hooks): Adjust.
15532 (struct die_struct): Add with_offset flag.
15533 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15534 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15535 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15536 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15537 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15538 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15539 defining section names for the early LTO debug variants.
15540 (reset_indirect_string): New helper.
15541 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15542 (print_dw_val): Add support for offsetted symbol references.
15543 (get_ultimate_context): Split out from is_cxx.
15544 (is_cxx): Use get_ultimate_context.
15545 (is_fortran): Add decl overload.
15546 (compute_comp_unit_symbol): Split out worker from
15547 compute_section_prefix.
15548 (compute_section_prefix): Call compute_comp_unit_symbol and
15549 set comdat_type_p here.
15550 (output_die): Skip DIE symbol output for the LTO added one.
15551 Handle DIE symbol references with offset.
15552 (output_comp_unit): Guard section name mangling properly.
15553 For LTO debug sections emit a symbol at the section beginning
15554 which we use to refer to its DIEs.
15555 (add_abstract_origin_attribute): For DIEs registered via
15556 dwarf2out_register_external_die directly refer to the early
15557 DIE rather than indirectly through the shadow one we created.
15558 Remove obsolete call to dwarf2out_abstract_function for
15559 non-function/block origins.
15560 (gen_array_type_die): When generating early LTO debug do
15561 not emit DW_AT_string_length.
15562 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15563 late when in LTO. As suggested place a gcc_unreachable for
15564 the DECL_ABSTRACT_P case.
15565 (gen_subprogram_die): Avoid another specification DIE
15566 for early built declarations/definitions for the late LTO case.
15567 (gen_variable_die): Add type references for late duplicated VLA dies
15569 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15570 we have the abstract instance already.
15571 (process_scope_var): Adjust decl DIE contexts in LTO which
15572 first puts them in limbo.
15573 (gen_decl_die): Do not generate type DIEs late apart from
15574 types for VLAs or for decls we do not yet have a DIE. Do not
15575 call dwarf2out_abstract_function late.
15576 (dwarf2out_early_global_decl): Make sure to create DIEs
15577 for abstract instances of a decl first.
15578 (dwarf2out_late_global_decl): Adjust comment.
15579 (output_macinfo_op): With multiple macro sections use
15580 macinfo_label_base to distinguish labels.
15581 (output_macinfo): Likewise. Update macinfo_label_base.
15582 Pass in the line info label.
15583 (note_variable_value_in_expr): When generating LTO resolve
15584 all variable values here by generating DIEs as needed.
15585 (init_sections_and_labels): Add early LTO debug flag parameter
15586 and generate different sections and names if set. Add generation
15587 counter for the labels so we can have multiple of them.
15588 (reset_dies): Helper to allow DIEs to be output multiple times.
15589 (dwarf2out_finish): When outputting DIEs to the fat part of an
15590 LTO object first reset DIEs.
15591 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15592 (modified_type_die): Check for decl_ultimate_origin being self
15594 (gen_type_die_with_usage): Likewise.
15595 (gen_typedef_die): Allow decl_ultimate_origin being self.
15596 (set_decl_abstract_flags): Remove.
15597 (set_block_abstract_flags): Likewise.
15598 (dwarf2out_abstract_function): Treat the early generated DIEs
15599 as the abstract copy and only add DW_AT_inline and
15600 DW_AT_artificial here and call set_decl_origin_self.
15601 If the DIE has an abstract origin don't do anything.
15602 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15603 if we have none yet (Go fails to build one, PR78628).
15604 (variably_modified_type_p): Prevent endless recursion for Ada
15605 cyclic pointer types.
15606 * lto-streamer-in.c: Include debug.h.
15607 (dref_queue): New global.
15608 (lto_read_tree_1): Stream in DIE references.
15609 (lto_input_tree): Register DIE references.
15610 (input_function): Stream DECL_DEBUG_ARGS.
15611 * lto-streamer-out.c: Include debug.h.
15612 (lto_write_tree_1): Output DIE references.
15613 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15614 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15615 (output_function): Stream DECL_DEBUG_ARGS.
15616 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15617 Stream DECL_ABSTRACT_ORIGIN.
15618 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15619 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15620 DECL_CONTEXT for file-scope decls.
15621 * lto-streamer.h (struct dref_entry): Declare.
15622 (dref_queue): Likewise.
15623 * cfgexpand.c (pass_expand::execute): Do not call the
15624 outlining_inline_function hook here.
15625 * lto-wrapper.c (debug_obj): New global.
15626 (tool_cleanup): Unlink it if required.
15627 (debug_objcopy): New function.
15628 (run_gcc): Handle early debug sections in the IL files by
15629 extracting them to separate files, partially linkin them and
15630 feeding the result back as result to the linker.
15631 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15632 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15633 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15634 sections into a separate segment.
15635 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15637 (darwin_asm_dwarf_section): Likewise.
15638 (darwin_asm_output_dwarf_offset): Likewise.
15639 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15641 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15642 Alan Hayward <alan.hayward@arm.com>
15643 David Sherwood <david.sherwood@arm.com>
15645 * read-md.h (md_reader::record_potential_iterator_use): Replace
15646 pointer argument with an rtx and an index.
15647 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15648 (apply_mode_iterator): Likewise.
15649 (apply_code_iterator): Likewise.
15650 (apply_int_iterator): Likewise.
15651 (apply_subst_iterator): Likewise.
15652 (record_iterator_use): Likewise.
15653 (record_attribute_use): Likewise.
15654 (md_reader::record_potential_iterator_use): Likewise. Update calls
15655 to record_iterator_use and apply_iterator.
15656 (iterator_use): Replace ptr with x and index.
15657 (attribute_use): Likewise.
15658 (apply_attribute_uses): Update calls to apply_iterator.
15659 (apply_iterators): Likewise. Update initialization of iterator_use.
15660 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15661 and record_potential_iterator_use.
15662 (rtx_reader::read_rtx_operand): Likewise.
15664 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15665 Alan Hayward <alan.hayward@arm.com>
15666 David Sherwood <david.sherwood@arm.com>
15668 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15671 2017-08-21 Richard Biener <rguenther@suse.de>
15673 PR middle-end/81884
15674 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15675 at struct end conservatively when comparing common bases.
15677 2017-08-21 Richard Biener <rguenther@suse.de>
15679 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15680 (mem_ref_in_stmt): Remove.
15681 (determine_max_movement): Use ref index to get at the reference.
15682 (invariantness_dom_walker::before_dom_children): Deal with
15683 lim data already initialized.
15684 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15686 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15688 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15689 (TARGET_ISA_ROUND): Ditto.
15690 (TARGET_ROUND): Ditto.
15691 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15692 * config/i386/i386.md: Ditto.
15693 * config/i386/sse.md: Ditto.
15694 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15695 with OPTION_MASK_ISA_SSE4_1.
15697 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15700 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15701 __builtin_ia32_lzcnt_u16.
15703 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15706 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15707 (rs6000_set_current_function): Rewrite function to use it.
15709 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15712 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15713 and TYPE_WARN_IF_NOT_ALIGN.
15714 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15715 (handle_warn_if_not_align): New.
15716 (place_union_field): Call handle_warn_if_not_align.
15717 (place_field): Call handle_warn_if_not_align.
15718 Copy TYPE_WARN_IF_NOT_ALIGN.
15719 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15720 (layout_type): Likewise.
15721 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15723 (tree_decl_common): Add warn_if_not_align.
15724 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15725 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15726 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15727 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15728 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15729 * doc/extend.texi: Document warn_if_not_aligned attribute.
15730 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15732 2017-08-17 Martin Liska <mliska@suse.cz>
15735 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15736 (get_data_dependence): Use it as pointer type.
15737 (distribute_loop): Likewise.
15739 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15741 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15742 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15743 (p8_vmrgow_<mode>_direct): New define_insn.
15744 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15745 handle endianness for vmrgew and vmrgow permute patterns.
15747 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15749 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15750 * config/rs6000/rs6000-cpus.def: Remove comment.
15751 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15752 (POWERPC_MASKS): Likewise.
15753 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15754 use of TARGET_VSX_TIMODE.
15755 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15756 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15757 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15758 (rs6000_option_override_internal): Remove dead code.
15759 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15760 (rs6000_legitimize_reload_address): Likewise.
15761 (rs6000_legitimate_address_p): Likewise.
15762 (rs6000_opt_masks): Delete "vsx-timode".
15763 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15764 from function comment.
15765 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15766 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15767 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15769 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15770 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15771 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15772 (VSX_TI): Likewise.
15774 (define_peephole2): Likewise.
15776 2017-08-17 Martin Sebor <msebor@redhat.com>
15779 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15780 past the end of an array.
15781 (test_pp_format): Add test cases.
15783 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15785 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15786 missing ECF_NOTHROW flags.
15788 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15791 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15792 operands residing in integer registers.
15793 (*vsx_le_perm_load_<mode>): Likewise.
15794 (*vsx_le_perm_store_<mode>): Likewise.
15795 (define_peephole2): Add peepholes to optimize the above.
15797 2017-08-17 Marek Polacek <polacek@redhat.com>
15799 PR middle-end/81814
15800 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15801 to mimic what shorten_compare did. Change the return type to bool.
15802 (fold_cond_expr_with_comparison): Update call to
15803 operand_equal_for_comparison_p.
15804 (fold_ternary_loc): Likewise.
15806 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15808 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15810 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15812 2017-08-17 Richard Biener <rguenther@suse.de>
15814 * tree-ssa-structalias.c (solve_graph): When propagating
15815 to successors update the graphs succ edges and avoid duplicate work.
15817 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15820 * config/i386/i386.c (ix86_option_override_internal): Save target
15821 specific options after ix86_stack_protector_guard_reg was changed.
15823 2017-08-17 Richard Biener <rguenther@suse.de>
15825 PR tree-optimization/81827
15826 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15827 (new_var_info): Initialize it conservatively.
15828 (get_call_vi): Mark register vars.
15829 (new_scalar_tmp_constraint_exp): Likewise.
15830 (handle_rhs_call): Likewise.
15831 (handle_const_call): Likewise.
15832 (create_function_info_for): Likewise.
15833 (solve_constraints): Sort varinfos to separate register from
15834 non-register vars to pack points-to solution bitmaps during
15837 2017-08-17 Marek Polacek <polacek@redhat.com>
15839 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15841 2017-08-17 Richard Biener <rguenther@suse.de>
15843 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15844 to true when overflow is undefined and we saturated the result.
15846 2017-08-17 Alan Modra <amodra@gmail.com>
15849 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15850 Don't use store multiple if only one reg needs saving.
15851 (interesting_frame_related_regno): New function.
15852 (rs6000_frame_related): Don't emit frame info for regs that
15854 (rs6000_emit_epilogue): Likewise.
15856 2017-08-16 Nathan Sidwell <nathan@acm.org>
15858 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15859 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15860 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15861 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15862 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15863 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15864 (verify_type): Adjust for TYPE_BINFO move.
15865 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15866 process TYPE_BINFO directly.
15867 (hash_tree): Likewise.
15868 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15870 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15873 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15875 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15877 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15880 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15881 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15882 Add ix86_binary_operator_ok to insn constraint.
15883 (*iordi_1_bts): Ditto.
15884 (*xordi_1_btc): Ditto.
15885 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15886 Update corresponding peephole2 pattern.
15890 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15892 PR tree-optimization/81832
15893 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15894 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15896 2017-08-16 Marek Polacek <polacek@redhat.com>
15899 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15900 regarding size_zero_node.
15902 2017-08-16 Martin Liska <mliska@suse.cz>
15905 * config.gcc: Respect previously set extra_objs in case
15908 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15910 PR tree-optimization/81835
15911 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15912 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15913 not depend on the phi.
15915 2017-08-16 Alan Modra <amodra@gmail.com>
15917 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15920 2017-08-16 Alan Modra <amodra@gmail.com>
15922 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15923 (save_reg_p): ..into this. Update all callers.
15924 (first_reg_to_save): Simplify.
15926 2017-08-16 Alan Modra <amodra@gmail.com>
15928 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15931 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15935 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15936 constant count if that count is less than 32.
15938 2017-08-15 Nathan Sidwell <nathan@acm.org>
15940 * gcc.c (execute): Emit friendlier message if inferior is killed
15941 by an external cause.
15943 2017-08-15 Richard Biener <rguenther@suse.de>
15945 PR tree-optimization/81790
15946 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15947 CONSTRUCTORs from simplifying and VN.
15949 2017-08-14 Martin Sebor <msebor@redhat.com>
15951 * builtin-attrs.def: Add comments.
15953 2017-08-14 Martin Sebor <msebor@redhat.com>
15956 * doc/extend.texi (attribute nonstring): Document new attribute.
15958 2017-08-14 Martin Sebor <msebor@redhat.com>
15961 * tree-diagnostic.c (default_tree_printer): Handle %G.
15962 * gimple-pretty-print.h (percent_G_format): Declare new function.
15963 * gimple-pretty-print.c (percent_G_format): Define.
15964 * tree-pretty-print.c (percent_K_format): Add argument.
15966 2017-08-14 Martin Sebor <msebor@redhat.com>
15968 PR translation/79998
15969 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15970 Remove a stray space.
15972 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15975 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15976 (*iordi_1_bts): Ditto.
15977 (*xordi_1_btc): Ditto.
15979 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15982 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15983 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15985 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15986 quoted strings, and make more translator-friendly.
15987 (darwin_rs6000_override_options): Likewise.
15988 (rs6000_option_override_internal): Likewise.
15989 (rs6000_return_in_memory): Fix overlong line.
15990 (init_cmulative_args): Use quoted strings, and make more
15991 translator-friendly.
15992 (rs6000_pass_by_reference): Fix overlong line.
15993 (def_builtin): Use quoted strings.
15994 (altivec_expand_predicate_builtin): Use quoted strings, and make
15995 more translator-friendly.
15996 (htm_expand_builtin): Use quoted strings.
15997 (cpu_expand_builtin): Use quoted strings, and make more
15998 translator-friendly.
15999 (altivec_expand_builtin): Likewise.
16000 (paired_expand_predicate_builtin): Likewise.
16001 (rs6000_invalid_builtin): Likewise.
16002 (builtin_function_type): Use quoted strings.
16003 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16004 more translator-friendly.
16005 (rs6000_trampoline_init): Likewise.
16006 (rs6000_handle_altivec_attribute): Likewise.
16007 (rs6000_inner_target_options): Use quoted strings.
16008 (rs6000_disable_incompatible_switches): Likewise.
16009 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16010 strings, and make more translator-friendly.
16011 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16013 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16015 PR tree-optimization/81799
16016 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16017 cond_expr to simple gimple operand.
16019 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16021 PR middle-end/46932
16022 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16024 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16028 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16029 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16030 TARGET_GASISR_PROLOGUES.
16031 * config/avr/avr.c (avr_option_override): Same.
16032 (avr_pass_pre_proep::execute): Same.
16034 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16037 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16038 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16040 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16042 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16043 define_insn_and_split. Split to a memory load from 0 in
16044 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16045 using PTR mode iterator.
16046 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16047 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16048 (*add_tp_<mode>): Redefine as define_insn_and_split.
16049 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16050 address space. Merge with *add_tp_x32 using PTR mode iterator.
16051 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16052 Split to an add with a memory load from 0 in
16053 DEFAULT_TLS_SEG_REG address space.
16055 2017-08-12 Andrew Pinski <apinski@cavium.com>
16057 * config/aarch64/aarch64-option-extensions.def (rdma):
16058 Fix feature string to what Linux prints out in /proc/cpuinfo.
16060 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16063 * dwarf2out.c (modified_type_die): For C typedef types that have
16064 an ultimate origin, process the ultimate origin instead of the
16066 (gen_typedef_die): Assert that input DECLs have no ultimate
16068 (gen_type_die_with_usage): For typedef variants that have an
16069 ultimate origin, just call gen_decl_die on the original DECL.
16070 (process_scope_var): Avoid creating DIEs for local typedefs and
16071 concrete static variables.
16073 2017-08-12 Alan Modra <amodra@gmail.com>
16077 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16078 match gnu-user.h startfile.
16079 (ENDFILE_LINUX_SPEC): Similarly.
16081 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16084 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16086 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16088 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16089 * config/aarch64/aarch64.md (mov<mode>): Change.
16090 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16091 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16092 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16094 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16096 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16097 for storage order barriers.
16099 2017-08-11 Martin Liska <mliska@suse.cz>
16102 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16103 variables of void type.
16105 2017-08-11 Martin Liska <mliska@suse.cz>
16107 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16108 TARGET_SUPPORTS_ALIASES.
16109 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16110 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16111 (optimize_weakref): Likewise.
16112 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16113 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16114 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16116 2017-08-11 Martin Liska <mliska@suse.cz>
16119 * config/i386/i386.c (make_resolver_func): Do complete
16120 refactoring of the function.
16122 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16125 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16126 * config/i386/i386.c (ix86_stack_protect_guard): Use
16127 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16128 * doc/invoke.texi (x86 Options): Document
16129 -mstack-protector-guard-symbol= option.
16131 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16133 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16134 * config/i386/i386.c (ix86_split_stack_guard): New function.
16135 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16136 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16137 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16138 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16139 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16140 (split_stack_space_check): Call ix86_split_stack_guard.
16142 2017-08-10 Martin Sebor <msebor@redhat.com>
16144 * print-tree.c (print_node): Print location using the established
16146 Replace spaces with colons.
16147 (debug_raw, debug): Ditto.
16149 2017-08-10 Martin Sebor <msebor@redhat.com>
16152 * pretty-print.c (pp_format): Correct the handling of %s precision.
16154 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16157 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16159 (ix86_finalize_stack_frame_flags): This. Also clear
16160 frame_pointer_needed if -fno-omit-frame-pointer is used without
16162 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16163 with ix86_finalize_stack_frame_flags.
16164 (ix86_expand_epilogue): Likewise.
16165 (ix86_expand_split_stack_prologue): Likewise.
16166 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16168 2017-08-10 Martin Liska <mliska@suse.cz>
16171 * c-attribs.c (handle_target_attribute):
16172 Report warning for an empty string argument of target attribute.
16174 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16177 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16179 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16180 or DECL_NONLOCAL labels.
16181 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16182 or DECL_NONLOCAL labels here.
16184 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16187 to indicate when early gimple folding has been disabled.
16188 (rs6000_gimple_fold_builtin): Add debug content.
16189 (rs6000_invalid_builtin): Fix whitespace.
16190 (rs6000_expand_builtin): Fix whitespace.
16191 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16193 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16196 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16197 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16199 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16201 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16202 (qdf24xx): Likewise.
16203 * config/aarch64/aarch64.md: Include falkor.md.
16204 * config/aarch64/falkor.md: New.
16206 2017-08-09 Marek Polacek <polacek@redhat.com>
16209 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16210 * diagnostic.c (emit_diagnostic): Add a comment.
16211 (emit_diagnostic_valist): New function.
16213 2017-08-09 Marek Polacek <polacek@redhat.com>
16216 * input.c (make_location): New overload.
16217 * input.h (make_location): Declare.
16219 2017-08-08 Alan Modra <amodra@gmail.com>
16220 H.J. Lu <hongjiu.lu@intel.com>
16223 * gcc.c (NO_PIE_SPEC): Delete.
16224 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16226 (LINK_PIE_SPEC): ..to here.
16227 (LINK_COMMAND_SPEC): Support -no-pie.
16228 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16229 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16231 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16232 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16233 (ENDFILE_CRTEND_SPEC): Similarly.
16235 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16238 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16239 (mstack-protector-guard-offset=): Ditto.
16240 * config/i386/i386.c (ix86_option_override): Handle
16241 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16243 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16244 ix86_stack_protect_guard_offset variables.
16245 (TARGET_STACK_PROTECT_GUARD): Always define.
16246 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16247 and -mstack-protector-guard-offset= options.
16249 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16251 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16252 boundary case for the last candidate.
16254 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16256 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16257 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16259 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16261 PR middle-end/19706
16262 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16263 * config/aarch64/aarch64-builtins.c
16264 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16265 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16266 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16268 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16269 Andrew Pinski <pinskia@gmail.com>
16271 PR middle-end/19706
16272 * internal-fn.def (XORSIGN): New.
16273 * optabs.def (xorsign_optab): New.
16274 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16275 (convert_expand_mult_copysign): New.
16276 (pass_optimize_widening_mul::execute): Call
16277 convert_expand_mult_copysign.
16279 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16281 PR tree-optimization/81354
16282 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16283 Insert on edges rather than explicitly creating landing pads.
16284 (analyze_candidates_and_replace): Commit edge inserts.
16286 2017-08-08 Richard Biener <rguenther@suse.de>
16288 PR middle-end/81719
16289 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16290 (expand_simple_operations): Also look through ADDR_EXPRs with
16291 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16293 2017-08-08 Richard Biener <rguenther@suse.de>
16295 PR tree-optimization/81723
16296 * tree-vect-slp.c (struct bst_traits): New hash traits.
16297 (bst_fail): New global.
16298 (vect_build_slp_tree_2): New worker, split out from ...
16299 (vect_build_slp_tree): ... this now wrapping it with using
16300 bst_fail set to cache SLP tree build fails. Properly handle
16302 (vect_analyze_slp_instance): Allocate and free bst_fail.
16304 2017-08-08 Martin Liska <mliska@suse.cz>
16307 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16308 LABEL_DECLs that can be from a different function.
16310 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16312 PR tree-optimization/81744
16313 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16314 loop's number of iterations.
16316 2017-08-08 Martin Liska <mliska@suse.cz>
16318 * asan.c: Include header files.
16319 * attribs.c (build_decl_attribute_variant): New function moved
16321 (build_type_attribute_qual_variant): Likewise.
16322 (cmp_attrib_identifiers): Likewise.
16323 (simple_cst_list_equal): Likewise.
16324 (omp_declare_simd_clauses_equal): Likewise.
16325 (attribute_value_equal): Likewise.
16326 (comp_type_attributes): Likewise.
16327 (build_type_attribute_variant): Likewise.
16328 (lookup_ident_attribute): Likewise.
16329 (remove_attribute): Likewise.
16330 (merge_attributes): Likewise.
16331 (merge_type_attributes): Likewise.
16332 (merge_decl_attributes): Likewise.
16333 (merge_dllimport_decl_attributes): Likewise.
16334 (handle_dll_attribute): Likewise.
16335 (attribute_list_equal): Likewise.
16336 (attribute_list_contained): Likewise.
16337 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16338 (lookup_attribute_by_prefix): Likewise.
16339 * bb-reorder.c: Include header files.
16340 * builtins.c: Likewise.
16341 * calls.c: Likewise.
16342 * cfgexpand.c: Likewise.
16343 * cgraph.c: Likewise.
16344 * cgraphunit.c: Likewise.
16345 * convert.c: Likewise.
16346 * dwarf2out.c: Likewise.
16347 * final.c: Likewise.
16348 * fold-const.c: Likewise.
16349 * function.c: Likewise.
16350 * gimple-expr.c: Likewise.
16351 * gimple-fold.c: Likewise.
16352 * gimple-pretty-print.c: Likewise.
16353 * gimple.c: Likewise.
16354 * gimplify.c: Likewise.
16355 * hsa-common.c: Likewise.
16356 * hsa-gen.c: Likewise.
16357 * internal-fn.c: Likewise.
16358 * ipa-chkp.c: Likewise.
16359 * ipa-cp.c: Likewise.
16360 * ipa-devirt.c: Likewise.
16361 * ipa-fnsummary.c: Likewise.
16362 * ipa-inline.c: Likewise.
16363 * ipa-visibility.c: Likewise.
16365 * lto-cgraph.c: Likewise.
16366 * omp-expand.c: Likewise.
16367 * omp-general.c: Likewise.
16368 * omp-low.c: Likewise.
16369 * omp-offload.c: Likewise.
16370 * omp-simd-clone.c: Likewise.
16371 * opts-global.c: Likewise.
16372 * passes.c: Likewise.
16373 * predict.c: Likewise.
16374 * sancov.c: Likewise.
16375 * sanopt.c: Likewise.
16376 * symtab.c: Likewise.
16377 * toplev.c: Likewise.
16378 * trans-mem.c: Likewise.
16379 * tree-chkp.c: Likewise.
16380 * tree-eh.c: Likewise.
16381 * tree-into-ssa.c: Likewise.
16382 * tree-object-size.c: Likewise.
16383 * tree-parloops.c: Likewise.
16384 * tree-profile.c: Likewise.
16385 * tree-ssa-ccp.c: Likewise.
16386 * tree-ssa-live.c: Likewise.
16387 * tree-ssa-loop.c: Likewise.
16388 * tree-ssa-sccvn.c: Likewise.
16389 * tree-ssa-structalias.c: Likewise.
16390 * tree-ssa.c: Likewise.
16391 * tree-streamer-in.c: Likewise.
16392 * tree-vectorizer.c: Likewise.
16393 * tree-vrp.c: Likewise.
16394 * tsan.c: Likewise.
16395 * ubsan.c: Likewise.
16396 * varasm.c: Likewise.
16397 * varpool.c: Likewise.
16398 * tree.c: Remove functions moved to attribs.[ch].
16399 * tree.h: Likewise.
16400 * config/aarch64/aarch64.c: Add attrs.h header file.
16401 * config/alpha/alpha.c: Likewise.
16402 * config/arc/arc.c: Likewise.
16403 * config/arm/arm.c: Likewise.
16404 * config/avr/avr.c: Likewise.
16405 * config/bfin/bfin.c: Likewise.
16406 * config/c6x/c6x.c: Likewise.
16407 * config/cr16/cr16.c: Likewise.
16408 * config/cris/cris.c: Likewise.
16409 * config/darwin.c: Likewise.
16410 * config/epiphany/epiphany.c: Likewise.
16411 * config/fr30/fr30.c: Likewise.
16412 * config/frv/frv.c: Likewise.
16413 * config/ft32/ft32.c: Likewise.
16414 * config/h8300/h8300.c: Likewise.
16415 * config/i386/winnt.c: Likewise.
16416 * config/ia64/ia64.c: Likewise.
16417 * config/iq2000/iq2000.c: Likewise.
16418 * config/lm32/lm32.c: Likewise.
16419 * config/m32c/m32c.c: Likewise.
16420 * config/m32r/m32r.c: Likewise.
16421 * config/m68k/m68k.c: Likewise.
16422 * config/mcore/mcore.c: Likewise.
16423 * config/microblaze/microblaze.c: Likewise.
16424 * config/mips/mips.c: Likewise.
16425 * config/mmix/mmix.c: Likewise.
16426 * config/mn10300/mn10300.c: Likewise.
16427 * config/moxie/moxie.c: Likewise.
16428 * config/msp430/msp430.c: Likewise.
16429 * config/nds32/nds32-isr.c: Likewise.
16430 * config/nds32/nds32.c: Likewise.
16431 * config/nios2/nios2.c: Likewise.
16432 * config/nvptx/nvptx.c: Likewise.
16433 * config/pa/pa.c: Likewise.
16434 * config/pdp11/pdp11.c: Likewise.
16435 * config/powerpcspe/powerpcspe.c: Likewise.
16436 * config/riscv/riscv.c: Likewise.
16437 * config/rl78/rl78.c: Likewise.
16438 * config/rx/rx.c: Likewise.
16439 * config/s390/s390.c: Likewise.
16440 * config/sh/sh.c: Likewise.
16441 * config/sol2.c: Likewise.
16442 * config/sparc/sparc.c: Likewise.
16443 * config/spu/spu.c: Likewise.
16444 * config/stormy16/stormy16.c: Likewise.
16445 * config/tilegx/tilegx.c: Likewise.
16446 * config/tilepro/tilepro.c: Likewise.
16447 * config/v850/v850.c: Likewise.
16448 * config/vax/vax.c: Likewise.
16449 * config/visium/visium.c: Likewise.
16450 * config/xtensa/xtensa.c: Likewise.
16452 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16455 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16456 constraints since the -mupper-regs-* switches have been
16458 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16459 into a vector from a double word element that was extracted from
16460 another vector, and eliminate extra XXPERMDI instructions.
16461 (vsx_concat_<mode>_2): Likewise.
16462 (vsx_concat_<mode>_3): Likewise.
16463 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16464 concat to allow optimizing inserts from previous extracts.
16466 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16468 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16469 memory reference to a SSP offset in TLS address space.
16470 (ix86_print_operand) <case '@'>: Remove.
16471 (ix86_print_operand_punct_valid_p): Remove '@' code.
16472 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16473 UNSPEC_SP_TLS_TEST.
16474 (stack_tls_protect_set_<mode>): Remove.
16475 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16476 (stack_tls_protect_test_<mode>): Remove.
16477 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16479 2017-08-07 Olivier Hainque <hainque@adacore.com>
16482 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16484 2017-08-07 Douglas Rupp <rupp@adacore.com>
16486 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16487 variable was referenced as multidir in command.
16489 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16492 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16495 2017-08-07 Martin Liska <mliska@suse.cz>
16497 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16498 * config/rl78/rl78.c: Add include of attribs.h.
16499 * config/sh/sh.c: Likewise.
16500 * config/v850/v850.c: Likewise.
16502 2017-08-07 Tom de Vries <tom@codesourcery.com>
16504 PR middle-end/78266
16505 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16507 2017-08-07 Martin Liska <mliska@suse.cz>
16509 * config/mips/mips.c: Include attribs.h.
16511 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16514 * doc/invoke.texi: Document change in behvaior for -Ofast for
16517 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16519 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16521 (aarch64_pop_regs): Likewise.
16522 (aarch64_gen_load_pair): Likewise.
16523 (aarch64_save_callee_saves): Likewise.
16524 (aarch64_restore_callee_saves): Likewise.
16526 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16528 * config/i386/i386.c: Revert the last change.
16530 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16533 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16535 (ix86_finalize_stack_frame_flags): This. Also clear
16536 frame_pointer_needed if -fno-omit-frame-pointer is used without
16538 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16539 with ix86_finalize_stack_frame_flags.
16540 (ix86_expand_epilogue): Likewise.
16541 (ix86_expand_split_stack_prologue): Likewise.
16543 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16546 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16547 to P_AES for Westmere.
16549 2017-08-07 Jonathan Yong <10walls@gmail.com>
16551 * config/i386/mingw.opt (fset-stack-executable): Removed.
16552 * config/i386/cygming.opt (fset-stack-executable): Moved
16554 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16556 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16558 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16560 2017-08-07 Marek Polacek <polacek@redhat.com>
16562 PR middle-end/81737
16563 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16565 2017-08-07 Martin Liska <mliska@suse.cz>
16567 * attribs.h (canonicalize_attr_name): New function.
16568 (cmp_attribs): Move from c-format.c and adjusted.
16569 (is_attribute_p): Moved from tree.h.
16570 * tree-inline.c: Add new includes.
16571 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16572 (private_is_attribute_p): Remove.
16573 (private_lookup_attribute): Likewise.
16574 (private_lookup_attribute_by_prefix): Simplify.
16575 (remove_attribute): Use is_attribute_p.
16576 * tree.h: Remove removed declarations.
16578 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16580 PR middle-end/81698
16581 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16582 instead of computing it in the function. Formatting fix.
16583 (expand_case): Don't rely on default_edge being the first edge,
16584 clear it if removing it, pass default_edge to
16585 emit_case_dispatch_table.
16586 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16589 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16591 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16592 insn in the function, emit NOP after the insn.
16594 2017-08-06 Tom de Vries <tom@codesourcery.com>
16596 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16599 2017-08-06 Tom de Vries <tom@codesourcery.com>
16601 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16604 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16606 PR tree-optimization/57371
16607 * match.pd: New pattern.
16609 2017-08-04 Marek Polacek <polacek@redhat.com>
16611 PR middle-end/81695
16612 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16613 perform the computation in offset_int.
16615 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16617 PR tree-optimization/81136
16618 * tree-vectorizer.h: Include tree-hash-traits.h.
16619 (vec_base_alignments): New typedef.
16620 (vec_info): Add a base_alignments field.
16621 (vect_record_base_alignments): Declare.
16622 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16624 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16625 (create_data_ref): Add an is_conditional_in_stmt argument.
16626 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16627 the is_conditional_in_stmt field.
16628 (data_ref_loc): Add an is_conditional_in_stmt field.
16629 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16630 (find_data_references_in_stmt): Update call to create_data_ref.
16631 (graphite_find_data_references_in_stmt): Likewise.
16632 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16633 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16634 (vect_record_base_alignment): New function.
16635 (vect_record_base_alignments): Likewise.
16636 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16637 for nested statements even if we fail to compute a misalignment.
16638 Use pooled base alignments for unconditional references.
16639 (vect_find_same_alignment_drs): Compare base addresses instead
16641 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16642 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16644 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16646 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16647 Add an explicit name for the enum. Use auto_vec for slp_instances
16648 and grouped_stores.
16649 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16651 (_bb_vec_info): Add a constructor and destructor.
16652 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16653 (destroy_loop_vec_info): Delete.
16654 (vect_destroy_datarefs): Likewise.
16655 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16656 (vec_info::vec_info): New function.
16657 (vec_info::~vec_info): Likewise.
16658 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16659 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16660 destroy_loop_vec_info.
16661 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16662 (_loop_vec_info::_loop_vec_info): ...this.
16663 (destroy_loop_vec_info): Replace with...
16664 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16665 the stmt_vec_infos. Leave handling of vec_info information to its
16666 destructor. Remove explicit vector releases.
16667 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16668 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16669 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16670 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16671 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16672 (destroy_bb_vec_info): Replace with...
16673 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16674 information to its destructor.
16675 (vect_slp_analyze_bb_1): Use new and delete instead of
16676 new_bb_vec_info and destroy_bb_vec_info.
16677 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16680 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16682 * tree-data-ref.h (subscript): Add access_fn field.
16683 (data_dependence_relation): Add could_be_independent_p.
16684 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16685 (same_access_functions): Move to tree-data-ref.c.
16686 * tree-data-ref.c (ref_contains_union_access_p): New function.
16687 (access_fn_component_p): Likewise.
16688 (access_fn_components_comparable_p): Likewise.
16689 (dr_analyze_indices): Add a reference to access_fn_component_p.
16690 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16692 (constant_access_functions): Likewise.
16693 (add_other_self_distances): Likewise.
16694 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16695 (initialize_data_dependence_relation): Use XCNEW and remove
16696 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16697 of access functions that have the same type. Allow the
16698 subsequence to end with different bases in some circumstances.
16699 Record the chosen access functions in SUB_ACCESS_FN.
16700 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16701 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16702 (subscript_dependence_tester_1): Likewise dra and drb.
16703 (build_classic_dist_vector): Update calls accordingly.
16704 (subscript_dependence_tester): Likewise.
16705 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16706 DDR_COULD_BE_INDEPENDENT_P.
16707 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16708 comp_alias_ddrs instead of may_alias_ddrs.
16709 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16711 (vect_analyze_data_ref_dependence): Use it if
16712 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16713 distance vectors if that fails.
16714 (dependence_distance_ge_vf): New function.
16715 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16716 LOOP_VINFO_MAY_ALIAS_DDRS.
16718 2017-08-04 Richard Biener <rguenther@suse.de>
16720 PR middle-end/81705
16721 * fold-const.c (fold_binary_loc): Properly restrict
16722 minus_var0 && minus_var1 case when associating undefined overflow
16725 2017-08-04 Tom de Vries <tom@codesourcery.com>
16727 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16729 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16731 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16732 Don't start diagnostic messages with a capital letter.
16733 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16735 (rs6000_invalid_builtin): Likewise.
16736 (rs6000_trampoline_init): Likewise.
16738 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16741 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16742 after setting changeable df flags.
16744 2017-08-03 Richard Biener <rguenther@suse.de>
16746 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16747 up if the use is in USE - X.
16749 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16751 * toplev.c (dumpfile.h): New include.
16752 (internal_error_reentered): New static function. Use it...
16753 (internal_error_function): ...here to handle reentered internal_error.
16755 2017-08-03 Richard Biener <rguenther@suse.de>
16757 PR middle-end/81148
16758 * fold-const.c (split_tree): Add minus_var and minus_con
16759 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16760 here but always use minus_*.
16761 (associate_trees): Assert we never associate with MINUS_EXPR
16762 and NULL first operand. Do not recurse for PLUS_EXPR operands
16763 when associating as MINUS_EXPR either.
16764 (fold_binary_loc): Track minus_var and minus_con.
16766 2017-08-03 Tom de Vries <tom@codesourcery.com>
16769 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16770 ACCEL_COMPILER, apply finish_options on
16771 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16773 2017-08-03 Tom de Vries <tom@codesourcery.com>
16776 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16777 function_entry_patch_area_size > 0.
16779 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16782 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16783 instead of 10??LU, perform unit multiplication in wide_int,
16784 don't change alloc_object_size_limit if the limit is larger
16787 PR tree-optimization/81655
16788 PR tree-optimization/81588
16789 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16790 the case when ranges[i].low and high are 1 for unsigned type with
16793 PR middle-end/81052
16794 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16795 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16797 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16799 * tree-vrp.h: Add include guard.
16801 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16804 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16805 (ud2): New insn pattern.
16806 * config/i386/i386.c (ix86_expand_epilogue):
16807 For naked functions, generate ud2 instead of trap insn.
16809 2017-08-02 Marek Polacek <polacek@redhat.com>
16812 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16814 2017-08-02 Tom de Vries <tom@codesourcery.com>
16815 Cesar Philippidis <cesar@codesourcery.com>
16817 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16818 Add missing edge probabilities.
16820 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16822 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16823 Correct endianness.
16825 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16827 PR middle-end/79499
16828 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16829 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16830 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16831 prologue_seq sequences - if any.
16833 2017-08-02 Richard Biener <rguenther@suse.de>
16835 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16836 via vectors if supported, integer extracts via punning if supported
16837 or otherwise vector extracts.
16839 2017-08-02 Richard Biener <rguenther@suse.de>
16841 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16843 (bitmap_insert_into_set): ... this.
16845 2017-08-02 Richard Biener <rguenther@suse.de>
16847 PR tree-optimization/81633
16849 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16851 PR tree-optimization/71752
16852 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16854 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16856 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16857 (machine_function::call_ms2sysv_pad_out): Remove field.
16858 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16859 (ix86_compute_frame_layout): Likewise.
16861 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16864 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16865 attribute with interrupt attribute.
16867 2017-08-01 Andrew Pinski <apinski@cavium.com>
16869 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16870 BIT_INSERT_EXPR's operand 1
16871 to see if the types precision matches.
16873 2017-08-01 Martin Liska <mliska@suse.cz>
16875 PR middle-end/70140
16876 * builtins.c (expand_builtin_memcpy_args): Remove.
16877 (expand_builtin_memcpy): Call newly added function
16878 expand_builtin_memory_copy_args.
16879 (expand_builtin_memcpy_with_bounds): Likewise.
16880 (expand_builtin_mempcpy): Remove last argument.
16881 (expand_builtin_mempcpy_with_bounds): Likewise.
16882 (expand_builtin_memory_copy_args): New function created from
16883 expand_builtin_mempcpy_args with small modifications.
16884 (expand_builtin_mempcpy_args): Remove.
16885 (expand_builtin_stpcpy): Remove unused argument.
16886 (expand_builtin): Likewise.
16887 (expand_builtin_with_bounds): Likewise.
16889 2017-08-01 Martin Liska <mliska@suse.cz>
16892 Make mempcpy more optimal (PR middle-end/70140).
16894 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16897 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16898 __builtin_vec_cmpne verify both arguments are compatible vectors
16899 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16900 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16901 move computation of aligned to after checking the argument types.
16905 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16906 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16909 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16910 Doug Rupp <rupp@adacore.com>
16911 Olivier Hainque <hainque@adacore.com>
16913 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16914 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16916 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16917 for TARGET_OS_CPP_BUILTIN.
16918 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16919 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16921 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16922 passing required abi options to the assembler for EABI configurations.
16923 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16924 of .text.hot and .text.unlikely sections for kernel modules when
16925 using ARM style exceptions.
16926 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16927 options. Add EXTRA_CC1_SPEC.
16928 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16930 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16932 (ARM_TARGET2_DWARF_FORMAT): Define.
16933 * config/arm/t-vxworks: Adjust multilib control to removal of the
16934 Diab command line options.
16936 2017-08-01 Martin Liska <mliska@suse.cz>
16938 PR gcov-profile/81561
16939 * gcov.c (unblock): Make unblocking safe as we need to preserve
16940 index correspondence of blocks and block_lists.
16942 2017-08-01 Richard Biener <rguenther@suse.de>
16944 PR tree-optimization/81181
16945 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16946 (compute_antic): ... end of iteration here.
16948 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16950 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16951 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16952 (ftree-slp-vectorize): Likewise.
16953 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16954 can no longer be set independent of flag_tree_loop_vectorize.
16955 * omp-general.c (emp_max_vf): Likewise.
16956 * opts.c (enable_fdo_optimizations): Remove references to
16957 flag_tree_vectorize, these are now implicit.
16958 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16959 and leave it for the options machinery.
16961 2017-08-01 Martin Liska <mliska@suse.cz>
16963 PR middle-end/70140
16964 * builtins.c (expand_builtin_memcpy_args): Remove.
16965 (expand_builtin_memcpy): Call newly added function
16966 expand_builtin_memory_copy_args.
16967 (expand_builtin_memcpy_with_bounds): Likewise.
16968 (expand_builtin_mempcpy): Remove last argument.
16969 (expand_builtin_mempcpy_with_bounds): Likewise.
16970 (expand_builtin_memory_copy_args): New function created from
16971 expand_builtin_mempcpy_args with small modifications.
16972 (expand_builtin_mempcpy_args): Remove.
16973 (expand_builtin_stpcpy): Remove unused argument.
16974 (expand_builtin): Likewise.
16975 (expand_builtin_with_bounds): Likewise.
16977 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16980 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16981 print "ds:" only for immediates in generic address space.
16983 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16986 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16987 (ix86_function_ok_for_sibcall): Return false for naked functions.
16989 2017-08-01 Richard Biener <rguenther@suse.de>
16991 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16992 (compute_antic): Seed worklist with exit block predecessors.
16993 * cfganal.c (dfs_find_deadend): For a cycle return the source
16994 of the edge closing it.
16996 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16998 * config/aarch64/aarch64.c
16999 (aarch64_can_const_movi_rtx_p): Move 0 check.
17001 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17003 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17004 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17006 * match.pd: Ditto in address comparison pattern.
17008 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17010 PR tree-optimization/81627
17011 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17012 closed ssa form for store-store chain.
17014 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17016 PR tree-optimization/81620
17017 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17018 for store-store chain.
17020 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17022 PR tree-optimization/81588
17023 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17024 ranges[i].in_p, invert comparison code ccode. For >/>=,
17025 swap rhs1 and rhs2 and comparison code unconditionally,
17026 for </<= don't do that. Don't swap rhs1/rhs2 again if
17027 ranges[i].in_p, instead invert comparison code ccode if
17028 opcode or oe->rank is BIT_IOR_EXPR.
17031 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17032 a direct optab to conversion optab.
17033 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17034 with GET_MODE_INNER as last argument instead of optab_handler.
17035 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17036 vector extraction if possible and optab is available.
17037 * expr.c (store_constructor): Use convert_optab_handler instead
17038 of optab_handler. Use vector initialization from smaller
17039 vectors if possible and optab is available.
17040 * tree-vect-stmts.c (vectorizable_load): Likewise.
17041 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17042 now have two modes.
17043 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17044 of vec_init from half-sized vectors with the same element mode.
17045 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17046 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17047 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17048 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17049 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17050 after mode in gen_vec_extract* calls.
17051 (vec_extract<mode>): Renamed to ...
17052 (vec_extract<mode><ssescalarmodelower>): ... this.
17053 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17054 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17055 element mode after mode in gen_vec_init* calls.
17056 (VEC_INIT_HALF_MODE): New mode iterator.
17057 (vec_init<mode>): Renamed to ...
17058 (vec_init<mode><ssescalarmodelower>): ... this.
17059 (vec_init<mode><ssehalfvecmodelower>): New expander.
17060 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17061 (vec_extractv2sfsf): ... this.
17062 (vec_initv2sf): Renamed to ...
17063 (vec_initv2sfsf): ... this.
17064 (vec_extractv2si): Renamed to ...
17065 (vec_extractv2sisi): ... this.
17066 (vec_initv2si): Renamed to ...
17067 (vec_initv2sisi): ... this.
17068 (vec_extractv4hi): Renamed to ...
17069 (vec_extractv4hihi): ... this.
17070 (vec_initv4hi): Renamed to ...
17071 (vec_initv4hihi): ... this.
17072 (vec_extractv8qi): Renamed to ...
17073 (vec_extractv8qiqi): ... this.
17074 (vec_initv8qi): Renamed to ...
17075 (vec_initv8qiqi): ... this.
17076 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17077 (vec_init<mode>): Renamed to ...
17078 (vec_init<mode><VEC_base_l>): ... this.
17079 (vec_extract<mode>): Renamed to ...
17080 (vec_extract<mode><VEC_base_l>): ... this.
17081 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17082 (vec_initv2sfsf): ... this.
17083 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17084 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17085 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17086 element mode after mode in gen_vec_init* calls.
17087 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17088 (vec_init<mode><Vel>): ... this.
17089 (vec_extract<mode>): Renamed to ...
17090 (vec_extract<mode><Vel>): ... this.
17091 * config/aarch64/iterators.md (Vel): New mode attribute.
17092 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17093 Add element mode after mode in gen_vec_extract* calls.
17094 * config/s390/vector.md (non_vec_l): New mode attribute.
17095 (vec_extract<mode>): Renamed to ...
17096 (vec_extract<mode><non_vec_l>): ... this.
17097 (vec_init<mode>): Renamed to ...
17098 (vec_init<mode><non_vec_l>): ... this.
17099 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17100 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17102 * config/arm/iterators.md (V_elem_l): New mode attribute.
17103 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17104 (vec_extract<mode><V_elem_l>): ... this.
17105 (vec_extractv2di): Renamed to ...
17106 (vec_extractv2didi): ... this.
17107 (vec_init<mode>): Renamed to ...
17108 (vec_init<mode><V_elem_l>): ... this.
17109 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17110 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17111 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17112 Add element mode after gen_vec_extract* calls.
17113 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17114 (vec_init<mode><unitmode>): ... this.
17115 (vec_extract<mode>): Renamed to ...
17116 (vec_extract<mode><unitmode>): ... this.
17117 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17118 (vec_init<mode><unitmode>): ... this.
17119 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17120 (vec_initv2sfsf): ... this.
17121 (vec_extractv2sf): Renamed to ...
17122 (vec_extractv2sfsf): ... this.
17123 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17124 Add element mode after gen_vec_extract* calls.
17125 * config/mips/mips.md (unitmode): New mode iterator.
17126 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17127 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17128 * config/spu/spu.md (inner_l): New mode attribute.
17129 (vec_init<mode>): Renamed to ...
17130 (vec_init<mode><inner_l>): ... this.
17131 (vec_extract<mode>): Renamed to ...
17132 (vec_extract<mode><inner_l>): ... this.
17133 * config/sparc/sparc.md (veltmode): New mode iterator.
17134 (vec_init<VMALL:mode>): Renamed to ...
17135 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17136 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17137 (vec_initv2sisi): ... this.
17138 (vec_initv2sf): Renamed to ...
17139 (vec_initv2sfsf): ... this.
17140 (vec_extractv2sf): Renamed to ...
17141 (vec_extractv2sfsf): ... this.
17142 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17143 (vec_init<mode>): Renamed to ...
17144 (vec_init<mode><VEC_base_l>): ... this.
17145 (vec_extract<mode>): Renamed to ...
17146 (vec_extract<mode><VEC_base_l>): ... this.
17147 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17148 (vec_initv2sfsf): ... this.
17149 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17150 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17151 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17152 gen_vec_init* calls.
17154 2017-08-01 Richard Biener <rguenther@suse.de>
17156 PR tree-optimization/81297
17157 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17158 TREE_OVERFLOW from INTEGER_CSTs.
17160 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17162 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17164 2017-07-31 Carl Love <cel@us.ibm.com>
17166 * config/rs6000/rs6000-c: Add support for built-in functions
17167 vector signed char vec_xl_be (signed long long, signed char *);
17168 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17169 vector signed int vec_xl_be (signed long long, signed int *);
17170 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17171 vector signed long long vec_xl_be (signed long long, signed long long *);
17172 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17173 vector signed short vec_xl_be (signed long long, signed short *);
17174 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17175 vector double vec_xl_be (signed long long, double *);
17176 vector float vec_xl_be (signed long long, float *);
17177 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17178 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17179 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17181 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17182 (altivec_expand_builtin): Add switch statement to call
17183 altivec_expand_xl_be for each builtin.
17184 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17185 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17186 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17187 __builtin_vsx_le_be_v16qi.
17188 * doc/extend.texi: Update the built-in documentation file for the
17189 new built-in functions.
17191 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17194 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17196 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17198 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17200 * config.gcc: Add z14.
17201 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17202 CPU model numbers for z13s and z14.
17203 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17205 * config/s390/s390-opts.h (enum processor_type): Rename
17206 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17207 * config/s390/s390.c (processor_table): Add field for CPU name to
17208 be passed to Binutils.
17209 (s390_asm_output_machine_for_arch): Use the new field in
17210 processor_table for Binutils.
17211 (s390_expand_builtin): Replace arch12 with z14.
17212 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17213 (s390_get_sched_attrmask): Likewise.
17214 (s390_get_unit_mask): Likewise.
17215 * config/s390/s390.opt: Add z14 to processor_type enum.
17217 2017-07-31 Martin Jambor <mjambor@suse.cz>
17220 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17221 regardless of optimization level.
17223 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17224 Martin Liska <mliska@suse.cz>
17226 * predict.def: Remove old comment and adjust probability.
17227 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17228 PREDICT statements.
17230 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17233 * config/i386/i386.c (ix86_function_naked): New function.
17234 (ix86_can_use_return_insn_p): Return false for naked functions.
17235 (ix86_expand_prologue): Skip prologue for naked functions.
17236 (ix86_expand_epilogue): Skip epilogue for naked functions
17237 and emit trap instruction.
17238 (ix86_warn_func_return): New function.
17239 (ix86_attribute_table): Add "naked" attribute specification.
17240 (TARGET_WARN_FUNC_RETURN): Define.
17241 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17243 2017-07-31 Martin Liska <mliska@suse.cz>
17245 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17246 (dump_gimple_bb_header): Always dump BB info.
17247 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17249 2017-07-31 Martin Liska <mliska@suse.cz>
17252 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17253 also with current_function_decl non-null equality.
17255 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17258 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17259 change type to the element type, instead add eltype variable and
17260 use it where we are interested in the element type.
17262 PR tree-optimization/81603
17263 * ipa-polymorphic-call.c
17264 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17265 offset arithmetic in offset_int, bail out if the resulting bit offset
17266 doesn't fit into shwi.
17268 2017-07-31 Martin Liska <mliska@suse.cz>
17270 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17271 (gimplify_save_expr): Fix comment.
17273 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17276 * config/i386/i386.c (ix86_function_arg): Update arguments for
17278 (ix86_compute_frame_layout): Set the initial stack offset to
17279 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17280 INCOMING_FRAME_SP_OFFSET.
17281 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17282 stack before exception handler returns.
17283 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17284 the 'ERROR_CODE' for exception handler.
17286 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17288 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17289 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17290 (ASM_OUTPUT_REG_POP): Ditto.
17291 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17292 instead of asm_fprintf to output pure string.
17294 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17296 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17297 to imported_module_or_decl hook.
17298 (debug_nothing_tree_tree_tree_bool): Remove.
17299 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17300 * debug.c (do_nothing_debug_hooks): Use
17301 debug_nothing_tree_tree_tree_bool_bool instead of
17302 debug_nothing_tree_tree_tree_bool.
17303 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17304 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17305 * sdbout.c (sdb_debug_hooks): Likewise.
17306 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17307 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17309 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17310 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17311 attribute, don't add anything.
17313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17315 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17316 (fold_build2_stat_loc): Likewise.
17317 (fold_build3_stat_loc): Likewise.
17318 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17319 (fold_build1_loc): Remove macro.
17320 (fold_build2_loc): Likewise.
17321 (fold_build3_loc): Likewise.
17323 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17325 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17326 (gimple_build_debug_bind_source_stat): Likewise.
17327 * gimple.h (gimple_build_debug_bind): Remove macro.
17328 (gimple_build_debug_bind_source): Likewise.
17330 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17332 * bitmap.c (bitmap_alloc): Adjust.
17333 (bitmap_gc_alloc): Likewise.
17334 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17336 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17338 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17339 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17340 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17341 (bitmap_gc_alloc_stat): Likewise.
17342 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17344 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17346 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17347 * rtl.h (shallow_copy_rtx): Remove macro.
17349 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17351 * emit-rtl.c (gen_raw_REG): Adjust.
17352 * gengenrtl.c (gendef): Likewise.
17353 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17354 * rtl.h (rtx_alloc): Remove macro.
17356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17358 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17359 (build_tree_list_stat): Likewise.
17360 * tree.h (build_tree_list): Remove macro.
17361 (build_tree_list_vec): Likewise.
17363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17365 * tree.c (make_vector_stat): Remove _stat from name.
17366 (build_vector_stat): Likewise.
17367 * tree.h (make_vector_stat): Remove macro.
17368 (build_vector_stat): Likewise.
17370 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17372 * tree.h (build_var_debug_value): Remove prototype.
17374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17376 * tree.c (tree_cons_stat): Remove _stat from name.
17377 * tree.h (tree_cons): Remove macro.
17379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17381 * tree.c (build_vl_exp_stat): Remove _stat from name.
17382 * tree.h (build_vl_exp): Remove macro.
17384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17386 * tree.c (build_decl_stat): Remove _stat from name.
17387 * tree.h (build_decl): Remove macro.
17389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17391 * gimple.c (gimple_build_with_ops_stat): Adjust.
17392 (gimple_alloc_stat): Remove _stat from name.
17393 * gimple.h (gimple_alloc): Remove macro.
17395 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17397 * tree.c (make_tree_vec_stat): Remove _stat from name.
17398 (grow_tree_vec_stat): Likewise.
17399 * tree.h (make_tree_vec_stat): Adjust prototype.
17400 (grow_tree_vec_stat): Likewise.
17401 (make_tree_vec): Remove macro.
17402 (grow_tree_vec): Likewise.
17404 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17406 * fold-const.c (fold_build1_stat_loc): Adjust.
17407 (fold_build2_stat_loc): Likewise.
17408 (fold_build3_stat_loc): Likewise.
17409 * tree.c (build0_stat): Remove _stat from name.
17410 (build1_stat): Likewise.
17411 (build2_stat): Likewise.
17412 (build3_stat): Likewise.
17413 (build4_stat): Likewise.
17414 (build5_stat): Likewise.
17415 * tree.h (build1_loc): Remove macro, and rename _stat function
17417 (build2_loc): Likewise.
17418 (build3_loc): Likewise.
17419 (build4_loc): Likewise.
17420 (build5_loc): Likewise.
17422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17424 * tree.c (make_int_cst_stat): Remove _stat from name.
17425 * tree.h (make_int_cst_stat): Adjust prototype.
17426 (make_int_cst): Remove macro.
17428 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17430 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17431 * tree.h (make_tree_binfo_stat): Adjust prototype.
17432 (make_tree_binfo): Remove.
17434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17436 * tree.c (copy_node_stat): Rename to copy_node.
17437 (build_distinct_type_copy): Adjust.
17438 * tree.h (copy_node_stat): Adjust prototype.
17439 (copy_node): Remove macro.
17441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17443 * tree.c (make_node_stat): rename to make_node.
17444 (build_tree_list_stat): Adjust.
17445 (build0_stat): Likewise.
17446 (build2_stat): Likewise.
17447 (build3_stat): Likewise.
17448 (build4_stat): Likewise.
17449 (build5_stat): Likewise.
17450 (build_decl_stat): Likewise.
17451 * tree.h (make_node_stat): Adjust prototype.
17452 (make_node): remove macro.
17454 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17456 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17457 (PPC_FEATURE2_SCV): Likewise.
17458 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17460 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17462 * config/aarch64/aarch64.c
17463 (aarch64_internal_mov_immediate): Add new special pattern.
17464 * config/aarch64/aarch64.md (*movdi_aarch64):
17465 Add reg/32bit const mov case.
17467 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17468 Richard Sandiford <richard.sandiford@linaro.org>
17470 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17471 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17472 Add integer and movi cases.
17473 (movi-split-hf-df-sf split, fp16): New.
17474 (enabled): Added TARGET_FP_F16INST.
17475 * config/aarch64/iterators.md (GPF_HF): New.
17476 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17478 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17480 * config/aarch64/aarch64.c
17481 (aarch64_simd_container_mode): Add prototype.
17482 (aarch64_expand_mov_immediate): Add HI support.
17483 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17484 (aarch64_can_const_movi_rtx_p): New.
17485 (aarch64_preferred_reload_class):
17486 Remove restrictions of using FP registers for certain SIMD operations.
17487 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17488 (aarch64_valid_floating_const): Add integer move validation.
17489 (aarch64_simd_imm_scalar_p): Remove.
17490 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17491 (aarch64_legitimate_constant_p): Expand list of supported cases.
17492 * config/aarch64/aarch64-protos.h
17493 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17494 (aarch64_reinterpret_float_as_int): New.
17495 (aarch64_simd_imm_scalar_p): Remove.
17496 * config/aarch64/constraints.md (Uvi): New.
17497 (Dd): Split into Ds and new Dd.
17498 * config/aarch64/aarch64.md (*movsi_aarch64):
17500 (*movdi_aarch64): Add SIMD mov case.
17502 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17504 * tree-predcom.c: (struct chain): Handle store-store chain in which
17505 stores for elimination only store loop invariant values.
17506 (execute_pred_commoning_chain): Ditto.
17507 (prepare_initializers_chain_store_elim): Ditto.
17508 (prepare_finalizers): Ditto.
17509 (is_inv_store_elimination_chain): New function.
17510 (initialize_root_vars_store_elim_1): New function.
17512 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17514 * tree-predcom.c: Revise general description of the pass.
17515 (enum chain_type): New enum type for store elimination.
17516 (struct chain): New field supporting store elimination.
17517 (struct component): Ditto.
17518 (dump_chain): Dump store-stores chain.
17519 (release_chain): Release resources.
17520 (split_data_refs_to_components): Compute and create component
17521 contains only stores for elimination.
17522 (get_chain_last_ref_at): New function.
17523 (make_invariant_chain): Initialization.
17524 (make_rooted_chain): Specify chain type in parameter and record it.
17525 (add_looparound_copies): Skip for store-stores chain.
17526 (determine_roots_comp): Compute type of chain and pass it to
17528 (initialize_root_vars_store_elim_2): New function.
17529 (finalize_eliminated_stores): New function.
17530 (remove_stmt): Handle store for elimination.
17531 (execute_pred_commoning_chain): Execute predictive commoning on
17532 store-store chains.
17533 (determine_unroll_factor): Skip unroll for store-stores chain.
17534 (prepare_initializers_chain_store_elim): New function.
17535 (prepare_initializers_chain): Hanlde store-store chain.
17536 (prepare_finalizers_chain, prepare_finalizers): New function.
17537 (tree_predictive_commoning_loop): Return integer value indicating
17538 if loop is unrolled or lcssa form is corrupted.
17539 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17541 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17543 * tree-predcom.c (initialize_root): Delete.
17544 (execute_pred_commoning_chain): Initialize root vars and replace
17545 reference of non-combined chain directly, rather than call above
17548 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17550 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17551 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17553 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17555 * tree-predcom.c (struct chain): New field init_seq.
17556 (release_chain): Release init_seq.
17557 (prepare_initializers_chain): Record intialization stmts in above
17559 (insert_init_seqs): New function.
17560 (tree_predictive_commoning_loop): Call insert_init_seqs.
17562 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17564 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17566 2017-07-28 Richard Biener <rguenther@suse.de>
17568 * match.pd: Remove superfluous :c.
17569 * genmatch.c (simplify::id): Add member.
17570 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17572 (current_id): New global.
17573 (dt_node::parent): Move from ...
17574 (dt_operand::parent): ... here. Add for_id member.
17575 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17576 (decision_tree::find_node): Relax order requirement when
17577 merging DT_TRUE nodes to ones inbetween the current simplify
17578 and the one we try to merge with. Add diagnostic whenever
17579 we need to enforce pattern order by not merging.
17580 (decision_tree::insert): Set current_id.
17581 (decision_tree::print_node): Dump parent node and for_id.
17582 (parser::last_id): Add member.
17583 (parser::push_simplify): Assign unique id.
17584 (parser::parser): Initialize last_id.
17586 2017-07-28 Martin Liska <mliska@suse.cz>
17589 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17590 gimple_build_debug_bind.
17592 2017-07-28 Richard Biener <rguenther@suse.de>
17594 PR tree-optimization/81502
17595 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17597 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17599 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17600 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17601 for BIT_FIELD_REF args.
17602 * fold-const.c (make_bit_field_ref): Likewise.
17603 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17605 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17608 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17609 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17610 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17611 Or it into SANITIZER_UNDEFINED.
17612 * ubsan.c: Include gimple-fold.h and varasm.h.
17613 (ubsan_expand_ptr_ifn): New function.
17614 (instrument_pointer_overflow): New function.
17615 (maybe_instrument_pointer_overflow): New function.
17616 (instrument_object_size): Formatting fix.
17617 (pass_ubsan::execute): Call instrument_pointer_overflow
17618 and maybe_instrument_pointer_overflow.
17619 * internal-fn.c (expand_UBSAN_PTR): New function.
17620 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17621 * sanitizer.def (__ubsan_handle_pointer_overflow,
17622 __ubsan_handle_pointer_overflow_abort): New builtins.
17623 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17624 * internal-fn.def (UBSAN_PTR): New internal function.
17625 * opts.c (sanitizer_opts): Add pointer-overflow.
17626 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17627 * fold-const.c (build_range_check): Compute pointer range check in
17628 integral type if pointer arithmetics would be needed. Formatting
17631 2017-07-28 Martin Liska <mliska@suse.cz>
17634 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17635 parameters that are of a variable-length.
17637 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17639 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17641 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17642 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17643 (CRT_CALL_STATIC_FUNCTION): Likewise.
17644 (ASM_DEFAULT_SPEC): New define.
17645 (ASM_SPEC32): Likewise.
17646 (ASM_SPEC64): Likewise.
17647 (ASM_SPEC_COMMON): Likewise.
17648 (ASM_SPEC): Likewise.
17649 (INVALID_64BIT): Likewise.
17650 (LINK_OS_DEFAULT_SPEC): Likewise.
17651 (LINK_OS_SPEC32): Likewise.
17652 (LINK_OS_SPEC64): Likewise.
17653 (POWERPC_LINUX): Likewise.
17654 (PTRDIFF_TYPE): Likewise.
17655 (RESTORE_FP_PREFIX): Likewise.
17656 (RESTORE_FP_SUFFIX): Likewise.
17657 (SAVE_FP_PREFIX): Likewise.
17658 (SAVE_FP_SUFFIX): Likewise.
17659 (SIZE_TYPE): Likewise.
17660 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17661 (TARGET_64BIT): Likewise.
17662 (TARGET_64BIT): Likewise.
17663 (TARGET_AIX): Likewise.
17664 (WCHAR_TYPE_SIZE): Likewise.
17665 (WCHAR_TYPE): Undefine.
17666 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17667 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17668 (CPP_OS_RTEMS_SPEC): Delete.
17669 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17670 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17672 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17674 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17676 PR tree-optimization/81578
17677 * tree-parloops.c (build_new_reduction): Bail out if
17678 reduction_code isn't one of the standard OpenMP reductions.
17679 Move the details printing after that decision.
17681 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17683 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17684 related to reload_in_progress.
17685 (splat_input_operand): Likewise.
17686 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17688 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17690 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17691 (TARGET_INSTANTIATE_DECLS): Likewise.
17692 (legitimate_indexed_address_p): Delete reload_in_progress code.
17693 (rs6000_debug_legitimate_address_p): Likewise.
17694 (rs6000_eliminate_indexed_memrefs): Likewise.
17695 (rs6000_emit_le_vsx_store): Likewise.
17696 (rs6000_emit_move_si_sf_subreg): Likewise.
17697 (rs6000_emit_move): Likewise.
17698 (register_to_reg_type): Likewise.
17699 (rs6000_pre_atomic_barrier): Likewise.
17700 (rs6000_machopic_legitimize_pic_address): Likewise.
17701 (rs6000_allocate_stack_temp): Likewise.
17702 (rs6000_address_for_fpconvert): Likewise.
17703 (rs6000_address_for_altivec): Likewise.
17704 (rs6000_secondary_memory_needed_rtx): Delete function.
17705 (rs6000_check_sdmode): Likewise.
17706 (rs6000_alloc_sdmode_stack_slot): Likewise.
17707 (rs6000_instantiate_decls): Likewise.
17708 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17709 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17710 Delete reload_in_progress.
17711 (*vec_reload_and_plus_<mptrsize>): Likewise.
17712 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17713 (vsx_div_v2di): Likewise.
17714 (vsx_udiv_v2di): Likewise.
17716 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17718 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17719 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17720 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17721 (rs6000_debug_reg_global): Delete print of LRA status.
17722 (rs6000_option_override_internal): Delete dead LRA related code.
17723 (rs6000_lra_p): Delete function.
17724 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17726 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17728 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17729 * config/riscv/rtems.h: New file.
17731 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17732 Sudakshina Das <sudi.das@arm.com>
17734 * config/aarch64/aarch64.md
17735 (define_split for and<mode>3nr_compare): Move
17736 non aarch64_logical_operand to a register.
17737 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17738 register immediate operand to a register.
17739 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17741 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17743 PR middle-end/81564
17744 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17746 2017-07-27 Richard Biener <rguenther@suse.de>
17748 PR tree-optimization/81573
17749 PR tree-optimization/81494
17750 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17751 multi defuse cycle case.
17753 2017-07-27 Richard Biener <rguenther@suse.de>
17755 PR tree-optimization/81571
17756 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17759 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17761 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17762 earlier and only if MASK_FPU is set. Adjust formatting.
17764 2017-07-27 Martin Liska <mliska@suse.cz>
17766 * opt-functions.awk: Add validation of value of Init.
17767 * optc-gen.awk: Pass new argument.
17769 2017-07-27 Martin Liska <mliska@suse.cz>
17771 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17772 Fix wrong condition.
17774 2017-07-27 Martin Liska <mliska@suse.cz>
17776 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17777 BBs and edges seen by autoFDO.
17779 2017-07-27 Richard Biener <rguenther@suse.de>
17781 PR tree-optimization/81502
17782 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17783 with incompatible but same sized type.
17784 (execute_update_addresses_taken): Likewise.
17786 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17788 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17789 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17791 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17794 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17795 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17796 Change s_operand to memory_operand.
17798 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17800 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17801 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17802 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17803 Emit instructions rather than returning an expression. Handle TFmode
17804 and KFmode by casting to TImode.
17805 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17806 (rs6000_emit_le_vsx_store): Likewise.
17807 * config/rs6000/vsx.md (VSX_TI): New iterator.
17808 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17809 (*vsx_le_undo_permute_<mode>): Likewise.
17810 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17811 emit the split sequence.
17812 (*vsx_le_perm_store_<mode>): Likewise.
17814 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17816 PR tree-optimization/81555
17817 PR tree-optimization/81556
17818 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17819 if true, force CHANGED for the recursive invocation.
17820 (reassociate_bb): Remember original length of ops array, pass
17821 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17823 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17824 attributes for noipa attribute. For naked attribute use
17825 lookup_attribute first before lookup_attribute_spec.
17826 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17828 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17829 for functions with noipa attribute.
17830 (cgraph_externally_visible_p): Return true for functions with noipa
17832 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17833 for functions with noipa attribute.
17834 * doc/extend.texi: Document noipa function attribute.
17835 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17836 also for functions with noipa attribute.
17837 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17839 2017-07-26 Andrew Pinski <apinski@cavium.com>
17841 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17842 vec_unalign_load_cost and vec_unalign_store_cost.
17844 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17846 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17847 -mvsx-small-integer option.
17848 (ISA_3_0_MASKS_IEEE): Likewise.
17849 (OTHER_VSX_VECTOR_MASKS): Likewise.
17850 (POWERPC_MASKS): Likewise.
17851 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17852 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17853 code, only testing for DImode being allowed in non-VSX floating
17855 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17856 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17858 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17859 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17860 TARGET_P8_VECTOR test.
17861 (rs6000_secondary_reload_simple_move): Likewise.
17862 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17863 since TARGET_P9_VECTOR was already tested.
17864 (rs6000_opt_masks): Remove -mvsx-small-integer.
17865 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17866 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17868 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17869 test for TARGET_VEXTRACTUB was used, and that uses
17871 (p9 extract splitter): Likewise.
17872 (vsx_extract_<mode>_di_p9): Likewise.
17873 (vsx_extract_<mode>_store_p9): Likewise.
17874 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17875 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17876 the elimination of TARGET_VSX_SMALL_INTEGER.
17877 (vsx_extract_<mode>_p8): Likewise.
17878 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17879 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17880 (vsx_set_<mode>_p9): Likewise.
17881 (vsx_set_v4sf_p9): Likewise.
17882 (vsx_set_v4sf_p9_zero): Likewise.
17883 (vsx_insert_extract_v4sf_p9): Likewise.
17884 (vsx_insert_extract_v4sf_p9_2): Likewise.
17885 * config/rs6000/rs6000.md (sign extend splitter): Change
17886 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17887 (floatsi<mode>2_lfiwax_mem): Likewise.
17888 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17889 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17890 since a test for TARGET_P9_VECTOR was used.
17891 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17892 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17893 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17894 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17895 TARGET_P8_VECTOR test.
17896 (fix_trunc<mode>si2_stfiwx): Likewise.
17897 (fix_trunc<mode>si2_internal): Likewise.
17898 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17899 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17901 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17902 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17903 TARGET_P8_VECTOR test.
17904 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17905 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17906 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17908 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17909 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17910 since a test for TARGET_P9_VECTOR was used.
17911 (splitter for loading small constants): Likewise.
17913 2017-07-26 Andrew Pinski <apinski@cavium.com>
17915 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17918 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17921 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17922 (fp_valid_at): Likewise.
17924 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17926 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17927 (qdf24xx_addrcost_table): Likewise.
17928 (cortexa57_tunings): Update to use generic_branch_cost.
17929 (cortexa72_tunings): Likewise.
17930 (cortexa73_tunings): Likewise.
17931 (qdf24xx_tunings): Likewise.
17933 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17935 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17936 (thunderx2t99_branch_cost): Likewise.
17937 (cortexa35_tunings): Update to use generic_branch_cost.
17938 (cortexa53_tunings): Likewise.
17939 (cortexa57_tunings): Likewise.
17940 (cortexa72_tunings): Likewise.
17941 (cortexa73_tunings): Likewise.
17942 (thunderx2t99_tunings): Likewise.
17944 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17946 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17947 (sparc_option_override): Honour MASK_FSMULD.
17948 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17949 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17950 * config/sparc/sparc.opt (mfsmuld): New option.
17951 * doc/invoke.texi (mfsmuld): Document option.
17953 2017-07-26 Marek Polacek <polacek@redhat.com>
17955 PR middle-end/70992
17956 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17958 2017-07-26 Richard Biener <rguenther@suse.de>
17960 * gimple-match-head.c (do_valueize): Return OP if valueize
17962 (get_def): New helper to get at the def stmt of a SSA name
17963 if valueize allows.
17964 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17965 do_valueize to get at the def stmt.
17966 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17968 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17970 PR middle-end/46932
17971 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17973 2017-07-26 Martin Liska <mliska@suse.cz>
17976 * function.c (expand_function_start): Make expansion of
17977 nonlocal_goto_save_area after parm_birth_insn.
17979 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17981 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17982 from all CPU target flags enable members.
17984 2017-07-26 Richard Biener <rguenther@suse.de>
17986 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17987 (decision_tree::gen): Make 'type' const.
17988 (write_predicate): Likewise.
17990 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17992 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17993 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17994 (rs6000_option_override_internal): Likewise.
17995 (rs6000_expand_vector_set): Likewise.
17996 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17997 (TARGET_UPPER_REGS_SF): Likewise.
17998 (TARGET_UPPER_REGS_DI): Likewise.
17999 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18000 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18001 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18002 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18003 (Splitters for DI constants in Altivec registers): Likewise.
18004 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18005 (vsx_set_v4sf_p9): Likewise.
18006 (vsx_set_v4sf_p9_zero): Likewise.
18007 (vsx_insert_extract_v4sf_p9): Likewise.
18008 (vsx_insert_extract_v4sf_p9_2): Likewise.
18010 2017-07-25 Carl Love <cel@us.ibm.com>
18012 * doc/extend.texi: Update the built-in documentation file for the
18013 existing built-in functions
18014 vector signed char vec_cnttz (vector signed char);
18015 vector unsigned char vec_cnttz (vector unsigned char);
18016 vector signed short vec_cnttz (vector signed short);
18017 vector unsigned short vec_cnttz (vector unsigned short);
18018 vector signed int vec_cnttz (vector signed int);
18019 vector unsigned int vec_cnttz (vector unsigned int);
18020 vector signed long long vec_cnttz (vector signed long long);
18021 vector unsigned long long vec_cnttz (vector unsigned long long);
18023 2017-07-25 Andrew Pinski <apinski@cavium.com>
18025 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18026 accesses where the use is for the first operand of a BIT_INSERT.
18028 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18031 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18032 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18034 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18036 * config/i386/gstabs.h: Delete.
18037 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18039 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18041 * config/i386/i386.c (ix86_decompose_address): Do not check for
18042 register RTX when looking at index_reg or base_reg.
18043 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18045 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18047 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18048 to update EH info here.
18050 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18052 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18054 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18056 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18058 2017-07-25 Torsten Duwe <duwe@suse.de>
18060 * common.opt: Introduce -fpatchable-function-entry
18061 command line option, and its variables function_entry_patch_area_size
18062 and function_entry_patch_area_start.
18063 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18064 including a two-value parser.
18065 * target.def (print_patchable_function_entry): New target hook.
18066 * targhooks.h (default_print_patchable_function_entry): New function.
18067 * targhooks.c (default_print_patchable_function_entry): Likewise.
18068 * toplev.c (process_options): Switch off IPA-RA if
18069 patchable function entries are being generated.
18070 * varasm.c (assemble_start_function): Look at the
18071 patchable-function-entry command line switch and current
18072 function attributes and maybe generate NOP instructions by
18073 calling the print_patchable_function_entry hook.
18074 * doc/extend.texi: Document patchable_function_entry attribute.
18075 * doc/invoke.texi: Document -fpatchable_function_entry
18076 command line option.
18077 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18079 * doc/tm.texi: Re-generate.
18081 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18084 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18085 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18087 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18089 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18091 2017-07-25 Richard Biener <rguenther@suse.de>
18093 PR tree-optimization/81455
18094 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18095 not walk in cycles when looking for guards.
18097 2017-07-25 Richard Biener <rguenther@suse.de>
18099 PR tree-optimization/81529
18100 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18101 when optimizing backedge uses.
18103 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18105 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18107 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18108 to dl_section_ref. On AIX, append an expression to subtract
18109 the size of the section length to dl_section_ref.
18111 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18113 * configure.ac: If any of the config.* scripts fail, exit 1.
18114 * configure: Regenerate.
18116 2017-07-25 Richard Biener <rguenther@suse.de>
18118 PR middle-end/81546
18119 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18120 of immediate uses, be more verbose on errors.
18122 2017-07-25 Richard Biener <rguenther@suse.de>
18124 PR tree-optimization/81510
18125 * tree-vect-loop.c (vect_is_simple_reduction): When the
18126 reduction stmt is not inside the loop bail out.
18128 2017-07-25 Richard Biener <rguenther@suse.de>
18130 PR tree-optimization/81303
18131 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18132 profitability check against LOOP_VINFO_NITERSM1.
18134 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18136 * domwalk.c (cmp_bb_postorder): Simplify.
18137 (sort_bbs_postorder): New function. Use it...
18138 (dom_walker::walk): ...here to optimize common cases.
18140 2017-07-25 Martin Liska <mliska@suse.cz>
18143 * ipa-visibility.c (function_and_variable_visibility): Make the
18144 redirection just on target that supports aliasing.
18145 Fix GNU coding style.
18147 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18150 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18152 * config/arm/rtems.h: Likewise.
18153 * config/bfin/rtems.h: Likewise.
18154 * config/i386/rtemself.h: Likewise.
18155 * config/lm32/rtems.h: Likewise.
18156 * config/m32c/rtems.h: Likewise.
18157 * config/m68k/rtemself.h: Likewise.
18158 * config/microblaze/rtems.h: Likewise.
18159 * config/mips/rtems.h: Likewise.
18160 * config/moxie/rtems.h: Likewise.
18161 * config/nios2/rtems.h: Likewise.
18162 * config/powerpcspe/rtems.h: Likewise.
18163 * config/rs6000/rtems.h: Likewise.
18164 * config/rtems.h: Likewise.
18165 * config/sh/rtems.h: Likewise.
18166 * config/sh/rtemself.h: Likewise.
18167 * config/sparc/rtemself.h: Likewise.
18169 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18172 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18173 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18174 * tree-ssa-structalias.c (alias_get_name): Same.
18176 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18179 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18180 instructions if no du chain is found.
18182 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18184 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18186 2017-07-25 Richard Biener <rguenther@suse.de>
18188 PR middle-end/81505
18189 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18192 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18194 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18195 upper-regs options.
18196 (ISA_2_7_MASKS_SERVER): Likewise.
18197 (ISA_3_0_MASKS_IEEE): Likewise.
18198 (OTHER_P8_VECTOR_MASKS): Likewise.
18199 (OTHER_VSX_VECTOR_MASKS): Likewise.
18200 (POWERPC_MASKS): Likewise.
18201 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18202 duplicate list of options.
18203 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18204 explicit -mupper-regs options.
18205 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18206 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18207 alias for -mupper-regs-df.
18208 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18209 (rs6000_init_hard_regno_mode_ok): Likewise.
18210 (rs6000_option_override_internal): Likewise.
18211 (rs6000_opt_masks): Likewise.
18212 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18213 options in terms of whether -mvsx or -mpower8-vector was used.
18214 (TARGET_UPPER_REGS_DI): Likewise.
18215 (TARGET_UPPER_REGS_SF): Likewise.
18216 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18217 -mupper-regs-* options.
18219 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18221 * passes.c (emergency_dump_function): Print some empty lines and a
18222 header before the RTL dump.
18224 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18226 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18228 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18231 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18232 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18234 2017-07-24 Carl Love <cel@us.ibm.com>
18236 * config/rs6000/rs6000-c.c: Add support for built-in functions
18237 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18238 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18239 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18240 vec_extract_fp_from_shortl): Add defines for the two builtins.
18241 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18242 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18244 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18245 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18246 * doc/extend.texi: Update the built-in documentation file for the
18247 new built-in function.
18249 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18252 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18254 * doc/generic.texi: Likewise.
18255 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18256 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18258 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18260 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18261 (aarch64_mls_elt_merge<mode>): Likewise.
18263 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18265 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18266 having __cxa_atexit.
18268 2017-07-23 Michael Collison <michael.collison@arm.com>
18270 * config/arm/arm.c (arm_option_override): Deprecate
18271 use of -mstructure-size-boundary.
18272 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18273 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18275 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18278 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18279 Reduce cost estimate for direct moves.
18281 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18284 * config/i386/i386.c (ix86_option_override_internal): Disable
18285 BMI, BMI2 and TBM instructions for -m16.
18287 2017-07-21 Carl Love <cel@us.ibm.com>
18289 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18290 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18291 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18292 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18293 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18294 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18295 VMULOSW): New enum "unspec" values.
18296 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18297 altivec_vmulosw): New patterns.
18298 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18299 VMULOSW): Add definitions.
18301 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18303 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18304 (qdf24xx): Likewise.
18305 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18306 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18307 (AARCH64_FL_V8_1): Renumber.
18308 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18309 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18310 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18311 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18312 rdma to feature modifiers list.
18314 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18316 PR middle-end/56727
18317 * ipa-visibility (function_and_variable_visibility): Convert
18318 recursive PLT call to direct call if appropriate.
18320 2017-07-21 Andrew Pinski <apinski@cavium.com>
18322 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18323 operand 1 to see if the types precision matches.
18324 * fold-const.c (operand_equal_p): Likewise.
18326 2017-07-21 Richard Biener <rguenther@suse.de>
18328 PR tree-optimization/81303
18329 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18330 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18331 (vect_peeling_hash_get_lowest_cost): Adjust.
18332 (vect_enhance_data_refs_alignment): Likewise. Use
18333 vect_get_peeling_costs_all_drs to compute the penalty for no
18334 peeling to match up costs.
18336 2017-07-21 Richard Biener <rguenther@suse.de>
18338 PR tree-optimization/81500
18339 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18340 we didn't identify a reduction path.
18342 2017-07-21 Tom de Vries <tom@codesourcery.com>
18343 Cesar Philippidis <cesar@codesourcery.com>
18345 PR gcov-profile/81442
18346 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18349 2017-07-21 Tom de Vries <tom@codesourcery.com>
18352 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18354 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18355 nvptx_override_options_after_change.
18357 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18359 * dwarf2out.c (output_file_names): Avoid double testing for
18360 dwarf_version >= 5.
18362 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18364 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18366 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18368 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18370 (try_crossjump_to_edge): Do not punt on partitioned functions.
18372 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18374 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18375 Put all BBs reachable only via paths crossing cold region to cold
18377 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18379 2016-07-21 Richard Biener <rguenther@suse.de>
18381 PR tree-optimization/81303
18382 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18383 into account prologue and epilogue iterations when raising
18384 min_profitable_iters to sth at least covering one vector iteration.
18386 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18388 * config/arm/arm.c (arm_test_cpu_arch_dat):
18391 2017-07-20 Nathan Sidwell <nathan@acm.org>
18393 Remove TYPE_METHODS.
18394 * tree.h (TYPE_METHODS): Delete.
18395 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18396 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18397 (dbxout_type_methods): Scan TYPE_FIELDS.
18398 (dbxout_type): Don't check TYPE_METHODS here.
18399 * function.c (use_register_for_decl): Always ignore register for
18400 class types when not optimizing.
18401 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18402 * tree.c (free_lang_data_in_type): Stitch out member functions and
18403 templates from TYPE_FIELDS.
18404 (build_distinct_type_copy, verify_type_variant,
18405 verify_type): Member fns are on TYPE_FIELDS.
18406 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18407 * tree-pretty-print.c (dump_generic_node): Likewise.
18409 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18412 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18413 V2TImode and V4TImode.
18414 (ix86_expand_vector_extract): Likewise.
18415 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18416 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18417 (ssescalarmode): Handle V4TImode and V2TImode.
18418 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18419 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18420 (VEXTRACTI128_MODE): New mode iterator.
18421 (splitter for *vec_extractv?ti first element): New.
18422 (VEC_INIT_MODE): New mode iterator.
18423 (vec_init<mode>): Consolidate 3 expanders into one using
18424 VEC_INIT_MODE mode iterator.
18426 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18428 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18429 non_spilled_static_chain_regno_p.
18431 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18433 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18435 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18437 * bb-reorder.c (connect_traces): Allow copying of blocks within
18440 2017-07-20 Richard Biener <rguenther@suse.de>
18442 * gimple.h (gimple_phi_result): Add gphi * overload.
18443 (gimple_phi_result_ptr): Likewise.
18444 (gimple_phi_arg): Likewise. Adjust index assert to only
18445 allow actual argument accesses rather than all slots available
18447 (gimple_phi_arg_def): Add gphi * overload.
18448 * tree-phinodes.c (make_phi_node): Initialize only actual
18450 (resize_phi_node): Clear memory not covered by old node,
18451 do not initialize excess argument slots.
18452 (reserve_phi_args_for_new_edge): Initialize new argument slot
18455 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18457 PR tree-optimization/81388
18459 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18461 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18462 by removing computation of may_be_zero.
18464 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18465 Tom de Vries <tom@codesourcery.com>
18467 PR middle-end/81030
18468 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18469 when gimple level profile disagrees with what RTL expander did.
18471 2017-07-20 Richard Biener <rguenther@suse.de>
18473 PR tree-optimization/61171
18474 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18475 (vect_analyze_stmt): Add slp instance parameter.
18476 (vectorizable_reduction): Likewise.
18477 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18478 (vect_is_simple_reduction): Deal with chains not detected
18479 as SLP reduction chain, specifically not properly associated
18480 chains containing a mix of plus/minus.
18481 (get_reduction_op): Remove.
18482 (get_initial_defs_for_reduction): Simplify, pass in whether
18483 this is a reduction chain, pass in the SLP node for the PHIs.
18484 (vect_create_epilog_for_reduction): Get the SLP instance as
18486 (vectorizable_reduction): Get the SLP instance as arg.
18487 During analysis remember the SLP node with the PHIs in the
18488 instance. Simplify getting at the vectorized reduction PHIs.
18489 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18490 through SLP instance.
18491 (vect_slp_analyze_operations): Likewise.
18492 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18493 (vect_transform_stmt): Likewise.
18495 2017-07-20 Tom de Vries <tom@codesourcery.com>
18497 PR tree-optimization/81489
18498 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18499 read of phi arg location to before loop that modifies phi.
18501 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18503 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18506 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18508 PR middle-end/81331
18509 * except.c (execute): Fix ordering issue.
18511 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18513 PR rtl-optimization/81423
18514 * combine.c (make_compound_operation_int): Don't try to optimize
18515 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18517 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18519 PR rtl-optimization/81423
18520 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18521 with a constant that is -1 in the truncated to mode.
18523 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18525 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18526 (determine_unlikely_bbs): ... here.
18527 * predict.h (propagate_unlikely_bbs_forward): Declare.
18528 * cfgexpand.c (pass_expand::execute): Use it.
18529 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18531 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18532 propagate_unlikely_bbs_forward.
18534 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18536 PR middle-end/81331
18537 * except.c (maybe_add_nop_after_section_switch): New function.
18540 2017-07-19 Tom de Vries <tom@codesourcery.com>
18542 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18544 2017-07-19 Tom de Vries <tom@codesourcery.com>
18546 * gimple.h (gimple_phi_arg): Make assert more strict.
18548 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18550 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18551 * config/rs6000/mmintrin.h: New file.
18552 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18554 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18556 PR tree-optimization/81346
18557 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18559 2017-07-19 Tom de Vries <tom@codesourcery.com>
18561 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18563 2017-07-19 Tom de Vries <tom@codesourcery.com>
18565 * config/nvptx/nvptx-modes.def: Add V2DImode.
18566 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18567 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18568 (nvptx_output_mov_insn): Handle lack of mov.b128.
18569 (nvptx_print_operand): Handle 'H' and 'L' codes.
18570 (nvptx_vector_mode_supported): Allow V2DImode.
18571 (nvptx_preferred_simd_mode): New function.
18572 (nvptx_data_alignment): New function.
18573 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18574 nvptx_preferred_simd_mode.
18575 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18577 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18579 2017-07-19 Tom de Vries <tom@codesourcery.com>
18581 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18582 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18583 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18584 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18585 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18586 (mov<VECIM>_insn): New define_insn.
18587 (define_expand "mov<VECIM>): New define_expand.
18589 2017-07-19 Tom de Vries <tom@codesourcery.com>
18591 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18593 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18595 PR tree-optimization/81346
18596 * fold-const.h (fold_div_compare, range_check_type): Declare.
18597 * fold-const.c (range_check_type): New function.
18598 (build_range_check): Use range_check_type.
18599 (fold_div_compare): No longer static, rewritten into
18600 a match.pd helper function.
18601 (fold_comparison): Don't call fold_div_compare here.
18602 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18603 as helper function.
18605 2017-07-19 Nathan Sidwell <nathan@acm.org>
18607 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18608 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18609 * tree.c (find_decls_types_r, verify_type): Use
18610 TYPE_{MIN,MAX}_VALUE_RAW.
18611 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18612 (hash_tree): Likewise.
18613 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18615 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18618 2017-07-18 Tom de Vries <tom@codesourcery.com>
18620 PR middle-end/81464
18621 * omp-expand.c (expand_omp_for_static_chunk): Handle
18622 equal-argument loop exit phi.
18624 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18627 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18628 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18629 operand 2 predicate.
18630 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18631 operand 2 predicate.
18632 (ror,rol -> rorx splitters): Use const_int_operand as
18633 operand 2 predicate.
18635 2017-06-18 Richard Biener <rguenther@suse.de>
18637 PR tree-optimization/81410
18638 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18639 the gap in the ! slp_perm SLP case after each group.
18641 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18643 PR middle-end/81463
18644 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18647 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18649 PR middle-end/81462
18650 * predict.c (set_even_probabilities): Cleanup; do not affect
18651 probabilities that are already known.
18652 (combine_predictions_for_bb): Call even when count is set.
18654 2017-07-18 Nathan Sidwell <nathan@acm.org>
18656 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18659 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18662 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18663 optimization for loop niter analysis.
18665 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18668 * config/avr/avr.c (avr_optimize_casesi): Don't use
18669 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18671 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18673 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18674 body_cost_vec from _vect_peel_extended_info.
18675 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18676 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18679 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18681 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18683 2017-07-18 Richard Biener <rguenther@suse.de>
18685 PR tree-optimization/80620
18686 PR tree-optimization/81403
18687 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18688 info when re-using a VN table entry.
18690 2017-07-18 Richard Biener <rguenther@suse.de>
18692 PR tree-optimization/81418
18693 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18694 vectype_in. Verify that with lane-reducing reduction operations
18695 we have a single def-use cycle.
18697 2017-07-17 Carl Love <cel@us.ibm.com>
18699 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18701 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18702 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18703 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18704 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18705 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18706 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18707 VMULOSW): New enum "unspec" values.
18708 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18709 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18710 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18711 altivec_vmulosw): New patterns.
18712 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18713 VMULOSW): Add definitions.
18714 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18716 * config/alpha/alpha.c: Include predict.h.
18718 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18720 * tree-vrp.c (compare_assert_loc): Fix comparison function
18721 to return predictable results.
18723 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18725 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18727 (subdi3): Likewise.
18728 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18729 * doc/invoke.texi (mexpand-adddi): Update text.
18731 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18733 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18734 that also clobbers the CC register. The old expand code is moved
18736 (*arc_clzsi2): ... here.
18737 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18738 the CC register. The old expand code is moved to ...
18739 (arc_ctzsi2): ... here.
18741 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18743 * config/arc/arc.opt (mindexed-loads): Use initial value
18744 TARGET_INDEXED_LOADS_DEFAULT.
18745 (mauto-modify-reg): Use initial value
18746 TARGET_AUTO_MODIFY_REG_DEFAULT.
18747 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18748 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18749 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18750 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18752 2017-07-17 Martin Liska <mliska@suse.cz>
18755 * opts.c (finish_options): Do not allow -fgnu-tm
18756 w/ -fsanitize={kernel-,}address. Say sorry.
18758 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18761 * tree-loop-distribution.c (classify_partition): Only assert on
18762 numer of iterations.
18763 (merge_dep_scc_partitions): Delete prameter. Update function call.
18764 (distribute_loop): Remove code handling loop with unknown niters.
18765 (pass_loop_distribution::execute): Skip loop with unknown niters.
18767 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18770 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18771 function sort_partitions_by_post_order.
18773 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18775 PR tree-optimization/81374
18776 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18777 the max index of basic blocks, rather than number of basic blocks.
18779 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18781 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18783 (arc_legitimate_pic_operand_p): Likewise.
18784 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18786 (arc_needs_pcl_p): Likewise.
18787 (arc_legitimate_pc_offset_p): Likewise.
18788 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18789 function is also used in constrains.md.
18790 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18791 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18792 PLUS. Only return true/false in known cases, otherwise assert.
18793 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18794 is already called in arc_legitimate_constant_p.
18795 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18797 (LEGITIMATE_PIC_OPERAND_P): Use
18798 arc_raw_symbolic_reference_mentioned_p function.
18799 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18804 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18805 Andrew Burgess <andrew.burgess@embecosm.com>
18807 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18808 (arc_return_address_register): New function.
18809 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18810 (arc_handle_fndecl_attribute): Add naked attribute.
18811 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18812 (TARGET_WARN_FUNC_RETURN): Likewise.
18813 (arc_allocate_stack_slots_for_args): New function.
18814 (arc_warn_func_return): Likewise.
18815 (machine_function): Change type fn_type.
18816 (arc_compute_function_type): Consider new naked function type,
18817 change function return type.
18818 (arc_must_save_register): Adapt to handle new
18819 arc_compute_function_type's return type.
18820 (arc_expand_prologue): Likewise.
18821 (arc_expand_epilogue): Likewise.
18822 (arc_return_address_regs): Delete.
18823 (arc_return_address_register): New function.
18824 (arc_epilogue_uses): Use above function.
18825 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18826 (arc_function_type): Change encoding, add naked type.
18827 (ARC_INTERRUPT_P): Change to handle the new encoding.
18828 (ARC_FAST_INTERRUPT_P): Likewise.
18829 (ARC_NORMAL_P): Define.
18830 (ARC_NAKED_P): Likewise.
18831 (arc_compute_function_type): Delete prototype.
18832 * config/arc/arc.md (in_ret_delay_slot): Use
18833 arc_return_address_register function.
18834 (simple_return): Likewise.
18835 (p_return_i): Likewise.
18837 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18839 PR tree-optimization/81428
18840 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18841 can't be built for those types.
18843 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18845 Remove stuff dead since r239246.
18847 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18848 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18849 (avr_inform_devices): Remove dead stuff.
18851 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18853 * config/arm/arm_neon.h: Fix softp typo.
18855 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18857 PR tree-optimization/81365
18858 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18859 aggregate moves onto bb predecessor edges, make sure there are no
18860 loads that could alias the lhs in between the start of bb and the
18863 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18866 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18867 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18868 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18870 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18872 PR tree-optimization/81396
18873 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18874 (init_symbolic_number): Initialize it to 1.
18875 (perform_symbolic_merge): Add n_ops from both operands into the new
18877 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18878 without base_addr as useless if they need more than one operation.
18879 (bswap_replace): Handle !bswap case for NULL base_addr.
18881 2017-07-17 Tom de Vries <tom@codesourcery.com>
18884 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18887 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18889 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18890 conditional builtin define __FIX_LEON3FT_B2BST.
18892 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18894 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18895 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18898 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18900 PR rtl-optimization/81424
18901 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18902 to remove potential trapping from operands if -fnon-call-exceptions.
18904 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18906 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18907 profile_proability for scalling.
18908 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18910 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18912 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18914 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18916 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18917 fixpoint arithmetics.
18919 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18921 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18922 fixpoint arithmetics.
18924 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18926 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18927 fixpoint arithmetics.
18929 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18931 * profile-count.h (profile_probability::from_reg_br_prob_note,
18932 profile_probability::to_reg_br_prob_note): New functions.
18933 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18934 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18935 * predict.c (probability_reliable_p): Update.
18936 (edge_probability_reliable_p): Update.
18937 (br_prob_note_reliable_p): Update.
18938 (invert_br_probabilities): Update.
18939 (add_reg_br_prob_note): New function.
18940 (combine_predictions_for_insn): Update.
18941 * asan.c (asan_clear_shadow): Update.
18942 * cfgbuild.c (compute_outgoing_frequencies): Update.
18943 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18944 (update_br_prob_note): Update.
18945 (rtl_verify_edges): Update.
18946 (purge_dead_edges): Update.
18947 (fixup_reorder_chain): Update.
18948 * emit-rtl.c (try_split): Update.
18949 * ifcvt.c (cond_exec_process_insns): Update.
18950 (cond_exec_process_if_block): Update.
18951 (dead_or_predicable): Update.
18952 * internal-fn.c (expand_addsub_overflow): Update.
18953 (expand_neg_overflow): Update.
18954 (expand_mul_overflow): Update.
18955 * loop-doloop.c (doloop_modify): Update.
18956 * loop-unroll.c (compare_and_jump_seq): Update.
18957 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18958 * predict.h: Update.
18959 * reorg.c (mostly_true_jump): Update.
18961 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18962 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18963 * config/arc/arc.c: (emit_unlikely_jump): Update.
18964 * config/arm/arm.c: (emit_unlikely_jump): Update.
18965 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18966 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18967 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18968 (ix86_print_operand): Update.
18969 (ix86_split_fp_branch): Update.
18970 (predict_jump): Update.
18971 * config/ia64/ia64.c (ia64_print_operand): Update.
18972 * config/mmix/mmix.c (mmix_print_operand): Update.
18973 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18974 (rs6000_expand_split_stack_prologue): Update.
18975 * config/rs6000/rs6000.c: Update.
18976 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18977 (s390_expand_vec_movstr): Update.
18978 (s390_expand_cs_tdsi): Update.
18979 (s390_expand_split_stack_prologue): Update.
18980 * config/sh/sh.c (sh_print_operand): Update.
18981 (expand_cbranchsi4): Update.
18982 (expand_cbranchdi4): Update.
18983 * config/sparc/sparc.c (output_v9branch): Update.
18984 * config/spu/spu.c (get_branch_target): Update.
18985 (ea_load_store_inline): Update.
18986 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18987 * config/tilepro/tilepro.c: Update.
18989 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18991 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18992 (gimplify_save_expr): Likewise.
18994 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18996 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18998 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19000 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19002 * timevar.def (TV_IPA_FNSUMMARY): Define.
19004 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19006 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19007 to back store errata sensitive sequence from being generated.
19008 (sqrtdf2_fix): Likewise.
19010 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19012 * tree-ssa-threadupdate.c (compute_path_counts,
19013 update_joiner_offpath_counts): Use profile_probability.
19015 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19018 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19020 * config/arm/arm-c.c (arm_cpu_builtins): Define
19021 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19023 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19025 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19026 array entries to represent __ieee128 versions of the
19027 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19028 scalar_extract_sig, and scalar_insert_exp built-in functions.
19029 (altivec_resolve_overloaded_builtin): Add special case handling
19030 for the __builtin_scalar_insert_exp function, as represented by
19031 the P9V_BUILTIN_VEC_VSIEDP constant.
19032 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19033 exponent support for __ieee128 argument.
19034 (VSESQP): Add scalar extract signature support for __ieee128
19036 (VSTDCNQP): Add scalar test negative support for __ieee128
19038 (VSIEQP): Add scalar insert exponent support for __int128 argument
19039 with __ieee128 result.
19040 (VSIEQPF): Add scalar insert exponent support for __ieee128
19041 argument with __ieee128 result.
19042 (VSTDCQP): Add scalar test data class support for __ieee128
19044 (VSTDCNQP): Add overload support for scalar test negative with
19045 __ieee128 argument.
19046 (VSTDCQP): Add overload support for scalar test data class
19047 __ieee128 argument.
19048 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19049 UNSPEC_VSX_SXSIGDP.
19050 (UNSPEC_VSX_SIEXPQP): New constant.
19051 (xsxexpqp): New insn for VSX scalar extract exponent quad
19053 (xsxsigqp): New insn for VSX scalar extract significand quad
19055 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19056 precision with floating point argument.
19057 (xststdcqp): New expand for VSX scalar test data class quad
19059 (xststdcnegqp): New expand for VSX scalar test negative quad
19061 (xststdcqp): New insn to match expansions for VSX scalar test data
19062 class quad precision and VSX scalar test negative quad precision.
19063 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19064 special case operand checking to enforce that second operand of
19065 VSX scalar test data class with quad precision argument is a 7-bit
19067 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19068 prototypes and descriptions of __ieee128 versions of
19069 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19070 scalar_test_data_class, and scalar_test_neg built-in functions.
19072 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19074 PR tree-optimization/81162
19075 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19076 replace a negate with an add.
19078 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19080 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19082 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19084 * config/arm/arm-c.c (arm_cpu_builtins): Define
19085 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19087 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19089 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19090 (armv8-r): Set ARM Cortex-R52 as default CPU.
19091 * config/arm/arm-tables.opt: Regenerate.
19092 * config/arm/arm-tune.md: Regenerate.
19093 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19095 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19096 extension for -mcpu=cortex-r52.
19098 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19100 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19101 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19102 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19103 (fp-armv8): Define it as FP_ARMv8 only.
19104 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19105 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19107 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19108 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19109 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19110 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19111 than TARGET_FPU_ARMV8.
19112 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19113 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19114 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19116 * config/arm/neon.md (neon_vrint): Likewise.
19117 (neon_vcvt): Likewise.
19118 (neon_<fmaxmin_op><mode>): Likewise.
19119 (<fmaxmin><mode>3): Likewise.
19120 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19121 * config/arm/predicates.md (arm_cond_move_operator): Check against
19122 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19124 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19126 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19127 to top of function.
19129 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19131 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19132 loop in comment with memset.
19134 2017-07-14 Martin Liska <mliska@suse.cz>
19136 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19137 * dwarf2out.c (is_java): Remove the function.
19138 (output_pubname): Remove usage of the function.
19139 (lower_bound_default): Remove usage of DW_LANG_Java.
19140 (gen_compile_unit_die): Likewise.
19141 * gcc.c: Remove compiler defaults for .java and .zip files.
19142 * gimple-expr.c (remove_suffix): Change as there's no longer
19143 extension than 4-letter one.
19144 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19145 (gimplify_save_expr): Likewise.
19146 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19147 as it's possible even for other languages than Java.
19148 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19149 * lto-opts.c (lto_write_options): Remove reference to Java.
19150 * opts.c (strip_off_ending): Update file extension handling.
19151 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19152 * tree-eh.c (lower_resx): Likewise.
19153 * tree.c (free_lang_data_in_type): Remove dead code.
19154 (find_decls_types_r): Likewise.
19155 (build_common_builtin_nodes): Remove Java from a comment.
19156 (verify_type): Remove dead code.
19157 * varasm.c (assemble_external): Remove Java from a comment.
19159 2017-07-14 Martin Liska <mliska@suse.cz>
19161 * opts.c (finish_options): Add quotes.
19162 (common_handle_option): Likewise.
19164 2017-07-14 Martin Liska <mliska@suse.cz>
19166 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19167 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19168 Remove N_SO_PASCAL.
19169 * dwarf2out.c (lower_bound_default): Do not handle
19171 (gen_compile_unit_die): Likewise.
19172 * gcc.c: Remove default extension binding for GNU Pascal.
19173 * stmt.c: Remove Pascal language from a comment.
19174 * xcoffout.c: Likewise.
19176 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19179 * diagnostic-show-locus.c (fixit_cmp): New function.
19180 (layout::layout): Sort m_fixit_hints.
19181 (column_range::column_range): Assert that the values are valid.
19182 (struct char_span): New struct.
19183 (correction::overwrite): New method.
19184 (struct source_line): New struct.
19185 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19186 calls in terms of classes source_line and char_span, and
19187 correction::overwrite.
19188 (selftest::test_overlapped_fixit_printing_2): New function.
19189 (selftest::diagnostic_show_locus_c_tests): Call it.
19191 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19193 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19194 early if there is no lhs.
19196 2017-07-13 Martin Liska <mliska@suse.cz>
19198 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19199 (gen_reference_type_die): Likewise.
19200 * stor-layout.c: Remove Pascal-related comment.
19202 2017-07-13 Martin Liska <mliska@suse.cz>
19204 * opts.c (finish_options): Add quotes to error messages.
19205 (parse_sanitizer_options): Likewise.
19207 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19209 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19211 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19213 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19215 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19217 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19219 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19221 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19224 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19225 provides the hardware capability bits, define the macro
19226 __BUILTIN_CPU_SUPPORTS__.
19227 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19228 if GLIBC does not provide the hardware capability bits. Add a
19229 gcc_unreachable call if the built-in cpu function is neither
19230 __builtin_cpu_is nor __builtin_cpu_supports.
19231 (rs6000_get_function_versions_dispatcher): Change the warning
19232 that an old GLIBC is used which does not export the capability
19233 bits to be an error.
19234 * doc/extend.texi (target_clones attribute): Document the
19235 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19236 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19237 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19238 the macros defined by GCC if the newer GLIBC is available.
19240 2017-07-12 Jeff Law <law@redhat.com>
19242 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19243 remaining includes slightly.
19244 * config/riscv/riscv-builtins.c: Include profile-count.h.
19246 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19249 * config/avr/avr.c (avr_set_current_function): In diagnostic
19250 messages: Quote keywords and (parts of) identifiers.
19251 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19254 2017-07-12 Carl Love <cel@us.ibm.com>
19256 * config/rs6000/rs6000-c.c: Add support for built-in functions
19257 vector bool char vec_revb (vector bool char);
19258 vector bool short vec_revb (vector short char);
19259 vector bool int vec_revb (vector bool int);
19260 vector bool long long vec_revb (vector bool long long);
19261 * doc/extend.texi: Update the built-in documentation file for the
19262 new built-in functions.
19264 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19266 * config/s390/s390.md: Remove movcc splitter.
19268 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19270 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19271 load/store on condition.
19273 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19276 * config/avr/avr.c (avr_encode_section_info)
19277 [progmem && !TREE_READONLY]: Error if progmem object needs
19280 2017-07-11 Michael Collison <michael.collison@arm.com>
19282 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19285 2017-07-11 Carl Love <cel@us.ibm.com>
19287 * config/rs6000/rs6000-c.c: Add support for builtins
19288 vector unsigned int vec_parity_lsbb (vector signed int);
19289 vector unsigned int vec_parity_lsbb (vector unsigned int);
19290 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19291 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19292 vector unsigned long long vec_parity_lsbb (vector signed long long);
19293 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19294 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19295 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19296 * doc/extend.texi: Update the built-in documentation file for the
19297 new built-in functions.
19299 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19301 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19302 (layout::m_primary_loc): New field.
19303 (layout::layout): Initialize new field. Move location filtering
19304 logic from here to...
19305 (layout::maybe_add_location_range): ...this new method. Add
19306 support for filtering to just the lines already specified by other
19308 (layout::will_show_line_p): New method.
19309 (gcc_rich_location::add_location_if_nearby): New method.
19310 (selftest::test_add_location_if_nearby): New test function.
19311 (selftest::diagnostic_show_locus_c_tests): Call it.
19312 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19315 2017-07-11 Tom de Vries <tom@codesourcery.com>
19317 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19318 (bb_first_real_insn): New function.
19319 (nvptx_single): Add extra initialization of broadcasted condition
19322 2017-07-11 Nathan Sidwell <nathan@acm.org>
19324 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19326 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19328 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19329 Binutils doc as TEXI will mess them up.
19330 * doc/invoke.texi (AVR Options): Same here.
19332 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19334 * config/sparc/sparc.opt (mfix-ut700): New option.
19335 (mfix-gr712rc): Likewise.
19336 (sparc_fix_b2bst): New variable.
19337 * doc/invoke.texi (SPARC options): Document them.
19338 (ARM options): Fix warnings.
19339 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19340 instructions to prevent sequences that can trigger the store-store
19341 errata for certain LEON3FT processors.
19342 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19343 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19344 * config/sparc/sparc.md (fix_b2bst): New attribute.
19345 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19347 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19350 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19352 (*rsqrtsf2_sse): Ditto.
19354 (div<mode>3): Macroize insn from divdf3 and divsf3
19355 using MODEF mode iterator.
19357 2017-07-10 Martin Sebor <msebor@redhat.com>
19359 PR tree-optimization/80397
19360 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19361 instead of testing for equality to INTEGER_TYPE.
19363 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19365 * config.gcc: Remove uclibc from arc target spec.
19367 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19369 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19371 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19374 * lto-wrapper.c (remove_option): New function.
19375 (merge_and_complain): Merge PIC/PIE options more realistically.
19377 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19379 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19383 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19384 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19385 * config.in: Regenerate.
19386 * configure: Regenerate.
19387 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19388 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19389 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19390 (TARGET_GASISR_PROLOGUES): ...target mask.
19391 * common/config/avr/avr-common.c
19392 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19393 Set -mgas-isr-prologues.
19394 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19395 INSERT_PASS_BEFORE for it.
19396 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19397 * config/avr/avr.c (avr_option_override)
19398 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19399 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19400 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19401 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19402 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19404 (make_avr_pass_pre_proep): New function.
19405 (emit_push_sfr) <treg>: Add argument to function and use it
19406 instead of TMP_REG.
19407 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19408 and set machine->gasisr.yes.
19409 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19410 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19411 __gcc_isr.n_pushed to .L__stack_usage.
19412 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19413 (avr_asm_final_postscan_insn): ...this new static function.
19414 * config/avr/avr.h (machine_function)
19415 <is_no_gccisr, use_L__stack_usage>: New fields.
19416 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19417 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19418 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19419 (gasisr, *gasisr): New expander and insn.
19420 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19421 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19422 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19424 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19426 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19429 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19431 Move jump-tables out of .text again.
19434 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19435 (ASM_OUTPUT_ADDR_VEC): New function.
19436 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19437 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19438 INSN_ADDRESSes as asm comment.
19439 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19440 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19441 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19442 * config/avr/avr.md (*tablejump): Adjust comment.
19443 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19444 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19446 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19447 (avr_output_addr_vec): New proto.
19448 (avr_log_t) <insn_addresses>: New field.
19450 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19453 * config/i386/i386.c (ix86_function_arg_advance): Set
19454 outgoing_args_on_stack to true if there are outgoing arguments
19456 (ix86_function_arg): Likewise.
19457 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19458 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19459 * config/i386/i386.h (machine_function): Add
19460 outgoing_args_on_stack.
19462 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19464 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19465 supporting pthreds.
19466 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19468 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19470 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19471 (REAL_H): Remove $(MACHMODE_H).
19472 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19474 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19475 $(MACHMODE_H) and double-int.h.
19476 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19478 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19481 2017-07-07 Andrew Pinski <apinski@cavium.com>
19483 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19484 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19486 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19488 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19489 Add warning if GCC was not configured to link against a GLIBC that
19490 exports the hardware capability bits.
19491 (make_resolver_func): Make resolver function private and not a
19492 COMDAT function. Create the name with clone_function_name instead
19493 of make_unique_name.
19496 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19497 correct operand in doing the split.
19499 2017-07-07 Carl Love <cel@us.ibm.com>
19501 * config/rs6000/rs6000-c: Add support for built-in function
19502 vector unsigned short vec_pack_to_short_fp32 (vector float,
19504 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19505 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19506 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19507 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19508 (convert_4f32_8i16): Add define_expand.
19509 * doc/extend.texi: Update the built-in documentation file for the
19510 new built-in function.
19512 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19514 * config/sparc/m8.md: New file.
19515 * config/sparc/sparc.md: Include m8.md.
19517 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19519 * config/sparc/sparc.opt: New option -mvis4b.
19520 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19521 (sparc_option_override): Handle VIS4B.
19522 (enum sparc_builtins): Define
19523 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19524 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19525 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19526 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19527 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19528 (check_constant_argument): New function.
19529 (sparc_vis_init_builtins): Define builtins
19530 __builtin_vis_dictunpack{8,16,32},
19531 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19532 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19533 __builtin_vis_fpcmpde{8,16,32}shl and
19534 __builtin_vis_fpcmpur{8,16,32}shl.
19535 (sparc_expand_builtin): Check that the constant operands to
19536 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19537 constant and in range.
19538 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19540 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19541 (SPARC_IMM5_P): Likewise.
19542 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19543 (enabled): Handle vis4b.
19544 (UNSPEC_DICTUNPACK): New unspec.
19545 (UNSPEC_FPCMPSHL): Likewise.
19546 (UNSPEC_FPUCMPSHL): Likewise.
19547 (UNSPEC_FPCMPDESHL): Likewise.
19548 (UNSPEC_FPCMPURSHL): Likewise.
19549 (cpu_feature): New CPU feature `vis4b'.
19550 (dictunpack{8,16,32}): New insns.
19551 (FPCSMODE): New mode iterator.
19552 (fpcscond): New code iterator.
19553 (fpcsucond): Likewise.
19554 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19555 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19556 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19557 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19558 * config/sparc/constraints.md: Define constraints `q' for unsigned
19559 2-bit integer constants and `t' for unsigned 5-bit integer
19561 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19563 (imm5_operand_dictunpack16): Likewise.
19564 (imm5_operand_dictunpack32): Likewise.
19565 (imm2_operand): Likewise.
19566 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19567 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19568 ditunpack* and fpcmp*shl builtins.
19570 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19572 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19573 * config.in: Add HAVE_AS_SPARC6 define.
19574 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19576 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19578 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19579 (CPP_CPU_SPEC): Handle m8.
19580 (ASM_CPU_SPEC): Likewise.
19581 * config/sparc/sparc-opts.h (enum processor_type): Add
19583 * config/sparc/sparc.c (m8_costs): New struct.
19584 (sparc_option_override): Handle TARGET_CPU_m8.
19585 (sparc32_initialize_trampoline): Likewise.
19586 (sparc64_initialize_trampoline): Likewise.
19587 (sparc_issue_rate): Likewise.
19588 (sparc_register_move_cost): Likewise.
19589 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19590 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19591 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19592 (CPP_CPU_SPEC): Handle M8.
19593 (ASM_CPU_SPEC): Likewise.
19594 (AS_M8_FLAG): Define.
19595 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19596 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19597 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19599 * configure: Regenerate.
19600 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19603 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19605 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19607 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19608 ("*movdi_insn_sp32"): Do not set v3pipe.
19609 ("*movsi_insn"): Likewise.
19610 ("*movdi_insn_sp64"): Likewise.
19611 ("*movsf_insn"): Likewise.
19612 ("*movdf_insn_sp32"): Likewise.
19613 ("*movdf_insn_sp64"): Likewise.
19614 ("*zero_extendsidi2_insn_sp64"): Likewise.
19615 ("*sign_extendsidi2_insn"): Likewise.
19616 ("*mov<VM32:mode>_insn"): Likewise.
19617 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19618 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19619 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19620 ("<vlop:code><VL:mode>3"): Likewise.
19621 ("*not_<vlop:code><VL:mode>3"): Likewise.
19622 ("*nand<VL:mode>_vis"): Likewise.
19623 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19624 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19625 ("one_cmpl<VL:mode>2"): Likewise.
19626 ("faligndata<VM64:mode>_vis"): Likewise.
19627 ("alignaddrsi_vis"): Likewise.
19628 ("alignaddrdi_vis"): Likweise.
19629 ("alignaddrlsi_vis"): Likewise.
19630 ("alignaddrldi_vis"): Likewise.
19631 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19632 ("bmaskdi_vis"): Likewise.
19633 ("bmasksi_vis"): Likewise.
19634 ("bshuffle<VM64:mode>_vis"): Likewise.
19635 ("cmask8<P:mode>_vis"): Likewise.
19636 ("cmask16<P:mode>_vis"): Likewise.
19637 ("cmask32<P:mode>_vis"): Likewise.
19638 ("pdistn<P:mode>_vis"): Likewise.
19639 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19641 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19643 * config/sparc/sparc.md ("subtype"): New insn attribute.
19644 ("*wrgsr_sp64"): Set insn subtype.
19645 ("*rdgsr_sp64"): Likewise.
19646 ("alignaddrsi_vis"): Likewise.
19647 ("alignaddrdi_vis"): Likewise.
19648 ("alignaddrlsi_vis"): Likewise.
19649 ("alignaddrldi_vis"): Likewise.
19650 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19651 ("fexpand_vis"): Likewise.
19652 ("fpmerge_vis"): Likewise.
19653 ("faligndata<VM64:mode>_vis"): Likewise.
19654 ("bshuffle<VM64:mode>_vis"): Likewise.
19655 ("cmask8<P:mode>_vis"): Likewise.
19656 ("cmask16<P:mode>_vis"): Likewise.
19657 ("cmask32<P:mode>_vis"): Likewise.
19658 ("fchksm16_vis"): Likewise.
19659 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19660 ("fmean16_vis"): Likewise.
19661 ("fp<plusminus_insn>64_vis"): Likewise.
19662 ("<plusminus_insn>v8qi3"): Likewise.
19663 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19664 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19665 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19666 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19667 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19668 ("*movqi_insn"): Likewise.
19669 ("*movhi_insn"): Likewise.
19670 ("*movsi_insn"): Likewise.
19671 ("movsi_pic_gotdata_op"): Likewise.
19672 ("*movdi_insn_sp32"): Likewise.
19673 ("*movdi_insn_sp64"): Likewise.
19674 ("movdi_pic_gotdata_op"): Likewise.
19675 ("*movsf_insn"): Likewise.
19676 ("*movdf_insn_sp32"): Likewise.
19677 ("*movdf_insn_sp64"): Likewise.
19678 ("*zero_extendhisi2_insn"): Likewise.
19679 ("*zero_extendqihi2_insn"): Likewise.
19680 ("*zero_extendqisi2_insn"): Likewise.
19681 ("*zero_extendqidi2_insn"): Likewise.
19682 ("*zero_extendhidi2_insn"): Likewise.
19683 ("*zero_extendsidi2_insn_sp64"): Likewise.
19684 ("ldfsr"): Likewise.
19685 ("prefetch_64"): Likewise.
19686 ("prefetch_32"): Likewise.
19687 ("tie_ld32"): Likewise.
19688 ("tie_ld64"): Likewise.
19689 ("*tldo_ldub_sp32"): Likewise.
19690 ("*tldo_ldub1_sp32"): Likewise.
19691 ("*tldo_ldub2_sp32"): Likewise.
19692 ("*tldo_ldub_sp64"): Likewise.
19693 ("*tldo_ldub1_sp64"): Likewise.
19694 ("*tldo_ldub2_sp64"): Likewise.
19695 ("*tldo_ldub3_sp64"): Likewise.
19696 ("*tldo_lduh_sp32"): Likewise.
19697 ("*tldo_lduh1_sp32"): Likewise.
19698 ("*tldo_lduh_sp64"): Likewise.
19699 ("*tldo_lduh1_sp64"): Likewise.
19700 ("*tldo_lduh2_sp64"): Likewise.
19701 ("*tldo_lduw_sp32"): Likewise.
19702 ("*tldo_lduw_sp64"): Likewise.
19703 ("*tldo_lduw1_sp64"): Likewise.
19704 ("*tldo_ldx_sp64"): Likewise.
19705 ("*mov<VM32:mode>_insn"): Likewise.
19706 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19707 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19709 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19711 * config/sparc/sparc.md ("type"): New insn type viscmp.
19712 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19714 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19715 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19716 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19717 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19719 ("n7_vis_logical_11cycle"): Likewise.
19720 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19721 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19722 * config/sparc/niagara.md ("niag_vis"): Likewise.
19723 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19724 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19726 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19728 * config/sparc/sparc.md: New instruction type `bmask'.
19729 (bmaskdi_vis): Use the `bmask' type.
19730 (bmasksi_vis): Likewise.
19731 * config/sparc/ultra3.md (us3_array): Likewise.
19732 * config/sparc/niagara7.md (n7_array): Likewise.
19733 * config/sparc/niagara4.md (n4_array): Likewise.
19734 * config/sparc/niagara2.md (niag2_vis): Likewise.
19735 (niag3_vis): Likewise.
19736 * config/sparc/niagara.md (niag_vis): Likewise.
19738 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19740 * ipa-comdats.c: Remove optimize check from gate.
19741 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19742 for functions not optimized.
19743 (ipa_fn_summary_read): Skip optimize check.
19744 (ipa_fn_summary_write): Likewise.
19745 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19747 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19749 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19750 (check_callers): Give up on caller which is not optimized.
19751 (inline_small_functions): Likewise.
19752 (ipa_inline): Do not give up when not optimizing.
19753 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19754 away unoptimizes cdtors.
19755 (whole_program_function_and_variable_visibility): Do
19756 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19757 * ipa.c (process_references): Do not check optimize.
19758 (symbol_table::remove_unreachable_nodes): Update optimize check.
19759 (set_writeonly_bit): Update optimize check.
19760 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19761 (pass_ipa_single_use::gate): Remove.
19763 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19765 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19766 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19767 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19768 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19769 permute_load, permute_store, adjust_extract, adjust_splat,
19770 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19771 replace_swap_with_copy, dump_swap_insn_table,
19772 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19773 recombine_lvx_pattern, recombine_stvx_pattern,
19774 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19775 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19776 to file rs6000-p8swap.c.
19777 * config/rs6000/rs6000-p8swap.c: New file.
19778 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19779 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19780 and rs6000*-*-* targets.
19782 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19784 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19786 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19788 * lto-wrapper.c (merge_and_complain): Do not merge
19789 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19790 fsigned_zeros, ftrapping_math, fwrapv.
19791 (append_compiler_options): Do not track these options.
19792 (append_linker_options): Likewie
19794 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19796 * cgraphunit.c (cgraph_node::finalize_function): When
19797 !flag_toplevel_reorde set no_reorder flag.
19798 (varpool_node::finalize_decl): Likewise.
19799 (symbol_table::compile): Drop no toplevel reorder path.
19801 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19803 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19804 edges; zero probability is not better than uninitialized.
19806 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19808 * asan.h (asan_sanitize_allocas_p): Declare.
19809 * asan.c (asan_sanitize_allocas_p): New function.
19810 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19811 (handle_builtin_alloca): Likewise.
19812 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19813 if !asan_sanitize_allocas_p.
19814 * params.def (asan-instrument-allocas): Add new option.
19815 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19816 * opts.c (common_handle_option): Disable allocas sanitization for
19819 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19821 * asan.c: Include gimple-fold.h.
19822 (get_last_alloca_addr): New function.
19823 (handle_builtin_stackrestore): Likewise.
19824 (handle_builtin_alloca): Likewise.
19825 (asan_emit_allocas_unpoison): Likewise.
19826 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19827 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19828 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19829 (instrument_builtin_call): Pass gimple iterator to
19830 get_mem_refs_of_builtin_call.
19831 (last_alloca_addr): New global.
19832 * asan.h (asan_emit_allocas_unpoison): Declare.
19833 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19834 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19835 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19836 if function calls alloca.
19837 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19838 * gimple-fold.h (replace_call_with_value): Declare.
19839 * internal-fn.c: Include asan.h.
19840 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19841 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19843 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19845 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19846 (C_SELFTEST_FLAGS): New.
19847 (CPP_SELFTEST_FLAGS): New.
19848 (SELFTEST_DEPS): New, from deps of s-selftest.
19849 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19850 (CPP_SELFTEST_DEPS): New.
19851 (selftest): Add dependency on s-selftest-c++.
19852 (s-selftest): Rename to...
19853 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19854 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19855 than SELFTEST_FLAGS.
19856 (selftest-gdb): Rename to...
19857 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19859 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19860 (selftest-valgrind): Rename to...
19861 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19863 (selftest-valgrind): Reintroduce as an alias for
19864 selftest-c-valgrind.
19865 (s-selftest-c++): New.
19866 (selftest-c++-gdb): New.
19867 (selftest-c++-valgrind): New.
19869 2017-07-06 Olivier Hainque <hainque@adacore.com>
19871 * gcc.c (process_command): When deciding if undefined variables
19872 should be ignored when processing specs, accept "gcc -v" as well.
19874 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19876 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19877 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19879 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19881 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19882 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19883 * config/arm/arm-tables.opt: Regenerate.
19884 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19886 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19888 2017-07-06 Carl Love <cel@us.ibm.com>
19890 * ChangeLog: Clean up from mid air collision
19892 2017-07-06 Carl Love <cel@us.ibm.com>
19894 * config/rs6000/rs6000-c.c: Add support for built-in functions
19895 vector signed int vec_subc (vector signed int, vector signed int);
19896 vector signed __int128 vec_subc (vector signed __int128,
19897 vector signed __int128);
19898 vector unsigned __int128 vec_subc (vector unsigned __int128,
19899 vector unsigned __int128);
19900 vector signed int vec_sube (vector signed int, vector signed int,
19901 vector signed int);
19902 vector unsigned int vec_sube (vector unsigned int,
19903 vector unsigned int,
19904 vector unsigned int);
19905 vector signed __int128 vec_sube (vector signed __int128,
19906 vector signed __int128,
19907 vector signed__int128);
19908 vector unsigned __int128 vec_sube (vector unsigned __int128,
19909 vector unsigned __int128,
19910 vector unsigned __int128);
19911 vector signed int vec_subec (vector signed int, vector signed int,
19912 vector signed int);
19913 vector unsigned int vec_subec (vector unsigned int,
19914 vector unsigned int,
19915 vector unsigned int);
19916 vector signed __int128 vec_subec (vector signed __int128,
19917 vector signed __int128,
19918 vector signed__int128);
19919 vector unsigned __int128 vec_subec (vector unsigned __int128,
19920 vector unsigned __int128,
19921 vector unsigned __int128);
19922 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19923 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19924 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19925 BU_ALTIVEC_OVERLOAD_X definitions.
19926 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19927 * doc/extend.texi: Update the built-in documentation file for the new
19928 built-in functions.
19930 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19933 * diagnostic-show-locus.c (layout::layout): Use start and finish
19934 spelling location for the start and finish of each range.
19935 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19936 Add unused aspect param.
19937 * input.c (expand_location_1): Add "aspect" param, and use it
19938 to access the correct part of the location.
19939 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19941 (expand_location_to_spelling_point): Likewise.
19942 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19943 param, and pass it to expand_location_1.
19945 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19947 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19948 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19949 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19950 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19951 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19952 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19953 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19954 _mm_maskz_getmant_ss): New intrinsics.
19955 (__builtin_ia32_getexpss128_mask): Changed to ...
19956 __builtin_ia32_getexpss128_round ... this.
19957 (__builtin_ia32_getexpsd128_mask): Changed to ...
19958 __builtin_ia32_getexpsd128_round ... this.
19959 * config/i386/i386-builtin-types.def
19960 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19961 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19962 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19963 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19964 __builtin_ia32_getmantss_mask_round): New builtins.
19965 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19966 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19967 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19968 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19969 * config/i386/sse.md
19970 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19971 avx512f_sgetexp<mode><mask_scalar_name>
19972 <round_saeonly_scalar_name> ... this.
19973 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19974 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19975 vgetexp<ssescalarmodesuffix>
19976 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19977 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19978 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19979 avx512f_vgetmant<mode><mask_scalar_name>
19980 <round_saeonly_scalar_name> ... this.
19981 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19982 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19983 vgetmant<ssescalarmodesuffix>
19984 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19985 %0<mask_scalar_operand4>, %1, %2
19986 <round_saeonly_scalar_mask_op4>, %3} ... this.
19987 * config/i386/subst.md (mask_scalar_operand4,
19988 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19989 round_saeonly_scalar_nimm_predicate): New subst attributes.
19991 2017-07-06 Julia Koval <julia.koval@intel.com>
19993 * config/i386/i386.c (ix86_erase_embedded_rounding):
19994 Remove code for old rounding pattern.
19996 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19998 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20000 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20002 * doc/sourcebuild.texi (Test Directives, Variants of
20003 dg-require-support): Add documentation for dg-require-stack-check.
20005 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20007 * config/i386/subst.md (mask_scalar, round_scalar,
20008 round_saeonly_scalar): New meta-templates.
20009 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20010 round_scalar_mask_operand3, round_scalar_mask_op3,
20011 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20012 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20013 round_saeonly_scalar_constraint,
20014 round_saeonly_scalar_prefix): New subst attribute.
20015 * config/i386/sse.md
20016 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20017 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20018 <round_scalar_name> ... this.
20019 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20020 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20021 <round_scalar_name> ... this.
20022 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20023 <sse>_vm<code><mode>3<mask_scalar_name>
20024 <round_saeonly_scalar_name> ... this.
20025 (v<plusminus_mnemonic><ssescalarmodesuffix>
20026 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20027 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20028 v<plusminus_mnemonic><ssescalarmodesuffix>
20029 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20030 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20031 (v<multdiv_mnemonic><ssescalarmodesuffix>
20032 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20033 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20034 v<multdiv_mnemonic><ssescalarmodesuffix>
20035 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20036 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20037 (v<maxmin_float><ssescalarmodesuffix>
20038 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20039 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20040 v<maxmin_float><ssescalarmodesuffix>
20041 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20042 %0<mask_scalar_operand3>, %1, %<iptr>2
20043 <round_saeonly_scalar_mask_op3>} ... this.
20045 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20047 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20048 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20050 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20051 Alan Hayward <alan.hayward@arm.com>
20052 David Sherwood <david.sherwood@arm.com>
20054 * combine.c (simplify_if_then_else): Remove "enum" before
20056 * compare-elim.c (can_eliminate_compare): Likewise.
20057 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20059 (aarch64_lookup_simd_builtin_type): Likewise.
20060 (aarch64_simd_builtin_type): Likewise.
20061 (aarch64_init_simd_builtin_types): Likewise.
20062 (aarch64_simd_expand_args): Likewise.
20063 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20065 (aarch64_reverse_mask): Likewise.
20066 (aarch64_simd_emit_reg_reg_move): Likewise.
20067 (aarch64_gen_adjusted_ldpstp): Likewise.
20068 (aarch64_ccmp_mode_to_code): Likewise.
20069 (aarch64_operands_ok_for_ldpstp): Likewise.
20070 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20071 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20073 (aarch64_min_divisions_for_recip_mul): Likewise.
20074 (aarch64_reassociation_width): Likewise.
20075 (aarch64_get_condition_code_1): Likewise.
20076 (aarch64_simd_emit_reg_reg_move): Likewise.
20077 (aarch64_simd_attr_length_rglist): Likewise.
20078 (aarch64_reverse_mask): Likewise.
20079 (aarch64_operands_ok_for_ldpstp): Likewise.
20080 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20081 (aarch64_gen_adjusted_ldpstp): Likewise.
20082 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20084 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20085 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20086 (arm_lookup_simd_builtin_type): Likewise.
20087 (arm_simd_builtin_type): Likewise.
20088 (arm_init_simd_builtin_types): Likewise.
20089 (arm_expand_builtin_args): Likewise.
20090 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20091 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20092 (ft32_setup_incoming_varargs): Likewise.
20093 (ft32_function_arg): Likewise.
20094 (ft32_function_arg_advance): Likewise.
20095 (ft32_pass_by_reference): Likewise.
20096 (ft32_arg_partial_bytes): Likewise.
20097 (ft32_valid_pointer_mode): Likewise.
20098 (ft32_addr_space_pointer_mode): Likewise.
20099 (ft32_addr_space_legitimate_address_p): Likewise.
20100 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20102 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20103 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20104 (iamcu_alignment): Likewise.
20105 (canonicalize_vector_int_perm): Likewise.
20106 (ix86_noce_conversion_profitable_p): Likewise.
20107 (ix86_mpx_bound_mode): Likewise.
20108 (ix86_operands_ok_for_move_multiple): Likewise.
20109 * config/microblaze/microblaze-protos.h
20110 (microblaze_expand_conditional_branch_reg): Likewise.
20111 * config/microblaze/microblaze.c
20112 (microblaze_expand_conditional_branch_reg): Likewise.
20113 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20115 (rs6000_reassociation_width): Likewise.
20116 (rs6000_invalid_binary_op): Likewise.
20117 (fusion_p9_p): Likewise.
20118 (emit_fusion_p9_load): Likewise.
20119 (emit_fusion_p9_store): Likewise.
20120 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20122 (riscv_hard_regno_mode_ok_p): Likewise.
20123 (riscv_address_insns): Likewise.
20124 (riscv_split_symbol): Likewise.
20125 (riscv_legitimize_move): Likewise.
20126 (riscv_function_value): Likewise.
20127 (riscv_hard_regno_nregs): Likewise.
20128 (riscv_expand_builtin): Likewise.
20129 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20130 (riscv_build_integer): Likewise.
20131 (riscv_split_integer): Likewise.
20132 (riscv_legitimate_constant_p): Likewise.
20133 (riscv_cannot_force_const_mem): Likewise.
20134 (riscv_regno_mode_ok_for_base_p): Likewise.
20135 (riscv_valid_base_register_p): Likewise.
20136 (riscv_valid_offset_p): Likewise.
20137 (riscv_valid_lo_sum_p): Likewise.
20138 (riscv_classify_address): Likewise.
20139 (riscv_legitimate_address_p): Likewise.
20140 (riscv_address_insns): Likewise.
20141 (riscv_load_store_insns): Likewise.
20142 (riscv_force_binary): Likewise.
20143 (riscv_split_symbol): Likewise.
20144 (riscv_force_address): Likewise.
20145 (riscv_legitimize_address): Likewise.
20146 (riscv_move_integer): Likewise.
20147 (riscv_legitimize_const_move): Likewise.
20148 (riscv_legitimize_move): Likewise.
20149 (riscv_address_cost): Likewise.
20150 (riscv_subword): Likewise.
20151 (riscv_output_move): Likewise.
20152 (riscv_canonicalize_int_order_test): Likewise.
20153 (riscv_emit_int_order_test): Likewise.
20154 (riscv_function_arg_boundary): Likewise.
20155 (riscv_pass_mode_in_fpr_p): Likewise.
20156 (riscv_pass_fpr_single): Likewise.
20157 (riscv_pass_fpr_pair): Likewise.
20158 (riscv_get_arg_info): Likewise.
20159 (riscv_function_arg): Likewise.
20160 (riscv_function_arg_advance): Likewise.
20161 (riscv_arg_partial_bytes): Likewise.
20162 (riscv_function_value): Likewise.
20163 (riscv_pass_by_reference): Likewise.
20164 (riscv_setup_incoming_varargs): Likewise.
20165 (riscv_print_operand): Likewise.
20166 (riscv_elf_select_rtx_section): Likewise.
20167 (riscv_save_restore_reg): Likewise.
20168 (riscv_for_each_saved_reg): Likewise.
20169 (riscv_register_move_cost): Likewise.
20170 (riscv_hard_regno_mode_ok_p): Likewise.
20171 (riscv_hard_regno_nregs): Likewise.
20172 (riscv_class_max_nregs): Likewise.
20173 (riscv_memory_move_cost): Likewise.
20174 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20175 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20176 (rl78_addr_space_address_mode): Likewise.
20177 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20179 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20180 (rs6000_reassociation_width): Likewise.
20181 (rs6000_invalid_binary_op): Likewise.
20182 (fusion_p9_p): Likewise.
20183 (emit_fusion_p9_load): Likewise.
20184 (emit_fusion_p9_store): Likewise.
20185 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20186 (ok_for_simple_move_operands): Likewise.
20187 (ok_for_simple_move_strict_operands): Likewise.
20188 (ok_for_simple_arith_logic_operands): Likewise.
20189 (visium_legitimize_reload_address): Likewise.
20190 (visium_select_cc_mode): Likewise.
20191 (output_cbranch): Likewise.
20192 (visium_split_double_move): Likewise.
20193 (visium_expand_copysign): Likewise.
20194 (visium_expand_int_cstore): Likewise.
20195 (visium_expand_fp_cstore): Likewise.
20196 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20197 (visium_function_arg): Likewise.
20198 (visium_function_arg_advance): Likewise.
20199 (visium_libcall_value): Likewise.
20200 (visium_setup_incoming_varargs): Likewise.
20201 (visium_legitimate_constant_p): Likewise.
20202 (visium_legitimate_address_p): Likewise.
20203 (visium_legitimize_address): Likewise.
20204 (visium_secondary_reload): Likewise.
20205 (visium_register_move_cost): Likewise.
20206 (visium_memory_move_cost): Likewise.
20207 (prepare_move_operands): Likewise.
20208 (ok_for_simple_move_operands): Likewise.
20209 (ok_for_simple_move_strict_operands): Likewise.
20210 (ok_for_simple_arith_logic_operands): Likewise.
20211 (visium_function_value_1): Likewise.
20212 (rtx_ok_for_offset_p): Likewise.
20213 (visium_legitimize_reload_address): Likewise.
20214 (visium_split_double_move): Likewise.
20215 (visium_expand_copysign): Likewise.
20216 (visium_expand_int_cstore): Likewise.
20217 (visium_expand_fp_cstore): Likewise.
20218 (visium_split_cstore): Likewise.
20219 (visium_select_cc_mode): Likewise.
20220 (visium_split_cbranch): Likewise.
20221 (output_cbranch): Likewise.
20222 (visium_print_operand_address): Likewise.
20223 * expmed.c (flip_storage_order): Likewise.
20224 * expmed.h (emit_cstore): Likewise.
20225 (flip_storage_order): Likewise.
20226 * genrecog.c (validate_pattern): Likewise.
20227 * hsa-gen.c (gen_hsa_addr): Likewise.
20228 * internal-fn.c (expand_arith_overflow): Likewise.
20229 * ira-color.c (allocno_copy_cost_saving): Likewise.
20230 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20231 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20232 (process_invariant_for_inheritance): Likewise.
20233 * lra-eliminations.c (move_plus_up): Likewise.
20234 * omp-low.c (lower_oacc_reductions): Likewise.
20235 * simplify-rtx.c (simplify_subreg): Likewise.
20236 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20237 (TARGET_CHKP_BOUND_MODE): Likewise..
20238 * targhooks.c (default_chkp_bound_mode): Likewise.
20239 (default_setup_incoming_vararg_bounds): Likewise.
20240 * targhooks.h (default_chkp_bound_mode): Likewise.
20241 (default_setup_incoming_vararg_bounds): Likewise.
20242 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20243 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20244 (have_whole_vector_shift): Likewise.
20245 * tree-vect-stmts.c (vectorizable_load): Likewise.
20246 * doc/tm.texi: Regenerate.
20248 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20250 Graceful degrade if Binutils PR21472 is not available.
20253 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20254 .rodata in flash test fails.
20255 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20256 * confgure: Regenerate.
20257 * config.in: Regenerate.
20258 * config/avr/avr.c (avr_asm_named_section)
20259 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20260 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20261 (avr_asm_init_sections): Same.
20263 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20265 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20266 (fma<VH:mode>4_intrinsic): Likewise.
20267 (*fmsub<VCVTF:mode>4): Likewise.
20268 (*fmsub<VH:mode>4_intrinsic): Likewise.
20270 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20273 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20274 Don't depend on "optimize > 0".
20275 (out_movhi_r_mr, out_movqi_mr_r): Same.
20276 (out_movhi_mr_r, out_movqi_r_mr): Same.
20277 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20278 io_address_operand on "optimize > 0".
20280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20282 * tree-loop-distribution.c: Add general explanantion on the pass.
20283 (generate_loops_for_partition): Mark distributed loop.
20284 (pg_add_dependence_edges): New parameter. Handle alias data
20285 dependence specially and record it in the parameter if asked.
20286 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20287 (init_partition_graph_vertices, add_partition_graph_edge): New.
20288 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20289 (free_partition_graph_vdata, build_partition_graph): New.
20290 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20291 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20292 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20293 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20294 (version_for_distribution_p, finalize_partitions): New.
20295 (distribute_loop): Handle alias data dependence specially. Factor
20296 out loop fusion code as functions and call these functions.
20298 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20300 * tree-loop-distribution.c (classify_partition): New parameter and
20301 better handle reduction statement.
20302 (rdg_build_partitions): Revise comment.
20303 (distribute_loop): Compute statements in all partitions and pass it
20304 to classify_partition.
20306 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20308 * tree-loop-distribution.c (enum partition_type): New.
20309 (struct partition): New field type.
20310 (partition_merge_into): Add parameter. Update partition type.
20311 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20312 (build_rdg_partition_for_vertex): Compute partition type.
20313 (rdg_build_partitions): Dump partition type.
20314 (distribute_loop): Update calls to partition_merge_into.
20316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20318 * tree-loop-distribution.c (struct ddr_hasher): New.
20319 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20321 (classify_partition): Call get_data_dependence.
20322 (pg_add_dependence_edges): Ditto.
20323 (distribute_loop): Release data dependence hash table.
20325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20327 * tree-loop-distribution.c (ref_base_address): Delete.
20328 (similar_memory_accesses): Rename ...
20329 (share_memory_accesses): ... to this. Check if partitions access
20330 the same memory reference.
20331 (distribute_loop): Call share_memory_accesses.
20333 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20335 * tree-loop-distribution.c (struct partition): New field recording
20336 its data reference.
20337 (partition_alloc, partition_free): Init and release data refs.
20338 (partition_merge_into): Merge data refs.
20339 (build_rdg_partition_for_vertex): Collect data refs for partition.
20340 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20342 (distribute_loop): Remve data refs from vertice data of partition
20345 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20347 * tree-loop-distribution.c (params.h): Include header file.
20348 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20349 (datarefs_vec): New global var.
20350 (create_rdg_vertices): Use datarefs_vec directly.
20351 (free_rdg): Don't free data references.
20352 (build_rdg): Update use. Don't free data references.
20353 (distribute_loop): Compute global variable for data references.
20354 Bail out if there are too many data references.
20356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20358 * tree-loop-distribution.c (loop_nest): New global var.
20359 (build_rdg): Use loop directly, rather than loop nest.
20360 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20362 (distribute_loop): Compute global variable loop nest. Update use.
20364 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20366 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20367 (partition_merge_into): New parameter. Dump reason for fusion.
20368 (distribute_loop): Update use of partition_merge_into.
20370 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20372 * tree-loop-distribution.c (bb_top_order_index): New.
20373 (bb_top_order_index_size, bb_top_order_cmp): New.
20374 (stmts_from_loop): Use topological order.
20375 (pass_loop_distribution::execute): Compute and release topological
20376 order for basic blocks.
20378 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20380 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20383 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20385 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20386 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20387 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20388 * internal-fn.def (LOOP_DIST_ALIAS): New.
20389 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20390 (fold_loop_internal_call): ... this.
20391 (vect_loop_dist_alias_call): New function.
20392 (set_uid_loop_bbs): Call fold_loop_internal_call.
20393 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20396 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20399 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20400 Require dead FLAGS_REG at the beginning of a peephole.
20402 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20405 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20406 arguments in the call to __builtin_ia32_sbb_u32.
20407 (_subborrow_u64): Swap _X and _Y arguments in the call to
20408 __builtin_ia32_sbb_u64.
20410 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20413 * tree-vrp.c (compare_assert_loc): Turn into a function template
20414 with stable template parameter. Only test if a->e is NULL,
20415 !a->e == !b->e has been verified already. Use e == NULL or
20416 e != NULL instead of e or ! e tests. If stable is true, don't use
20417 iterative_hash_expr, on the other side allow a or b or both NULL
20418 and sort the NULLs last.
20419 (process_assert_insertions): Sort using compare_assert_loc<false>
20420 instead of compare_assert_loc, later sort using
20421 compare_assert_loc<true> before calling process_assert_insertions_for
20422 in a loop. Use break instead of continue once seen NULL pointer.
20424 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20426 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20427 Cortex-R7 and Cortex-R8 processors.
20429 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20431 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20432 uninitialized while src is not.
20434 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20436 * common/config/arm/arm-common.c: Adjust include path for
20438 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20439 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20440 (arm-cpu-data.h): Likewise.
20441 (arm-cpu-cdata.h): Likewise.
20442 * config/arm/arm-cpu.h: Delete.
20443 * config/arm/arm-cpu-cdata.h: Delete.
20444 * config/arm/arm-cpu-data.h: Delete.
20446 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20448 * config/arm/arm-cpus.in (cortex-a55): New.
20449 (cortex-a75): Likewise.
20450 (cortex-a75.cortex-a55): Likewise.
20451 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20453 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20454 * config/arm/arm-cpu-cdata.h: Regenerate.
20455 * config/arm/arm-cpu-data.h: Regenerate.
20456 * config/arm/arm-cpu.h: Regenerate.
20457 * config/arm/arm-tables.opt: Regenerate.
20458 * config/arm/arm-tune.md: Regenerate.
20460 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20462 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20464 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20466 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20469 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20471 PR tree-optimization/81292
20472 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20473 full_string_p, also call adjust_related_strinfos if the adjustment
20474 is simple, otherwise invalidate related strinfos.
20476 2017-07-04 Martin Liska <mliska@suse.cz>
20479 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20480 newly created variable as DECL_IGNORED_P.
20482 2017-07-04 Martin Liska <mliska@suse.cz>
20485 * ipa-inline.c (inline_small_functions):
20486 Use xstrdup_for_dump.
20488 2017-07-04 Tom de Vries <tom@codesourcery.com>
20490 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20492 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20495 * config/darwin.c (darwin_function_switched_text_sections):
20498 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20500 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20502 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20504 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20506 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20508 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20509 min_profitable_iters, and th as inclusive lower bounds.
20510 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20511 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20512 for min_profitable_iters and min_profitable_estimate.
20513 (vect_transform_loop): Treat th as an inclusive lower bound.
20514 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20516 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20519 * config/darwin.c (darwin_function_switched_text_sections):
20520 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20521 in two pieces, and suppress the use of buf.
20523 2017-07-03 Nathan Sidwell <nathan@acm.org>
20525 * hash-table.h (hash_table_mod1): Fix indentation.
20527 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20529 PR middle-end/81290
20530 * predict.c (force_edge_cold): Be more careful about propagation
20532 * profile-count.h (profile_probability::guessed,
20533 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20535 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20537 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20539 * doc/invoke.texi (rcpc architecture extension): Document it.
20541 2017-07-03 Richard Biener <rguenther@suse.de>
20543 PR tree-optimization/60510
20544 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20545 the scalar reduction PHI and use it.
20546 (vectorizable_reduction): Properly guard the single_defuse_cycle
20547 path for non-SLP reduction chains where we cannot use it.
20548 Rework reduc_def/index and vector type deduction. Rework
20549 vector operand gathering during reduction op code-gen.
20550 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20551 chains dissolve the chain and leave it to non-SLP reduction
20554 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20556 * tree-data-ref.h (dr_alignment): Declare.
20557 * tree-data-ref.c (dr_alignment): New function.
20558 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20559 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20561 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20563 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20565 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20566 and base_misalignment fields.
20567 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20568 * tree-data-ref.c: Include builtins.h.
20569 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20570 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20571 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20572 * tree-vect-data-refs.c: Include tree-cfg.h.
20573 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20574 fields instead of calculating an alignment here.
20575 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20576 innermost_loop_behavior fields.
20578 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20580 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20582 (DR_STEP_ALIGNMENT): New macro.
20583 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20584 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20585 (create_data_ref): Print it.
20586 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20587 to tell whether the step preserves vector (mis)alignment.
20588 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20589 Move the check for an integer step and generalise to all INTEGER_CST.
20590 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20591 Print the outer step alignment.
20593 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20595 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20596 with offset_alignment.
20597 (DR_ALIGNED_TO): Delete.
20598 (DR_OFFSET_ALIGNMENT): New macro.
20599 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20600 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20601 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20602 (create_data_ref): Likewise.
20603 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20604 (vect_analyze_data_refs): Likewise.
20605 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20606 creating dummy innermost behavior.
20608 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20610 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20611 with a "innermost_loop_behavior *" and refeence tree.
20612 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20613 (create_data_ref): Update call accordingly.
20614 * tree-predcom.c (find_looparound_phi): Likewise.
20616 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20618 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20619 fields with dr_wrt_vec_loop.
20620 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20621 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20622 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20623 (vect_dr_behavior): New function.
20624 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20625 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20626 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20627 track whether the step preserves the misalignment.
20628 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20629 Use vect_dr_behavior.
20630 (vect_setup_realignment): Update call accordingly.
20631 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20632 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20633 call to vect_create_addr_base_for_vector_ref.
20634 (vect_create_cond_for_align_checks): Likewise.
20635 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20636 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20637 (vect_recog_mask_conversion_pattern): Likewise.
20638 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20639 (new_stmt_vec_info): Remove redundant zeroing.
20641 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20643 * common/config/arm/arm-common.c (arm_be8_option): New function.
20644 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20645 (ISA_ARMv6): Add isa_bit_be8.
20646 * config/arm/arm.h (arm_be8_option): Add prototype.
20647 (BE8_SPEC_FUNCTION): New define.
20648 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20649 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20650 (mlittle-endian): Similarly.
20651 (mbe8, mbe32): New options.
20652 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20653 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20655 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20657 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20659 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20661 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20662 (cleanup_tree_cfg_bb): Use it.
20663 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20665 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20667 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20670 * loop-doloop.c (add_test): Update profile.
20672 2017-07-03 Martin Liska <mliska@suse.cz>
20675 * sanopt.c (rewrite_usage_of_param): New function.
20676 (sanitize_rewrite_addressable_params): Likewise.
20677 (pass_sanopt::execute): Call rewrite_usage_of_param.
20679 2017-07-03 Richard Biener <rguenther@suse.de>
20681 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20682 back to using VIEW_CONVERT_EXPR.
20684 2017-07-03 Martin Liska <mliska@suse.cz>
20687 * doc/extend.texi: Document when a resolver function is
20688 generated for target_clones.
20690 2017-07-03 Martin Liska <mliska@suse.cz>
20692 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20693 and shadow memory of auto variables which are subject of
20694 use-after-scope sanitization.
20695 (asan_expand_mark_ifn): Add do set only when is_poison.
20697 2016-07-03 Richard Biener <rguenther@suse.de>
20699 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20701 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20702 (vectorizable_reduction): Perform reduction PHI creation when
20703 visiting a reduction PHI and adjust and simplify code generation
20704 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20705 (vect_transform_loop): Visit reduction PHIs.
20706 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20707 defs into the SLP tree.
20708 (vect_build_slp_tree): Reduction defs terminate the recursion.
20709 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20711 (vect_get_vec_defs_for_stmt_copy): Export.
20712 (vect_get_vec_defs): Likewise.
20713 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20715 (vect_get_vec_defs_for_stmt_copy): Declare.
20716 (vect_get_vec_defs): Likewise.
20718 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20720 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20721 parameter with a "loop" parameter and use it instead of the
20722 loop containing DR_STMT. Don't check simple_iv when doing
20723 BB analysis. Describe the two analysis modes in the comment.
20725 2017-07-03 Tom de Vries <tom@codesourcery.com>
20727 PR tree-optimization/69468
20728 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20729 (find_same_succ_bb): Handle ignore_edge_flags.
20731 2017-07-03 Tom de Vries <tom@codesourcery.com>
20733 PR tree-optimization/81192
20734 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20736 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20738 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20740 2017-07-03 Tom de Vries <tom@codesourcery.com>
20742 PR tree-optimization/81192
20743 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20744 BB_SAME_SUCC (bb) == NULL.
20746 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20748 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20751 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20753 * dumpfile.c: Include profile-count.h
20754 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20756 (insert_cond_bb): Update profile.
20757 * tree-cfg.h (insert_cond_bb): Update prototype.
20758 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20759 * tree-dump.c: Do not include tree-cfg.
20761 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20763 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20765 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20767 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20770 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20772 * tree-complex.c (expand_complex_div_wide): update profile.
20774 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20775 Alan Hayward <alan.hayward@arm.com>
20776 David Sherwood <david.sherwood@arm.com>
20778 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20779 (CORETYPES_H): New define.
20780 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20781 (insn-modes-inline.h, s-modes-inline-h): New rules.
20782 (generated_files): Add insn-modes-inline.h.
20783 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20784 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20785 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20786 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20787 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20788 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20789 (build/gencodes.o, build/genconditions.o): Likewise.
20790 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20791 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20792 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20793 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20794 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20795 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20796 * coretypes.h: Include everything up to real.h for generators.
20797 Include insn-modes.h first. Include wide-int-print.h after
20798 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20799 * machmode.h: Don't include insn-modes.h here.
20800 * function-tests.c: Remove includes of signop.h, machmode.h,
20801 double-int.h and wide-int.h.
20803 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20805 * optc-save-gen.awk: Likewise.
20806 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20807 * godump.c: Remove include of wide-int-print.h.
20808 * pretty-print.h: Likewise.
20809 * wide-int-print.cc: Likewise.
20810 * wide-int.cc: Likewise.
20811 * hash-map-tests.c: Remove include of signop.h.
20812 * hash-set-tests.c: Likewise.
20813 * rtl-tests.c: Likewise.
20814 * mkconfig.sh: Remove include of machmode.h.
20815 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20817 (emit_insn_modes_inline_h): ...this new function. Emit the code
20818 into an insn-modes-inline.h header file, adding appropriate
20819 include guards and end comments.
20820 (emit_insn_modes_c_header): Remove include of machmode.h.
20821 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20823 (main): Handle -i flag and call emit_insn_modes_inline_h when
20826 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20828 * tree-ssa-strlen.c (strinfo): Rename the length field to
20829 nonzero_chars. Add a full_string_p field.
20830 (compare_nonzero_chars, zero_length_string_p): New functions.
20831 (get_addr_stridx): Add an offset_out parameter.
20832 Use compare_nonzero_chars.
20833 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20834 (new_strinfo): Update after above changes to strinfo.
20835 (set_endptr_and_length): Set full_string_p.
20836 (get_string_length): Update after above changes to strinfo.
20837 (unshare_strinfo): Update call to new_strinfo.
20838 (maybe_invalidate): Likewise.
20839 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20840 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20841 as a uhwi instead of an shwi. Update after above changes to
20842 strinfo and new_strinfo.
20843 (zero_length_string): Assert that chainsi contains full strings.
20844 Use zero_length_string_p. Update call to new_strinfo.
20845 (adjust_related_strinfos): Update after above changes to strinfo.
20846 Copy full_string_p from origsi.
20847 (adjust_last_stmt): Use zero_length_string_p.
20848 (handle_builtin_strlen): Update after above changes to strinfo and
20849 new_strinfo. Install the lhs as the string length if the previous
20850 entry didn't describe a full string.
20851 (handle_builtin_strchr): Update after above changes to strinfo
20853 (handle_builtin_strcpy): Likewise.
20854 (handle_builtin_strcat): Likewise.
20855 (handle_builtin_malloc): Likewise.
20856 (handle_pointer_plus): Likewise.
20857 (handle_builtin_memcpy): Likewise. Track nonzero characters
20858 that aren't necessarily followed by a nul terminator.
20859 (handle_char_store): Likewise.
20861 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20863 PR tree-optimization/80769
20864 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20865 for malloc and calloc. Document the new invariant that all related
20866 strinfos have delayed lengths or none do.
20867 (verify_related_strinfos): Move earlier in file.
20868 (set_endptr_and_length): New function, split out from...
20869 (get_string_length): ...here. Also set the lengths of related
20871 (zero_length_string): Assert that chainsi has known (rather than
20873 (adjust_related_strinfos): Likewise.
20875 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20877 PR tree-optimization/81136
20878 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20879 assert that two references with the same misalignment have the same
20880 compile-time misalignment if those compile-time misalignments
20883 2017-07-01 Andi Kleen <ak@linux.intel.com>
20885 * print-tree.c (print_node): Print all attributes.
20887 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20889 * cfg.c (scale_bbs_frequencies): New function.
20890 * cfg.h (scale_bbs_frequencies): Declare it.
20891 * cfgloopanal.c (single_likely_exit): Cleanup.
20892 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20894 (scale_loop_profile): Likewise.
20895 (loop_version): Likewise.
20896 (create_empty_loop_on_edge): Update.
20897 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20898 scale_loop_frequencies, scale_loop_profile, loopify,
20899 loop_version): Update prototypes.
20900 * modulo-sched.c (sms_schedule): Update.
20901 * predict.c (unlikely_executed_edge_p): Also check probability.
20902 (probably_never_executed_edge_p): Fix typo.
20903 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20904 * tree-parloops.c (gen_parallel_loop): Update.
20905 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20906 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20907 * tree-ssa-loop-split.c (split_loop): Update.
20908 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20909 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20910 (vect_loop_versioning): Update.
20911 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20913 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20915 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20917 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20919 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20920 to keep profile consistent.
20922 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20924 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20925 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20926 * profile-count.h (max_safe_multiplier): Make unsigned.
20927 (profile_count::guessed_zero): New.
20929 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20931 * bb-reorder.c (fix_up_crossing_landing_pad,
20932 fix_crossing_conditional_branches): Use make_single_succ_edge
20933 to keep profile consistent.
20935 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20937 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20940 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20943 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20944 the right scopes, make sure cond_jump isn't preserved between multiple
20945 iterations. Search for fallthru edge whenever there are 3+ edges and
20946 use find_fallthru_edge for it.
20948 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20950 Patch by Alexander Monakov <amonakov@ispras.ru>
20951 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20952 probabilities consistently.
20954 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20956 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20957 emit_cmp_and_jump_insns.
20959 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20962 * tree-inline.c (expand_call_inline): Combine profile statuses.
20964 2017-06-30 Andrew Pinski <apinski@cavium.com>
20966 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20967 fold_stmt returned true.
20969 2017-06-30 Nathan Sidwell <nathan@acm.org>
20971 * ggc.h (empty_string): Delete.
20972 * cfgexpand.c (expand_asm_stmt): Use plain "".
20973 * optabs.c (expand_asm_memory_barrier): Likewise.
20974 * stringpool.c (empty_string): Delete.
20975 (digit_vector, digit_string): Delete.
20976 (ggc_alloc_string): Use plain "", don't optimize single digit
20977 strings. Use ggc_alloc_atomic.
20979 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20981 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20982 comparison set and one other set, use the cost of the non-comparison
20985 2017-06-30 Nathan Sidwell <nathan@acm.org>
20987 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20990 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20992 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20993 loops. Remove now unneeded calls to gimple_switch_set_label() that
20994 just set removed labels to NULL_TREE.
20996 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20998 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20999 (set_range_info): ...here. Only call set_range_info_raw if domain
21001 (set_nonzero_bits): Call set_range_info_raw.
21002 * tree-ssanames.h (set_range_info_raw): New.
21004 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21007 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21008 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21009 of nonimmediate_operand and <store_mask_constraint> instead of m
21010 for the input operand. For V8FI iterator, always split if input
21011 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21012 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21013 <store_mask_predicate> instead of register_operand and
21014 <store_mask_constraint> instead of v for the input operand. Make
21015 sure both operands aren't MEMs for if not <mask_applied>.
21017 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21019 * lto-wrapper.c (copy_file) Close both file descriptors before
21022 2017-06-30 Martin Liska <mliska@suse.cz>
21025 * multiple_target.c (create_dispatcher_calls): Make ifunc
21026 also for function that don't have calls or are not referenced.
21028 2017-06-30 Richard Biener <rguenther@suse.de>
21030 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21031 analyze the first scalar stmt. Move vector type computation
21032 for the BB case here from ...
21033 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21034 live operation processing in the SLP case properly.
21036 2017-06-30 Richard Biener <rguenther@suse.de>
21038 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21040 2017-06-30 Martin Liska <mliska@suse.cz>
21043 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21044 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21046 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21048 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21051 2017-06-30 Richard Biener <rguenther@suse.de>
21053 PR tree-optimization/81249
21054 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21055 condition reduction result to original scalar type.
21057 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21059 * profile-count.h (enum profile_quality): Fix typos and whitespace
21062 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21064 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21065 type for branch probabilities.
21067 2017-06-29 Julian Brown <julian@codesourcery.com>
21068 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21070 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21071 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21072 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21073 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21075 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21077 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21078 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21079 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21080 CC usage from generic code to here.
21081 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21082 CC usage into the target macros.
21084 2017-06-29 Maya Rashish <coypu@sdf.org>
21086 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21089 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21091 * arm/arm-builtins.c: Include profile-count.h
21092 * except.c (sjlj_emit_function_enter): Use
21093 profile_probability::unlikely.
21095 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21097 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21098 and tocrel_offset be pointer args rather than implicitly using
21100 (legitimate_constant_pool_address_p, rs6000_emit_move,
21101 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21102 tocrel_offset and use in toc_relative_expr_p call.
21103 (print_operand, print_operand_address): Use static tocrel_base_oac
21104 and tocrel_offset_oac.
21105 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21108 2017-06-29 Maya Rashish <coypu@sdf.org>
21110 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21112 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21114 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21115 objects, take into account only the alignment of 'op0' and 'mode1' if
21118 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21120 * ccmp.c (ccmp_tree_comparison_p): New function.
21121 (ccmp_candidate_p): Update to use above function.
21122 (get_compare_parts): New function.
21123 (expand_ccmp_next): Update to use new functions.
21124 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21126 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21127 take mode as argument.
21128 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21129 * expr.c (expand_expr_real_1): Pass mode as argument.
21131 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21133 * combine.c (combine_instructions): Print insns to dump_file, together
21136 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21138 * asan.c (asan_emit_stack_protection): Update.
21139 (create_cond_insert_point): Update.
21140 * auto-profile.c (afdo_propagate_circuit): Update.
21141 * basic-block.h (struct edge_def): Turn probability to
21142 profile_probability.
21143 (EDGE_FREQUENCY): Update.
21144 * bb-reorder.c (find_traces_1_round): Update.
21145 (better_edge_p): Update.
21146 (sanitize_hot_paths): Update.
21147 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21148 (make_single_succ_edge): Update.
21149 (check_bb_profile): Update.
21150 (dump_edge_info): Update.
21151 (update_bb_profile_for_threading): Update.
21152 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21153 probabilitycount to 0.
21154 * cfgbuild.c (compute_outgoing_frequencies): Update.
21155 * cfgcleanup.c (try_forward_edges): Update.
21156 (outgoing_edges_match): Update.
21157 (try_crossjump_to_edge): Update.
21158 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21159 (expand_gimple_tailcall): Update.
21160 (construct_init_block): Use make_single_succ_edge.
21161 (construct_exit_block): Use make_single_succ_edge.
21162 * cfghooks.c (verify_flow_info): Update.
21163 (redirect_edge_succ_nodup): Update.
21164 (split_edge): Update.
21165 (account_profile_record): Update.
21166 * cfgloopanal.c (single_likely_exit): Update.
21167 * cfgloopmanip.c (scale_loop_profile): Update.
21168 (set_zero_probability): Remove.
21169 (duplicate_loop_to_header_edge): Update.
21170 * cfgloopmanip.h (loop_version): Update prototype.
21171 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21172 (force_nonfallthru_and_redirect): Update.
21173 (update_br_prob_note): Update.
21174 (rtl_verify_edges): Update.
21175 (purge_dead_edges): Update.
21176 (rtl_lv_add_condition_to_bb): Update.
21177 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21178 * cgraphunit.c (init_lowered_empty_function): Update.
21179 (cgraph_node::expand_thunk): Update.
21180 * cilk-common.c: Include profile-count.h
21181 * dojump.c (inv): Remove.
21182 (jumpifnot): Update.
21183 (jumpifnot_1): Update.
21184 (do_jump_1): Update.
21186 (do_jump_by_parts_greater_rtx): Update.
21187 (do_compare_rtx_and_jump): Update.
21188 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21189 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21190 * dwarf2cfi.c: Include profile-count.h
21191 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21192 (sjlj_emit_dispatch_table): Likewise.
21193 * explow.c: Include profile-count.h
21194 * expmed.c (emit_store_flag_force): Update.
21195 (do_cmp_and_jump): Update.
21196 * expr.c (compare_by_pieces_d::generate): Update.
21197 (compare_by_pieces_d::finish_mode): Update.
21198 (emit_block_move_via_loop): Update.
21199 (store_expr_with_bounds): Update.
21200 (store_constructor): Update.
21201 (expand_expr_real_2): Update.
21202 (expand_expr_real_1): Update.
21203 * expr.h (try_casesi, try_tablejump): Update prototypes.
21204 * gimple-pretty-print.c (dump_probability): Update.
21205 (dump_profile): New.
21206 (dump_gimple_label): Update.
21207 (dump_gimple_bb_header): Update.
21208 * graph.c (draw_cfg_node_succ_edges): Update.
21209 * hsa-gen.c (convert_switch_statements): Update.
21210 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21211 (find_if_case_1): Update.
21212 (find_if_case_2): Update.
21213 * internal-fn.c (expand_arith_overflow_result_store): Update.
21214 (expand_addsub_overflow): Update.
21215 (expand_neg_overflow): Update.
21216 (expand_mul_overflow): Update.
21217 (expand_vector_ubsan_overflow): Update.
21218 * ipa-cp.c (good_cloning_opportunity_p): Update.
21219 * ipa-split.c (split_function): Use make_single_succ_edge.
21220 * ipa-utils.c (ipa_merge_profiles): Update.
21221 * loop-doloop.c (add_test): Update.
21222 (doloop_modify): Update.
21223 * loop-unroll.c (compare_and_jump_seq): Update.
21224 (unroll_loop_runtime_iterations): Update.
21225 * lra-constraints.c (lra_inheritance): Update.
21226 * lto-streamer-in.c (input_cfg): Update.
21227 * lto-streamer-out.c (output_cfg): Update.
21228 * mcf.c (adjust_cfg_counts): Update.
21229 * modulo-sched.c (sms_schedule): Update.
21230 * omp-expand.c (expand_omp_for_init_counts): Update.
21231 (extract_omp_for_update_vars): Update.
21232 (expand_omp_ordered_sink): Update.
21233 (expand_omp_for_ordered_loops): Update.
21234 (expand_omp_for_generic): Update.
21235 (expand_omp_for_static_nochunk): Update.
21236 (expand_omp_for_static_chunk): Update.
21237 (expand_cilk_for): Update.
21238 (expand_omp_simd): Update.
21239 (expand_omp_taskloop_for_outer): Update.
21240 (expand_omp_taskloop_for_inner): Update.
21241 * omp-simd-clone.c (simd_clone_adjust): Update.
21242 * optabs.c (expand_doubleword_shift): Update.
21243 (expand_abs): Update.
21244 (emit_cmp_and_jump_insn_1): Update.
21245 (expand_compare_and_swap_loop): Update.
21246 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21247 * predict.c (predictable_edge_p): Update.
21248 (edge_probability_reliable_p): Update.
21249 (set_even_probabilities): Update.
21250 (combine_predictions_for_insn): Update.
21251 (combine_predictions_for_bb): Update.
21252 (propagate_freq): Update.
21253 (estimate_bb_frequencies): Update.
21254 (force_edge_cold): Update.
21255 * profile-count.c (profile_count::dump): Add missing space into dump.
21256 (profile_count::debug): Add newline.
21257 (profile_count::differs_from_p): Explicitly convert to unsigned.
21258 (profile_count::stream_in): Update.
21259 (profile_probability::dump): New member function.
21260 (profile_probability::debug): New member function.
21261 (profile_probability::differs_from_p): New member function.
21262 (profile_probability::differs_lot_from_p): New member function.
21263 (profile_probability::stream_in): New member function.
21264 (profile_probability::stream_out): New member function.
21265 * profile-count.h (profile_count_quality): Rename to ...
21266 (profile_quality): ... this one.
21267 (profile_probability): New.
21268 (profile_count): Update.
21269 * profile.c (compute_branch_probabilities): Update.
21270 * recog.c (peep2_attempt): Update.
21271 * sched-ebb.c (schedule_ebbs): Update.
21272 * sched-rgn.c (find_single_block_region): Update.
21273 (compute_dom_prob_ps): Update.
21274 (schedule_region): Update.
21275 * sel-sched-ir.c (compute_succs_info): Update.
21276 * stmt.c (struct case_node): Update.
21277 (do_jump_if_equal): Update.
21278 (get_outgoing_edge_probs): Update.
21279 (conditional_probability): Update.
21280 (emit_case_dispatch_table): Update.
21281 (expand_case): Update.
21282 (expand_sjlj_dispatch_table): Update.
21283 (emit_case_nodes): Update.
21284 * targhooks.c: Update.
21285 * tracer.c (better_p): Update.
21286 (find_best_successor): Update.
21287 * trans-mem.c (expand_transaction): Update.
21288 * tree-call-cdce.c: Update.
21289 * tree-cfg.c (gimple_split_edge): Upate.
21290 (move_sese_region_to_fn): Upate.
21291 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21292 * tree-eh.c (lower_resx): Upate.
21293 (cleanup_empty_eh_move_lp): Upate.
21294 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21295 * tree-inline.c (copy_edges_for_bb): Update.
21296 (copy_cfg_body): Update.
21297 * tree-parloops.c (gen_parallel_loop): Update.
21298 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21299 (gimple_gen_time_profiler): Update.
21300 * tree-ssa-dce.c (remove_dead_stmt): Update.
21301 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21302 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21303 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21304 (unloop_loops): Update.
21305 (try_peel_loop): Update.
21306 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21307 * tree-ssa-loop-split.c (connect_loops): Update.
21308 (split_loop): Update.
21309 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21310 (hoist_guard): Update.
21311 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21312 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21313 (value_replacement): Update.
21314 * tree-ssa-reassoc.c (branch_fixup): Update.
21315 * tree-ssa-tail-merge.c (replace_block_by): Update.
21316 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21317 (create_edge_and_update_destination_phis): Update.
21318 (compute_path_counts): Update.
21319 (recompute_probabilities): Update.
21320 (update_joiner_offpath_counts): Update.
21321 (freqs_to_counts_path): Update.
21322 (duplicate_thread_path): Update.
21323 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21324 (struct switch_conv_info): Update.
21325 (gen_inbound_check): Update.
21326 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21327 (vect_do_peeling): Update.
21328 (vect_loop_versioning): Update.
21329 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21330 (optimize_mask_stores): Update.
21331 * ubsan.c (ubsan_expand_null_ifn): Update.
21332 * value-prof.c (gimple_divmod_fixed_value): Update.
21333 (gimple_divmod_fixed_value_transform): Update.
21334 (gimple_mod_pow2): Update.
21335 (gimple_mod_pow2_value_transform): Update.
21336 (gimple_mod_subtract): Update.
21337 (gimple_mod_subtract_transform): Update.
21338 (gimple_ic): Update.
21339 (gimple_stringop_fixed_value): Update.
21340 (gimple_stringops_transform): Update.
21341 * value-prof.h: Update.
21343 2017-06-29 Carl Love <cel@us.ibm.com>
21345 * config/rs6000/rs6000-c.c: Add support for built-in functions
21346 vector signed int vec_signed (vector float);
21347 vector signed long long vec_signed (vector double);
21348 vector signed int vec_signed2 (vector double, vector double);
21349 vector signed int vec_signede (vector double);
21350 vector signed int vec_signedo (vector double);
21351 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21352 instruction generator.
21353 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21354 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21355 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21357 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21358 vunsignede_v2df): Add define_expands.
21359 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21360 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21361 VEC_UNSIGNEDO): Add definitions.
21362 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21363 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21364 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21365 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21366 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21367 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21368 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21369 vec_unsignede, vec_unsignedo): Add builtin defines.
21370 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21372 * doc/extend.texi: Update the built-in documentation file for the
21373 new built-in functions.
21375 2017-06-29 Richard Biener <rguenther@suse.de>
21377 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21378 reduction chains to LOOP_VINFO_REDUCTIONS.
21379 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21380 SLP reductions after processing reduction chains.
21382 2017-06-29 Nathan Sidwell <nathan@acm.org>
21384 * builtins.c (fold_builtin_FUNCTION): Use
21385 lang_hooks.decl_printable_name.
21387 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21389 PR middle-end/81194
21390 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21391 with only one label.
21392 * stmt.c (expand_case): Assert NCASES is greater than one.
21394 2017-06-29 Richard Biener <rguenther@suse.de>
21396 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21398 (group_case_labels): Likewise.
21399 (find_taken_edge): Push sanity checking on val to workers...
21400 (find_taken_edge_cond_expr): ... here
21401 (find_taken_edge_switch_expr): ... and here, handle cases
21402 with just a default label.
21403 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21404 (group_case_labels): Likewise.
21405 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21406 group_case_labels does anything cleanup the CFG again.
21408 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21410 PR tree-optimization/81196
21411 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21412 exit condition comparing two IVs.
21414 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21416 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21417 profile to the dummy entry at the end of the list of architectures.
21418 * config/arm/arm-cpu-cdata.h: Regenerated.
21420 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21421 Michael Collison <michael.collison@arm.com>
21424 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21426 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21427 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21428 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21429 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21430 of shift when the shift amount is masked with constant equal to
21431 the size of the mode.
21432 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21435 2017-06-29 Martin Liska <mliska@suse.cz>
21437 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21439 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21442 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21445 (edge_growth_cache_entry::edge_growth_cache_entry): New
21447 (reset_edge_growth_cache): Update to use constructor.
21449 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21451 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21452 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21453 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21455 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21457 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21458 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21460 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21462 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21463 (*-linux-uclibc*): Add t-uclibc tmake_file.
21464 * config/t-musl: New.
21465 * config/t-uclibc: New.
21467 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21469 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21471 (gen_comm_data): Emit architectural setting of arch_prof.
21472 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21474 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21475 (armv8-m.base, armv8-m.main): Likewise.
21476 * arm-protos.h (arm_build_target): Add profile field.
21477 (arch_option): Likewise.
21478 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21480 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21481 arm_active_target.profile.
21483 2017-06-28 Richard Biener <rguenther@suse.de>
21485 PR middle-end/81227
21486 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21487 TYPE_OVERFLOW_WRAPS.
21488 * match.pd (negate_expr_p): Likewise.
21489 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21490 fold_build2, not fold_binary.
21492 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21494 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21495 Convert memory address to Pmode.
21496 (aarch64_print_operand): Assert MEM operands are always Pmode.
21498 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21501 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21502 Remove redundant if.
21503 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21504 * config/arm/aarch-common-protos.h
21505 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21506 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21508 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21511 * multiple_target.c (create_dispatcher_calls): Set the default
21512 clone to be static, not public.
21514 2017-06-28 Richard Biener <rguenther@suse.de>
21516 * tree-vect-loop.c (vectorizable_reduction): Move special
21517 cond reduction IV var creation ...
21518 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21519 parameter. Use STMT_VINFO_VECTYPE.
21520 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21523 2017-06-28 Martin Liska <mliska@suse.cz>
21526 * ipa-visibility.c (non_local_p): Handle visibility.
21528 2017-06-28 Martin Liska <mliska@suse.cz>
21531 * common.opt: Add IntegerRange to various options.
21532 * opt-functions.awk (integer_range_info): New function.
21533 * optc-gen.awk: Add integer_range_info to cl_options struct.
21534 * opts-common.c (decode_cmdline_option): Handle
21535 CL_ERR_INT_RANGE_ARG.
21536 (cmdline_handle_error): Likewise.
21537 * opts.c (print_filtered_help): Show valid interval in
21538 when --help is provided.
21539 * opts.h (struct cl_option): Add range_min and range_max fields.
21540 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21542 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21544 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21545 (x * C EQ/NE y * C): New transformation.
21547 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21549 * genmultilib (combination_space): Accept '+' in option names.
21551 2017-06-28 Martin Liska <mliska@suse.cz>
21554 * asan.c (instrument_derefs): Bail out inner references
21555 that are hard register variables.
21557 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21560 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21561 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21563 2017-06-28 Richard Biener <rguenther@suse.de>
21565 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21566 (vect_get_slp_defs): Adjust.
21567 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21569 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21571 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21572 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21573 (vectorizable_reduction): Adjust.
21574 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21576 (vect_get_slp_defs): Likewise.
21577 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21578 (vectorizable_bswap): Adjust.
21579 (vectorizable_call): Likewise.
21580 (vectorizable_conversion): Likewise.
21581 (vectorizable_assignment): Likewise.
21582 (vectorizable_shift): Likewise.
21583 (vectorizable_operation): Likewise.
21584 (vectorizable_store): Likewise.
21585 (vectorizable_condition): Likewise.
21586 (vectorizable_comparison): Likewise.
21588 2017-06-28 Michael Collison <michael.collison@arm.com>
21591 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21593 (arm_gen_movmemqi): Removed unused variable 'i'.
21594 Convert 'for' loop into 'while' loop.
21595 (arm_expand_prologue): Remove last unnecessary set of insn.
21596 (thumb_pop): Remove unused variable 'pushed_words'.
21597 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21599 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21601 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21602 * config/s390/s390-protos.h: Add prototype of
21603 s390_rel_address_ok_p.
21604 * config/s390/s390.c (s390_got_symbol): New function.
21605 (s390_rel_address_ok_p): New function.
21606 (legitimize_pic_address): Use s390_rel_address_ok_p.
21607 (s390_load_got): Use s390_got_symbol.
21608 (s390_option_override): Issue error if
21609 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21610 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21612 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21614 2017-06-27 Andrew Pinski <apinski@cavium.com>
21616 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21617 (X * copysign (1.0, X)): New pattern.
21618 (X * copysign (1.0, -X)): New pattern.
21619 (copysign (-1.0, CST)): New pattern.
21621 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21623 * genmultilib (combination_space): Remove variable.
21624 Validate reuse rules against regular expression for any sequence
21625 of multilib options in any order.
21627 2017-06-27 Michael Collison <michael.collison@arm.com>
21629 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21630 call aarch64_split_simd_combine.
21631 * (aarch64_combine_internal<mode>): Delete pattern.
21632 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21633 Allow register and subreg operands.
21635 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21637 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21638 specific need, just fallback on defaults.
21639 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21641 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21642 Olivier Hainque <hainque@adacore.com>
21644 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21646 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21647 targets. Pick a default if no particular attempt applied.
21648 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21651 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21653 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21654 (x86_64-wrs-vxworks7): Likewise.
21656 2017-06-27 Marek Polacek <polacek@redhat.com>
21659 * ubsan.c (instrument_null): Check get_base_address's result for null.
21661 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21663 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21665 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21667 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21668 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21669 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21670 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21671 New function types.
21672 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21673 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21674 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21675 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21676 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21677 BUILT_IN_FEUPDATEENV): New builtins.
21678 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21679 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21680 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21681 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21683 (builtin_structptr_types): Adjust size.
21684 * tree.c (builtin_structptr_types): Add four entries.
21686 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21687 Olivier Hainque <hainque@adacore.com>
21689 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21690 (TLS_SYM): New local macro, forcing reference to __tls__ on
21691 link command lines for VxWorks 7 RTPs, triggering initialization
21693 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21694 OS features TLS support, true for RTPs on VxWorks 7.
21695 * config/vxworks.c (vxworks_override_options): Setup emutls
21698 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21700 * predict.c (test_prediction_value_range): Use -1U instead of -1
21701 to avoid narrowing conversion warning.
21702 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21703 to avoid narrowing conversion warning.
21704 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21706 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21708 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21710 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21711 64bit configurations.
21712 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21713 (SIZE_TYPE): Likewise.
21714 * config/vxworks.c (vxworks_emutls_var_fields): Use
21715 long_unsigned_type_node instead of unsigned_type_node as the offset
21716 field type, which is "pointer" mode in emutls.c.
21718 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21721 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21723 PR middle-end/81207
21724 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21725 gimple_vuse copying separately from gimple_vdef copying.
21727 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21729 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21732 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21733 Olivier Hainque <hainque@adacore.com>
21735 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21736 for all vxworks7 targets.
21737 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21738 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21739 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21740 variations for VX6/VX7 and 32/64bits later on in ...
21741 (VXWORKS_LIB_SPEC): Leverage new macros.
21742 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21743 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21745 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21747 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21748 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21750 2017-06-26 Carl Love <cel@us.ibm.com>
21752 * config/rs6000/rs6000-c.c: Add support for built-in functions
21753 vector bool char vec_reve (vector bool char);
21754 vector signed char vec_reve (vector signed char);
21755 vector unsigned char vec_reve (vector unsigned char);
21756 vector bool int vec_reve (vector bool int);
21757 vector signed int vec_reve (vector signed int);
21758 vector unsigned int vec_reve (vector unsigned int);
21759 vector bool long long vec_reve (vector bool long long);
21760 vector signed long long vec_reve (vector signed long long);
21761 vector unsigned long long vec_reve (vector unsigned long long);
21762 vector bool short vec_reve (vector bool short);
21763 vector signed short vec_reve (vector signed short);
21764 vector double vec_reve (vector double);
21765 vector float vec_reve (vector float);
21766 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21767 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21768 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21769 (altivec_vreve): New pattern.
21770 * config/rs6000/altivec.h (vec_reve): New define.
21771 * doc/extend.texi (vec_rev): Update the built-in documentation file
21772 for the new built-in functions.
21774 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21776 PR tree-optimization/71815
21777 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21779 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21781 (slsr_process_phi): Likewise.
21782 (replace_uncond_cands_and_profitable_phis): Don't replace a
21783 multiply candidate with a stride of 1 (copy or cast).
21784 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21786 (lowest_cost_path): Likewise.
21787 (total_savings): Likewise.
21789 2017-06-26 Richard Biener <rguenther@suse.de>
21792 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21793 Use def_builtin_pure for all gather builtins.
21795 2017-06-26 Richard Biener <rguenther@suse.de>
21797 PR tree-optimization/81203
21798 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21799 non-dominating BBs.
21801 2017-06-26 Marek Polacek <polacek@redhat.com>
21804 * doc/invoke.texi: Document -Wmultistatement-macros.
21806 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21808 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21809 arm_neon_ok_no_float_abi effective target.
21811 2017-06-26 Richard Biener <rguenther@suse.de>
21813 PR tree-optimization/80928
21814 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21815 (copy_bbs): Set BB_DUPLICATED flag early.
21816 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21818 (execute_on_shrinking_pred): Likewise.
21819 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21820 BB_DUPLICATED blocks.
21821 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21822 iterate over all PHIs considering removal of *gsi.
21824 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21826 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21829 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21831 * config/rs6000/rs6000-string.c: (expand_block_clear,
21832 do_load_for_compare, select_block_compare_mode,
21833 compute_current_alignment, expand_block_compare,
21834 expand_strncmp_align_check, expand_strn_compare,
21835 expand_block_move, rs6000_output_load_multiple)
21836 Move functions related to string/block move/compare
21837 to a separate file.
21838 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21839 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21840 for this function which is now used in two files.
21841 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21842 * config.gcc: Add rs6000-string.o to extra_objs for
21843 targets powerpc*-*-* and rs6000*-*-*.
21845 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21848 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21849 32-bit, since indexed is not valid for DImode.
21850 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21851 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21852 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21853 (define_peephole2 for Altivec d-form store): Likewise.
21856 * multiple_target.c (create_dispatcher_calls): Only create the
21857 dispatcher call if the function is the default clone of a
21858 versioned function.
21860 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21862 PR middle-end/80902
21863 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21864 a call, force the call to not be a tail call.
21866 2017-06-23 Jeff Law <law@redhat.com>
21868 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21871 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21873 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21874 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21875 handling for early expansion of vector shifts (sl,sr,sra,rl).
21876 (builtin_function_type): Add vector shift right instructions
21877 to the unsigned argument list.
21879 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21881 rtl-optimizatoin/79286
21882 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21883 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21884 trap. PIC register plus a const unspec without offset can never trap.
21886 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21888 * tree.h (builtin_structptr_type): New type.
21889 (builtin_structptr_types): Declare new array.
21890 * tree.c (builtin_structptr_types): New array.
21891 (free_lang_data, build_common_tree_nodes): Use it.
21893 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21896 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21899 2017-06-22 Matt Turner <mattst88@gmail.com>
21901 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21902 Lake models to skylake case. Assume skylake for unknown
21903 models with clflushopt.
21905 2017-06-22 Jeff Law <law@redhat.com>
21907 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21908 frame sizes that do not satisfy aarch64_uimm12_shift.
21910 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21912 * profile-count.h (apply_probability,
21913 apply_scale, probability_in): Fix checks for zero.
21915 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21917 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21918 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21920 2016-06-22 Richard Biener <rguenther@suse.de>
21922 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21923 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21924 REDUC_MAX_EXPR support.
21925 (vectorizable_reduction): Likewise.
21926 (vect_create_epilog_for_reduction): Likewise.
21928 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21930 * match.pd (A / (1 << B) -> A >> B): New.
21931 * generic-match-head.c: Include optabs-tree.h.
21932 * gimple-match-head.c: Likewise.
21933 * optabs-tree.h (target_supports_op_p): New.
21934 * optabs-tree.c (target_supports_op_p): New.
21936 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21938 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21939 $gcc_cv_ld --help output.
21940 (gcc_cv_ld_demangle): Likewise.
21941 (gcc_cv_ld_eh_frame_hdr): Likewise.
21942 (gcc_cv_ld_pie): Likewise.
21943 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21944 (gcc_cv_ld_buildid): Likewise.
21945 (gcc_cv_ld_sysroot): Likewise.
21946 (ld_bndplt_support): Likewise.
21947 (ld_pushpopstate_support): Likewise.
21948 * configure: Regenerate.
21949 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21951 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21954 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21955 operands indexes to avoid gap between operands and match_dups.
21957 2017-06-21 Andrew Pinski <apinski@cavium.com>
21959 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21960 Increment Arith_shift and Arith_shift_reg by 1.
21961 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21963 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21964 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21965 (aarch64_strip_extend): Add new argument and test for it.
21966 (aarch64_cheap_mult_shift_p): New function.
21967 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21968 add a cost if it is true.
21969 Update calls to aarch64_strip_extend.
21970 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21972 2017-06-21 Andrew Pinski <apinski@cavium.com>
21974 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21976 (thunderxt88): Likewise.
21977 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21978 (thunderx_prefetch_tune): New variable.
21979 (thunderx2t99_prefetch_tune): Update for the correct values.
21980 (thunderxt88_tunings): New variable.
21981 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21982 generic_prefetch_tune.
21983 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21985 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21987 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21988 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21989 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21990 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21991 (aarch64_atomic_cas<mode>, GPI): Likewise.
21993 2017-06-21 Martin Liska <mliska@suse.cz>
21995 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21996 statements on cold and hot labels.
21997 * predict.c (tree_estimate_probability_bb): Remove the
21998 prediction from this place.
22000 2017-06-21 Martin Liska <mliska@suse.cz>
22002 PR tree-optimization/79489
22003 * gimplify.c (maybe_add_early_return_predict_stmt): New
22005 (gimplify_return_expr): Call the function.
22006 * predict.c (tree_estimate_probability_bb): Remove handling
22008 * predict.def: Update comment about early return predictor.
22009 * gimple-predict.h (is_gimple_predict): New function.
22010 * predict.def: Change default value of early return to 66.
22011 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22013 * passes.def: Put pass_strip_predict_hints to the beginning of
22016 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22018 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22019 FUNCTION_DECL declarations.
22020 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22022 (dwaf2out_decl): Likewise.
22023 * godump.c (go_early_global_decl): Skip call to the real debug hook
22024 for FUNCTION_DECL declarations.
22025 * passes.c (rest_of_decl_compilation): Skip call to the
22026 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22027 -fdump-go-spec is passed.
22029 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22031 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22032 Reorder for compactness.
22033 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22034 (def_builtin_pure, def_builtin_pure2): New functions.
22035 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22037 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22039 * match.pd (nop_convert): New predicate.
22040 ((A +- CST1) +- CST2): Allow some NOP conversions.
22042 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22045 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22046 with ctors/dtors if GOVD_SHARED is set.
22048 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22050 * config/aarch64/aarch64.md (movti_aarch64):
22051 Emit mov rather than orr.
22052 (movtf_aarch64): Likewise.
22053 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22054 Emit mov rather than orr.
22056 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22058 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22059 Swap alternatives, make integer dup more expensive.
22061 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22063 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22064 Return true for non-tls symbols.
22066 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22068 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22069 (cortex-a75): Likewise.
22070 (cortex-a75.cortex-a55): Likewise.
22071 * config/aarch64/aarch64-tune.md: Regenerate.
22072 * doc/invoke.texi (-mtune): Document new values for -mtune.
22074 2017-06-21 Tom de Vries <tom@codesourcery.com>
22076 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22077 stack_size feature.
22078 (Effective-Target Keywords, Other attributes): Suggest using
22079 dg-add-options stack_size feature to get stack limit in stack_size
22080 effective target documentation.
22082 2017-06-21 Julian Brown <julian@codesourcery.com>
22083 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22085 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22086 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22087 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22089 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22090 attribute type list for neon_multiply.
22091 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22092 attribute type list for neon_multiply.
22093 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22094 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22095 attribute type list for neon_multiply.
22096 * config/arm/types.md (crypto_pmull): Add.
22097 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22098 attribute type list.
22100 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22102 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22105 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22107 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22108 to make sure not to dereference a NULL cost_classes_ptr pointer.
22110 2017-06-20 Carl Love <cel@us.ibm.com>
22112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22113 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22114 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22115 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22116 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22117 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22118 VMULOSW): New enum "unspec" values.
22119 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22120 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22121 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22122 altivec_vmulosw): New patterns.
22123 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22124 VMULOSW): Add definitions.
22126 2017-06-20 Julia Koval <julia.koval@intel.com>
22128 * config/i386/i386.c: Fix rounding expand for new pattern.
22129 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22131 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22133 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22134 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22136 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22138 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22141 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22143 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22144 architecture, then by implementer ID.
22145 * config/aarch64/aarch64-tune.md: Regenerate.
22147 2017-06-20 Richard Biener <rguenther@suse.de>
22149 PR middle-end/81097
22150 * fold-const.c (split_tree): Fold to type before negating.
22152 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22154 * diagnostic-show-locus.c
22155 (selftest::test_fixit_deletion_affecting_newline): New function.
22156 (selftest::diagnostic_show_locus_c_tests): Call it.
22158 2017-06-20 Andreas Schwab <schwab@suse.de>
22161 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22164 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22166 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22167 __ARM_FEATURE_COPROC according to support.
22169 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22171 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22172 Rewritten to avoid overflow for > 32-bit pointers.
22175 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22176 by removing enum keyword.
22177 (ubsan_type_descriptor): Likewise. Formatting fix.
22180 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22181 splitter): Require TARGET_SSE2 in the condition.
22183 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22186 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22187 for doing vector set of SFmode on ISA 3.0.
22188 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22189 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22191 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22192 SFmode value into a V4SF variable that was extracted from another
22193 V4SF variable without converting the element to double precision
22194 and back to single precision vector format.
22195 (vsx_insert_extract_v4sf_p9_2): Likewise.
22197 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22199 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22200 in UWHI to avoid undefined overflow.
22203 * ubsan.h (enum ubsan_encode_value_phase): New.
22204 (ubsan_encode_value): Change second argument to
22205 enum ubsan_encode_value_phase with default value of
22206 UBSAN_ENCODE_VALUE_GENERIC.
22207 * ubsan.c (ubsan_encode_value): Change second argument to
22208 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22209 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22210 create_tmp_var_raw instead of create_tmp_var and use a
22212 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22213 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22214 ubsan_encode_value callers.
22217 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22218 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22219 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22221 2017-06-19 Richard Biener <rguenther@suse.de>
22223 PR middle-end/81118
22224 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22225 estimates if we changed anything.
22227 2017-06-19 Richard Biener <rguenther@suse.de>
22229 PR tree-optimization/80887
22230 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22231 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22232 simplified lookups, then reset mprts_hook.
22233 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22235 (try_to_simplify): Likewise.
22237 2017-06-19 Martin Liska <mliska@suse.cz>
22240 * gimplify.c (gimplify_switch_expr):
22241 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22243 2017-06-19 Martin Liska <mliska@suse.cz>
22245 * doc/install.texi: Document that PGO runs in 4 stages.
22247 2017-06-19 Martin Liska <mliska@suse.cz>
22250 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22251 to dispatcher function name.
22252 * multiple_target.c (replace_function_decl): New function.
22253 (create_dispatcher_calls): Redirect both edges and references.
22255 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22257 * profile-count.c (profile_count::dump): Dump quality.
22258 (profile_count::differs_from_p): Update for unsigned val.
22259 * profile-count.h (profile_count_quality): New enum.
22260 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22262 2017-06-19 Richard Biener <rguenther@suse.de>
22264 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22265 struct function as arg.
22266 (estimate_numbers_of_iterations): Export overload with loop arg.
22267 (free_numbers_of_iterations_estimates_loop): Use an overload of
22268 free_numbers_of_iterations_estimates instead.
22269 * tree-cfg.c (remove_bb): Adjust.
22270 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22271 * tree-parloops.c (gen_parallel_loop): Likewise.
22272 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22274 (tree_unroll_loops_completely): Likewise.
22275 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22276 Use an overload instead and export.
22277 (estimated_loop_iterations): Adjust.
22278 (max_loop_iterations): Likewise.
22279 (likely_max_loop_iterations): Likewise.
22280 (estimate_numbers_of_iterations): Take struct function as arg
22282 (loop_exits_before_overflow): Adjust.
22283 (free_numbers_of_iterations_estimates_loop): Use an overload.
22284 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22285 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22287 2017-06-19 Richard Biener <rguenther@suse.de>
22290 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22291 RANGE_EXPR conservatively.
22293 2017-06-16 Carl Love <cel@us.ibm.com>
22295 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22296 definitions for vec_float, vec_float2, vec_floato,
22297 vec_floate built-ins.
22298 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22299 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22301 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22302 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22303 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22304 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22305 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22306 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22307 vec_floato): Add builtin defines.
22308 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22309 Update the built-in documentation file for the new built-in
22312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22314 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22315 (mthumb): Mark as the negative of -marm.
22317 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22319 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22321 (ARM Options, -mtune): Document that this accepts the same
22322 extension options as -mcpu.
22323 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22325 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22327 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22328 permitted extensions.
22330 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22332 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22333 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22334 (armv8-m.main): Add option +nodsp.
22335 * config/arm/arm-cpu-cdata.h: Regenerated.
22337 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22339 * config/arm/t-fuchsia: New file.
22340 * config.gcc (arm*-*-fuchsia*): Use it.
22342 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22344 * config/arm/t-symbian: Rewrite for new option infrastructure.
22346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22348 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22349 (MULTILIB_REQUIRED): Likewise.
22351 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22353 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22354 (MULTILIB_RESUE): Likewise.
22355 (MULTILIB_MATCHES): Likewise.
22356 (MULTLIB_REQUIRED): Likewise.
22358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22360 * config/arm/t-rtems: Rewrite for new option framework.
22362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22364 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22365 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22366 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22367 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22368 * config/arm/t-multilib: ... here.
22369 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22370 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22371 armv7-a and armv8*-a when A-profile libraries have not been built.
22372 * config/arm/t-rmprofile: Rewrite.
22374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22376 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22377 with a backslash. Remove the backslash after substituting unescaped
22379 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22383 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22384 the driver to pass through the default setting of -mfloat-abi.
22385 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22387 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22388 (all_feat_combs): New rule.
22389 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22391 * config/arm/t-aprofile: Rewrite.
22393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22395 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22396 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22397 fpu is not specified by the user/command-line.
22398 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22399 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22400 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22401 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22402 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22403 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22407 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22408 * config/arm/t-arm-elf: Rewritten.
22410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22412 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22413 have some floating-point instructions.
22414 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22415 (TARGET_MAYBE_HARD_FLOAT): New macro.
22416 * config/arm/arm-builtins.c (arm_init_builtins): Use
22417 TARGET_MAYBE_HARD_FLOAT.
22418 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22422 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22423 (configargs.h): Include it.
22424 (arm_print_hint_for_fpu_option): New function.
22425 (arm_parse_fpu_option): New function.
22426 (candidate_extension): New class.
22427 (arm_canon_for_multilib): New function.
22428 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22429 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22430 (ARCH_CANONICAL_SPECS): New macro.
22431 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22435 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22436 are set after handling multilib fragments. Set target_cpu_default2
22439 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22441 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22443 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22444 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22446 (check_arch): Likewise.
22447 * config/arm/arm.c (arm_configure_build_target): Handle
22448 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22449 options in the default.
22451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22453 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22454 when an option is an alias of another.
22455 * config/arm/parsecpu.awk (optalias): New parser token.
22456 (gen_comm_data): Mark non-alias options as such. Emit entries
22457 for extension aliases.
22458 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22459 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22460 (armv6kz, armv6zk, armv6t2): Likewise.
22461 (armv7): Make vfpv3-d16 an alias.
22462 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22464 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22465 Sort in canonical order.
22466 (armv8-a): Sort in canonical order.
22467 (armv8.1-a, armv8.2-a): Likewise.
22468 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22470 (cortex-a9): Sort in canonical order.
22471 * config/arm/arm.c (selftests.h): Include it.
22472 (arm_test_cpu_arch_data): New function.
22473 (arm_run_self_tests): New function.
22474 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22475 (targetm): Move declaration to the end of the file.
22476 * arm-cpu-cdata.h: Regenerated.
22478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22480 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22481 call to target_mode_check describing the type of option passed.
22482 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22483 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22484 arm_parse_cpu_option_name to match parameters against list of
22486 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22487 arm_arch_core_flags data structure.
22488 * config/arm/arm-cpu_cdata.h: Regenerated.
22490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22492 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22494 (arm_print_hint_for_cpu_option): Likewise.
22495 (arm_print_hint_for_arch_option): Likewise.
22496 (arm_parse_cpu_option_name): Likewise.
22497 (arm_parse_arch_option_name): Likewise.
22498 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22499 of entries in the all_fpus list.
22500 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22501 (arm_parse_cpu_option_name): Declare.
22502 (arm_parse_arch_option_name): Declare.
22503 (arm_parse_option_features): Declare.
22504 (arm_intialize_isa): Declare.
22505 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22507 (gen_comm_data): ... here. Make definitions non-static.
22508 * config/arm/arm-cpu-data.h: Regenerated.
22509 * config/arm/arm-cpu-cdata.h: Regenerated.
22511 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22513 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22514 (cpu_arch_extension): New structure.
22515 (cpu_arch_option, arch_option, cpu_option): New structures.
22516 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22517 architecture types.
22518 (gen_data): Generate new format data tables.
22519 * config/arm/arm.c (cpu_tune): New structure.
22520 (cpu_option, processors): Delete.
22521 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22522 (arm_print_hint_for_cpu_option): ... this and ...
22523 (arm_print_hint_for_arch_option): ... this.
22524 (arm_parse_arch_cpu_name): Delete. Replace with ...
22525 (arm_parse_cpu_option_name): ... this and ...
22526 (arm_parse_arch_option_name): ... this.
22527 (arm_unrecognized_feature): Change type of target parameter to
22529 (arm_parse_arch_cpu_features): Delete. Replace with ...
22530 (arm_parse_option_features): ... this.
22531 (arm_configure_build_target): Rework to use new configuration data
22533 (arm_print_tune_info): Rework for new configuration data tables.
22534 * config/arm/arm-cpu-data.h: Regenerated.
22535 * config/arm/arm-cpu.h: Regenerated.
22537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22539 * Makefile.in (OBJS): Move sbitmap.o from here ...
22540 (OBJS-libcommon): ... to here.
22542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22544 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22545 (ISA_ALL_CRYPTO): New macro.
22546 (ISA_ALL_SIMD): New macro
22547 (ISA_ALL_FP): New macro.
22548 * config/arm/arm.c (fpu_bitlist): Update initializer.
22549 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22551 (arm9e): Add fpu. Add option for nofp
22552 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22553 (arm926ej-s, arm1026ej-s): Likewise.
22554 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22555 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22556 neon-fp16, neon-vfpv4, nofp and nosimd.
22557 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22558 (cortex-a8): Add fpu. Add option for nofp.
22559 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22560 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22561 (cortex-r4f): Add fpu.
22562 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22563 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22565 (cortex-r8): Likewise.
22566 (cortex-m4): Add fpu. Add option for nofp.
22567 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22568 (cortex-a17.cortex-a7): Likewise.
22569 (cortex-a32): Add fpu. Add options for crypto and nofp.
22570 (cortex-a35, cortex-a53): Likewise.
22571 (cortex-a57): Add fpu. Add option for crypto.
22572 (cortex-a72, cortex-a73): Likewise.
22573 (exynos-m1): Likewise.
22574 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22575 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22576 (cortex-m33): Add fpu. Add option for nofp.
22577 * config/arm/arm-cpu-cdata.h: Regenerated
22578 * config/arm/arm-cpu-data.h: Regenerated.
22580 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22582 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22583 (armv5te, armv5tej): Likewise.
22584 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22585 (armv7): Add options fp and vfpv3-d16.
22586 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22587 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22589 (armv7ve): Likewise.
22590 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22591 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22592 (armv8-a): Add nocrypto option.
22593 (armv8.1-a, armv8.2-a): Likewise.
22594 (armv8-m.main): add options fp, fp.dp and nofp.
22596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22598 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22600 (armv8-a+crc): Delete.
22601 (armv8.1-a): Add options simd, crypto and nofp.
22602 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22603 (armv8.2-a+fp16): Delete.
22604 (armv8-m.main): Add option dsp.
22605 (armv8-m.main+dsp): Delete.
22606 (cortex-a8): Add fpu. Add nofp option.
22607 (cortex-a9): Add fpu. Add nofp and nosimd options.
22608 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22609 link to main cpu and architecture data structures.
22610 (gen_comm_data): Only put isa attributes from the main architecture
22612 (option): New statement for architecture and CPU entries.
22613 * arm.c (struct cpu_option): New structure.
22614 (struct processors): Add entry for options.
22615 (arm_unrecognized_feature): New function.
22616 (arm_parse_arch_cpu_name): Ignore any characters after the first
22618 (arm_parse_arch_cpu_feature): New function.
22619 (arm_configure_build_target): Separate out any CPU and architecture
22620 features and parse separately. Don't error out if -mfpu=auto is
22621 used with only an architecture string.
22622 (arm_print_asm_arch_directives): New function.
22623 (arm_file_start): Call it.
22624 * config/arm/arm-cpu-cdata.h: Regenerated.
22625 * config/arm/arm-cpu-data.h: Likewise.
22626 * config/arm/arm-tables.opt: Likewise.
22628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22630 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22631 assembler when it is not -mfpu=auto.
22633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22635 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22636 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22637 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22638 (ASM_CPU_SPEC): Rewrite.
22639 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22640 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22641 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22642 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22643 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22644 copied string is NUL-terminated. Also strip any characters prefixed
22646 (arm_rewrite_selected_arch): New function.
22647 (arm_rewrite_march): New function.
22649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22651 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22652 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22653 (march, mcpu, mtune): Convert to string-based options.
22654 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22655 (arm_parse_arch_cpu_name): New function.
22656 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22657 identify selected architecture or CPU.
22658 (arm_option_save): New function.
22659 (TARGET_OPTION_SAVE): Redefine.
22660 (arm_option_restore): Restore string options.
22661 (arm_option_print): Print string options.
22663 2017-06-16 Martin Sebor <msebor@redhat.com>
22665 PR tree-optimization/80933
22666 PR tree-optimization/80934
22667 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22668 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22669 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22670 (gimple_fold_builtin): Call them.
22672 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22674 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22675 as unlikely; update profile.
22677 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22679 * predict.c (force_edge_cold): Handle declaring edges impossible
22682 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22684 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22686 (try_unroll_loop_completely): Fix reporting.
22688 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22690 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22692 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22695 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22696 if given a non-constant argument for an intrinsic which requires a
22699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22701 * profile.c (compare_freqs): New function.
22702 (branch_prob): Sort edge list.
22703 (find_spanning_tree): Assume that the list is priority sorted.
22705 2017-06-16 Richard Biener <rguenther@suse.de>
22707 PR tree-optimization/81090
22708 * passes.def (pass_record_bounds): Remove.
22709 * tree-pass.h (make_pass_record_bounds): Likewise.
22710 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22711 make_pass_record_bounds): Likewise.
22712 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22713 not free niter estimates at the beginning but at the end.
22714 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22716 2017-06-16 Richard Biener <rguenther@suse.de>
22718 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22719 initializer to workaround ICE in host GCC 4.8.
22721 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22723 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22725 (clone_inlined_nodes): Update.
22727 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22729 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22730 prefetch settings, and enable prefetching by default at -O3.
22732 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22734 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22735 Set flag_prefetch_loop_arrays according to tuning data.
22737 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22739 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22740 New tune structure.
22741 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22742 [Unrelated to main purpose of the patch] Place the pointer field last
22743 to enable type checking errors when tune structure are wrongly merged.
22744 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22745 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22746 (thunderx2t99_prefetch_tune): New tune constants.
22747 (tune_params *_tunings): Update all tunings (no functional change).
22748 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22749 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22750 from tunings structures.
22752 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22755 * ubsan.c (instrument_null): Add T argument, use it instead
22756 of computing it based on IS_LHS.
22757 (instrument_object_size): Likewise.
22758 (pass_ubsan::execute): Adjust instrument_null and
22759 instrument_object_size callers to pass gimple_get_lhs or
22760 gimple_assign_rhs1 result to it. Use instrument_null instead of
22761 calling get_base_address and instrument_mem_ref. Handle
22762 aggregate call arguments for object-size sanitization.
22764 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22766 PR tree-optimization/81089
22767 * tree-vrp.c (is_masked_range_test): Validate operands of
22770 2017-06-15 Martin Sebor <msebor@redhat.com>
22773 * dumpfile.c (dump_register): Avoid calling memset to initialize
22774 a class with a default ctor.
22775 * gcc.c (struct compiler): Remove const qualification.
22776 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22777 * hash-table.h: Ditto.
22778 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22780 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22781 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22783 * params.h (struct param_info): Make struct members non-const.
22784 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22785 with default initialization.
22786 * vec.h (vec_copy_construct, vec_default_construct): New helper
22788 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22789 with vec_copy_construct.
22790 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22791 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22792 * doc/invoke.texi (-Wclass-memaccess): Document.
22794 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22796 * emit-rtl.h (is_leaf): Update comment about local
22797 register allocator.
22799 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22802 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22803 for a variable to have a section before checking if the section has a
22805 Set section to.persistent if persistent attribute is set.
22806 Warn if .persistent attribute is used on an automatic variable.
22808 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22810 PR rtl-optimization/80474
22811 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22813 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22815 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22818 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22820 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22821 (casesi): Emit bounds checking as RTL.
22822 (casesi_internal_mips16_<mode>): Remove bounds checking.
22824 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22826 * config/xtensa/xtensa.c (xtensa_option_override): Append
22827 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22828 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22829 xtensa_doloop_hooks): Define unconditionally.
22830 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22832 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22833 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22834 for it in xtensa_option_override.
22835 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22836 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22838 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22840 * doc/cppopts.texi: Document '-' special value to -MF.
22842 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22844 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22845 (cortex_a53_fconst): Likewise.
22846 (cortex_a53_fpmul): Likewise.
22847 (cortex_a53_f_load_64): Likewise.
22848 (cortex_a53_f_load_many): Likewise.
22849 (cortex_a53_advsimd_alu): Likewise.
22850 (cortex_a53_advsimd_alu_q): Likewise.
22851 (cortex_a53_advsimd_mul): Likewise.
22852 (cortex_a53_advsimd_mul_q): Likewise.
22853 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22854 Add missing fmul, r2f_cvt and fconst cases.
22856 2017-06-14 Richard Biener <rguenther@suse.de>
22858 PR middle-end/81088
22859 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22861 (fold_binary_loc): When associating do not treat pre-existing
22862 TREE_OVERFLOW on literal constants as a reason to allow
22863 TREE_OVERFLOW on associated literal constants.
22865 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22867 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22868 (MASK_FEATURES): New macro.
22869 * config/sparc/sparc.c (sparc_option_override): Remove the special
22870 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22872 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22874 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22875 a division of 0 if non-call exceptions are enabled.
22877 2017-06-14 Andrew Pinski <apinski@cavium.com>
22878 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22881 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22882 Improve vector initialization code gen for only variable case.
22884 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22886 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22888 2017-06-14 Richard Biener <rguenther@suse.de>
22890 PR tree-optimization/81083
22891 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22894 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22896 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22897 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22898 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22899 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22900 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22901 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22903 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22905 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22906 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22908 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22910 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22912 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22914 * config/rs6000/t-rtems: Don't handle SPE.
22916 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22918 * config/rs6000/t-linux: Don't handle SPE.
22920 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22922 * config/rs6000/eabispe.h: Delete file.
22924 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22926 * config/rs6000/t-spe: Delete file.
22928 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22930 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22931 (rs6000_legitimate_offset_address_p): Return false for anything in
22932 V2SImode or V2SFmode.
22934 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22936 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22937 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22938 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22940 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22941 (rs6000_legitimate_offset_address_p): Ditto.
22942 (rs6000_emit_move): Ditto.
22943 (rs6000_init_builtins): Remove V4HI_type_node.
22945 2017-06-13 Martin Liska <mliska@suse.cz>
22948 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22949 (gate_asan): Likewise.
22950 * asan.h (asan_no_sanitize_address_p): Remove the function.
22951 (sanitize_flags_p): New function.
22952 * builtins.def: Fix coding style.
22953 * common.opt: Use renamed enum value.
22954 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22955 * doc/extend.texi: Document no_sanitize attribute.
22956 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22957 to SANITIZE_UNDEFINED_NONDEFAULT.
22958 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22959 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22960 Use sanitize_flags_p.
22961 * gimplify.c (gimplify_function_tree): Likewise.
22962 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22963 * opts.c (parse_no_sanitize_attribute): New function.
22964 (common_handle_option): Use renamed enum value.
22965 * opts.h (parse_no_sanitize_attribute): Declare.
22966 * tree.c (sanitize_flags_p): New function.
22967 * tree.h: Declared here.
22968 * tsan.c: Use sanitize_flags_p.
22969 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22970 (instrument_mem_ref): Likewise.
22971 (instrument_bool_enum_load): Likewise.
22972 (do_ubsan_in_current_function): Remove the function.
22973 (pass_ubsan::execute): Use sanitize_flags_p.
22974 * ubsan.h: Remove do_ubsan_in_current_function
22975 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22976 (dump_function_to_file): Use it here.
22978 2017-06-13 Martin Jambor <mjambor@suse.cz>
22980 PR tree-optimization/80803
22981 PR tree-optimization/81063
22982 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22983 (propagate_subaccesses_across_link): Enqueue subtree whenever
22984 necessary instead of relying on the caller.
22986 2017-06-13 Martin Jambor <mjambor@suse.cz>
22988 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22989 that have a first_link.
22990 (sort_and_splice_var_accesses): Do not check first_link before
22992 (subtree_mark_written_and_enqueue): Likewise.
22993 (propagate_all_subaccesses): Likewise and do not stop at first
22994 parent with a first_link.
22996 2017-06-13 Martin Jambor <mjambor@suse.cz>
22998 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23001 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23003 * match.pd: New pattern.
23005 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23007 * tree-vrp.c (is_masked_range_test): New function.
23008 (register_edge_assert_for): Determine ranges for
23011 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23013 PR tree-optimization/67328
23014 * fold-const.c (maskable_range_p): New function.
23015 (build_range_check): Generate bittests if possible.
23017 2017-06-13 Martin Liska <mliska@suse.cz>
23019 * gimple-pretty-print.c (dump_probability): Add new argument.
23020 (dump_edge_probability): Dump both probability and count.
23021 (dump_gimple_label): Likewise.
23022 (dump_gimple_bb_header): Likewise.
23024 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23027 * config/avr/avr-devices.c: Fix indentation.
23028 * config/avr/gen-avr-mmcu-specs.c: Dito.
23030 2017-06-13 Richard Biener <rguenther@suse.de>
23032 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23033 instead get vector type from stmt_info.
23034 (vectorizable_reduction): Adjust. Remove dead code.
23036 2017-06-13 Richard Biener <rguenther@suse.de>
23038 PR middle-end/81065
23039 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23040 case of C * (x * C2 + C3).
23041 (fold_addr_of_array_ref_difference): Properly fold index difference.
23043 2017-06-12 David S. Miller <davem@davemloft.net>
23046 * config/sparc/sparc.md (return expander): Emit frame blockage if
23047 function uses alloca.
23049 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23051 * combine.c (make_field_assignment): Check len rather than the mode
23052 precision when calling force_to_mode.
23054 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23056 Support multilibs and devices that see flash in RAM address range.
23059 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23060 (avr_mcu_t) <flash_pm_offset>: New field.
23061 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23062 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23063 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23064 (AVR_TINY_PM_OFFSET): Remove macro.
23065 * config/avr/avr.opt (-mshort-calls): New option.
23066 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23067 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23068 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23069 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23070 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23071 instead of avr_arch->have_jmp_call.
23072 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23073 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23074 avr_arch->flash_pm_offset to define.
23075 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23076 new field flash_pm_offset. Add entry for avrxmega3.
23077 (avr_texinfo): Add entry for avrxmega3.
23078 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23079 attiny212, attiny214,
23080 attiny412, attiny414, attiny416, attiny417,
23081 attiny814, attiny816, attiny817,
23082 attiny1614, attiny1616, attiny1617,
23083 attiny3214, attiny3216, attiny3217.
23084 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23085 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23086 (avr_print_operand_address) [AVR_TINY]: Same.
23087 (avr_asm_init_sections) <readonly_data_section>: Only patch
23088 callback if avr_arch->flash_pm_offset = 0.
23089 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23090 for rodata if avr_arch->flash_pm_offset != 0.
23091 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23092 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23093 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23094 (m_options): Append opt_rcall.
23095 (m_dirnames): Append dir_rcall.
23096 * config/avr/t-multilib: Regenerate.
23098 * configure.ac [target=avr]: Check whether avrxmega3 default
23099 linker description file works as needed.
23100 * configure: Regenerate.
23101 * doc/avr-mmcu.texi: Regenerate.
23102 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23103 <__AVR_ARCH__>: Document avrxmega3 and 103.
23104 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23105 <__AVR_SHORT_CALLS__>: Document it.
23106 <__AVR_PM_BASE_ADDRESS__>: Document it.
23107 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23108 (AVR Variable Attributes) <progmem>: Document this is
23109 not needed for avrxmega3.
23110 (AVR Named Address Spaces) <__flash>: Dito.
23112 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23114 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23116 2017-06-12 Doug Rupp <rupp@adacore.com>
23118 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23119 Append vxworks-stdint.h to the tm_file list.
23120 * config/vxworks-stdint.h: New file.
23122 2017-06-12 Martin Liska <mliska@suse.cz>
23124 PR tree-optimization/81041
23125 * tree-profile.c (gimple_gen_ic_func_profiler):
23126 Create an extra BB in profile-generate
23127 (gimple_gen_time_profiler): Likewise.
23129 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23131 PR tree-optimization/81003
23132 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23133 (update_range_test): Use it instead of force_gimple_operand_gsi.
23135 2017-06-12 Richard Biener <rguenther@suse.de>
23137 PR tree-optimization/81053
23138 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23139 with backedge value not defined in loop. Simplify def stmt
23142 2017-06-11 Tom de Vries <tom@codesourcery.com>
23145 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23147 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23148 nvptx_cannot_force_const_mem.
23150 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23152 * opts.c (finish_options): Move test for flag_split_stack after
23153 it has been initialized.
23155 2017-06-11 Jason Merrill <jason@redhat.com>
23157 * tree.h (id_equal): New.
23158 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23159 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23160 instead of strcmp of IDENTIFIER_POINTER.
23162 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23164 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23165 (mark_all_inlined_calls_cdtor): Fix formating.
23166 (inline_transform): Rescale profile before inlining.
23168 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23170 * cgraph.h (cgraph_edge::clone): Update prototype.
23171 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23172 (cgraph_node::create_clone): Update.
23173 (cgraph_node::create_version_clone): Update.
23174 * tree-inline.c (copy_bb): Update.
23175 (expand_call_inline): Update.
23177 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23179 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23180 factored out from ...
23181 (rs6000_emit_prologue): ... here.
23183 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23185 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23186 factored out from ...
23187 (rs6000_emit_prologue): ... here.
23189 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23191 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23193 (handle_missing_profiles): Fix computation of tp_first_run.
23194 (counts_to_freqs): Do not touch freqs when count is 0.
23196 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23198 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23201 2017-06-10 Tom de Vries <tom@codesourcery.com>
23203 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23204 attributes): Document signal effective target.
23206 2017-06-10 Tom de Vries <tom@codesourcery.com>
23208 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23209 Document effective target stack_size.
23211 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23213 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23214 to the edit_context if they can be auto-applied.
23216 2017-06-9 Ian Lance Taylor <iant@golang.org>
23218 * opts.c (finish_options): If -fsplit-stack, disable implicit
23219 -forder-blocks-and-partition.
23220 * doc/invoke.texi (Optimize Options): Document that when using
23221 -fsplit-stack -forder-blocks-and-partition is not implicitly
23224 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23226 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23227 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23228 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23229 * builtins.def (abort, trap, unreachable): Declare cold.
23230 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23231 * tree-core.h (ECF_COLD): New.
23232 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23233 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23235 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23237 * predict.c (unlikely_executed_stmt_p): Cleanup.
23239 2017-06-09 Richard Biener <rguenther@suse.de>
23241 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23242 model if the ref is always written to.
23244 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23246 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23248 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23250 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23253 2017-06-09 Tom de Vries <tom@codesourcery.com>
23256 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23257 "target cannot support label values" when encountering LABEL_REF.
23259 2017-06-09 Martin Liska <mliska@suse.cz>
23261 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23262 (gimple_gen_ic_func_profiler): Emit direct comparison
23263 of __gcov_indirect_call_callee with NULL.
23264 (gimple_gen_time_profiler): Change probability from
23265 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23267 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23269 * profile.c (edge_gcov_counts): Turn to pointer.
23270 (compute_branch_probabilities, compute_branch_probabilities): Update.
23271 (branch_prob): Do not clear edge_gcov_count.
23272 * profile.h (edge_gcov_counts): Turn to pointer.
23273 (edge_gcov_count): Update.
23275 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23277 * gimple.h (gimple_check_failed): Mark cold.
23279 2017-06-09 Richard Biener <rguenther@suse.de>
23281 PR tree-optimization/66623
23282 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23283 refactor check_reduction into two parts, properly computing
23284 whether we have to check reduction validity for outer loop
23287 2017-06-09 Richard Biener <rguenther@suse.de>
23289 PR tree-optimization/79483
23290 * graphite-scop-detection.c (order): New global.
23291 (get_order): Compute bb to order mapping that satisfies code
23292 generation constraints.
23293 (cmp_pbbs): New helper.
23294 (build_scops): Start domwalk at entry block, sort generated
23297 2017-06-09 Richard Biener <rguenther@suse.de>
23299 PR middle-end/81007
23300 * ipa-polymorphic-call.c
23301 (ipa_polymorphic_call_context::restrict_to_inner_class):
23302 Skip FIELD_DECLs with error_mark_node type.
23303 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23306 2017-06-09 Martin Liska <mliska@suse.cz>
23308 * predict.c (struct branch_predictor): New struct.
23309 (test_prediction_value_range): New test.
23310 (predict_c_tests): New function.
23311 * selftest-run-tests.c (selftest::run_tests): Run the function.
23312 * selftest.h: Declare new tests.
23314 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23317 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23318 gen_add3_insn did not fail.
23319 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23320 r0, construct that number in a temporary reg and add that reg to r0.
23321 If asked to put the result in r0 as well, fail.
23323 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23325 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23326 for early expansion of vec_eqv.
23328 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23330 PR middle-end/81005
23331 * ubsan.c (instrument_null): Avoid pointless code temporary.
23332 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23334 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23338 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23341 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23344 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23346 * predict.c (unlikely_executed_edge_p): Move ahead.
23347 (probably_never_executed_edge_p): Use it.
23349 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23351 PR middle-end/79988
23352 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23353 gimple_call_builtin_p call.
23355 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23357 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23358 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23359 rtl_check_failed_type2, rtl_check_failed_code1,
23360 rtl_check_failed_code2, rtl_check_failed_code_mode,
23361 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23362 rtvec_check_failed_bounds, rtl_check_failed_flag,
23363 _fatal_insn_not_found, _fatal_insn): Likewise.
23364 * tree.h (tree_contains_struct_check_failed,
23365 tree_check_failed, tree_not_check_failed,
23366 tree_class_check_failed, tree_range_check_failed,
23367 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23368 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23369 tree_operand_check_failed, omp_clause_check_failed,
23370 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23373 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23375 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23376 flag_branch_probabilities.
23377 * ipa-inline.c (edge_badness): Likewise.
23378 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23379 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23380 * predict.c (maybe_hot_frequency_p): Likewise.
23381 (probably_never_executed): Likewise.
23382 * sched-ebb.c (schedule_ebbs): Likewise.
23383 * sched-rgn.c (find_single_block_region): Likewise.
23384 * tracer.c (tail_duplicate): Likewise.
23386 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23388 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23389 longer requires x_flag_profile_use.
23391 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23393 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23394 instead of flag_reorder_blocks_and_partition.
23395 * dbxout.c (dbxout_function_end): Likewise.
23396 * dwarf2out.c (gen_subprogram_die): Likewise.
23397 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23398 * hw-doloop.c (reorg_loops): Likewise.
23399 * varasm.c (assemble_start_function,
23400 assemble_end_function): Likewise.
23401 (decide_function_section): Do not check for
23402 flag_reorder_blocks_and_partition.
23404 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23406 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23408 (chkp_get_hard_register_fake_addr_expr): Ditto.
23409 (chkp_build_addr_expr): Add check for hard reg case.
23410 (chkp_parse_array_and_component_ref): Ditto.
23411 (chkp_find_bounds_1): Ditto.
23412 (chkp_process_stmt): Don't generate bounds store for
23415 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23417 * predict.c (maybe_hot_bb_p): Do not check profile status.
23418 (maybe_hot_edge_p): Likewise.
23419 (probably_never_executed): Check for zero counts even if profile
23421 (unlikely_executed_edge_p): New function.
23422 (unlikely_executed_stmt_p): New function.
23423 (unlikely_executed_bb_p): New function.
23424 (set_even_probabilities): Use unlikely predicates.
23425 (combine_predictions_for_bb): Likewise.
23426 (predict_paths_for_bb): Likewise.
23427 (predict_paths_leading_to_edge): Likewise.
23428 (determine_unlikely_bbs): New function.
23429 (estimate_bb_frequencies): Use it.
23430 (compute_function_frequency): Use zero counts even if profile is
23432 * profile-count.h: Fix typo.
23434 2017-08-08 Julia Koval <julia.koval@intel.com>
23436 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23437 _mm512_mask_cvtsepi16_storeu_epi8,
23438 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23439 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23440 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23441 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23442 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23443 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23444 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23445 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23446 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23447 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23448 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23449 __builtin_ia32_pmovuswb256mem_mask,
23450 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23451 __builtin_ia32_pmovwb512mem_mask): New builtins.
23453 2017-08-08 Julia Koval <julia.koval@intel.com>
23455 PR target/73350,80862
23456 * config/i386/subst.md (round): Fix round pattern.
23457 * config/i386/i386.c (ix86_erase_embedded_rounding):
23458 Fix erasing rounding for the fixed pattern.
23460 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23462 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23464 2017-06-08 Martin Liska <mliska@suse.cz>
23466 PR gcov-profile/80911
23467 * gcov.c (block_info::block_info): New constructor.
23469 2017-06-07 Carl Love <cel@us.ibm.com>
23471 * config/rs6000/rs6000-c: The return type of the following
23472 built-in functions was implemented as int not long long. Fix sign
23473 of return value for the unsigned version of vec_mulo and vec_mule.
23474 vector unsigned long long vec_bperm (vector unsigned long long,
23475 vector unsigned char)
23476 vector signed long long vec_mule (vector signed int,
23478 vector unsigned long long vec_mule (vector unsigned int,
23479 vector unsigned int)
23480 vector signed long long vec_mulo (vector signed int,
23482 vector unsigned long long vec_mulo (vector unsigned int,
23483 vector unsigned int)
23484 * doc/extend.texi: Fix the documentation for the built-in
23487 2017-06-07 Carl Love <cel@us.ibm.com>
23490 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23493 2017-06-07 Carl Love <cel@us.ibm.com>
23495 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23496 support, Generate doublehv for signed int/float for BE case only.
23498 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23500 * doc/invoke.texi (mcx16): Rewrite.
23502 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23504 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23505 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23506 *mov<mode>_softfloat, and an anonymous splitter): Use
23507 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23509 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23511 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23513 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23514 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23515 (rs6000_debug_reg_global): Adjust.
23516 (rs6000_init_hard_regno_mode_ok): Adjust.
23517 (rs6000_dbx_register_number): Adjust.
23518 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23519 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23520 Remove SPE_ACC and SPEFSCR.
23521 (REG_ALLOC_ORDER): Ditto.
23522 (FRAME_POINTER_REGNUM): Change to 111.
23523 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23524 (REG_CLASS_NAMES): Ditto.
23525 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23526 (REGISTER_NAMES): Ditto.
23527 (ADDITIONAL_REG_NAMES): Ditto.
23528 (rs6000_reg_names): Ditto.
23529 * config/rs6000/rs6000.md: Renumber some register number
23532 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23534 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23536 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23537 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23539 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23540 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23541 Delete the SPE high registers.
23542 (REG_ALLOC_ORDER): Ditto.
23543 (enum reg_class): Remove SPE_HIGH_REGS.
23544 (REG_CLASS_NAMES): Ditto.
23545 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23546 (REGISTER_NAMES): Ditto.
23547 (rs6000_reg_names): Ditto.
23548 * doc/tm.texi.in: Remove SPE as example.
23549 * doc/tm.texi: Regenerate.
23551 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23553 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23554 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23555 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23556 * config/rs6000/rs6000.md (type): Remove "brinc".
23558 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23560 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23561 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23562 * config/rs6000/linuxspe.h: Delete file.
23563 * config/rs6000/rs6000.md: Don't include spe.md.
23564 * config/rs6000/spe.h: Delete file.
23565 * config/rs6000/spe.md: Delete file.
23566 * config/rs6000/t-rs6000: Remove spe.md.
23568 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23570 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23571 (reg_or_none500mem_operand): Delete.
23572 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23573 instead of reg_or_none500mem_operand.
23575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23578 handling of SPE flags.
23579 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23581 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23583 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23585 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23586 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23587 paired_divv2sf3): Similar.
23588 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23589 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23590 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23592 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23593 Rename the paired_* instruction patterns.
23594 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23596 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23597 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23598 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23599 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23600 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23601 PAIRED_VECTOR_MODE.
23602 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23603 (spe_func_has_64bit_regs_p): Delete.
23604 (spe_expand_predicate_builtin): Delete.
23605 (spe_expand_evsel_builtin): Delete.
23606 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23607 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23608 (invalid_e500_subreg): Delete.
23609 (rs6000_legitimize_address): Always force_reg op2 as well, for
23610 paired single memory accesses.
23611 (rs6000_member_type_forces_blk): Delete.
23612 (rs6000_spe_function_arg): Delete.
23613 (rs6000_expand_unop_builtin): Delete SPE handling.
23614 (rs6000_expand_binop_builtin): Ditto.
23615 (spe_expand_stv_builtin): Delete.
23616 (bdesc_2arg_spe): Delete.
23617 (spe_expand_builtin): Delete.
23618 (spe_expand_predicate_builtin): Delete.
23619 (spe_expand_evsel_builtin): Delete.
23620 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23621 (spe_init_builtins): Delete.
23622 (spe_func_has_64bit_regs_p): Delete.
23623 (savres_routine_name): Delete "info" parameter. Adjust callers.
23624 (rs6000_emit_stack_reset): Ditto.
23625 (rs6000_dwarf_register_span): Delete.
23626 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23627 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23628 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23630 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23632 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23633 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23634 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23635 mulv2sf3, divv2sf3): Delete expanders.
23637 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23639 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23641 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23643 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23644 * config/rs6000/rs6000.c: Ditto.
23646 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23648 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23649 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23650 comparison_operator.
23652 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23654 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23655 * config/rs6000/rs6000.opt: Ditto.
23656 * config/rs6000/t-rtems: Ditto.
23658 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23660 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23661 TARGET_E500_SINGLE by 0, simplify.
23662 * config/rs6000/rs6000.c: Ditto.
23663 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23664 (spe_build_register_parallel): Delete.
23665 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23666 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23667 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23668 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23669 (E500_CONVERT): Delete.
23670 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23672 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23674 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23675 * config/rs6000/dfp.md: Ditto.
23676 (negdd2, *negdd2_fpr): Merge.
23677 (absdd2, *absdd2_fpr): Merge.
23678 (negtd2, *negtd2_fpr): Merge.
23679 (abstd2, *abstd2_fpr): Merge.
23680 * config/rs6000/e500.h: Delete file.
23681 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23682 TARGET_FPRS by 1 and simplify.
23683 * config/rs6000/rs6000-c.c: Ditto.
23684 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23685 TARGET_DF_SPE by 0.
23686 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23688 * config/rs6000/rs6000.md: Ditto.
23689 (floatdidf2, *floatdidf2_fpr): Merge.
23690 (move_from_CR_gt_bit): Delete.
23691 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23692 (E500_CR_IOR_COMPARE): Delete.
23693 (All patterns that require !TARGET_FPRS): Delete.
23694 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23696 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23698 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23700 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23702 * graphds.c (add_edge): Intitialize edge's attached data.
23703 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23704 pointer parameter. Call pointed function on each edge during
23705 graph traversing. Skip traversing the edge when the function
23707 (graphds_dfs, graphds_scc): Ditto.
23708 (for_each_edge): New parameter. Pass the new parameter to callback
23710 * graphds.h (skip_edge_callback): New function pointer type.
23711 (graphds_dfs, graphds_scc): New function pointer parameter.
23712 (graphds_edge_callback, for_each_edge): New parameter.
23714 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23716 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23717 out code checking if runtime alias check is possible to below ...
23718 Call the new function.
23719 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23720 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23722 2017-06-07 Marek Polacek <polacek@redhat.com>
23725 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23726 TYPE_OVERFLOW_WRAPS check.
23728 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23730 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23731 if versioning is required.
23732 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23733 peeling with the check for versioning.
23735 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23737 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23738 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23739 Set true to new parameter if new ssa variable is defined.
23740 (vect_gen_vector_loop_niters): Refactor. Set range information
23741 for the new vector loop bound variable.
23742 (vect_do_peeling): Ditto.
23744 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23746 * tree-affine.c (ssa.h): Include header file.
23747 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23748 has wrapping overflow behavior.
23750 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23752 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23754 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23756 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23757 (tree_to_aff_combination): ... here.
23759 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23761 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23762 reg_pressure model function.
23763 (ivopts_global_cost_for_size): Delete.
23764 (determine_set_costs, iv_ca_recount_cost): Call new model function
23765 ivopts_estimate_reg_pressure.
23767 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23769 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23770 expensive than udiv. Remove floating point cases from mod.
23772 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23774 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23775 Increase idiv cost.
23777 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23779 * config/aarch64/aarch64.md
23780 (copysignsf3): Fix mask generation.
23782 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23784 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23786 (class dump_manager): Add register_dumps method.
23787 * dumpfile.c: Include langhooks.h.
23788 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23789 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23790 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23791 (dump_manager::dump_register): Start with 512 entries instead of 32.
23792 (dump_manager::register_dumps): New method.
23793 * toplev.c (general_init): Instead of invoking register_dumps
23794 langhook, invoke register_dumps method on the dump manager.
23795 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23798 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23800 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23801 Say that # requires an associated define_split to exist, and that
23802 the define_split must be suitable for use after register allocation.
23804 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23806 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23807 (compute_outgoing_frequencies): Also initialize zero counts.
23808 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23809 around loops; preserve more of profile when nothing changes.
23811 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23813 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23815 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23816 * config/arm/arm-cpu-cdata.h: Regenerate.
23817 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23818 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23819 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23820 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23821 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23823 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23825 * config/arm/t-rmprofile: Likewise.
23826 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23828 2017-06-06 David S. Miller <davem@davemloft.net>
23831 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23832 blockage if function uses alloca.
23834 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23836 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23837 New "uid" fields to hold pretty-print IDs of group and ref.
23838 Memory references are now identified as <group_id>:<ref_id>
23839 instead of using [random] addresses.
23840 (dump_mem_details): Simplify, no functional change.
23841 (dump_mem_ref): Simplify and make output more concise.
23842 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23843 (find_or_create_group): Initialize group uid.
23844 (record_ref): Initialize ref uid. Improve debug output.
23845 (prune_group_by_reuse, should_issue_prefetch_p,)
23846 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23847 (mark_nontemporal_store, determine_loop_nest_reuse):
23848 Improve debug output.
23850 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23852 * dbgcnt.def (prefetch): New debug counter.
23853 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23854 (schedule_prefetches): Stop issueing prefetches if debug counter
23857 2017-06-06 Tom de Vries <tom@codesourcery.com>
23859 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23860 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23862 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23864 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23865 Use aarch64_reg_or_zero predicate for operand 4.
23866 (aarch64_compare_and_swap<mode> define_insn_and_split):
23867 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23868 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23870 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23872 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23873 (arm_compute_save_core_reg_mask): This.
23874 (thumb1_compute_save_reg_mask): Rename into ...
23875 (thumb1_compute_save_core_reg_mask): This.
23876 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23877 (arm_compute_frame_layout): Likewise.
23879 2017-06-06 Richard Biener <rguenther@suse.de>
23881 PR tree-optimization/80974
23882 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23883 keep or clear leaders SSA info.
23885 2017-06-06 Tom de Vries <tom@codesourcery.com>
23887 * config/nvptx/nvptx.c (split_mode_p): New function.
23888 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23890 2017-06-06 Tom de Vries <tom@codesourcery.com>
23892 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23894 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23897 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23900 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23902 * shrink-wrap.c (handle_simple_exit): Update profile.
23903 (try_shrink_wrapping): Upate profile.
23905 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23907 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23908 (tree_guess_outgoing_edge_probabilities): New.
23909 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23910 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23912 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23914 * ipa-split.c (split_function): Initialize return bb profile.
23916 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23918 * profile.c (compute_branch_probabilities): Also initialize
23919 EXIT_BLOCK profile.
23921 2017-06-06 Richard Biener <rguenther@suse.de>
23923 PR tree-optimization/80928
23924 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23925 (vect_analyze_loop_operations): Properly guard analysis for
23927 (vect_transform_loop): Likewise.
23928 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23929 (vect_model_induction_cost): Do not cost for pure SLP.
23930 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23931 of induction in inner loop vectorization.
23932 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23933 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23934 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23936 (vect_analyze_slp_cost_1): Cost induction.
23937 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23938 (vect_get_slp_vect_defs): Likewise.
23939 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23940 (vect_transform_stmt): Handle SLP reductions.
23941 * tree-vectorizer.h (vectorizable_induction): Adjust.
23943 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23945 * config/rs6000/rs6000.c (make_resolver_func): Update
23946 init_lowered_empty_function call.
23948 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23950 * doc/invoke.texi: Document the -fprofile-abs-path option.
23951 * common.opt (fprofile-abs-path): New option.
23952 * gcov-io.h (gcov_write_filename): Declare.
23953 * gcov-io.c (gcov_write_filename): New function.
23954 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23955 * profile.c (output_location): Likewise.
23957 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23959 * shring-wrap.c: Revert accidental commit.
23961 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23963 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23965 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23967 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23969 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23970 profile in callgraph edge.
23971 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23972 (apply_scale): Likewise.
23973 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23974 Also scale profile when inlining function with zero profile.
23975 (initialize_cfun): Update exit block profile even when it is zero.
23976 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23977 when profile is read.
23979 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23981 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23982 (CLONE_*): New constants to define the processors we can generate
23983 code for with the target_clone attribute.
23984 (rs6000_clone_map): New array to identify which clone processors
23985 the current program is running on.
23986 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23987 target_clone attribute.
23988 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23989 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23990 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23991 (cpu_expand_builtin): Add support for target_clone attribute.
23992 (rs6000_valid_attribute_p): Allow "default" attribute.
23993 (get_decl_name): New debug function to simplify printing the
23994 current function name in debugging statements.
23995 (rs6000_clone_priority): New functions to support the target_clone
23996 attribute, and be able to generate code to switch between ISA 2.05
23997 through ISA 3.0 (power6 through power9).
23998 (rs6000_compare_version_priority): Likewise.
23999 (rs6000_get_function_versions_dispatcher): Likewise.
24000 (make_resolver_func): Likewise.
24001 (add_condition_to_bb): Likewise.
24002 (dispatch_function_versions): Likewise.
24003 (rs6000_generate_version_dispatcher_body): Likewise.
24004 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24005 (fusion_gpr_load_p): Fix a spacing issue.
24006 * doc/extend.texi (Common Function Attributes): Document that the
24007 PowerPC supports the target_clone attribute.
24009 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24011 * config/arm/arm.h: explain F symbol found in description of ARM
24012 register allocation in its legend.
24014 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24016 * config/mips/frame-header-opt.c: Include profile-count.h.
24017 * config/riscv/riscv.c: Include profile-count.h
24019 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24021 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24023 (sm_set_flag_if_changed): Add bbs field.
24024 (execute_sm_if_changed_flag_set): Pass BBS.
24025 (execute_sm): Update.
24027 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24029 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24032 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24034 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24035 (peephole2): New peephole2 to emit the above.
24036 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24038 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24040 * config/aarch64/aarch64.c (define_peephole2 above
24041 *sub_<shift>_<mode>): New peephole.
24043 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24045 * config/i386/i386.c (make_resolver_func): Update.
24046 * Makefile.in: Add profile-count.h and profile-count.o
24047 * auto-profile.c (afdo_indirect_call): Update to new API.
24048 (afdo_set_bb_count): Update.
24049 (afdo_propagate_edge): Update.
24050 (afdo_propagate_circuit): Update.
24051 (afdo_calculate_branch_prob): Update.
24052 (afdo_annotate_cfg): Update.
24053 * basic-block.h: Include profile-count.h
24054 (struct edge_def): Turn count to profile_count.
24055 (struct basic_block_def): Likewie.
24056 (REG_BR_PROB_BASE): Move to profile-count.h
24057 (RDIV): Move to profile-count.h
24058 * bb-reorder.c (max_entry_count): Turn to profile_count.
24059 (find_traces): Update.
24060 (rotate_loop):Update.
24061 (connect_traces):Update.
24062 (sanitize_hot_paths):Update.
24063 * bt-load.c (migrate_btr_defs): Update.
24064 * cfg.c (RDIV): Remove.
24065 (init_flow): Use alloc_block.
24066 (alloc_block): Uninitialize count.
24067 (unchecked_make_edge): Uninitialize count.
24068 (check_bb_profile): Update.
24069 (dump_edge_info): Update.
24070 (dump_bb_info): Update.
24071 (update_bb_profile_for_threading): Update.
24072 (scale_bbs_frequencies_int): Update.
24073 (scale_bbs_frequencies_gcov_type): Update.
24074 (scale_bbs_frequencies_profile_count): New.
24075 * cfg.h (update_bb_profile_for_threading): Update.
24076 (scale_bbs_frequencies_profile_count): Declare.
24077 * cfgbuild.c (compute_outgoing_frequencies): Update.
24078 (find_many_sub_basic_blocks): Update.
24079 * cfgcleanup.c (try_forward_edges): Update.
24080 (try_crossjump_to_edge): Update.
24081 * cfgexpand.c (expand_gimple_tailcall): Update.
24082 (construct_exit_block): Update.
24083 * cfghooks.c (verify_flow_info): Update.
24084 (dump_bb_for_graph): Update.
24085 (split_edge): Update.
24086 (make_forwarder_block): Update.
24087 (duplicate_block): Update.
24088 (account_profile_record): Update.
24089 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24090 (get_estimated_loop_iterations): Update.
24091 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24092 (single_likely_exit): Update.
24093 * cfgloopmanip.c (scale_loop_profile): Update.
24095 (set_zero_probability): Update.
24096 (lv_adjust_loop_entry_edge): Update.
24097 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24098 (purge_dead_edges): Update.
24099 (rtl_account_profile_record): Update.
24100 * cgraph.c (cgraph_node::create): Uninitialize count.
24101 (symbol_table::create_edge): Uninitialize count.
24102 (cgraph_update_edges_for_call_stmt_node): Update.
24103 (cgraph_edge::dump_edge_flags): Update.
24104 (cgraph_node::dump): Update.
24105 (cgraph_edge::maybe_hot_p): Update.
24106 * cgraph.h: Include profile-count.h
24107 (create_clone), create_edge, create_indirect_edge): Update.
24108 (cgraph_node): Turn count to profile_count.
24109 (cgraph_edge0: Likewise.
24110 (make_speculative, clone): Update.
24111 (create_edge): Update.
24112 (init_lowered_empty_function): Update.
24113 * cgraphclones.c (cgraph_edge::clone): Update.
24114 (duplicate_thunk_for_node): Update.
24115 (cgraph_node::create_clone): Update.
24116 * cgraphunit.c (cgraph_node::analyze): Update.
24117 (cgraph_node::expand_thunk): Update.
24118 * final.c (dump_basic_block_info): Update.
24119 * gimple-streamer-in.c (input_bb): Update.
24120 * gimple-streamer-out.c (output_bb): Update.
24121 * graphite.c (print_global_statistics): Update.
24122 (print_graphite_scop_statistics): Update.
24123 * hsa-brig.c: Include basic-block.h.
24124 * hsa-dump.c: Include basic-block.h.
24125 * hsa-gen.c (T sum_slice): Update.
24126 (convert_switch_statements):Update.
24127 * hsa-regalloc.c: Include basic-block.h.
24128 * ipa-chkp.c (chkp_produce_thunks): Update.
24129 * ipa-cp.c (struct caller_statistics): Update.
24130 (init_caller_stats): Update.
24131 (gather_caller_stats): Update.
24132 (ipcp_cloning_candidate_p): Update.
24133 (good_cloning_opportunity_p): Update.
24134 (get_info_about_necessary_edges): Update.
24135 (dump_profile_updates): Update.
24136 (update_profiling_info): Update.
24137 (update_specialized_profile): Update.
24138 (perhaps_add_new_callers): Update.
24139 (decide_about_value): Update.
24140 (ipa_cp_c_finalize): Update.
24141 * ipa-devirt.c (struct odr_type_warn_count): Update.
24142 (struct decl_warn_count): Update.
24143 (struct final_warning_record): Update.
24144 (possible_polymorphic_call_targets): Update.
24145 (ipa_devirt): Update.
24146 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24147 * ipa-icf.c (sem_function::merge): Update.
24148 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24149 * ipa-inline.c (compute_uninlined_call_time): Update.
24150 (compute_inlined_call_time): Update.
24151 (want_inline_small_function_p): Update.
24152 (want_inline_self_recursive_call_p): Update.
24153 (edge_badness): Update.
24154 (lookup_recursive_calls): Update.
24155 (recursive_inlining): Update.
24156 (inline_small_functions): Update.
24157 (dump_overall_stats): Update.
24158 (dump_inline_stats): Update.
24159 * ipa-profile.c (ipa_profile_generate_summary): Update.
24160 (ipa_propagate_frequency): Update.
24161 (ipa_profile): Update.
24162 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24163 * ipa-utils.c (ipa_merge_profiles): Update.
24164 * loop-doloop.c (doloop_modify): Update.
24165 * loop-unroll.c (report_unroll): Update.
24166 (unroll_loop_runtime_iterations): Update.
24167 * lto-cgraph.c (lto_output_edge): Update.
24168 (lto_output_node): Update.
24169 (input_node): Update.
24170 (input_edge): Update.
24171 (merge_profile_summaries): Update.
24172 * lto-streamer-in.c (input_cfg): Update.
24173 * lto-streamer-out.c (output_cfg): Update.
24174 * mcf.c (create_fixup_graph): Update.
24175 (adjust_cfg_counts): Update.
24176 (sum_edge_counts): Update.
24177 * modulo-sched.c (sms_schedule): Update.
24178 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24179 * predict.c (maybe_hot_count_p): Update.
24180 (probably_never_executed): Update.
24181 (dump_prediction): Update.
24182 (combine_predictions_for_bb): Update.
24183 (propagate_freq): Update.
24184 (handle_missing_profiles): Update.
24185 (counts_to_freqs): Update.
24186 (rebuild_frequencies): Update.
24187 (force_edge_cold): Update.
24188 * predict.h: Include profile-count.h
24189 (maybe_hot_count_p, counts_to_freqs): UPdate.
24190 * print-rtl-function.c: Do not include cfg.h
24191 * print-rtl.c: Include basic-block.h
24192 * profile-count.c: New file.
24193 * profile-count.h: New file.
24194 * profile.c (is_edge_inconsistent): Update.
24195 (correct_negative_edge_counts): Update.
24196 (is_inconsistent): Update.
24197 (set_bb_counts): Update.
24198 (read_profile_edge_counts): Update.
24199 (compute_frequency_overlap): Update.
24200 (compute_branch_probabilities): Update; Initialize and deinitialize
24202 (branch_prob): Update.
24203 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24204 (edge_gcov_count): New.
24205 (bb_gcov_count): New.
24206 * shrink-wrap.c (try_shrink_wrapping): Update.
24207 * tracer.c (better_p): Update.
24208 * trans-mem.c (expand_transaction): Update.
24209 (ipa_tm_insert_irr_call): Update.
24210 (ipa_tm_insert_gettmclone_call): Update.
24211 * tree-call-cdce.c: Update.
24212 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24213 (gimple_duplicate_sese_tail): Update.
24214 (gimple_account_profile_record): Update.
24215 (execute_fixup_cfg): Update.
24216 * tree-inline.c (copy_bb): Update.
24217 (copy_edges_for_bb): Update.
24218 (initialize_cfun): Update.
24219 (freqs_to_counts): Update.
24220 (copy_cfg_body): Update.
24221 (expand_call_inline): Update.
24222 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24223 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24224 (try_unroll_loop_completely): Update.
24225 (try_peel_loop): Update.
24226 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24227 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24228 * tree-ssa-loop-split.c (connect_loops): Update.
24229 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24230 * tree-ssa-reassoc.c (branch_fixup): Update.
24231 * tree-ssa-tail-merge.c (replace_block_by): Update.
24232 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24233 (compute_path_counts): Update.
24234 (update_profile): Update.
24235 (recompute_probabilities): Update.
24236 (update_joiner_offpath_counts): Update.
24237 (estimated_freqs_path): Update.
24238 (freqs_to_counts_path): Update.
24239 (clear_counts_path): Update.
24240 (ssa_fix_duplicate_block_edges): Update.
24241 (duplicate_thread_path): Update.
24242 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24243 (struct switch_conv_info): Update.
24244 * tree-tailcall.c (decrease_profile): Update.
24245 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24246 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24247 * value-prof.c (check_counter): Update.
24248 (gimple_divmod_fixed_value): Update.
24249 (gimple_mod_pow2): Update.
24250 (gimple_mod_subtract): Update.
24251 (gimple_ic_transform): Update.
24252 (gimple_stringop_fixed_value): Update.
24253 * value-prof.h (gimple_ic): Update.
24255 2017-06-02 Carl Love <cel@us.ibm.com>
24257 * config/rs6000/rs6000-c: Add support for built-in functions
24258 vector double vec_doublee (vector signed int);
24259 vector double vec_doublee (vector unsigned int);
24260 vector double vec_doublee (vector float);
24261 vector double vec_doubleh (vector signed int);
24262 vector double vec_doubleh (vector unsigned int);
24263 vector double vec_doubleh (vector float);
24264 vector double vec_doublel (vector signed int);
24265 vector double vec_doublel (vector unsigned int);
24266 vector double vec_doublel (vector float);
24267 vector double vec_doubleo (vector signed int);
24268 vector double vec_doubleo (vector unsigned int);
24269 vector double vec_doubleo (vector float);.
24270 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24271 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24273 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24274 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24275 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24277 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24278 vec_doublel, vec_doubleh.
24279 * doc/extend.texi: Update the built-in documentation file for the
24280 new built-in functions.
24282 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24285 * ipa-inline-analysis.c (free_growth_caches): Set
24286 edge_removal_hook_holder to NULL after removing it.
24288 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24290 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24291 comparision with zero.
24293 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24294 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24295 for early expansion of vec_min and vec_max builtins.
24296 (builtin_function_type): Add min/max unsigned variants to those
24297 identified as having unsigned arguments.
24299 2017-06-02 Olivier Hainque <hainque@adacore.com>
24301 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24303 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24305 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24306 Use VALL_F16 iterator rather than VALL.
24308 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24310 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24311 Emit CBNZ inside loop when doing a strong exchange and comparing
24312 against zero. Generate the CC flags after the loop.
24314 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24316 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24317 (dl_section_ref): New.
24318 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24319 On AIX, append an expression to subtract the size of the
24320 section length to dl_section_ref.
24322 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24324 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24325 for early expansion of vector absolute builtins.
24327 2017-06-02 Richard Biener <rguenther@suse.de>
24329 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24330 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24332 2017-06-02 Richard Biener <rguenther@suse.de>
24334 PR tree-optimization/80948
24335 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24336 stmt order as well.
24338 2017-06-02 Richard Biener <rguenther@suse.de>
24340 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24342 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24343 for inductions as relevant.
24345 2017-06-02 Richard Biener <rguenther@suse.de>
24347 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24348 (vectorizable_induction): ... this. Remove dead code.
24350 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24352 * builtins. (expand_builtin_alloca): Remove second parameter and
24353 infer its value from the first parameter instead.
24354 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24356 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24358 PR rtl-optimization/80903
24359 * loop-doloop.c (add_test): Unshare sequence.
24361 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24363 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24365 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24367 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24369 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24370 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24371 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24372 (xlogue_layout::get_stub_rtx): Make static.
24373 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24374 (xlogue_layout::compute_stub_managed_regs): Rename to...
24375 (xlogue_layout::count_stub_managed_regs): ...this.
24376 (xlogue_layout::is_stub_managed_reg): New function.
24377 (xlogue_layout::m_stub_names): Rename to...
24378 (xlogue_layout::s_stub_names): ...this, make static.
24379 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24380 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24381 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24382 xlogue_layout::s_stub_names): Instantiate statics.
24383 (stub_managed_regs): Remove.
24384 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24385 (disable_call_ms2sysv_xlogues): Rename to...
24386 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24387 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24389 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24390 change after reload_completed.
24391 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24393 (ix86_expand_prologue): Likewise.
24394 (ix86_expand_epilogue): Likewise.
24395 (ix86_expand_split_stack_prologue): Likewise.
24396 (ix86_compute_frame_layout): Remove frame parameter ...
24397 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24398 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24400 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24401 (ix86_frame): Move from here ...
24402 * config/i386/i386.h (ix86_frame): ... to here.
24403 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24404 complete ix86_frame data structure instead. Remove some_ld_name.
24406 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24408 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24409 symbols that hold a DECL_VALUE_EXPR.
24411 2017-06-01 Martin Jambor <mjambor@suse.cz>
24413 PR tree-optimization/80898
24414 * tree-sra.c (process_subtree_disqualification): Removed.
24415 (disqualify_candidate): Do not acll
24416 process_subtree_disqualification.
24417 (subtree_mark_written_and_enqueue): New function.
24418 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24419 RHS has been disqualified and re-queue LHS if necessary. Apart
24420 from that, ignore disqualified RHS.
24422 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24424 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24425 address fetch for z10 or later.
24427 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24429 * config/arc/arc.md (tst_movb): Add guard when splitting.
24431 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24433 * config/arc/arc.c (arc_can_eliminate): Test against
24434 arc_frame_pointer_needed.
24436 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24438 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24439 to prevent store reordering.
24440 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24441 (type): Add block type.
24442 (stack_tie): Define special instruction to be used in
24445 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24447 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24448 constraint. It is not valid for the pattern.
24449 (noncommutative_binary_comparison): Likewise.
24451 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24453 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24456 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24458 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24459 be used by the reg-alloc.
24461 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24463 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24464 reg-alloc when having mul64 or mul32x16 instructions.
24465 (mulsidi3): Likewise.
24466 (umulsidi3): Likewise.
24467 (mulsi32x16): New pattern.
24468 (mulsi64): Likewise.
24469 (mulsidi64): Likewise.
24470 (umulsidi64): Likewise.
24471 (MUL32x16_REG): Define.
24472 (mul64_600): Use MUL32x16_REG.
24473 (mac64_600): Likewise.
24474 (umul64_600): Likewise.
24475 (umac64_600): Likewise.
24477 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24479 * config/arc/arc.md (mulsi3_700): Make it commutative.
24481 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24483 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24485 (*sign_extendsidi2_insn): Likewise for movstosw.
24487 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24489 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24490 the type of the input discriminant value. Convert the
24491 discriminant value of signedness vary.
24493 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24495 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24496 Add to -Wall section.
24498 2017-06-01 Richard Biener <rguenther@suse.de>
24500 PR middle-end/66313
24501 * fold-const.c (fold_plusminus_mult_expr): If the factored
24502 factor may be zero use a wrapping type for the inner operation.
24503 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24504 and handle moved defs.
24505 (process_assignment): Properly guard the unary op case. Return a
24506 tri-state indicating that moving the stmt before the call may allow
24507 to continue. Pass through to_move.
24508 (find_tail_calls): Handle moving unrelated defs before
24511 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24514 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24515 splitter result in the canonical way.
24517 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24519 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24520 also for 32bit target. Update insn attributes.
24521 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24523 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24525 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24526 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24527 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24528 (_mm_maskz_min_ss): New intrinsics.
24530 2017-05-31 Martin Liska <mliska@suse.cz>
24532 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24533 Change comment style to one we normally use.
24534 (vectorizable_reduction): Likewise.
24535 (vectorizable_induction): Likewise.
24536 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24537 (vectorizable_call): Likewise.
24538 (vectorizable_simd_clone_call): Likewise.
24539 (vectorizable_conversion): Likewise.
24540 (vectorizable_assignment): Likewise.
24541 (vectorizable_shift): Likewise.
24542 (vectorizable_operation): Likewise.
24543 (vectorizable_store): Likewise.
24544 (vectorizable_load): Likewise.
24545 * tree-vectorizer.h: Likewise.
24547 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24549 * passes.c (emergency_dump_function): New.
24550 * tree-pass.h (emergency_dump_function): Declare.
24551 * plugin.c (plugins_internal_error_function): Remove.
24552 * plugin.h (plugins_internal_error_function): Remove declaration.
24553 * toplev.c (internal_error_function): New static function. Use it...
24554 (general_init): ...here.
24556 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24558 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24559 (arc_rtx_costs): Add costs for new patterns.
24560 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24561 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24563 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24565 * tree-ssa-strlen.c (get_next_strinfo): New function.
24566 (get_stridx_plus_constant): Use it.
24567 (zero_length_string): Likewise.
24568 (adjust_related_strinfos): Likewise.
24569 (adjust_last_stmt): Likewise.
24571 2017-05-31 Richard Biener <rguenther@suse.de>
24574 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24575 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24577 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24579 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24580 loop_vinfo argument and use of dependence distance vectors.
24581 Check instead whether the two references differ only in their
24582 initial value and assume that they have the same alignment if the
24583 difference is a multiple of the vector alignment.
24584 (vect_analyze_data_refs_alignment): Update call accordingly.
24586 2017-05-31 Martin Liska <mliska@suse.cz>
24589 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24591 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24593 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24594 (create_intersect_range_checks): Move from ...
24595 * tree-data-ref.c (create_intersect_range_checks_index)
24596 (create_intersect_range_checks): ... to here.
24597 (create_runtime_alias_checks): New function factored from ...
24598 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24599 here. Call above function.
24600 * tree-data-ref.h (create_runtime_alias_checks): New function.
24602 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24604 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24605 segment length for dr_b and compute it in wide_int.
24607 2017-05-31 Richard Biener <rguenther@suse.de>
24609 PR tree-optimization/80906
24610 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24611 and pass through iv_map.
24612 (copy_bb_and_scalar_dependences): Adjust.
24613 (translate_pending_phi_nodes): Likewise.
24614 (copy_loop_close_phi_args): Handle code-generating IVs instead
24617 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24619 * diagnostic-color.c (color_dict): Add "type-diff".
24620 (parse_gcc_colors): Update comment.
24621 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24622 -fdiagnostics-show-template-tree and -fno-elide-type.
24623 (GCC_COLORS): Add type-diff to example.
24625 (-fdiagnostics-show-template-tree): New.
24626 (-fno-elide-type): New.
24627 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24628 the pp_format_decoder callback. Call any m_format_postprocessor's
24630 (pretty_printer::pretty_printer): Initialize
24631 m_format_postprocessor.
24632 (pretty_printer::~pretty_printer): Delete any
24633 m_format_postprocessor.
24634 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24635 (class format_postprocessor): New class.
24636 (struct pretty_printer::format_decoder): Document the new parameters.
24637 (struct pretty_printer::m_format_postprocessor): New field.
24638 * tree-diagnostic.c (default_tree_printer): Update for new
24639 bool and const char ** params.
24640 * tree-diagnostic.h (default_tree_printer): Likewise.
24642 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24644 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24645 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24647 rs6000_gen_cell_microcode code.
24648 (rs6000_final_prescan_insn): Delete.
24649 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24650 "warn-cell-microcode" entries.
24651 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24652 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24653 throughout. Change cc_reg_not_micro_cr0_operand to
24654 cc_reg_not_cr0_operand throughout.
24655 (*extendhi<mode>2_noload): Delete.
24656 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24657 (mwarn-cell-microcode): Delete.
24658 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24659 -mgen-cell-microcode and -mwarn-cell-microcode.
24661 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24664 * config/i386/constraints.md (Yd): New constraint.
24666 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24667 and (?Yd, r) alternatives. Update insn attributes.
24668 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24669 and (?*Yd, r) alternatives. Update insn attributes.
24670 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24672 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24674 * gimplify.c (gimplify_modify_expr): Don't create a
24675 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24678 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24680 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24682 2017-05-30 Richard Biener <rguenther@suse.de>
24684 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24685 and reduc_def fields.
24686 (STMT_VINFO_REDUC_TYPE): New define.
24687 (STMT_VINFO_REDUC_DEF): Likewise.
24688 (vect_force_simple_reduction): Adjust prototype.
24689 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24690 (vect_is_simple_reduction): Remove check_reduction argument.
24691 (vect_force_simple_reduction): Adjust and set
24692 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24693 (vectorizable_reduction): Do not re-do reduction analysis
24694 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24695 * tree-parloops.c (gather_scalar_reductions): Adjust.
24697 2017-05-30 Richard Biener <rguenther@suse.de>
24699 PR middle-end/80901
24700 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24703 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24705 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24706 Introduce unknown_misalignment parameter and remove vf.
24707 (vect_peeling_hash_get_lowest_cost):
24708 Pass unknown_misalignment parameter.
24709 (vect_enhance_data_refs_alignment):
24710 Fix unsupportable data ref treatment.
24712 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24714 * tree-vect-data-refs.c (vect_get_data_access_cost):
24715 Workaround for SLP handling.
24716 (vect_enhance_data_refs_alignment):
24717 Compute costs for doing no peeling at all, compare to the best
24718 peeling costs so far and avoid peeling if cheaper.
24720 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24722 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24723 Return peeling info and set costs to zero for unlimited cost
24725 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24726 with unknown misalignment. Compute and costs for unknown
24727 misalignment, compare them to the costs for known misalignment
24728 and choose the cheapest for peeling.
24730 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24732 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24733 (vect_get_peeling_costs_all_drs): Create function.
24734 (vect_peeling_hash_get_lowest_cost):
24735 Use vect_get_peeling_costs_all_drs.
24736 (vect_peeling_supportable): Create function.
24737 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24739 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24741 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24742 DR_HAS_NEGATIVE_STEP.
24743 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24744 (vect_enhance_data_refs_alignment): Use.
24745 (vect_duplicate_ssa_name_ptr_info): Use.
24746 * tree-vectorizer.h (dr_misalignment): Use.
24747 (known_alignment_for_access_p): Use.
24749 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24752 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24754 (has_section_name): New function.
24756 2017-05-30 Martin Liska <mliska@suse.cz>
24759 * auto-profile.c (get_function_decl_from_block): Fix
24762 2017-05-30 Richard Biener <rguenther@suse.de>
24764 PR middle-end/80876
24765 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24767 2017-05-30 Martin Liska <mliska@suse.cz>
24769 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24770 * dumpfile.h (struct dump_file_info): Remove ctors.
24772 2017-05-30 Martin Liska <mliska@suse.cz>
24774 * predict.def: Fix GNU coding style.
24776 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24778 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24779 Mark 'to' argument with ATTRIBUTE_UNUSED.
24781 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24783 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24784 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24785 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24788 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24790 * doc/install.texi (Options specification): Restore entry of
24791 --enable-sjlj-exceptions.
24793 2017-05-27 Michael Eager <eager@eagercon.com>
24796 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24798 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24800 * config/microblaze/microblaze.h
24801 (FIXED_REGISTERS): Update in macro.
24802 (CALL_USED_REGISTERS): Update in macro.
24804 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24806 * doc/install.texi: Add links to macOS binary distributions.
24808 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24812 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24814 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24816 2017-05-26 Martin Liska <mliska@suse.cz>
24818 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24820 2017-05-26 Martin Liska <mliska@suse.cz>
24822 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24823 always leading ';; '.
24824 (dump_bb_info): Likewise.
24825 (brief_dump_cfg): Likewise.
24826 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24827 * dumpfile.c: Remove usage of TDF_VERBOSE.
24828 * dumpfile.h (enum dump_kind): Likewise.
24829 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24830 * print-tree.c (debug_verbose): Remove.
24831 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24832 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24833 * tree-diagnostic.c (default_tree_printer): Replace
24834 TDF_DIAGNOSTIC with TDF_SLIM.
24836 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24838 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24839 in parameter loop, rather than loop_vinfo.
24840 (create_intersect_range_checks): Ditto.
24841 (vect_create_cond_for_alias_checks): Update call to above functions.
24843 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24845 PR tree-optimization/80815
24846 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24847 for merging runtime alias checks. Handle negative DR_STEPs.
24849 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24851 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24853 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24854 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24855 out code pruning runtime alias checks.
24856 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24857 factored out from above.
24858 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24860 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24862 (prune_runtime_alias_test_list): New decalaration.
24864 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24866 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24867 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24868 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24869 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24870 (operator==, comp_dr_with_seg_len_pair): Ditto.
24871 (vect_prune_runtime_alias_test_list): Ditto.
24873 2017-05-26 Martin Liska <mliska@suse.cz>
24876 * params.def: Bound partial-inlining-entry-probability param.
24878 2017-05-26 Marek Polacek <polacek@redhat.com>
24881 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24884 2017-05-26 Richard Biener <rguenther@suse.de>
24886 PR tree-optimization/80842
24887 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24890 2017-05-26 Richard Biener <rguenther@suse.de>
24892 PR tree-optimization/80844
24893 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24895 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24897 * doc/md.texi (Machine Constraints): Update x86 family
24898 machine constraints section to match 'config/i386/constraints.md'.
24900 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24902 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24904 2017-05-25 Nathan Sidwell <nathan@acm.org>
24906 * doc/invoke.texi (--enable-languages): Update documentation.
24908 2017-05-25 Martin Liska <mliska@suse.cz>
24910 * dumpfile.c: Add TDF_FOLDING.
24911 * dumpfile.h (enum dump_kind): Likewise.
24912 * genmatch.c (dt_simplify::gen_1): Use it.
24914 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24916 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24918 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24920 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24921 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24923 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24925 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24926 * match.pd (X == C): Rewrite it here.
24927 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24928 with_certain_nonzero_bits2): New predicates.
24929 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24931 2017-05-24 Nathan Sidwell <nathan@acm.org>
24933 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24936 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24939 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24941 * config/powerpcspe: New port. Files are copied from the rs6000
24942 port, with "rs6000" in filenames replaced by "powerpcspe".
24944 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24946 PR rtl-optimization/80754
24947 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24949 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24951 * config/sparc/sparc.md (length): Return the correct value for -mflat
24952 sibcalls to match output_sibcall.
24954 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24958 * config/rs6000/rs6000.c (struct machine_function): Add new field
24960 (rs6000_get_separate_components): Init that field, use it.
24961 (rs6000_components_for_bb): Use the field.
24963 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24965 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24967 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24969 PR middle-end/80823
24970 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24972 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24975 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24976 address against address_operand predicate.
24977 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24979 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24981 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24982 parameters passed indirectly.
24984 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24986 * config/i386/i386.md (*movdi_internal): Remove SSE4
24987 alternative 18 (?r, *v). Update insn attributes.
24988 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24989 Update insn attributes.
24990 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24991 Update insn attributes.
24992 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24993 alternative 1 (r, v). Remove isa attribute.
24994 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24995 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24996 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24998 2017-05-23 Tom de Vries <tom@codesourcery.com>
25000 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25003 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25005 * cgraphunit.c (symbol_table::process_new_functions): Update.
25006 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25007 (inline_generate_summary): Rename to ...
25008 (ipa_fn_summary_generate): ... this one.
25009 (inline_read_summary): Rename to ...
25010 (ipa_fn_summary_read): ... this one.
25011 (inline_write_summary): Rename to ...
25012 (ipa_fn_summary_write): ... this one.
25013 (inline_free_summary): Rename to ...
25014 (ipa_free_fn_summary): ... this one.
25015 (pass_data_local_fn_summary, pass_local_fn_summary,
25016 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25017 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25018 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25019 make_pass_ipa_fn_summary): New.
25020 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25021 inline_write_summary, inline_free_summary): Remove.
25022 (ipa_free_fn_summary) : New.
25023 * ipa-inline.c (ipa_inline): Update.
25024 (pass_ipa_inline): Do not generate summaries.
25025 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25027 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25028 and add pass_ipa_fn_summary.
25029 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25031 (make_pass_inline_parameters): Remove.
25033 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25035 * omp-low.c (struct omp_context): Remove "default_kind" member.
25038 * omp-offload.c (execute_oacc_device_lower): Remove the
25039 parallelism dimensions function attributes for unparallelized
25040 OpenACC kernels constructs.
25042 2017-05-23 Martin Liska <mliska@suse.cz>
25044 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25046 (cgraph_edge::make_speculative): Likewise.
25047 (cgraph_edge::resolve_speculation): Likewise.
25048 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25049 (cgraph_node::dump): Likewise.
25050 * cgraph.h: Likewise.
25051 * cgraphunit.c (analyze_functions): Likewise.
25052 (symbol_table::compile): Likewise.
25053 * ipa-cp.c (print_all_lattices): Likewise.
25054 (determine_versionability): Likewise.
25055 (initialize_node_lattices): Likewise.
25056 (ipcp_verify_propagated_values): Likewise.
25057 (estimate_local_effects): Likewise.
25058 (update_profiling_info): Likewise.
25059 (create_specialized_node): Likewise.
25060 (perhaps_add_new_callers): Likewise.
25061 (decide_about_value): Likewise.
25062 (decide_whether_version_node): Likewise.
25063 (identify_dead_nodes): Likewise.
25064 (ipcp_store_bits_results): Likewise.
25065 * ipa-devirt.c (dump_targets): Likewise.
25066 (ipa_devirt): Likewise.
25067 * ipa-icf.c (sem_item::dump): Likewise.
25068 (sem_function::equals): Likewise.
25069 (sem_variable::equals): Likewise.
25070 (sem_item_optimizer::read_section): Likewise.
25071 (sem_item_optimizer::execute): Likewise.
25072 (congruence_class::dump): Likewise.
25073 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25074 (dump_inline_summary): Likewise.
25075 (estimate_node_size_and_time): Likewise.
25076 (inline_analyze_function): Likewise.
25077 * ipa-inline-transform.c (inline_call): Likewise.
25078 * ipa-inline.c (report_inline_failed_reason): Likewise.
25079 (want_early_inline_function_p): Likewise.
25080 (edge_badness): Likewise.
25081 (update_edge_key): Likewise.
25082 (inline_small_functions): Likewise.
25083 * ipa-profile.c (ipa_profile): Likewise.
25084 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25085 (ipa_make_edge_direct_to_target): Likewise.
25086 (remove_described_reference): Likewise.
25087 (ipa_impossible_devirt_target): Likewise.
25088 (propagate_controlled_uses): Likewise.
25089 (ipa_print_node_params): Likewise.
25090 (ipcp_transform_function): Likewise.
25091 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25092 (propagate_pure_const): Likewise.
25093 * ipa-reference.c (generate_summary): Likewise.
25094 (read_write_all_from_decl): Likewise.
25095 (propagate): Likewise.
25096 (ipa_reference_read_optimization_summary): Likewise.
25097 * ipa-utils.c (ipa_merge_profiles): Likewise.
25098 * ipa.c (walk_polymorphic_call_targets): Likewise.
25099 (symbol_table::remove_unreachable_nodes): Likewise.
25100 (ipa_single_use): Likewise.
25101 * passes.c (execute_todo): Likewise.
25102 * predict.c (drop_profile): Likewise.
25103 * symtab.c (symtab_node::get_dump_name): New function.
25104 (symtab_node::dump_name): Likewise.
25105 (symtab_node::dump_asm_name): Likewise.
25106 (symtab_node::dump_references): Likewise.
25107 (symtab_node::dump_referring): Likewise.
25108 (symtab_node::dump_base): Likewise.
25109 (symtab_node::debug_symtab): Likewise.
25110 * tree-sra.c (convert_callers_for_node): Likewise.
25111 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25112 * value-prof.c (init_node_map): Likewise.
25114 2017-05-23 Martin Liska <mliska@suse.cz>
25116 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25117 and symtab_node::debug_symtab to symbol_table::debug.
25118 * cgraphunit.c (analyze_functions): Use the renamed function.
25119 (symbol_table::compile): Likewise.
25120 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25121 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25122 * passes.c (execute_todo): Likewise.
25123 * symtab.c (symbol_table::dump): New function.
25124 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25126 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25128 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25129 that nonconst implies exec.
25131 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25133 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25134 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25135 (inline_edge_summary_vec): Turn into ...
25136 (ipa_call_summaries): ... this one.
25137 (redirect_to_unreachable, edge_set_predicate,
25138 evaluate_properties_for_edge, inline_summary_alloc,
25139 reset_ipa_call_summary, reset_inline_summary,
25140 inline_summary_t::duplicate): Update.
25141 (inline_edge_duplication_hook): Turn to ...
25142 (ipa_call_summary_t::duplicate): ... this one.
25143 (inline_edge_removal_hook): Turn to ...
25144 (ipa_call_summary_t::remove): ... this one.
25145 (dump_inline_edge_summary): Turn to ...
25146 (dump_ipa_call_summary): ... this one.
25147 (estimate_function_body_sizes): Update.
25148 (inline_update_callee_summaries): Update.
25149 (remap_edge_change_prob): Update.
25150 (remap_edge_summaries): Update.
25151 (inline_merge_summary): Update.
25152 (do_estimate_edge_time): Update.
25153 (inline_generate_summary): Update.
25154 (inline_read_section): Update.
25155 (inline_read_summary): Update.
25156 (inline_free_summary): Update.
25157 * ipa-inline.c (can_inline_edge_p): Update.
25158 (compute_inlined_call_time): Update.
25159 (want_inline_small_function_p): Update.
25160 (edge_badness): Update.
25161 (early_inliner): Update.
25162 * ipa-inline.h (inline_edge_summary): Turn to ...
25163 (ipa_call_summary): ... this one.
25164 (ipa_call_summary_t): New class.
25165 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25166 (ipa_call_summaries): New.
25167 (inline_edge_summary): Remove.
25168 (estimate_edge_growth): Update.
25169 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25170 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25171 * ipa-split.c (execute_split_functions): Update.
25172 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25174 2017-05-23 Tom de Vries <tom@codesourcery.com>
25176 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25177 attributes): Document rdrand effective target.
25179 2017-05-23 Tom de Vries <tom@codesourcery.com>
25181 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25182 attributes): Sort alphabetically.
25184 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25186 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25188 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25191 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25192 V2DF/V2DI splat into two separate patterns, one that handles
25193 registers, and the other that only handles memory. Drop support
25194 for splatting from a GPR on ISA 2.07 and then splitting the
25195 splat into direct move and splat.
25196 (vsx_splat_<mode>_reg): Likewise.
25197 (vsx_splat_<mode>_mem): Likewise.
25199 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25201 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25203 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25205 PR middle-end/80809
25206 * omp-low.c (finish_taskreg_remap): New function.
25207 (finish_taskreg_scan): If unit size of ctx->record_type
25208 is non-constant, unshare the size expression and replace
25209 decls in it with possible outer var refs.
25211 PR middle-end/80809
25212 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25213 GOVD_SHARED rather than GOVD_PRIVATE with it.
25214 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25215 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25217 PR middle-end/80853
25218 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25219 as last argument to build_outer_var_ref for pointer bases of array
25220 section reductions.
25222 2017-05-19 Martin Sebor <msebor@redhat.com>
25224 * print-tree.c (print_node): Print DECL_READ_P flag.
25226 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25228 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25229 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25230 * cgraph.c: Likewise.
25231 * cgraphunit.c: Likewise.
25232 * gengtype.c: Likewise.
25233 * ipa-cp.c: Likewise.
25234 * ipa-devirt.c: Likewise.
25235 * ipa-icf.c: Likewise.
25236 * ipa-predicate.c: Likewise.
25237 * ipa-profile.c: Likewise.
25238 * ipa-prop.c: Likewise.
25239 * ipa-split.c: Likewise.
25241 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25242 edge_predicate_pool, dump_inline_hints,
25243 inline_summary::account_size_time, redirect_to_unreachable,
25244 edge_set_predicate, set_hint_predicate,
25245 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25246 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25247 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25248 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25249 ipa_call_summary_t::remove, initialize_growth_caches,
25250 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25251 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25252 mark_modified, unmodified_parm_1, unmodified_parm,
25253 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25254 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25255 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25256 will_be_nonconstant_predicate, record_modified_bb_info,
25257 get_minimal_bb, record_modified, param_change_prob,
25258 phi_result_unknown_predicate, predicate_for_phi_result,
25259 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25260 estimate_function_body_sizes, compute_inline_parameters,
25261 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25262 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25263 inline_update_callee_summaries, remap_edge_change_prob,
25264 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25265 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25266 inline_analyze_function, inline_summary_t::insert,
25267 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25268 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25269 inline_free_summary): Move to ipa-fnsummary.h
25270 (predicate_t): Remove.
25271 * ipa-fnsummary.c: New file.
25272 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25273 (enum inline_hints_vals, inline_hints, agg_position_info,
25274 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25275 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25276 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25277 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25278 inline_read_summary, inline_write_summary, inline_free_summary,
25279 inline_analyze_function, initialize_inline_failed,
25280 inline_merge_summary, inline_update_overall_summary,
25281 compute_inline_parameters): Move to ipa-fnsummary.h
25282 * ipa-fnsummary.h: New file.
25283 * ipa-inline-transform.h: Include ipa-inline.h.
25284 * ipa-inline.c: LIkewise.
25286 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25288 * ipa-inline.c (edge_badness): Use inlined_time instead of
25289 inline_summaries->get.
25291 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25293 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25295 2017-05-22 Nathan Sidwell <nathan@acm.org>
25297 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25298 (fdump-lang): Document 'raw' option.
25299 * dumpfile.h (TDI_tu): Delete.
25300 * dumpfile.c (dump_files): Remove translation-unit.
25301 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25303 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25305 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25306 command option from $(AWK) call.
25307 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25308 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25309 [FORMAT]: Remove handling of variable.
25310 * config/avr/t-multilib: Regenerate.
25312 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25314 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25316 (dump_inline_summary): Do not print self_time.
25317 (estimate_function_body_sizes): Do not set self_time.
25318 (compute_inline_parameters): Likewise.
25319 (inline_read_section, inline_write_summary): Do not stream self_time.
25320 * ipa-inline.h (inline_summary): Drop self_time.
25322 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25324 * ipa-inline-analysis.c (account_size_time): Rename to ...
25325 (inline_summary::account_size_time): ... this one.
25326 (reset_ipa_call_summary): Turn to ...
25327 (ipa_call_summary::reset): ... this one.
25328 (reset_inline_summary): Turn to ...
25329 (inline_summary::reset): ... this one.
25330 (inline_summary_t::remove): Update.
25331 (inline_summary_t::duplicate): Update.
25332 (ipa_call_summary_t::remove): Update.
25333 (dump_inline_summary): Update.
25334 (estimate_function_body_sizes): Update.
25335 (compute_inline_parameters): Update.
25336 (estimate_node_size_and_time): Update.
25337 (inline_merge_summary): Update.
25338 (inline_update_overall_summary): Update.
25339 (inline_read_section): Update.
25340 (inline_write_summary): Update.
25341 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25342 add account_size_time and reset member functions.
25343 (ipa_call_summary): Add reset function.
25344 * ipa-predicate.h (predicate::operator &): Constify.
25346 2017-05-22 Richard Biener <rguenther@suse.de>
25348 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25350 2017-05-19 Jason Merrill <jason@redhat.com>
25352 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25354 2017-05-19 Marek Polacek <polacek@redhat.com>
25357 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25358 TYPE_OVERFLOW_WRAPS checks.
25360 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25362 * tree-core.h (enum omp_clause_default_kind): Add
25363 "OMP_CLAUSE_DEFAULT_PRESENT".
25364 * tree-pretty-print.c (dump_omp_clause): Handle it.
25365 * gimplify.c (enum gimplify_omp_var_data): Add
25366 "GOVD_MAP_FORCE_PRESENT".
25367 (gimplify_adjust_omp_clauses_1): Map it to
25368 "GOMP_MAP_FORCE_PRESENT".
25369 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25371 * gimplify.c (oacc_default_clause): Clarify.
25373 2017-05-19 Nathan Sidwell <nathan@acm.org>
25375 LANG_HOOK_REGISTER_DUMPS
25376 * toplev.c (general_init): Call register dump lang hook.
25377 * doc/invoke.texi: Document -fdump-lang option family.
25378 * dumpfile.c (dump_files): Remove class dump here.
25379 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25380 * dumpfile.h (tree_dump_index): Remove TDI_class.
25381 * langhooks-def.h (lhd_register_dumps): Declare.
25382 (LANG_HOOKS_REGISTER_DUMPS): Define.
25383 (LANG_HOOKS_INITIALIZER): Add it.
25384 * langhooks.c (lhd_register_dumps): Define.
25385 * langhooks.h (struct lang_hooks): Add register_dumps.
25387 2017-05-19 Nathan Sidwell <nathan@acm.org>
25389 * context.h (context::set_passes): New.
25390 * context.c (context::context): Do not create pass manager.
25391 * toplev.c (general_init): Create pass manager here.
25393 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25395 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25396 use this splitter if two add or or instructions would also work for
25397 the constant we want to generate.
25399 2017-05-19 Richard Biener <rguenther@suse.de>
25402 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25403 predicate evaluation.
25405 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25407 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25409 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25412 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25414 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25415 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25416 (inline_edge_summary_vec): Turn into ...
25417 (ipa_call_summaries): ... this one.
25418 (redirect_to_unreachable, edge_set_predicate,
25419 evaluate_properties_for_edge, inline_summary_alloc,
25420 reset_ipa_call_summary, reset_inline_summary,
25421 inline_summary_t::duplicate): Update.
25422 (inline_edge_duplication_hook): Turn to ...
25423 (ipa_call_summary_t::duplicate): ... this one.
25424 (inline_edge_removal_hook): Turn to ...
25425 (ipa_call_summary_t::remove): ... this one.
25426 (dump_inline_edge_summary): Turn to ...
25427 (dump_ipa_call_summary): ... this one.
25428 (estimate_function_body_sizes): Update.
25429 (inline_update_callee_summaries): Update.
25430 (remap_edge_change_prob): Update.
25431 (remap_edge_summaries): Update.
25432 (inline_merge_summary): Update.
25433 (do_estimate_edge_time): Update.
25434 (inline_generate_summary): Update.
25435 (inline_read_section): Update.
25436 (inline_read_summary): Update.
25437 (inline_free_summary): Update.
25438 * ipa-inline.c (can_inline_edge_p): Update.
25439 (compute_inlined_call_time): Update.
25440 (want_inline_small_function_p): Update.
25441 (edge_badness): Update.
25442 (early_inliner): Update.
25443 * ipa-inline.h (inline_edge_summary): Turn to ...
25444 (ipa_call_summary): ... this one.
25445 (ipa_call_summary_t): New class.
25446 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25447 (ipa_call_summaries): New.
25448 (inline_edge_summary): Remove.
25449 (estimate_edge_growth): Update.
25450 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25451 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25452 * ipa-split.c (execute_split_functions): Update.
25453 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25455 2017-05-19 Richard Biener <rguenther@suse.de>
25457 PR middle-end/80764
25458 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25460 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25462 * config/rs6000/rs6000.c (struct machine_function): Add field
25463 fpr_is_wrapped_separately.
25464 (rs6000_get_separate_components): Use 64 components. Handle the
25465 new FPR components.
25466 (rs6000_components_for_bb): Handle the FPR components.
25467 (rs6000_emit_prologue_components): Handle the FPR components.
25468 (rs6000_emit_epilogue_components): Handle the FPR components.
25469 (rs6000_set_handled_components): Handle the FPR components.
25470 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25471 that are already separately shrink-wrapped.
25472 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25473 that are already separately shrink-wrapped.
25475 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25478 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25481 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25482 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25483 cases where the register allocator uses a move and an offsettable
25484 memory operation to/from a FPR register on ISA 2.06/2.07.
25485 (define_peephole2 for Altivec d-form store): Likewise.
25487 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25490 * config/i386/mmx.md (*mov<mode>_internal): Enable
25491 alternatives 11, 12, 13 and 14 also for 32bit targets.
25492 Remove alternatives 15, 16, 17 and 18.
25493 * config/i386/sse.md (vec_concatv2di): Change
25494 alternative (!x, *y) to (x, ?!*Yn).
25496 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25498 * dumpfile.h (enum dump_kind): Remove stray comma.
25500 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25502 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25503 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25504 predicate::num_conditions
25505 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25506 (CHANGED): turn into predicate::changed.
25507 (agg_position_info): Move to ipa-predicate.h
25508 (add_condition, predicate::add_clause, predicate::operator &=,
25509 predicate::or_with, predicate::evaluate, predicate::probability,
25510 dump_condition, dump_clause, predicate::dump,
25511 predicate::remap_after_duplication, predicate::remap_after_inlining,
25512 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25513 (evaluate_conditions_for_known_args): Update.
25514 (set_cond_stmt_execution_predicate): Update.
25515 * ipa-inline.h: Include ipa-predicate.h
25516 (condition, inline_param_summary, conditions, agg_position_info,
25517 predicate): Move to ipa-predicate.h
25518 * ipa-predicate.c: New file.
25519 * ipa-predicate.h: New file.
25521 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25523 * final.c (leaf_function_p): Check we are not in a sequence.
25525 2017-05-18 Martin Liska <mliska@suse.cz>
25527 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25528 * dumpfile.c (dump_register): Use new enum dump_kind.
25529 (get_dump_file_name): Likewise.
25530 (dump_enable_all): Likewise.
25531 (dump_switch_p_1): Likewise.
25532 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25533 * dumpfile.h (enum dump_kind): New enum type.
25534 (struct dump_file_info): Create constructor and
25535 format fields and comments.
25536 * passes.c (pass_manager::register_one_dump_file):
25538 * statistics.c (statistics_early_init): Likewise.
25539 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25540 TDF_TREE with TDF_SLIM.
25541 (gather_memory_references_ref): Likewise.
25543 2017-05-18 Martin Liska <mliska@suse.cz>
25545 * vec.h (struct vnull): Use it.
25547 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25549 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25550 (true_predicate, false_predicate, true_predicate_p,
25551 false_predicate_p): Remove.
25552 (single_cond_predicate, not_inlined_predicate): Turn to member function
25554 (add_condition): Update.
25555 (add_clause): Turn to...
25556 (predicate::add_clause): ... this one; update; allow passing NULL
25558 (and_predicates): Turn to ...
25559 (predicate::operator &=): ... this one.
25560 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25561 (or_predicates): Turn to ...
25562 (predicate::or_with): ... this one.
25563 (evaluate_predicate): Turn to ...
25564 (predicate::evaluate): ... this one.
25565 (predicate_probability): Turn to ...
25566 (predicate::probability): ... this one.
25567 (dump_condition): Update.
25568 (dump_predicate): Turn to ...
25569 (predicate::dump): ... this one.
25570 (account_size_time): Update.
25571 (edge_set_predicate): Update.
25572 (set_hint_predicate): UPdate.
25573 (evaluate_conditions_for_known_args): Update.
25574 (evaluate_properties_for_edge): Update.
25575 (remap_predicate_after_duplication): Turn to...
25576 (predicate::remap_after_duplication): ... this one.
25577 (remap_hint_predicate_after_duplication): Update.
25578 (inline_summary_t::duplicate): UPdate.
25579 (dump_inline_edge_summary): Update.
25580 (dump_inline_summary): Update.
25581 (set_cond_stmt_execution_predicate): Update.
25582 (set_switch_stmt_execution_predicate): Update.
25583 (compute_bb_predicates): Update.
25584 (will_be_nonconstant_expr_predicate): Update.
25585 (will_be_nonconstant_predicate): Update.
25586 (phi_result_unknown_predicate): Update.
25587 (predicate_for_phi_result): Update.
25588 (array_index_predicate): Update.
25589 (estimate_function_body_sizes): Update.
25590 (estimate_node_size_and_time): Update.
25591 (estimate_ipcp_clone_size_and_time): Update.
25592 (remap_predicate): Rename to ...
25593 (predicate::remap_after_inlining): ... this one.
25594 (remap_hint_predicate): Update.
25595 (inline_merge_summary): Update.
25596 (inline_update_overall_summary): Update.
25597 (estimate_size_after_inlining): Update.
25598 (read_predicate): Rename to ...
25599 (predicate::stream_in): ... this one.
25600 (read_inline_edge_summary): Update.
25601 (write_predicate): Rename to ...
25602 (predicate::stream_out): ... this one.
25603 (write_inline_edge_summary): Update.
25604 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25605 (clause_t): Turn to uint32_t
25606 (predicate): Turn to class; implement constructor and operators
25608 (size_time_entry): Update.
25609 (inline_summary): Update.
25610 (inline_edge_summary): Update.
25612 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25614 * fold-const.c (fold_binary_loc): Move transformation...
25615 * match.pd (C - X CMP X): ... here.
25617 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25619 * config/sparc/sparc.c (sparc_option_override): Set function
25620 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25621 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25623 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25625 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25627 2017-05-18 Marek Polacek <polacek@redhat.com>
25630 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25631 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25632 gimple_assign_load_p.
25634 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25636 PR middle-end/80692
25637 * real.c (do_compare): Give decimal_do_compare preference over
25638 comparing just the signs.
25640 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25642 * doc/md.texi (Canonicalization of Instructions): Describe the
25643 canonical form of instructions that inherently set a condition
25646 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25648 PR middle-end/80775
25649 * tree-cfg.c: Move deletion of unreachable case statements to after
25650 the merging of consecutive case labels.
25652 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25654 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25655 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25656 restoring of callee-saved registers.
25658 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25660 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25661 * config/visium/visium.c (single_set_and_flags): Likewise.
25662 * config/visium/visium.md (Substitutions): Likewise.
25664 2017-05-17 Martin Liska <mliska@suse.cz>
25666 * cfg.c: Introduce dump_flags_t type and
25667 use it instead of int type.
25669 * cfghooks.c: Likewise.
25670 * cfghooks.h (struct cfg_hooks): Likewise.
25671 * cfgrtl.c: Likewise.
25672 * cfgrtl.h: Likewise.
25673 * cgraph.c (cgraph_node::get_body): Likewise.
25674 * coretypes.h: Likewise.
25675 * domwalk.c: Likewise.
25676 * domwalk.h: Likewise.
25677 * dumpfile.c (struct dump_option_value_info): Likewise.
25678 (dump_enable_all): Likewise.
25679 (dump_switch_p_1): Likewise.
25680 (opt_info_switch_p): Likewise.
25681 * dumpfile.h (enum tree_dump_index): Likewise.
25682 (struct dump_file_info): Likewise.
25683 * genemit.c: Likewise.
25684 * generic-match-head.c: Likewise.
25685 * gengtype.c (open_base_files): Likewise.
25686 * gimple-pretty-print.c: Likewise.
25687 * gimple-pretty-print.h: Likewise.
25688 * graph.c (print_graph_cfg): Likewise.
25689 * graphite-scop-detection.c (dot_all_sese): Likewise.
25690 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25691 * loop-unroll.c (report_unroll): Likewise.
25692 * passes.c (pass_manager::register_one_dump_file): Likewise.
25693 * print-tree.c: Likewise.
25694 * statistics.c: Likewise.
25695 * tree-cfg.c: Likewise.
25696 * tree-cfg.h: Likewise.
25697 * tree-dfa.c: Likewise.
25698 * tree-dfa.h: Likewise.
25699 * tree-dump.c (dump_function): Likewise.
25700 * tree-dump.h (struct dump_info): Likewise.
25701 * tree-pretty-print.c: Likewise.
25702 * tree-pretty-print.h: Likewise.
25703 * tree-ssa-live.c: Likewise.
25704 * tree-ssa-live.h: Likewise.
25705 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25706 * tree-vect-loop.c: Likewise.
25707 * tree-vect-slp.c: Likewise.
25709 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25710 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25712 PR tree-optimization/80457
25713 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25714 of all arguments to a statement as scalar_to_vec operations.
25715 (vectorizable_call): Adjust call to vect_model_simple_cost for
25717 (vectorizable_conversion): Likewise.
25718 (vectorizable_assignment): Likewise.
25719 (vectorizable_shift): Likewise.
25720 (vectorizable_operation): Likewise.
25721 (vectorizable_comparison): Likewise.
25722 (vect_is_simple_cond): Record the def types for operands.
25723 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25724 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25725 for statement argument count.
25727 2017-05-16 Carl Love <cel@us.ibm.com>
25729 * config/rs6000/rs6000-c: Add support for built-in functions
25730 vector unsigned long long vec_bperm (vector unsigned long long,
25731 vector unsigned char)
25732 vector signed long long vec_mule (vector signed int,
25734 vector unsigned long long vec_mule (vector unsigned int,
25735 vector unsigned int)
25736 vector signed long long vec_mulo (vector signed int,
25738 vector unsigned long long vec_mulo (vector unsigned int,
25739 vector unsigned int)
25740 vector signed char vec_sldw (vector signed char,
25741 vector signed char,
25743 vector unsigned char vec_sldw (vector unsigned char,
25744 vector unsigned char,
25746 vector signed short vec_sldw (vector signed short,
25747 vector signed short,
25749 vector unsigned short vec_sldw (vector unsigned short,
25750 vector unsigned short,
25752 vector signed int vec_sldw (vector signed int,
25755 vector unsigned int vec_sldw (vector unsigned int,
25756 vector unsigned int,
25758 vector signed long long vec_sldw (vector signed long long,
25759 vector signed long long,
25761 vector unsigned long long vec_sldw (vector unsigned long long,
25762 vector unsigned long long,
25764 * config/rs6000/rs6000-c: Add support for built-in functions
25765 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25766 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25767 * doc/extend.texi: Update the built-in documentation for the
25768 new built-in functions.
25770 2017-05-16 Marek Polacek <polacek@redhat.com>
25774 * tree.c (save_expr): Don't fold the expression.
25776 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25778 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25779 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25780 and (?*y,m). Update insn attributes.
25782 2017-05-16 Martin Liska <mliska@suse.cz>
25784 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25785 flags argument of print_gimple_stmt, print_gimple_expr,
25786 print_generic_stmt and print_generic_expr.
25787 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25788 * coretypes.h: Likewise.
25789 * except.c (dump_eh_tree): Likewise.
25790 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25791 * gimple-pretty-print.h: Likewise.
25792 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25793 (backprop::push_to_worklist): Likewise.
25794 (backprop::pop_from_worklist): Likewise.
25795 (backprop::process_use): Likewise.
25796 (backprop::intersect_uses): Likewise.
25797 (note_replacement): Likewise.
25798 * gimple-ssa-store-merging.c
25799 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25800 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25801 (pass_store_merging::execute): Likewise.
25802 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25803 (ssa_base_cand_dump_callback): Likewise.
25804 (dump_incr_vec): Likewise.
25805 (replace_refs): Likewise.
25806 (replace_mult_candidate): Likewise.
25807 (create_add_on_incoming_edge): Likewise.
25808 (create_phi_basis): Likewise.
25809 (insert_initializers): Likewise.
25810 (all_phi_incrs_profitable): Likewise.
25811 (introduce_cast_before_cand): Likewise.
25812 (replace_one_candidate): Likewise.
25813 * gimplify.c (gimplify_expr): Likewise.
25814 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25815 (set_rename): Likewise.
25816 (rename_uses): Likewise.
25817 (copy_loop_phi_nodes): Likewise.
25818 (add_close_phis_to_merge_points): Likewise.
25819 (copy_loop_close_phi_args): Likewise.
25820 (copy_cond_phi_args): Likewise.
25821 (graphite_copy_stmts_from_block): Likewise.
25822 (translate_pending_phi_nodes): Likewise.
25823 * graphite-poly.c (print_pdr): Likewise.
25824 (dump_gbb_cases): Likewise.
25825 (dump_gbb_conditions): Likewise.
25826 (print_scop_params): Likewise.
25827 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25828 (build_cross_bb_scalars_use): Likewise.
25829 (gather_bbs::before_dom_children): Likewise.
25830 * hsa-dump.c (dump_hsa_immed): Likewise.
25831 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25832 (get_replacement_map): Likewise.
25833 * ipa-inline-analysis.c (dump_condition): Likewise.
25834 (estimate_function_body_sizes): Likewise.
25835 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25836 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25837 * ipa-prop.c (ipa_dump_param): Likewise.
25838 (ipa_print_node_jump_functions_for_edge): Likewise.
25839 (ipa_modify_call_arguments): Likewise.
25840 (ipa_modify_expr): Likewise.
25841 (ipa_dump_param_adjustments): Likewise.
25842 (ipa_dump_agg_replacement_values): Likewise.
25843 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25844 * ipa-pure-const.c (check_stmt): Likewise.
25845 (pass_nothrow::execute): Likewise.
25846 * ipa-split.c (execute_split_functions): Likewise.
25847 * omp-offload.c (dump_oacc_loop_part): Likewise.
25848 (dump_oacc_loop): Likewise.
25849 * trans-mem.c (tm_log_emit): Likewise.
25850 (tm_memopt_accumulate_memops): Likewise.
25851 (dump_tm_memopt_set): Likewise.
25852 (dump_tm_memopt_transform): Likewise.
25853 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25854 (print_loop): Likewise.
25855 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25856 (chkp_gather_checks_info): Likewise.
25857 (chkp_get_check_result): Likewise.
25858 (chkp_remove_check_if_pass): Likewise.
25859 (chkp_use_outer_bounds_if_possible): Likewise.
25860 (chkp_reduce_bounds_lifetime): Likewise.
25861 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25862 (chkp_mark_completed_bounds): Likewise.
25863 (chkp_register_incomplete_bounds): Likewise.
25864 (chkp_mark_invalid_bounds): Likewise.
25865 (chkp_maybe_copy_and_register_bounds): Likewise.
25866 (chkp_build_returned_bound): Likewise.
25867 (chkp_get_bound_for_parm): Likewise.
25868 (chkp_build_bndldx): Likewise.
25869 (chkp_get_bounds_by_definition): Likewise.
25870 (chkp_generate_extern_var_bounds): Likewise.
25871 (chkp_get_bounds_for_decl_addr): Likewise.
25872 * tree-chrec.c (chrec_apply): Likewise.
25873 * tree-data-ref.c (dump_data_reference): Likewise.
25874 (dump_subscript): Likewise.
25875 (dump_data_dependence_relation): Likewise.
25876 (analyze_overlapping_iterations): Likewise.
25877 * tree-inline.c (expand_call_inline): Likewise.
25878 (tree_function_versioning): Likewise.
25879 * tree-into-ssa.c (dump_defs_stack): Likewise.
25880 (dump_currdefs): Likewise.
25881 (dump_names_replaced_by): Likewise.
25882 (dump_update_ssa): Likewise.
25883 (update_ssa): Likewise.
25884 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25885 * tree-parloops.c (build_new_reduction): Likewise.
25886 (try_create_reduction_list): Likewise.
25887 (ref_conflicts_with_region): Likewise.
25888 (oacc_entry_exit_ok_1): Likewise.
25889 (oacc_entry_exit_single_gang): Likewise.
25890 * tree-pretty-print.h: Likewise.
25891 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25892 (get_scalar_evolution): Likewise.
25893 (add_to_evolution): Likewise.
25894 (get_loop_exit_condition): Likewise.
25895 (analyze_evolution_in_loop): Likewise.
25896 (analyze_initial_condition): Likewise.
25897 (analyze_scalar_evolution): Likewise.
25898 (instantiate_scev): Likewise.
25899 (number_of_latch_executions): Likewise.
25900 (gather_chrec_stats): Likewise.
25901 (final_value_replacement_loop): Likewise.
25902 (scev_const_prop): Likewise.
25903 * tree-sra.c (dump_access): Likewise.
25904 (disqualify_candidate): Likewise.
25905 (create_access): Likewise.
25906 (reject): Likewise.
25907 (maybe_add_sra_candidate): Likewise.
25908 (create_access_replacement): Likewise.
25909 (analyze_access_subtree): Likewise.
25910 (analyze_all_variable_accesses): Likewise.
25911 (sra_modify_assign): Likewise.
25912 (initialize_constant_pool_replacements): Likewise.
25913 (find_param_candidates): Likewise.
25914 (decide_one_param_reduction): Likewise.
25915 (replace_removed_params_ssa_names): Likewise.
25916 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25917 * tree-ssa-copy.c (dump_copy_of): Likewise.
25918 (copy_prop_visit_cond_stmt): Likewise.
25919 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25920 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25921 (record_equivalences_from_stmt): Likewise.
25922 * tree-ssa-dse.c (compute_trims): Likewise.
25923 (delete_dead_call): Likewise.
25924 (delete_dead_assignment): Likewise.
25925 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25926 (forward_propagate_into_cond): Likewise.
25927 (pass_forwprop::execute): Likewise.
25928 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25929 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25931 (move_computations_worker): Likewise.
25932 (execute_sm): Likewise.
25933 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25934 (remove_exits_and_undefined_stmts): Likewise.
25935 (remove_redundant_iv_tests): Likewise.
25936 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25937 (adjust_iv_update_pos): Likewise.
25938 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25939 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25940 (value_replacement): Likewise.
25941 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25942 * tree-ssa-pre.c (print_pre_expr): Likewise.
25943 (get_representative_for): Likewise.
25944 (create_expression_by_pieces): Likewise.
25945 (insert_into_preds_of_block): Likewise.
25946 (eliminate_insert): Likewise.
25947 (eliminate_dom_walker::before_dom_children): Likewise.
25948 (eliminate): Likewise.
25949 (remove_dead_inserted_code): Likewise.
25950 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25951 * tree-ssa-reassoc.c (get_rank): Likewise.
25952 (eliminate_duplicate_pair): Likewise.
25953 (eliminate_plus_minus_pair): Likewise.
25954 (eliminate_not_pairs): Likewise.
25955 (undistribute_ops_list): Likewise.
25956 (eliminate_redundant_comparison): Likewise.
25957 (update_range_test): Likewise.
25958 (optimize_range_tests_var_bound): Likewise.
25959 (optimize_vec_cond_expr): Likewise.
25960 (rewrite_expr_tree): Likewise.
25961 (rewrite_expr_tree_parallel): Likewise.
25962 (linearize_expr): Likewise.
25963 (break_up_subtract): Likewise.
25964 (linearize_expr_tree): Likewise.
25965 (attempt_builtin_powi): Likewise.
25966 (attempt_builtin_copysign): Likewise.
25967 (transform_stmt_to_copy): Likewise.
25968 (transform_stmt_to_multiply): Likewise.
25969 (dump_ops_vector): Likewise.
25970 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25971 (print_scc): Likewise.
25972 (set_ssa_val_to): Likewise.
25973 (visit_reference_op_store): Likewise.
25974 (visit_use): Likewise.
25975 (sccvn_dom_walker::before_dom_children): Likewise.
25976 (run_scc_vn): Likewise.
25977 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25979 (expr_hash_elt::print): Likewise.
25980 (const_and_copies::pop_to_marker): Likewise.
25981 (const_and_copies::record_const_or_copy_raw): Likewise.
25982 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25983 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25984 (dump_predicates): Likewise.
25985 (find_uninit_use): Likewise.
25986 (warn_uninitialized_phi): Likewise.
25987 (pass_late_warn_uninitialized::execute): Likewise.
25988 * tree-ssa.c (verify_vssa): Likewise.
25989 (verify_ssa): Likewise.
25990 (maybe_optimize_var): Likewise.
25991 * tree-vrp.c (dump_value_range): Likewise.
25992 (dump_all_value_ranges): Likewise.
25993 (dump_asserts_for): Likewise.
25994 (register_edge_assert_for_2): Likewise.
25995 (vrp_visit_cond_stmt): Likewise.
25996 (vrp_visit_switch_stmt): Likewise.
25997 (vrp_visit_stmt): Likewise.
25998 (vrp_visit_phi_node): Likewise.
25999 (simplify_cond_using_ranges_1): Likewise.
26000 (fold_predicate_in): Likewise.
26001 (evrp_dom_walker::before_dom_children): Likewise.
26002 (evrp_dom_walker::push_value_range): Likewise.
26003 (evrp_dom_walker::pop_value_range): Likewise.
26004 (execute_early_vrp): Likewise.
26006 2017-05-16 Richard Biener <rguenther@suse.de>
26008 * dwarf2out.c (loc_list_from_tree_1): Do not create
26009 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26011 2017-05-16 Richard Biener <rguenther@suse.de>
26013 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26015 (note_variable_value_in_expr): If we resolved the decl ref
26016 do not push to the stack.
26018 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26020 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26021 operations in fast-math mode.
26022 (vaddq_f16): Likewise.
26023 (vmul_f16): Likewise.
26024 (vmulq_f16): Likewise.
26025 (vsub_f16): Likewise.
26026 (vsubq_f16): Likewise.
26027 * config/arm/neon.md (add<mode>3): New.
26029 (fma:<VH:mode>3): New. Also remove outdated comment.
26032 2017-05-16 Martin Liska <mliska@suse.cz>
26036 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26037 (odr_types_equivalent_p): Likewise.
26039 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26041 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26043 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26046 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26047 non-interunit SSE move alternatives with '?'.
26048 (zero-extendsidi peephole2): New peephole to skip intermediate
26049 general register in SSE zero-extend sequence.
26051 2017-05-15 Jeff Law <law@redhat.com>
26053 * reorg.c (relax_delay_slots): Create a new variable to hold
26054 the temporary target rather than clobbering TARGET_LABEL.
26056 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26057 missing argument to extract_bit_field call.
26058 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26060 2017-05-15 Martin Liska <mliska@suse.cz>
26063 * gcc.c (process_command): Do not allow empty argument of -o option.
26065 2017-05-15 Renlin Li <renlin.li@arm.com>
26067 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26068 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26069 * config/aarch64/constraints.md (Usf): Add long call check.
26070 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26071 (call_value): Likewise.
26072 (sibcall): Likewise.
26073 (sibcall_value): Likewise.
26075 (call_value_insn): New.
26076 (sibcall_insn): Update rtx pattern.
26077 (sibcall_value_insn): Likewise.
26078 (call_internal): Remove.
26079 (call_value_internal): Likewise.
26080 (sibcall_internal): Likewise.
26081 (sibcall_value_internal): Likewise.
26082 (call_reg): Likewise.
26083 (call_symbol): Likewise.
26084 (call_value_reg): Likewise.
26085 (call_value_symbol): Likewise.
26087 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26090 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26092 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26094 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26095 compatible with CCGOCmode and with CCZmode.
26097 2017-05-14 Martin Sebor <msebor@redhat.com>
26099 PR middle-end/77671
26100 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26101 (gimple_fold_builtin_snprintf): Same.
26102 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26103 (gimple_fold_builtin_snprintf): Same.
26104 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26105 of character types.
26106 (is_call_safe): New function.
26107 (try_substitute_return_value): Call it.
26108 (try_simplify_call): New function.
26109 (pass_sprintf_length::handle_gimple_call): Call it.
26111 2017-05-14 Martin Sebor <msebor@redhat.com>
26113 PR middle-end/80669
26114 * builtins.c (expand_builtin_stpncpy): Simplify.
26116 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26118 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26119 * config/i386/i386.h
26120 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26121 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26122 (struct machine_function): Add new members call_ms2sysv,
26123 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26124 (struct machine_frame_state): New fields sp_realigned and
26125 sp_realigned_offset.
26126 * config/i386/i386.c
26127 (enum xlogue_stub): New enum.
26128 (enum xlogue_stub_sets): New enum.
26129 (class xlogue_layout): New class.
26130 (struct ix86_frame): New fields stack_realign_allocate_offset,
26131 stack_realign_offset and outlined_save_offset. Modify comments to
26132 detail stack layout when using out-of-line stubs.
26133 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26134 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26135 -mcall-ms2sysv-xlogues.
26136 (stub_managed_regs): New static variable.
26137 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26138 registers managed by out-of-line stub.
26139 (disable_call_ms2sysv_xlogues): New function.
26140 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26141 m->call_ms2sysv when appropriate and compute frame layout for
26143 (sp_valid_at, fp_valid_at): New inline functions.
26144 (choose_basereg): New function.
26145 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26147 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26148 Use align parameter of choose_baseaddr to generated aligned SSE movs
26150 (pro_epilogue_adjust_stack): Modify to track
26151 machine_frame_state::sp_realigned.
26152 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26153 (ix86_nsaved_sseregs): Likewise.
26154 (ix86_emit_save_regs): Likewise.
26155 (ix86_emit_save_regs_using_mov): Likewise.
26156 (ix86_emit_save_sse_regs_using_mov): Likewise.
26157 (get_scratch_register_on_entry): Likewise.
26158 (gen_frame_set): New function.
26159 (gen_frame_load): Likewise.
26160 (gen_frame_store): Likewise.
26161 (emit_outlined_ms2sysv_save): Likewise.
26162 (emit_outlined_ms2sysv_restore): Likewise.
26163 (ix86_expand_prologue): Modify stack re-alignment code and call
26164 emit_outlined_ms2sysv_save when appropriate.
26165 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26166 parameter rtx_insn *insn, which allows the function to be used to only
26167 generate the notes.
26168 (ix86_expand_epilogue): Modify validity checks of frame and stack
26169 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26170 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26171 * config/i386/predicates.md
26172 (save_multiple): New predicate.
26173 (restore_multiple): Likewise.
26174 * config/i386/sse.md
26175 (save_multiple<mode>): New pattern.
26176 (save_multiple_realign<mode>): Likewise.
26177 (restore_multiple<mode>): Likewise.
26178 (restore_multiple_and_return<mode>): Likewise.
26179 (restore_multiple_leave_return<mode>): Likewise.
26180 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26182 2017-05-14 Julia Koval <julia.koval@intel.com>
26184 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26185 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26186 (__builtin_ia32_xsetbv): New builtins.
26187 * config/i386/i386.c (ix86_expand_special_args_builtin):
26189 (ix86_expand_builtin): Special expand for new intrinsics.
26190 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26191 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26192 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26194 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26196 * cfganal.c (inverted_post_order_compute): Change argument type
26198 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26199 * df-core.c (rest_of_handle_df_initialize): Adjust.
26200 (rest_of_handle_df_finish): Likewise.
26201 (df_analyze_1): Likewise.
26202 (df_analyze): Likewise.
26203 (loop_inverted_post_order_compute): Change argument to be a vec *.
26204 (df_analyze_loop): Adjust.
26205 (df_get_n_blocks): Likewise.
26206 (df_get_postorder): Likewise.
26207 * df.h (struct df_d): Change field to be a vec.
26208 * lcm.c (compute_laterin): Adjust.
26209 (compute_available): Likewise.
26210 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26211 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26212 * tree-ssa-pre.c (compute_antic): Likewise.
26214 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26216 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26217 (depth_first_search::depth_first_search): Change structure init
26218 function to this constructor.
26219 (depth_first_search::add_bb): Rename function to this member.
26220 (depth_first_search::execute): Likewise.
26221 (flow_dfs_compute_reverse_finish): Adjust.
26223 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26225 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26226 (longest_simple_path): Likewise.
26227 * shrink-wrap.c (spread_components): Likewise.
26228 (disqualify_problematic_components): Likewise.
26229 (emit_common_heads_for_components): Likewise.
26230 (emit_common_tails_for_components): Likewise.
26231 (insert_prologue_epilogue_for_components): Likewise.
26233 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26235 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26238 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26240 * df-core.c (df_set_blocks): Start using auto_bitmap.
26241 (df_compact_blocks): Likewise.
26242 * df-problems.c (df_rd_confluence_n): Likewise.
26243 * df-scan.c (df_insn_rescan_all): Likewise.
26244 (df_process_deferred_rescans): Likewise.
26245 (df_update_entry_block_defs): Likewise.
26246 (df_update_exit_block_uses): Likewise.
26247 (df_entry_block_bitmap_verify): Likewise.
26248 (df_exit_block_bitmap_verify): Likewise.
26249 (df_scan_verify): Likewise.
26250 * lra-constraints.c (lra_constraints): Likewise.
26251 (undo_optional_reloads): Likewise.
26252 (lra_undo_inheritance): Likewise.
26253 * lra-remat.c (calculate_gen_cands): Likewise.
26254 (do_remat): Likewise.
26255 * lra-spills.c (assign_spill_hard_regs): Likewise.
26256 (spill_pseudos): Likewise.
26257 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26258 (bitmap_set_subtract_values): Likewise.
26260 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26262 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26263 management with auto_bitmap.
26264 (fix_inter_tick): Likewise.
26265 (fix_recovery_deps): Likewise.
26266 * ira.c (add_store_equivs): Likewise.
26267 (find_moveable_pseudos): Likewise.
26268 (split_live_ranges_for_shrink_wrap): Likewise.
26269 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26270 (rtx_reuse_manager::seen_def_p): Likewise.
26271 (rtx_reuse_manager::set_seen_def): Likewise.
26272 * print-rtl.h (class rtx_reuse_manager): Likewise.
26274 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26276 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26278 (migrate_btr_def): Likewise.
26279 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26280 * df-core.c (loop_post_order_compute): Likewise.
26281 (loop_inverted_post_order_compute): Likewise.
26282 * hsa-common.h: Likewise.
26283 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26284 * init-regs.c (initialize_uninitialized_regs): Likewise.
26285 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26286 (inline_small_functions): Likewise.
26287 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26288 * ira.c (combine_and_move_insns): Likewise.
26289 (build_insn_chain): Likewise.
26290 * loop-invariant.c (find_invariants): Likewise.
26291 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26292 * predict.c (tree_predict_by_opcode): Likewise.
26293 (predict_paths_leading_to): Likewise.
26294 (predict_paths_leading_to_edge): Likewise.
26295 (estimate_loops_at_level): Likewise.
26296 (estimate_loops): Likewise.
26297 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26298 (spread_components): Likewise.
26299 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26300 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26301 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26302 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26303 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26304 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26305 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26306 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26307 (mark_threaded_blocks): Likewise.
26308 (thread_through_all_blocks): Likewise.
26309 * tree-ssa.c (verify_ssa): Likewise.
26310 (execute_update_addresses_taken): Likewise.
26311 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26313 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26315 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26317 (post_order_compute): Likewise.
26318 (inverted_post_order_compute): Likewise.
26319 (pre_and_rev_post_order_compute_fn): Likewise.
26321 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26323 * genrecog.c (int_set::int_set): Explicitly construct our
26324 auto_vec base class.
26325 * vec.h (auto_vec::auto_vec): New constructor.
26327 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26329 * bitmap.h (class auto_bitmap): New constructor taking
26330 bitmap_obstack * argument.
26332 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26334 * bitmap.h (class auto_bitmap): Change type of m_bits to
26335 bitmap_head, and adjust ctor / dtor and member operators.
26337 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26339 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26340 when returned register mode doesn't match original mode.
26342 2017-05-12 Jeff Law <law@redhat.com>
26343 Jakub Jelinek <jakub@redhat.com>
26345 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26346 we look for cc setter after the compare-elim changes.
26347 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26348 within the vector to match what compare-elim now expects.
26349 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26350 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26352 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26353 after the compare-elim changes.
26354 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26355 the vector to match what compare-elim now expects.
26356 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26357 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26358 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26359 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26360 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26362 * config/visium/visium.c (single_set_and_flags): Fix where
26363 we look for cc setter after the compare-elim changes.
26364 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26365 with the vector to match what compare-elim now expects.
26366 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26367 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26368 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26369 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26370 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26371 (neg<mode>2_insn_set_overflow): Likewise.
26373 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26375 PR middle-end/79794
26376 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26377 maybe_expand_insn call, set ops[0].target. If still set after call,
26378 set alt_rtl. Add extra arg to recursive calls.
26379 (extract_bit_field): Add alt_rtl argument. Pass to
26381 * expmed.h (extract_bit_field): Fix prototype.
26382 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26383 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26384 to extract_bit_field_calls.
26385 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26386 Pass alt_rtl to extract_bit_field calls.
26387 * calls.c (store_unaligned_arguments_into_psuedos)
26388 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26389 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26391 * optabs.h (struct expand_operand): Add target bitfield.
26393 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26395 * compare-elim.c (try_eliminate_compare): Canonicalize
26396 operation with embedded compare to
26397 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26398 (set (reg) (operation)].
26400 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26402 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26405 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26406 cost of adding a carry flag for ADC instruction.
26407 [case MINUS]: Ignore the cost of subtracting a carry flag
26408 for SBB instruction.
26410 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26412 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26414 * config/rs6000/bmiintrin.h: New file.
26415 * config/rs6000/bmi2intrin.h: New file.
26416 * config/rs6000/x86intrin.h: New file.
26418 2017-05-12 Jeff Law <law@redhat.com>
26420 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26423 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26425 PR middle-end/80707
26426 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26428 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26430 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26431 early expansion of vector divide builtins.
26432 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26433 builtins identified as having unsigned arguments.
26435 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26437 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26438 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26439 expansion of vector logical operations (and, andc, or, xor,
26442 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26444 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26445 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26447 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26449 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26450 early GIMPLE expansion of vector multiplies.
26452 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26454 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26455 TARGET_HAVE_MOVT conditional.
26456 (movt splitter): Likewise.
26458 2017-05-12 Richard Biener <rguenther@suse.de>
26460 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26461 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26462 Fold all stmts not inplace.
26464 2017-05-12 Richard Biener <rguenther@suse.de>
26466 PR tree-optimization/80713
26467 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26468 inserted_exprs bit for not removed stmts.
26470 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26472 PR middle-end/69921
26473 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26474 parallelized" attribute for parallelized OpenACC kernels.
26475 * omp-offload.c (execute_oacc_device_lower): Use it.
26477 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26478 Set "oacc kernels" attribute.
26479 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26480 parameter. Adjust all users.
26481 (oacc_fn_attrib_kernels_p): Remove function.
26482 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26483 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26484 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26485 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26486 assert "oacc kernels" attribute is set.
26488 2017-05-11 Carl Love <cel@us.ibm.com>
26490 * config/rs6000/rs6000-c: Add support for built-in functions
26491 vector unsigned char vec_popcnt (vector signed char)
26492 vector unsigned char vec_popcnt (vector unsigned char)
26493 vector unsigned short vec_popcnt (vector signed short)
26494 vector unsigned short vec_popcnt (vector unsigned short)
26495 vector unsigned int vec_popcnt (vector signed int)
26496 vector unsigned int vec_popcnt (vector unsigned int)
26497 vector unsigned long long vec_popcnt (vector signed long long)
26498 vector unsigned long long vec_popcnt (vector unsigned long long)
26499 vector signed long long vec_slo (vector signed long long,
26500 vector signed char)
26501 vector signed long long vec_slo (vector signed long long,
26502 vector unsigned char)
26503 vector unsigned long long vec_slo (vector unsigned long long,
26504 vector signed char)
26505 vector unsigned long long vec_slo (vector unsigned long long,
26506 vector unsigned char)
26507 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26508 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26509 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26510 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26511 * doc/extend.texi: Update the built-in documentation file for the
26512 new built-in functions.
26514 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26516 * attribs.h (sorted_attr_string): Move machine independent
26517 functions for target clone support from the i386 port to common
26518 code. Rename ix86_function_versions to common_function_versions.
26519 Rename make_name to make_unique_name.
26520 (common_function_versions): Likewise.
26521 (make_unique_name): Likewise.
26522 (make_dispatcher_decl): Likewise.
26523 (is_function_default_version): Likewise.
26524 * attribs.c (attr_strcmp): Likewise.
26525 (sorted_attr_string): Likewise.
26526 (common_function_versions): Likewise.
26527 (make_unique_name): Likewise.
26528 (make_dispatcher_decl): Likewise.
26529 (is_function_default_version): Likewise.
26530 * config/i386/i386.c (attr_strcmp): Likewise.
26531 (sorted_attr_string): Likewise.
26532 (ix86_function_versions): Likewise.
26533 (make_name): Likewise.
26534 (make_dispatcher_decl): Likewise.
26535 (is_function_default_version): Likewise.
26536 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26538 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26541 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26542 Account for direct move costs for vec_construct of integer
26545 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26548 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26549 (UNSPEC_STX_ATOMIC): Ditto.
26550 (loaddi_via_sse): New insn.
26551 (storedi_via_sse): Ditto.
26552 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26553 Update corresponding peephole2 patterns.
26554 (atomic_storedi_fpu): Ditto.
26556 2017-05-11 Julia Koval <julia.koval@intel.com>
26558 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26559 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26561 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26562 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26563 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26565 2017-05-11 Nathan Sidwell <nathan@acm.org>
26567 * graphite-poly.c: Include dumpfile.h.
26569 * dumpfle.h (dump_function): Declare here ...
26570 * tree-dump.h (dump_function): ... not here.
26571 * dumpfile.c: #include tree-cfg.h.
26572 (dump_function): Move here from ...
26573 * tree-dump.c (dump_function): ... here.
26574 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26575 * graphite-poly.c: Don't include tree-dump.h.
26576 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26577 * print-tree.c: Likewise.
26578 * stor-layout.c: Likewise.
26579 * tree-nested.c: Likewise.
26581 * dumpfile.c (dump_start): Use TDF_FLAGS.
26582 (dump_enable_all): Fix TDF_KIND check thinko.
26584 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26586 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26587 array entries to represent two legal parameterizations of the
26588 overloaded __builtin_cmpb function, as represented by the
26589 P6_OV_BUILTIN_CMPB constant.
26590 (altivec_resolve_overloaded_builtin): Add special case handling
26591 for the __builtin_cmpb function, as represented by the
26592 P6_OV_BUILTIN_CMPB constant.
26593 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26594 (BU_P6_64BIT_2): New macro.
26595 (BU_P6_OVERLOAD_2): New macro
26596 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26597 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26598 (CMPB): Add overload support to represent both 32-bit and 64-bit
26599 compare-bytes function.
26600 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26601 support for TARGET_CMPB.
26602 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26603 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26604 documentation of the __builtin_cmpb overloaded built-in function.
26606 2017-05-11 Richard Biener <rguenther@suse.de>
26608 PR tree-optimization/80705
26609 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26610 bases are not vectorizable.
26612 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26614 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26615 when counting register pressure.
26617 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26619 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26620 (iv_ca_more_deps): Renamed to ...
26621 (iv_ca_compare_deps): ... this.
26622 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26624 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26626 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26628 (determine_group_iv_costs): ... here.
26629 (find_inv_vars_cb): Record inv var if it's not recorded before.
26631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26633 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26634 (get_shiftadd_cost): Ditto.
26636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26638 * tree-ssa-address.c: Include header file.
26639 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26641 (add_to_parts): Refactor.
26642 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26643 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26646 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26648 PR tree-optimization/53090
26649 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26651 (extract_cond_operands): Detect condition with IV on both sides
26652 and return COMP_IV_EXPR_2.
26653 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26654 (rewrite_use_compare): Simplify by removing call to function
26655 extract_cond_operands.
26657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26659 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26660 (extract_cond_operands): Detect condition comparing against non-
26661 invariant bound and return appropriate enum value.
26662 (find_interesting_uses_cond): Update use of extract_cond_operands.
26663 Handle its return value accordingly.
26664 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26666 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26668 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26669 nonlinear iv_use computation in loop invariant sensitive way.
26671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26673 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26674 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26678 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26679 (dump_cand): Support iv_cand.inv_exprs.
26680 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26682 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26685 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26687 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26689 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26690 as local function. Include necessary header files.
26691 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26695 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26699 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26700 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26701 RSHIFT_EXPR and BIT_NOT_EXPR.
26703 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26705 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26706 (adjust_setup_cost): New parameter supporting round up adjustment.
26707 (struct address_cost_data): Delete.
26708 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26709 (split_address_cost, ptr_difference_cost): Delete.
26710 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26711 (struct ainc_cost_data): New struct.
26712 (get_address_cost_ainc): New function.
26713 (get_address_cost, get_computation_cost): Reimplement.
26714 (determine_group_iv_cost_address): Record inv_expr for all uses of
26716 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26717 (iv_ca_has_deps): Reimplemented to ...
26718 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26720 (iv_ca_extend): Call iv_ca_more_deps.
26722 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26724 * tree-ssa-address.c (struct mem_address): Move to header file.
26725 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26726 * tree-ssa-address.h (struct mem_address): Move from C file.
26727 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26729 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26731 * tree-affine.h (aff_combination_type): New interface.
26732 (aff_combination_zero_p): Remove static.
26733 (aff_combination_const_p): New interface.
26734 (aff_combination_singleton_var_p): New interfaces.
26736 2017-05-11 Richard Biener <rguenther@suse.de>
26738 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26739 Skip unreachable blocks and destinations.
26740 (eliminate): Move stmt removal and fixup ...
26741 (fini_eliminate): ... here. Skip inserted exprs.
26742 (pass_pre::execute): Move fini_pre after fini_eliminate.
26743 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26744 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26745 PRE to get rid of dead code that has invalid SSA form and
26746 split critical edges again.
26748 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26750 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26752 2017-05-11 Richard Biener <rguenther@suse.de>
26754 * passes.c (execute_function_todo): Verify loops if they are
26755 said to be up-to-date.
26756 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26757 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26759 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26762 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26763 handle calling assemble_external ourself.
26766 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26767 modes with zero size. Enhance comment.
26769 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26771 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26772 built-ins for vec_xl and vec_xst with short and char pointer
26775 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26777 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26778 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26779 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26780 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26781 (_mm_maskz_min_round_ss): New intrinsics.
26782 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26783 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26784 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26785 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26786 (__builtin_ia32_minss_mask_round): New builtins.
26787 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26788 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26789 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26791 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26792 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26794 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26797 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26799 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26800 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26801 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26802 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26803 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26804 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26805 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26806 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26807 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26808 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26809 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26810 (__builtin_ia32_mulss_mask_round): New builtins.
26811 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26812 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26813 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26815 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26816 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26818 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26821 2017-05-10 Julia Koval <julia.koval@intel.com>
26823 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26824 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26825 (_mm256_setr_m128i): New intrinsics.
26827 2017-05-10 Julia Koval <julia.koval@intel.com>
26829 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26830 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26831 (_mm_maskz_rcp14_ss): New intrinsics.
26832 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26833 (__builtin_ia32_rcp14ss_mask): New builtins.
26834 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26836 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26838 PR tree-optimization/51513
26839 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26840 (assert_unreachable_fallthru_edge_p): Use it.
26841 (group_case_labels_stmt): Likewise.
26842 * tree-cfg.h: Prototype it.
26843 * stmt.c: Include cfghooks.h and tree-cfg.h.
26844 (emit_case_dispatch_table) <gap_label>: New local variable.
26845 Use it to fill dispatch table gaps.
26846 Test for default_label before updating probabilities.
26847 (expand_case) <default_label>: Remove unneeded initialization.
26848 Test for unreachable default case statement and remove its edge.
26849 Set default_label accordingly.
26850 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26852 2017-05-10 Carl Love <cel@us.ibm.com>
26854 * config/rs6000/rs6000-c: Add support for built-in functions
26855 vector signed char vec_neg (vector signed char)
26856 vector signed short int vec_neg (vector short int)
26857 vector signed int vec_neg (vector signed int)
26858 vector signed long long vec_neg (vector signed long long)
26859 vector float vec_neg (vector float)
26860 vector double vec_neg (vector double)
26861 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26863 * config/rs6000/altivec.h: Add define for vec_neg
26864 * doc/extend.texi: Update the built-in documentation for the
26865 new built-in functions.
26867 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26869 PR tree-optimization/77644
26870 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26872 2017-05-10 Nathan Sidwell <nathan@acm.org>
26874 * dumpfile.h (TDI_lang_all): New.
26875 (TDF_KIND): New. Renumber others
26876 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26878 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26880 (get_dump_file_name): Adjust suffix generation.
26881 (dump_enable_all): Use TDF_KIND.
26882 * doc/invoke.texi (-fdump-lang-all): Document.
26884 * dumpfile.h: Tabify.
26886 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26889 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26890 Move member access before delete.
26892 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26894 * tree-inline.c (expand_call_inline): Split block at stmt
26897 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26900 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26901 are now unused after splitting mov{sf,sd}_hardfloat.
26902 (f32_lr2): Likewise.
26903 (f32_lm): Likewise.
26904 (f32_lm2): Likewise.
26905 (f32_li): Likewise.
26906 (f32_li2): Likewise.
26907 (f32_lv): Likewise.
26908 (f32_sr): Likewise.
26909 (f32_sr2): Likewise.
26910 (f32_sm): Likewise.
26911 (f32_sm2): Likewise.
26912 (f32_si): Likewise.
26913 (f32_si2): Likewise.
26914 (f32_sv): Likewise.
26915 (f32_dm): Likewise.
26916 (f32_vsx): Likewise.
26917 (f32_av): Likewise.
26918 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26919 For movsf, order stores so the VSX stores occur before the GPR
26920 store which encourages the register allocator to use a traditional
26921 FPR instead of a GPR. For movsd, order the stores so that the GPR
26922 store comes before the VSX stores to allow the power6 to work.
26923 This is due to the power6 not having a 32-bit integer store
26924 instruction from a FPR.
26925 (movsf_hardfloat): Likewise.
26926 (movsd_hardfloat): Likewise.
26928 2017-05-09 Martin Sebor <msebor@redhat.com>
26930 PR translation/80280
26931 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26934 PR translation/80280
26935 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26936 data member added in r247778.
26937 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26939 2017-05-09 Nathan Sidwell <nathan@acm.org>
26941 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26943 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26946 2017-05-09 Marek Polacek <polacek@redhat.com>
26948 * doc/invoke.texi: Fix typo.
26950 2017-05-09 Richard Biener <rguenther@suse.de>
26952 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26953 (vrp_val_is_min): Likewise.
26954 (set_value_range_to_value): Likewise.
26955 (set_value_range_to_nonnegative): Likewise.
26956 (gimple_assign_nonzero_p): Likewise.
26957 (gimple_stmt_nonzero_p): Likewise.
26958 (vrp_int_const_binop): Likewise. Remove unreachable case.
26959 (adjust_range_with_scev): Adjust comments.
26960 (compare_range_with_value): Likewise.
26961 (extract_range_from_phi_node): Likewise.
26962 (test_for_singularity): Likewise.
26964 2017-05-09 Richard Biener <rguenther@suse.de>
26966 * tree-vrp.c (get_single_symbol): Add assert that we don't
26967 get overflowed constants as invariant part.
26968 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26969 checks. Use wi::cmp instead of recursing for integer constants.
26970 (compare_values): Just ignore whether we assumed undefined
26971 overflow instead of failing the compare.
26972 (extract_range_for_var_from_comparison_expr): Add comment before the
26973 TREE_NO_WARNING sets.
26974 (test_for_singularity): Likewise.
26975 (extract_range_from_comparison): Do not disable optimization
26976 when we assumed undefined overflow.
26977 (extract_range_basic): Remove init of unused var.
26979 2017-05-09 Richard Biener <rguenther@suse.de>
26981 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26982 (extract_range_from_multiplicative_op_1): Adjust.
26983 (extract_range_from_binary_expr_1): Use int_const_binop.
26985 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26988 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26989 rs6000_store_data_bypass_p in seven define_bypass directives and
26990 in several comments.
26991 * config/rs6000/rs6000-protos.h: Add prototype for
26992 rs6000_store_data_bypass_p function.
26993 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26994 function implements slightly different (rs6000-specific) semantics
26995 than store_data_bypass_p, returning false rather than aborting
26996 with assertion error when arguments do not satisfy the
26997 requirements of store data bypass.
26998 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26999 rs6000_store_data_bypass_p.
27001 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27003 * config/xtensa/xtensa-protos.h
27004 (xtensa_initial_elimination_offset): New declaration.
27005 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27006 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27007 macro definition, add case for FRAME_POINTER_REGNUM when
27008 FRAME_GROWS_DOWNWARD.
27009 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27010 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27011 xtensa_initial_elimination_offset.
27013 2017-05-08 Nathan Sidwell <nathan@acm.org>
27015 * doc/invoke.texi: Alphabetize -fdump options.
27017 2017-05-08 Martin Sebor <msebor@redhat.com>
27019 PR translation/80280
27020 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27022 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27024 * target.def (compute_frame_layout): New optional target hook.
27025 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27026 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27027 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27029 * reload1.c (verify_initial_elim_offsets): Likewise.
27030 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27031 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27032 (arm_get_frame_offsets): Split up into this ...
27033 (arm_compute_frame_layout): ... and this function.
27035 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27037 * config/aarch64/constraints.md (Usa): New constraint.
27038 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27040 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27042 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27043 with_multilib_list after it has been checked.
27045 2017-05-08 Richard Biener <rguenther@suse.de>
27047 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27048 (bitmap_set_subtract_values): Likewise.
27050 2017-05-08 Richard Biener <rguenther@suse.de>
27052 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27053 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27055 (gimple_stmt_nonzero_warnv_p): Rename to ...
27056 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27058 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27059 (extract_range_basic): Adjust, do not disable propagation on
27060 strict overflow sensitive simplification.
27061 (vrp_visit_cond_stmt): Likewise.
27063 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27065 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27066 body size unconditionally.
27068 2017-05-07 Jeff Law <law@redhat.com>
27071 2017-05-06 Jeff Law <law@redhat.com>
27072 PR tree-optimization/78496
27073 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27076 PR tree-optimization/78496
27077 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27078 (simplify_stmt_using_ranges): Call it.
27079 (vrp_dom_walker::before_dom_children): Extract equivalences
27080 from an ASSERT_EXPR with an equality comparison against a
27083 2017-05-06 Jeff Law <law@redhat.com>
27085 PR tree-optimization/78496
27086 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27089 PR tree-optimization/78496
27090 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27091 (simplify_stmt_using_ranges): Call it.
27092 (vrp_dom_walker::before_dom_children): Extract equivalences
27093 from an ASSERT_EXPR with an equality comparison against a
27096 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27098 * lra-constraints.c (lra_copy_reg_equiv): New function.
27099 (split_reg): Use it to copy equivalence information from the
27100 original register to the spill register.
27102 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27104 PR rtl-optimization/75964
27105 * simplify-rtx.c (simplify_const_relational_operation): Remove
27106 invalid handling of comparisons of integer ABS.
27108 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27110 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27111 initialize to zero.
27112 (init_regs): Remove declaration.
27113 (function_arg_advance_32): Initialize error_p as boolean variable.
27115 2017-05-05 Nathan Sidwell <nathan@acm.org>
27117 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27118 lines. Use for (;;).
27120 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27122 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27123 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27124 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27125 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27126 VF=2 that require versioning.
27128 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27130 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27133 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27135 * diagnostic.h (diagnostic_override_option_index): Convert from
27136 macro to inline function.
27138 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27140 * diagnostic.c (last_module_changed_p): New function.
27141 (set_last_module): New function.
27142 (diagnostic_report_current_module): Convert macro usage to
27143 the above functions.
27144 * diagnostic.h (diagnostic_context::last_module): Strengthen
27145 from const line_map * to const line_map_ordinary *.
27146 (diagnostic_last_module_changed): Delete macro.
27147 (diagnostic_set_last_module): Delete macro.
27149 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27151 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27152 with diagnostic_report_diagnostic.
27153 (diagnostic_n_impl_richloc): Likewise.
27154 * diagnostic.h (report_diagnostic): Delete macro.
27155 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27156 with diagnostic_report_diagnostic.
27157 * substring-locations.c (format_warning_va): Likewise.
27159 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27161 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27162 save/restor of format_spec. Move option-printing code to...
27163 (print_option_information): ...this new function, and
27164 reimplement by simply printing to the pretty_printer,
27165 rather than appending to the format string.
27167 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27169 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27170 handling logic into...
27171 (update_effective_level_from_pragmas): ...this new function.
27173 2017-05-04 Andrew Waterman <andrew@sifive.com>
27175 * config/riscv/riscv.opt (mstrict-align): New option.
27176 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27177 (SLOW_UNALIGNED_ACCESS): Define.
27178 (riscv_slow_unaligned_access): Declare.
27179 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27181 (riscv_slow_unaligned_access): New variable.
27182 (rocket_tune_info): Set slow_unaligned_access to true.
27183 (optimize_size_tune_info): Set slow_unaligned_access to false.
27184 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27185 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27186 (riscv_option_override): Set riscv_slow_unaligned_access.
27187 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27189 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27191 * config/riscv/riscv.md: Unify indentation.
27193 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27198 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27200 (extendsi<mode>2): Add support for doing sign extension via
27201 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27202 don't have ISA 3.0 instructions.
27203 (extendsi<mode>2 splitter): Likewise.
27204 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27205 generate the normal insns since SImode can now go in vector
27206 registers. Disallow the special UNSPECs needed for previous
27207 machines to hide SImode being used. Add new insns
27208 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27209 (fix_trunc<mode>si2_stfiwx): Likewise.
27210 (fix_trunc<mode>si2_internal): Likewise.
27211 (fixuns_trunc<mode>si2): Likewise.
27212 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27213 (fctiw<u>z_<mode>_smallint): Likewise.
27214 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27215 of floating point to 32-bit integer from doing a direct move to
27216 the GPR registers to do a store.
27217 (fctiwz_<mode>): Break long line.
27219 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27221 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27222 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27223 (addr_list, addr_offset_valid_p): New.
27224 (split_address_groups): Check offset validity with above function.
27225 (gt-tree-ssa-loop-ivopts.h): Include header file.
27227 2017-05-05 Nathan Sidwell <nathan@acm.org>
27229 * config.gcc (arm*-*-*): Add missing 'fi'.
27231 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27233 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27234 included in -fopt-info does not matter.
27235 * doc/optinfo.texi (-fopt-info): Fix description of default
27236 behavour. Explicitly say order of options included in -fopt-info
27239 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27241 * config.gcc: Allow combinations of aprofile and rmprofile values for
27242 --with-multilib-list.
27243 * config/arm/t-multilib: New file.
27244 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27245 variables. Remove setting of ISA and floating-point ABI in
27246 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27247 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27248 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27249 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27251 * config/arm/t-rmprofile: Likewise except for the matches changes.
27252 * doc/install.texi (--with-multilib-list): Document the combination of
27253 aprofile and rmprofile values and warn about pitfalls in doing that.
27255 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27257 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27258 (movdi_aarch64): Likewise.
27260 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27262 PR tree-optimization/80632
27263 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27265 (build_arrays): Initialize it for virtual phis.
27266 (fix_phi_nodes): Use it for virtual phis.
27268 PR tree-optimization/80558
27269 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27270 [x, y] op z into [x op, y op z] for op & or | if conditions
27273 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27274 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27277 * config/arm/arm.md (use_literal_pool): Remove.
27278 (64-bit immediate split): No longer takes cost into consideration
27279 if arm_disable_literal_pool is enabled.
27280 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27281 used when arm_disable_literal_pool is enabled.
27282 (arm_max_const_double_inline_cost): Remove use of
27283 arm_disable_literal_pool.
27284 (push_minipool_fix): Add assert.
27285 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27286 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27287 (no_literal_pool_sf_immediate): New.
27289 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27291 PR tree-optimization/80613
27292 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27293 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27295 2017-05-05 Richard Biener <rguenther@suse.de>
27297 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27299 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27301 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27302 of this flag from insn conditions due to removal from r247495.
27304 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27306 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27308 (arm_early_store_addr_dep_ptr): Likewise.
27309 * config/arm/aarch-common-protos.h
27310 (arm_early_load_addr_dep_ptr): Add prototype.
27311 (arm_early_store_addr_dep_ptr): Likewise.
27312 * config/arm/cortex-a53.md: Add new bypasses.
27314 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27316 * tree.c (next_type_uid): Change type to unsigned.
27317 (type_hash_canon): Decrement back next_type_uid if
27318 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27319 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27322 2017-05-04 Martin Sebor <msebor@redhat.com>
27324 * builtins.c: Fix a trivial typo in a comment.
27326 PR middle-end/79234
27327 * builtins.c (check_sizes): Adjust to handle reading past the end.
27328 Avoid printing excessive upper bound of ranges. Use %E to print
27329 tree nodes instead of converting them to %wu.
27330 (expand_builtin_memchr): New function.
27331 (compute_dest_size): Rename...
27332 (compute_objsize): ...to this.
27333 (expand_builtin_memcpy): Adjust.
27334 (expand_builtin_mempcpy): Adjust.
27335 (expand_builtin_strcat): Adjust.
27336 (expand_builtin_strcpy): Adjust.
27337 (check_strncat_sizes): Adjust.
27338 (expand_builtin_strncat): Adjust.
27339 (expand_builtin_strncpy): Adjust and simplify.
27340 (expand_builtin_memset): Adjust.
27341 (expand_builtin_bzero): Adjust.
27342 (expand_builtin_memcmp): Adjust.
27343 (expand_builtin): Handle memcmp.
27344 (maybe_emit_chk_warning): Check strncat just once.
27346 2017-05-04 Martin Sebor <msebor@redhat.com>
27348 PR preprocessor/79214
27349 PR middle-end/79222
27350 PR middle-end/79223
27351 * builtins.c (check_sizes): Add inlining context and issue
27352 warnings even when -Wno-system-headers is set.
27353 (check_strncat_sizes): Same.
27354 (expand_builtin_strncat): Same.
27355 (expand_builtin_memmove): New function.
27356 (expand_builtin_stpncpy): Same.
27357 (expand_builtin): Handle memmove and stpncpy.
27359 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27361 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27362 which is not used any more.
27364 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27366 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27368 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27370 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27371 (cortexa53_tunings): Likewise.
27372 (cortexa57_tunings): Likewise.
27373 (cortexa72_tunings): Likewise.
27374 (cortexa73_tunings): Likewise.
27376 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27378 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27379 Set loop alignment to 8.
27381 2017-05-04 Martin Sebor <msebor@redhat.com>
27383 PR translation/80280
27384 * builtins.c (expand_builtin_object_size): Add missing quoting to
27385 %D and like directives.
27386 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27387 (hsa_type_for_tree_type): Same.
27388 (verify_function_arguments): Same.
27389 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27390 * varasm.c (get_section): Same.
27393 2017-05-04 Martin Sebor <msebor@redhat.com>
27395 PR translation/80280
27396 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27398 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27400 * config/aarch64/aarch64.c (generic_addrcost_table):
27401 Change HI/TI mode setting.
27403 2017-05-04 Martin Jambor <mjambor@suse.cz>
27405 PR tree-optimization/80622
27406 * tree-sra.c (comes_initialized_p): New function.
27407 (build_accesses_from_assign): Only set write lazily when
27408 comes_initialized_p is false.
27409 (analyze_access_subtree): Use comes_initialized_p.
27410 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27411 instead of testing for PARM_DECL.
27413 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27415 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27416 constraint on operand 0 to allow more general addressing modes.
27417 Adjust output template.
27418 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27420 * config/aarch64/aarch64-protos.h
27421 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27422 * config/aarch64/constraints.md (Dp): New address constraint.
27423 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27426 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27428 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27429 update use of estimate_ipcp_clone_size_and_time.
27430 (estimate_local_effects): Update use of
27431 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27432 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27433 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27434 Return nonspecialized time.
27436 2017-05-04 Richard Biener <rguenther@suse.de>
27438 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27439 for the last VUSE which def dominates the PHI. Directly call
27441 (get_continuation_for_phi_1): Remove.
27443 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27445 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27446 to explain the use of truncating division. Cap the number of
27447 iterations to the maximum given by nb_iterations_upper_bound,
27450 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27452 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27453 * configure: Regenerate.
27454 * config.in: Regenerate.
27455 * config/i386/driver-mingw32.c: new file.
27456 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27457 * config.host: Link driver-mingw32.o on MinGW host.
27458 * doc/install.texi: Document new --enable-mingw-wildcard configure
27461 2017-05-04 Marek Polacek <polacek@redhat.com>
27463 PR tree-optimization/80612
27464 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27466 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27467 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27469 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27470 (movt splitter): Likewise.
27471 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27472 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27473 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27474 block for Thumb-1 with MOVT.
27475 (thumb2_legitimate_address_p): Move code block ...
27476 (can_avoid_literal_pool_for_label_p): ... into this new function.
27477 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27479 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27480 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27481 "M-profile targets with the MOVT instruction".
27483 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27485 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27486 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27487 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27489 2017-05-04 Martin Liska <mliska@suse.cz>
27491 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27492 variable cond_code.
27494 2017-05-04 Richard Biener <rguenther@suse.de>
27496 * tree.c (array_at_struct_end_p): Handle arrays at struct
27497 end with flexarrays more conservatively. Refactor and treat
27498 arrays of arrays or aggregates more strict. Fix
27499 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27500 * tree.c (array_at_struct_end_p): Adjust prototype.
27501 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27502 * gimple-fold.c (get_range_strlen): Likewise.
27503 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27505 2017-05-04 Richard Biener <rguenther@suse.de>
27507 PR tree-optimization/31130
27508 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27510 (supports_overflow_infinity): Likewise.
27511 (is_negative_overflow_infinity): Likewise.
27512 (is_positive_overflow_infinity): Likewise.
27513 (is_overflow_infinity): Likewise.
27514 (stmt_overflow_infinity): Likewise.
27515 (overflow_infinity_range_p): Likewise.
27516 (usable_range_p): Remove as always returning true.
27517 (make_overflow_infinity): Remove.
27518 (negative_overflow_infinity): Likewise.
27519 (positive_overflow_infinity): Likewise.
27520 (avoid_overflow_infinity): Likewise.
27521 (set_value_range): Adjust accordingly.
27522 (set_value_range_to_nonnegative): Likewise, remove now unused
27523 overflow_infinity arg.
27524 (vrp_operand_equal_p): Adjust.
27525 (update_value_range): Likewise.
27526 (range_int_cst_singleton_p): Likewise.
27527 (operand_less_p): Likewise.
27528 (compare_values_warnv): Likewise.
27529 (extract_range_for_var_from_comparison_expr): Likewise.
27530 (vrp_int_const_binop): Likewise.
27531 (zero_nonzero_bits_from_vr): Likewise.
27532 (extract_range_from_multiplicative_op_1): Likewise.
27533 (extract_range_from_binary_expr_1): Likewise.
27534 (extract_range_from_unary_expr): Likewise.
27535 (extract_range_from_comparison): Likewise.
27536 (extract_range_basic): Likewise.
27537 (adjust_range_with_scev): Likewise.
27538 (compare_ranges): Likewise.
27539 (compare_range_with_value): Likewise.
27540 (dump_value_range): Likewise.
27541 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27543 (simplify_cond_using_ranges): Adjust.
27545 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27547 * brig-builtins.def: Added a builtin for class_f64.
27548 * builtin-types.def: Added a builtin type needed by class_f64.
27550 2017-05-03 Jason Merrill <jason@redhat.com>
27552 * timevar.def: Add TV_CONSTEXPR.
27554 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27556 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27558 2017-05-03 Martin Jambor <mjambor@suse.cz>
27560 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27561 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27562 * ipa-cp.c (ipcp_propagate_stage): Do not call
27563 ipa_update_after_lto_read.
27564 * ipa-inline.c (ipa_inline): Likewise.
27566 2017-05-03 Martin Jambor <mjambor@suse.cz>
27568 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27569 tag. Added a default constructor and a destructor.
27570 (ipa_edge_args_sum_t): New class;
27571 (ipa_edge_args_sum): Declare.
27572 (ipa_edge_args_vector): Remove declaration.
27573 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27574 (ipa_free_edge_args_substructures): Remove declaration.
27575 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27576 (ipa_edge_args_info_available_for_edge_p): Likewise.
27577 * ipa-prop.c (ipa_edge_args_vector): Removed.
27578 (edge_removal_hook_holder): Likewise.
27579 (edge_duplication_hook_holder): Likewise.
27580 (ipa_edge_args_sum): New variable.
27581 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27582 ipa_edge_args_vector.
27583 (ipa_free_edge_args_substructures): Likewise.
27584 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27585 ipa_edge_args_vector.
27586 (ipa_edge_removal_hook): Turned into method
27587 ipa_edge_args_sum_t::remove.
27588 (ipa_edge_duplication_hook): Turned into method
27589 ipa_edge_args_sum_t::duplicate.
27590 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27591 registering edge hooks.
27592 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27593 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27594 ipa_edge_args_sum instead of ipa_edge_args_vector.
27595 * ipa-profile.c (ipa_profile): Likewise.
27597 2017-05-03 Martin Jambor <mjambor@suse.cz>
27599 * symbol-summary.h (function_summary): New method exists.
27600 (function_summary::symtab_removal): Deallocate through release.
27601 (call_summary): New class.
27602 (gt_ggc_mx): New overload.
27603 (gt_pch_nx): Likewise.
27604 (gt_pch_nx): Likewise.
27606 2017-05-03 Jeff Law <law@redhat.com>
27608 PR tree-optimization/78496
27609 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27610 from simplify_cond_using_ranges. Split off code to walk
27611 backwards through casts into ...
27612 (simplify_cond_using_ranges_2): New function.
27613 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27614 (execute_vrp): After identifying jump threads, call
27615 simplify_cond_using_ranges_2.
27617 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27620 * ipa-inline.h (inline_summary): Add ctor.
27621 (create_ggc): Do not use ggc_cleared_alloc.
27623 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27624 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27626 * gcc.c (handle_braces): Support escaping in switch matching
27628 * doc/invoke.texi (Spec Files): Document it.
27629 Remove superfluous @code markup in items.
27631 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27633 * diagnostic-show-locus.c (struct column_range): New struct.
27634 (get_affected_columns): New function.
27635 (get_printed_columns): New function.
27636 (struct correction): New struct.
27637 (correction::ensure_capacity): New function.
27638 (correction::ensure_terminated): New function.
27639 (struct line_corrections): New struct.
27640 (line_corrections::~line_corrections): New dtor.
27641 (line_corrections::add_hint): New function.
27642 (layout::print_trailing_fixits): Reimplement in terms of the new
27644 (selftest::test_overlapped_fixit_printing): New function.
27645 (selftest::diagnostic_show_locus_c_tests): Call it.
27647 2017-05-03 Nathan Sidwell <nathan@acm.org>
27649 Canonicalize canonical type hashing
27650 * tree.h (type_hash_canon_hash): Declare.
27651 * tree.c (type_hash_list, attribute_hash_list): Move into
27652 type_hash_canon_hash.
27653 (build_type_attribute_qual_variant): Break out hash code calc into
27654 type_hash_canon_hash.
27655 (type_hash_canon_hash): New. Generic type hash computation.
27656 (build_range_type_1, build_array_type_1, build_function_type,
27657 build_method_type_directly, build_offset_type, build_complex_type,
27658 make_vector_type): Call it.
27660 2017-05-03 Richard Biener <rguenther@suse.de>
27662 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27663 When all DRs have unknown misaligned do not always peel
27664 when there is a store but apply the same costing model as if
27665 there were only loads.
27667 2017-05-03 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/80492
27671 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27672 compare_base_decls returning dont-know properly.
27674 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27676 * config/arm/iterators.md (CCSI): New mode iterator.
27677 (arch): New mode attribute.
27678 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27679 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27680 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27681 code iterator for success result mode.
27682 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27683 the corresponding new insn generators.
27685 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27688 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27689 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27691 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27693 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27694 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27695 (DDR_A): Wrap DDR argument in brackets.
27696 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27697 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27698 (DDR_REVERSED_P): Likewise.
27700 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27702 PR tree-optimization/79472
27703 * tree-switch-conversion.c (struct switch_conv_info): Add
27704 contiguous_range and default_case_nonstandard fields.
27705 (collect_switch_conv_info): Compute contiguous_range and
27706 default_case_nonstandard fields, don't clear final_bb if
27707 contiguous_range and only the default case doesn't have the required
27709 (check_all_empty_except_final): Set default_case_nonstandard instead
27710 of failing if contiguous_range and the default case doesn't have empty
27712 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27713 and only the default case doesn't have the required constants. Skip
27715 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27716 if default_case_nonstandard.
27717 (build_constructors): Build constant 1 just once. Assert that default
27718 values aren't inserted in between cases if contiguous_range. Skip
27720 (build_arrays): Skip virtual phis.
27721 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27722 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27723 Handle virtual phis.
27724 (gen_inbound_check): Handle default_case_nonstandard case.
27725 (process_switch): Adjust check_final_bb caller. Call
27726 gather_default_values with the first non-default case instead of
27727 default case if default_case_nonstandard.
27729 2017-05-02 Nathan Sidwell <nathan@acm.org>
27731 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27732 check. Fix formatting.
27734 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27736 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27737 errors when comparing specialized and unspecialized times.
27739 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27741 * diagnostic-show-locus.c
27742 (layout::should_print_annotation_line_p): Make private.
27743 (layout::print_annotation_line): Make private.
27744 (layout::annotation_line_showed_range_p): Make private.
27745 (layout::show_ruler): Make private.
27746 (layout::print_source_line): Make private. Pass in line and
27747 line_width, rather than calling location_get_source_line. Drop
27749 (layout::print_leading_fixits): New method.
27750 (layout::print_any_fixits): Rename to...
27751 (layout::print_trailing_fixits): ...this, and make private.
27752 Don't print newline fixits.
27753 (diagnostic_show_locus): Move logic for printing one row into...
27754 (layout::print_line): ...this new function. Move the
27755 location_get_source_line call and error-handling from
27756 print_source_line to here. Call print_leading_fixits, and rename
27757 print_any_fixits to print_trailing_fixits.
27758 (selftest::test_fixit_insert_containing_newline): Update now that
27759 newlines are partially supported.
27760 (selftest::test_fixit_insert_containing_newline_2): New test.
27761 (selftest::test_fixit_replace_containing_newline): Update comments.
27762 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27763 * edit-context.c (class added_line): New class.
27764 (class edited_line): Describe newline handling in comment.
27765 (edited_line::actually_edited_p): New method.
27766 (edited_line::print_content): Delete redundant decl.
27767 (edited_line::m_predecessors): New field.
27768 (edited_file::print_content): Call edited_line::print_content.
27769 (edited_file::print_diff): Update to support newlines.
27770 (edited_file::print_diff_hunk): Likewise.
27771 (edited_file::print_run_of_changed_lines): New function.
27772 (edited_file::print_diff_line): Convert to...
27773 (print_diff_line): ...this.
27774 (edited_file::get_effective_line_count): New function.
27775 (edited_line::edited_line): Initialize new field m_predecessors.
27776 (edited_line::~edited_line): Clean up m_predecessors.
27777 (edited_line::apply_fixit): Handle newlines.
27778 (edited_line::get_effective_line_count): New function.
27779 (edited_line::print_content): New function.
27780 (edited_line::print_diff_lines): New function.
27781 (selftest::test_applying_fixits_insert_containing_newline): New
27783 (selftest::test_applying_fixits_replace_containing_newline): New
27785 (selftest::insert_line): New function.
27786 (selftest::test_applying_fixits_multiple_lines): Add example of
27788 (selftest::edit_context_c_tests): Call the new tests.
27790 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27792 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27793 parameter cand. Update dump information.
27794 (get_computation_cost): Update uses.
27796 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27798 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27799 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27800 (get_computation_at, rewrite_use_address): Update use of
27801 get_computation_aff.
27803 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27805 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27806 (get_computation): Delete.
27807 (get_computation_cost): Implement like get_computation_cost_at.
27808 Use get_computation_at.
27809 (get_computation_cost_at): Delete.
27810 (rewrite_use_nonlinear_expr): Use get_computation_at.
27811 (rewrite_use_compare, remove_unused_ivs): Ditto.
27813 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27815 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27817 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27819 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27820 (ivopts_global_cost_for_size): Rename parameter and update uses.
27821 (iv_ca_recount_cost): Update uses.
27822 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27823 candidates seperately in n_invs and n_cands.
27824 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27826 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27828 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27829 (find_inv_vars_cb): New.
27830 (find_depends): Renamed to ...
27831 (find_inv_vars): ... this.
27832 (add_candidate_1, force_var_cost): Call find_inv_vars.
27833 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27837 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27838 inv_vars. Add inv_exprs.
27839 (struct iv_cand): Rename depends_on to inv_vars.
27840 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27841 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27842 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27843 (dump_cand): Dump inv_vars.
27844 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27845 (record_invariant, find_depends, add_candidate_1): Ditto.
27846 (set_group_iv_cost, force_var_cost): Ditto.
27847 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27848 (get_computation_cost_at, get_computation_cost): Ditto.
27849 (determine_group_iv_cost_generic): Ditto.
27850 (determine_group_iv_cost_address): Ditto.
27851 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27852 (determine_group_iv_costs): Ditto.
27853 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27854 (iv_ca_set_remove_invariants): Renamed to ...
27855 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27856 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27857 (iv_ca_set_add_invariants): Renamed to ...
27858 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27859 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27860 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27861 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27862 (create_new_ivs): Remove useless dump.
27864 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27866 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27868 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27869 (iv_ca_set_no_cp, create_new_iv): Ditto.
27871 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27873 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27875 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27877 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27878 function tree_check2.
27880 2017-05-02 Martin Liska <mliska@suse.cz>
27882 * doc/gcov.texi: Add missing preposition.
27883 * gcov.c (function_info::function_info): Properly fill up
27884 all member variables.
27886 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27888 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27890 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27892 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27894 2017-05-02 Martin Liska <mliska@suse.cz>
27897 * lto-streamer-in.c (lto_read_tree_1): Remove
27898 LTO_STREAMER_DEBUG.
27899 * lto-streamer.c (struct tree_hash_entry): Likewise.
27900 (struct tree_entry_hasher): Likewise.
27901 (tree_entry_hasher::hash): Likewise.
27902 (tree_entry_hasher::equal): Likewise.
27903 (lto_streamer_init): Likewise.
27904 (lto_orig_address_map): Likewise.
27905 (lto_orig_address_get): Likewise.
27906 (lto_orig_address_remove): Likewise.
27907 * lto-streamer.h: Likewise.
27908 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27909 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27911 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27913 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27914 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27915 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27916 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27917 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27918 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27919 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27920 (mm_maskz_sub_ss): New intrinsics.
27921 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27922 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27923 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27924 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27925 (__builtin_ia32_subss_mask_round): New builtins.
27926 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27927 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27928 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27930 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27931 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27933 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27936 2017-05-02 Martin Jambor <mjambor@suse.cz>
27938 PR tree-optimization/78687
27939 * tree-sra.c (access): New field parent.
27940 (process_subtree_disqualification): New function.
27941 (disqualify_candidate): Call it.
27942 (build_accesses_from_assign): Reset write flag if creating an
27944 (build_access_subtree): Fill in parent field and also prpagate
27945 down grp_write flag.
27946 (create_artificial_child_access): New parameter set_grp_write, set
27947 grp_write to its value.
27948 (propagate_subaccesses_across_link): Also propagate grp_write flag
27950 (propagate_all_subaccesses): Push the closest parent back to work
27951 queue if add_access_to_work_queue returned true.
27953 2017-05-02 Richard Biener <rguenther@suse.de>
27955 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27956 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27957 -fstrict-overflow documentation.
27958 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27959 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27960 flag_strict_overflow.
27961 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27962 * lto-opts.c (lto_write_options): Do not stream it.
27963 * lto-wrapper.c (merge_and_complain): Do not handle it.
27964 * opts.c (default_options_table): Do not set -fstrict-overflow.
27965 (finish_options): Likewise do not clear it when sanitizing.
27966 * simplify-rtx.c (simplify_const_relational_operation): Do not
27967 test flag_strict_overflow.
27969 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27971 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27972 using enabled attribute.
27973 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27974 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27975 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27976 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27977 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27978 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27979 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27980 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27981 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27982 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27984 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27986 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27988 2017-05-02 Richard Biener <rguenther@suse.de>
27990 PR tree-optimization/80591
27992 2017-04-10 Richard Biener <rguenther@suse.de>
27994 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27997 2017-05-02 Richard Biener <rguenther@suse.de>
27999 PR tree-optimization/80549
28000 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28001 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28002 headers if they do not have a preheader.
28004 2017-05-02 Martin Liska <mliska@suse.cz>
28007 * common.opt: Fix typo.
28008 * doc/invoke.texi: Likewise.
28010 2017-05-01 Jan Beulich <jbeulich@suse.com>
28012 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28013 swapping, add (x,x,m,x,n) alternative.
28015 2017-05-01 Nathan Sidwell <nathan@acm.org>
28017 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28018 unnecessary unadjusted_alignment check.
28020 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28023 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28025 * gimplify.c (gimplify_cilk_detach): New function.
28026 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28027 * tree-core.h: Document EXPR_CILK_SPAWN.
28028 * tree.h (EXPR_CILK_SPAWN): Define.
28030 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28032 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28033 to use new fixit_hint representation, using the "replace" logic.
28034 (get_line_span_for_fixit_hint): Likewise.
28035 (layout::print_any_fixits): Likewise.
28036 (selftest::test_one_liner_many_fixits): Rename to...
28037 (selftest::test_one_liner_many_fixits_1): ...this, and update
28038 comment and expected output to reflect that the multiple fix-it
28039 hints are now consolidated into one insertion.
28040 (selftest::test_one_liner_many_fixits_2): New test.
28041 (selftest::test_diagnostic_show_locus_one_liner): Update for
28043 (selftest::test_fixit_consolidation): Update for fix-it API
28045 * diagnostic.c (print_parseable_fixits): Likewise.
28046 * edit-context.c (edited_line::m_line_events): Convert from
28047 auto_vec <line_event *> to auto_vec <line_event>.
28048 (class line_event): Convert from abstract base class to a concrete
28049 class, taking over the role of replace_event.
28050 (class insert_event): Delete.
28051 (class replace_event): Rename to class line_event. Convert to
28053 (edit_context::add_fixits): Reimplement.
28054 (edit_context::apply_insert): Delete.
28055 (edit_context::apply_replace): Rename to...
28056 (edit_context::apply_fixit): ...this. Convert to half-open range.
28057 (edited_file::apply_insert): Delete.
28058 (edited_file::apply_replace): Rename to...
28059 (edited_file::apply_fixit): ...this.
28060 (edited_line::~edited_line): Drop deletion of events.
28061 (edited_line::apply_insert): Delete.
28062 (edited_line::apply_replace): Rename to...
28063 (edited_line::apply_fixit): ...this. Convert to half-open range.
28064 Update for change to type of m_line_events.
28065 * edit-context.h (edit_context::apply_insert): Delete.
28066 (edit_context::apply_replace): Rename to...
28067 (edit_context::apply_fixit): ...this.
28069 2017-05-01 Martin Sebor <msebor@redhat.com>
28071 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28074 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28077 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28078 __get_cpuid_max returns 0.
28079 (__get_cpuid_count): Ditto.
28081 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28083 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28084 replacement expression is another instance of one of its arguments.
28086 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28089 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28090 check for stack push/pop autoinc.
28091 * config/i386/i386.c (ix86_agi_dependent): Return false
28092 if the only reason why modified_in_p returned true is that
28093 addr is SP based and set_insn is a push or pop.
28095 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28097 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28100 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28103 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28104 (account_size_time): Use two predicates - exec_pred and
28106 (evaluate_conditions_for_known_args): Compute both clause and
28108 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28109 (inline_summary_t::duplicate): Update.
28110 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28112 (compute_inline_parameters): Likewise.
28113 (estimate_edge_size_and_time): Update caluclation of time.
28114 (estimate_node_size_and_time): Compute both time and nonspecialized
28116 (estimate_ipcp_clone_size_and_time): Update.
28117 (inline_merge_summary): Update.
28118 (do_estimate_edge_time): Update.
28119 (do_estimate_edge_size): Update.
28120 (do_estimate_edge_hints): Update.
28121 (inline_read_section, inline_write_summary): Stream both new predicates.
28122 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28124 (compute_inlined_call_time): Cleanup.
28125 (big_speedup_p): Update.
28126 (edge_badness): Update.
28127 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28128 (size_time_entry): Replace predicate by exec_predicate and
28129 nonconst_predicate.
28130 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28131 (estimate_edge_time): Return also nonspec_time.
28132 (reset_edge_growth_cache): Update.
28134 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28136 PR rtl-optimization/80491
28137 * ifcvt.c (noce_process_if_block): When looking for x setter
28138 with missing else_bb, don't check only the insn right before
28139 cond_earliest, but look for the last insn that x is modified in
28140 within the same bb.
28142 PR rtl-optimization/80491
28143 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28145 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28147 PR tree-optimization/80487
28148 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28150 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28152 PR tree-optimization/79697
28153 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28154 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28155 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28157 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28158 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28160 2017-04-28 Martin Sebor <msebor@redhat.com>
28162 PR tree-optimization/80523
28163 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28164 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28166 (maybe_warn, format_directive, parse_directive): Use new functions.
28167 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28169 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28171 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28173 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28175 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28176 target_header_dir): Set correctly.
28177 * configure: Regenerated.
28178 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28179 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28180 instead of SYSTEM_HEADER_DIR.
28182 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28184 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28185 (estimate_local_effects): Likewise.
28186 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28187 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28188 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28189 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28190 do_estimate_edge_time, estimate_edge_time): Likewise.
28191 * ipa-inline-analysis.c (estimate_node_size_and_time,
28192 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28193 (estimate_time_after_inlining): Remove.
28195 2017-04-28 Martin Liska <mliska@suse.cz>
28197 * doc/gcov.texi: Enhance documentation of gcov.
28199 2017-04-28 Martin Liska <mliska@suse.cz>
28201 * doc/gcov.texi: Sort options in alphabetic order.
28202 * doc/gcov-dump.texi: Likewise.
28203 * doc/gcov-tool.texi: Likewise.
28204 * gcov.c (print_usage): Likewise.
28205 * gcov-dump.c (print_usage): Likewise.
28206 * gcov-tool.c (print_merge_usage_message): Likewise.
28207 (print_rewrite_usage_message): Likewise.
28208 (print_overlap_usage_message): Likewise.
28210 2017-04-28 Martin Liska <mliska@suse.cz>
28212 PR gcov-profile/53915
28213 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28215 2017-04-28 Martin Liska <mliska@suse.cz>
28217 PR gcov-profile/79891
28218 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28219 is marked by compiler as living on a line.
28220 (get_cycles_count): Remove usage of the union.
28221 (output_intermediate_file): Likewise.
28222 (find_source): Fix GNU coding style.
28223 (accumulate_line_counts): Remove old non-all block mode.
28224 (output_lines): Remove usage of the union.
28225 * profile.c (output_location): Include all BBs, even if
28226 belonging to a same line (and file) as a previous BB.
28228 2017-04-28 Martin Liska <mliska@suse.cz>
28230 * gcov.c (process_args): Handle new argument 'w'.
28231 (read_graph_file): Assign ID to BBs.
28232 (output_branch_count): Display BB # if verbose flag is set.
28233 (output_lines): Likewise for arcs.
28234 (print_usage): Add '--verbose' option help.
28235 * doc/gcov.texi: Document --verbose (-w) option.
28237 2017-04-28 Martin Liska <mliska@suse.cz>
28239 * gcov.c (struct block_location_info): New struct.
28240 (process_file): Fill up the new structure.
28241 (read_graph_file): Replace usage of encoding by the newly added
28243 (add_line_counts): Likewise.
28244 (accumulate_line_counts): Remove usage of the union.
28245 (function_info::function_info): New function.
28246 (function_info::~function_info): Likewise.
28247 (process_file): Call delete instead of release_function.
28248 (release_function): Release the function.
28249 (release_structures): Call delete instead of release_function.
28250 (solve_flow_graph): Replace usage of num_blocks.
28251 (find_exception_blocks): Likewise.
28252 (output_lines): Fix GNU coding style.
28254 2017-04-28 Martin Liska <mliska@suse.cz>
28257 * coverage.c (coverage_remove_note_file): New function.
28258 * coverage.h: Declare the function.
28259 * toplev.c (finalize): Clean if an error has been seen.
28261 2017-04-28 Martin Liska <mliska@suse.cz>
28263 PR gcov-profile/80031
28264 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28265 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28266 * gcov.c (read_graph_file): Read just number of blocks.
28267 * profile.c (branch_prob): Do not stream 0 flags per a basic
28270 2017-04-28 Martin Liska <mliska@suse.cz>
28272 * gcov-dump.c (tag_*): Add new argument to declarations.
28273 (dump_gcov_file): Likewise.
28274 (tag_blocks): Add and use new argument depth.
28275 (tag_arcs): Likewise.
28276 (tag_lines): Likewise.
28277 (tag_counters): Likewise.
28278 (tag_summary): Likewise.
28279 (dump_working_sets): Use depth to do a proper indentation.
28281 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28284 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28285 * symtab.c (symtab_node::debug_symtab): Move definition here.
28287 2017-04-28 Richard Biener <rguenther@suse.de>
28289 * lto-streamer.h (LTO_major_version): Bump to 7.
28291 2017-04-28 Richard Biener <rguenther@suse.de>
28293 * tree-vrp.c (assert_info): New struct.
28294 (add_assert_info): New helper.
28295 (register_edge_assert_for_2): Refactor to add asserts to a vector
28297 (register_edge_assert_for_1): Likewise.
28298 (register_edge_assert_for): Likewise.
28299 (finish_register_edge_assert_for): New helper actually registering
28300 asserts where live on edge.
28301 (find_conditional_asserts): Adjust.
28302 (find_switch_asserts): Likewise.
28303 (evrp_dom_walker::try_find_new_range): Generalize.
28304 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28306 2017-04-27 Marek Polacek <polacek@redhat.com>
28309 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28310 arg10 and arg11 to itype.
28312 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28314 * doc/extend.texi (Object Size Checking): Improve grammar.
28316 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28319 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28320 that the logic for permitting reciprocal estimates matches that
28323 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28326 * tree.c (type_cache_hasher::equal): Only compare
28327 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28328 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28329 non-aggregate element types.
28330 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28331 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28333 2017-04-27 Richard Biener <rguenther@suse.de>
28335 PR middle-end/80533
28336 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28337 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28338 keeping a reference to a trailing array.
28340 2017-04-27 Richard Biener <rguenther@suse.de>
28342 PR middle-end/80539
28343 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28344 being in loop-closed SSA form conservatively.
28345 (chrec_fold_multiply_poly_poly): Likewise.
28347 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28349 PR middle-end/79665
28350 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28351 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28353 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28356 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28357 (aarch64_function_arg_alignment): Return unsigned int again, but still
28358 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28359 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28360 Don't emit -Wpsabi note.
28361 (aarch64_function_arg_boundary): Likewise.
28362 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28365 2017-04-26 Nathan Sidwell <nathan@acm.org>
28367 * tree.h (crc32_unsigned_n): Declare.
28368 (crc32_unsigned, crc32_unsigned): Make inline.
28369 * tree.c (crc32_unsigned_bits): Replace with ...
28370 (crc32_unsigned_n): ... this.
28371 (crc32_unsigned, crc32_byte): Remove.
28372 (crc32_string): Remove unnecessary braces.
28374 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28376 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28377 * ipa-inline-analysis.c (MAX_TIME): Remove.
28378 (account_size_time): Use sreal for time.
28379 (dump_inline_summary): Update.
28380 (estimate_function_body_sizes): Update.
28381 (estimate_edge_size_and_time): Update.
28382 (estimate_calls_size_and_time): Update.
28383 (estimate_node_size_and_time): Update.
28384 (inline_merge_summary): Update.
28385 (inline_update_overall_summary): Update.
28386 (estimate_time_after_inlining): Update.
28387 (inline_read_section): Update.
28388 (inline_write_summary): Update.
28389 * ipa-inline.c (compute_uninlined_call_time): Update.
28390 (compute_inlined_call_time): Update.
28391 (recursive_inlining): Update.
28392 (inline_small_functions): Update.
28393 (dump_overall_stats): Update.
28394 * ipa-inline.h: Include sreal.h.
28395 (size_time_entry): Turn time to sreal.
28396 (inline_summary): Turn self_time nad time to sreal.
28398 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28400 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28402 (sreal::stream_out, sreal::stream_in): New.
28403 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28405 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28407 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28410 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28413 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28415 (dimode_scalar_chain::compute_convert_gain): Ditto.
28416 (dimode_scalar_chain::make_vector_copies): Ditto.
28417 (dimode_scalar_chain::convert_reg): Ditto.
28418 (dimode_scalar_chain::convert_insn): Ditto.
28419 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28420 (VI248_AVX512BW_1): New mode iterator.
28421 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28422 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28425 2017-04-25 Martin Sebor <msebor@redhat.com>
28427 PR tree-optimization/80497
28428 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28429 constants are representable in HOST_WIDE_INT.
28430 (parse_directive): Ditto.
28432 2017-04-25 Martin Sebor <msebor@redhat.com>
28435 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28436 (new_zero_array): Adjust signature.
28437 (dom_info::dom_init): Used unsigned rather that size_t.
28438 (dom_info::dom_info): Same.
28440 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28441 Jakub Jelinek <jakub@redhat.com>
28444 * config/arm/arm.c: Include gimple.h.
28445 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28446 returns negative, increment ncrn only if it returned positive.
28447 (arm_needs_doubleword_align): Return int instead of bool,
28448 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28449 members, but if there is any such non-FIELD_DECL
28450 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28451 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28452 returns negative, increment nregs only if it returned positive.
28453 (arm_setup_incoming_varargs): Likewise.
28454 (arm_function_arg_boundary): Emit -Wpsabi note if
28455 arm_needs_doubleword_align returns negative, return
28456 DOUBLEWORD_ALIGNMENT only if it returned positive.
28458 2017-04-25 Marek Polacek <polacek@redhat.com>
28461 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28462 first argument to type.
28464 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28467 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28468 type checks to test for compatibility instead of equality.
28470 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28471 Jakub Jelinek <jakub@redhat.com>
28474 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28476 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28477 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28478 the alignment computation, but return their maximum in warn_alignment.
28479 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28480 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28482 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28483 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28486 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28488 * config/arc/simdext.md (dmpyh): Fix typo.
28490 2017-04-25 Richard Biener <rguenther@suse.de>
28492 PR tree-optimization/80492
28493 * alias.c (compare_base_decls): Handle registers with asm
28494 specification conservatively.
28495 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28496 compare_base_decls returning dont-know properly.
28498 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28500 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28501 (legitimate_offset_address_p): New function.
28502 (arc_legitimate_address_p): Use above function.
28504 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28506 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28508 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28510 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28511 ACCH registers whenever they are available.
28513 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28515 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28516 double regs fix when not used.
28518 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28520 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28522 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28523 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28525 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28527 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28528 when emitting short ADD instructions.
28530 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28532 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28534 (cmpsi_cc_c_insn): Likewise.
28535 (cbranchsi4_scratch): Compute proper instruction length using
28536 compact_hreg_operand.
28537 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28539 2017-04-25 Richard Biener <rguenther@suse.de>
28541 PR middle-end/80509
28542 * passes.c (pass_manager::pass_manager): Initialize
28543 m_name_to_pass_map.
28545 2017-04-25 Richard Biener <rguenther@suse.de>
28547 PR tree-optimization/79201
28548 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28550 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28553 * config/s390/vector.md: Split MEM->GPR vector moves for
28554 non-s_operand addresses.
28556 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28559 * config/s390/predicates.md (reload_const_wide_int_operand): New
28561 * config/s390/s390.md ("movti"): Remove d/P alternative.
28562 ("movti_bigconst"): New pattern definition.
28564 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28567 * s390-protos.h (s390_expand_cs_hqi): Removed.
28568 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28569 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28570 modes as well as CCZ1mode and CCZmode.
28571 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28572 signature of s390_emit_compare_and_swap.
28573 (s390_expand_cs_hqi): Likewise, make static.
28574 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28575 compare-and-swap, in some cases.
28576 (s390_expand_cs): Wrapper function.
28577 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28579 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28580 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28581 patterns for small and large integers. Forbid symref memory operands.
28582 Move expander to s390.c. Require cc register.
28583 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28584 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28585 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28586 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28587 symref memory operands. Remove CC mode and call s390_match_ccmode
28589 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28591 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28593 * config/s390/s390.md (define_peephole2): New peephole to help
28594 combining the load-and-test pattern with volatile memory.
28596 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28598 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28599 with CCZmode for TARGET_Z196.
28601 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28603 PR rtl-optimization/80501
28604 * combine.c (make_compound_operation_int): Set subreg_code to SET
28605 even for AND with mask of the sign bit of mode.
28607 PR rtl-optimization/80500
28608 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28609 sum's initial value.
28611 2017-04-25 Julian Brown <julian@codesourcery.com>
28612 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28614 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28616 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28618 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28620 2017-04-25 Julian Brown <julian@codesourcery.com>
28621 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28623 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28624 (thunderx2t99_sha): New Reservation.
28626 2017-04-25 Julian Brown <julian@codesourcery.com>
28627 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28629 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28630 type for 1-element load.
28632 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28634 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28636 2017-04-24 Martin Jambor <mjambor@suse.cz>
28638 PR tree-optimization/80293
28639 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28640 char arrays not totally scalarizable if it is false.
28641 (analyze_all_variable_accesses): Pass correct value in the new
28642 parameter. Add a statistics counter.
28644 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28646 PR middle-end/79931
28647 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28649 2017-04-24 Richard Biener <rguenther@suse.de>
28651 PR tree-optimization/80494
28652 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28653 out for complex types.
28655 2017-04-24 Richard Biener <rguenther@suse.de>
28657 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28658 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28659 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28660 (DFS): Adjust and never fail.
28661 (sccvn_dom_walker::fail): Remove.
28662 (sccvn_dom_walker::before_dom_children): Adjust.
28663 (run_scc_vn): Likewise and never fail.
28664 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28665 (pass_fre::execute): Likewise.
28667 2017-04-24 Richard Biener <rguenther@suse.de>
28669 PR tree-optimization/79725
28670 * tree-ssa-sink.c (statement_sink_location): Return whether
28671 failure reason was zero uses. Move that check later.
28672 (sink_code_in_bb): Deal with zero uses by removing the stmt
28675 2017-04-24 Richard Biener <rguenther@suse.de>
28678 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28679 pointer-based references.
28681 2017-04-24 Richard Biener <rguenther@suse.de>
28684 * pass_manager.h (pass_manager::operator new): Remove.
28685 (pass_manager::operator delete): Likewise.
28686 * passes.c (pass_manager::operator new): Remove.
28687 (pass_manager::operator delete): Likewise.
28688 (pass_manager::pass_manager): Zero individual pass members.
28690 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28693 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28694 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28695 Check "XEXP (src, 1)" operand here.
28696 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28697 Check "XEXP (src, 1)" operand here.
28698 (dimode_scalar_chain::make_vector_copies): Detect count register
28699 of a shift instruction. Zero extend count register from QImode
28700 to DImode to satisfy vector shift pattern count operand predicate.
28701 Substitute vector shift count operand with a DImode copy.
28702 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28705 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28707 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28708 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28709 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28710 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28711 (UNSPEC_NOREX_MEM): Remove definition.
28713 2017-04-21 Richard Biener <rguenther@suse.de>
28715 PR tree-optimization/79547
28716 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28717 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28718 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28719 without any constraints.
28721 2017-04-21 Richard Biener <rguenther@suse.de>
28723 PR tree-optimization/78847
28724 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28726 2017-04-21 Richard Biener <rguenther@suse.de>
28728 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28729 (build_distinct_type_copy): Likewise.
28730 (build_variant_type_copy): Likewise.
28731 * tree.c (build_qualified_type): Pass down mem-stat info.
28732 (build_distinct_type_copy): Likewise.
28733 (build_variant_type_copy): Likewise.
28735 2017-04-21 Richard Biener <rguenther@suse.de>
28737 PR tree-optimization/80237
28738 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28740 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28741 for a simplified result.
28743 2016-04-21 Richard Biener <rguenther@suse.de>
28745 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28746 sth as strict as a simple_iv but a chrec without symbols and an
28747 operand defined in the loop we are peeling (and not some subloop).
28748 (propagate_constants_for_unrolling): Propagate all constants.
28750 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28753 * config/i386/i386.c (print_reg): Remove assert for disalowed
28754 regno values, call output_operand_lossage instead.
28756 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28759 * config/i386/constraints.md (Yc): New register constraint.
28760 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28761 Use Yc constraint for alternative 2 of operand 0. Remove
28762 preferred_for_speed attribute.
28764 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28766 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28767 lastprivate clauses in SIMT case.
28769 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28771 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28773 2017-04-20 Richard Biener <rguenther@suse.de>
28775 PR tree-optimization/57796
28776 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28777 as N scalar stores.
28778 (vect_model_load_cost): Cost gathers as N scalar loads.
28780 2017-04-20 Richard Biener <rguenther@suse.de>
28782 * ggc-page.c (ggc_allocated_p): Rename to ...
28783 (safe_lookup_page_table_entry): ... this and return the lookup
28785 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28787 2017-04-20 Richard Biener <rguenther@suse.de>
28789 PR tree-optimization/80453
28790 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28791 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28792 from the conditions.
28793 (vn_phi_eq): Pass them down.
28794 (vn_phi_lookup): Record them.
28795 (vn_phi_insert): Likewise.
28797 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28799 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28800 uninitialized variable warning to avoid buffer overrun.
28802 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28805 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28806 is suppressed for '{ 0 }' in C.
28808 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28810 * BASE-VER: Set to 8.0.0.
28812 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28814 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28815 priority .init_array and .fini_array section with SECTION_NOTYPE
28818 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28820 PR middle-end/80423
28821 * tree.h (build_array_type): Add typeless_storage default argument.
28822 * tree.c (type_cache_hasher::equal): Also compare
28823 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28824 (build_array_type): Add typeless_storage argument, set
28825 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28827 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28828 (build_array_type): Likewise. Add typeless_storage argument.
28830 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28831 Jakub Jelinek <jakub@redhat.com>
28833 PR tree-optimization/80426
28834 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28835 operation on symbolic operands, also compute the overflow for the
28836 invariant part when the operation degenerates into a negation.
28838 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28841 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28842 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28845 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28847 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28850 * config/avr/avr.c (tree.h): Include it.
28851 (cgraph.h): Include it.
28852 (avr_encode_section_info): Don't warn for uninitialized progmem
28853 variable if it's just an alias.
28855 2017-04-19 Richard Biener <rguenther@suse.de>
28858 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28859 when needed by AutoPGO.
28861 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28864 * doc/lto.texi: Remove an extra 'that'.
28866 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28868 PR rtl-optimization/80429
28869 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28870 are only used in debug insns.
28872 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28873 Vladimir Makarov <vmakarov@redhat.com>
28875 * config/sparc/predicates.md (input_operand): Add comment. Return
28876 true for any memory operand when LRA is in progress.
28877 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28879 2017-04-18 Jeff Law <law@redhat.com>
28882 * mips.md ({return,simple_return}_internal): Do not overwrite
28885 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28887 PR tree-optimization/80443
28888 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28889 instead of adding 1, subtract -1 and similarly instead of subtracting
28892 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28894 PR rtl-optimization/80357
28895 * haifa-sched.c (tmp_bitmap): New variable.
28896 (model_recompute): Handle duplicate use records.
28897 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28898 (free_global_sched_pressure_data): Free it.
28900 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28903 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28904 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28905 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28906 instead of SYSTEM_HEADER_DIR.
28908 2017-04-18 Jeff Law <law@redhat.com>
28910 PR middle-end/80422
28911 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28912 predecessors after walking up the insn chain.
28914 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28917 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28918 sizetype type into debug info.
28920 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28923 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28924 unneeded test for TARGET_UPPER_REGS_SF.
28925 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28927 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28930 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28931 instead of gsi_after_labels.
28933 2017-04-18 Jeff Law <law@redhat.com>
28935 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28939 2017-04-13 Jeff Law <law@redhat.com>
28940 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28941 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28943 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28946 * config/avr/avr.c (intl.h): Include it.
28947 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28949 2017-04-18 Martin Liska <mliska@suse.cz>
28951 PR gcov-profile/78783
28952 * gcov-tool.c (gcov_output_files): Validate that destination
28953 file is either removed by the tool or by a user.
28955 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28956 Guy Benyei <guybe@mellanox.com>
28958 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28959 block, and do not negate it, the stored id is already negative.
28961 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28963 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28965 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28968 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28969 masks of options that should be turned off if the VSX vector
28970 options are turned off.
28971 (OTHER_P8_VECTOR_MASKS): Likewise.
28972 (OTHER_VSX_VECTOR_MASKS): Likewise.
28973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28974 rs6000_disable_incompatible_switches to validate no type switches
28976 (rs6000_incompatible_switch): New function to disallow turning on
28977 other vector options if -mno-vsx, -mno-power8-vector, or
28978 -mno-power9-vector are specified.
28980 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28982 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28984 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28986 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28987 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28988 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28989 (ARG_POINTER_CFA_OFFSET): Likewise.
28991 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28993 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28994 conditions to take advantage of various optimizations.
28996 2017-04-13 Jeff Law <law@redhat.com>
28998 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28999 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29000 (zero_extendsidi2_dext): Likewise.
29002 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29005 * fold-const.c (fold_ternary_loc): Revert
29006 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29009 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29011 PR rtl-optimization/80343
29012 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29013 new scratch pseudo.
29015 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29018 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29019 to ubsan_encode_value.
29021 2017-04-13 Jeff Law <law@redhat.com>
29023 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29024 appearing in DEBUG_INSNs.
29026 2017-04-13 Martin Liska <mliska@suse.cz>
29028 PR gcov-profile/80413
29029 * gcov-io.c (gcov_write_string): Copy to buffer just when
29030 allocated size is greater than zero.
29032 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29035 * dwarf2out.c (decls_for_scope): Ignore declarations of
29036 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29038 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29041 * ipa-visibility.c (non_local_p): Fix typos.
29042 (localize_node): When localizing symbol in same comdat group,
29043 dissolve the group only when we know external symbols are going
29045 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29047 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29049 PR tree-optimization/79390
29050 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29051 order does not result in usable sequence, retry with reversed operand
29057 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29058 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29059 op0 instead of fold_convert_loc (loc, type, arg0).
29061 2017-04-12 Jeff Law <law@redhat.com>
29063 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29064 has a delay slot in the generated code.
29066 * config/cris/cris.md (cris_preferred_reload_class): Return
29067 GENNONACR_REGS rather than GENERAL_REGS.
29069 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29072 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29073 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29074 signedness of the result type.
29076 2017-04-12 Richard Biener <rguenther@suse.de>
29077 Jeff Law <law@redhat.com>
29079 PR tree-optimization/80359
29080 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29081 trim stores to TARGET_MEM_REFs.
29083 2017-04-12 Richard Biener <rguenther@suse.de>
29085 PR tree-optimization/79390
29086 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29087 threading case even more.
29089 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29092 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29093 for quad_address_p for TImode, instead of just not indexed_address.
29095 2017-04-12 Richard Biener <rguenther@suse.de>
29096 Bernd Edlinger <bernd.edlinger@hotmail.de>
29098 PR middle-end/79671
29099 * alias.c (component_uses_parent_alias_set_from): Handle
29100 TYPE_TYPELESS_STORAGE.
29101 (get_alias_set): Likewise.
29102 * tree-core.h (tree_type_common): Add typeless_storage flag.
29103 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29104 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29105 for types containing members with TYPE_TYPELESS_STORAGE.
29106 (place_field): Likewise.
29107 (layout_type): Likewise for ARRAY_TYPE.
29108 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29109 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29110 TYPE_TYPELESS_STORAGE.
29111 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29113 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29116 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29117 first argument to type.
29119 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29123 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29124 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29125 (rs6000_expand_binop_builtin): Likewise.
29126 (rs6000_expand_ternop_builtin): Likewise; also add missing
29127 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29128 vshasigma built-ins.
29129 * doc/extend.texi: Document that vec_xxpermdi's third argument
29130 must be a constant.
29132 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29134 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29135 Use shift_const cost parameter when calculating gain of STV shifts.
29137 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29139 PR rtl-optimization/70478
29140 * lra-constraints.c (process_alt_operands): Check memory for
29141 disfavoring memory insn operand.
29143 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29145 PR middle-end/80100
29146 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29147 left shift in unsigned HOST_WIDE_INT type.
29149 PR rtl-optimization/80385
29150 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29151 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29154 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29155 if they have any depend clauses.
29157 2017-04-11 Martin Liska <mliska@suse.cz>
29160 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29161 * ipa-split.c (split_function): Create a local comdat symbol
29162 if caller is in a comdat group.
29164 2017-04-11 Martin Liska <mliska@suse.cz>
29167 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29170 2017-04-11 Martin Sebor <msebor@redhat.com>
29172 PR middle-end/80364
29173 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29174 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29176 (directive::set_width, directive::set_precision, format_character):
29178 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29181 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29184 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29185 conflict, set target->arch_name instead of target->cpu_name.
29187 2017-04-11 Richard Biener <rguenther@suse.de>
29189 PR tree-optimization/80374
29190 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29191 build_zero_cst, remove fold_convertible_p check again.
29193 2017-04-11 Martin Liska <mliska@suse.cz>
29196 * ubsan.c (instrument_object_size): Do not instrument register
29199 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29202 * config/i386/i386-builtin-types.def
29203 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29204 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29205 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29206 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29207 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29208 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29209 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29210 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29211 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29212 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29213 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29214 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29215 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29216 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29217 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29218 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29219 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29220 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29221 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29222 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29223 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29224 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29225 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29226 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29227 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29228 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29229 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29230 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29231 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29232 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29233 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29234 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29235 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29236 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29237 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29238 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29239 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29240 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29241 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29242 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29243 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29244 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29245 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29246 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29247 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29248 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29250 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29251 flag to second_arg_count, handle 4 argument function type _COUNT
29252 aliases, handle second_arg_count on second argument rather than last.
29254 2017-04-10 Jeff Law <law@redhat.com>
29256 PR tree-optimization/80374
29257 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29258 record anything if we can not convert integer_zero_node to the
29261 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29264 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29265 Enhance special handling given to the TARGET_P9_MINMAX option in
29266 relation to certain other options.
29268 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29270 PR tree-optimization/80153
29271 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29272 remove POINTER_PLUS_EXPR's base part directly, rather than through
29275 2017-04-10 Richard Biener <rguenther@suse.de>
29276 Bin Cheng <bin.cheng@arm.com>
29278 PR tree-optimization/80153
29279 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29280 the first element of pointer type aff_tree. Build result expr in
29282 (add_elt_to_tree): Convert to type unconditionally. Remove other
29283 fold_convert calls.
29284 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29285 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29287 2017-04-10 Richard Biener <rguenther@suse.de>
29289 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29292 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29294 PR rtl-optimization/70478
29295 * lra-constraints.c (curr_small_class_check): New.
29296 (update_and_check_small_class_inputs): New.
29297 (process_alt_operands): Update curr_small_class_check. Disfavor
29298 alternative insn memory operands. Check available regs for small
29301 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29304 * config/mips/mips.opt (-mvirt): Update description.
29305 * doc/invoke.texi (-mvirt): Likewise.
29307 2017-04-10 Richard Biener <rguenther@suse.de>
29309 PR middle-end/80362
29310 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29311 looking for NEGATE_EXPR in -A / -B to A / B folding.
29313 2017-04-10 Martin Liska <mliska@suse.cz>
29315 PR gcov-profile/80224
29316 * gcov.c (print_usage): Fix usage string.
29317 (get_gcov_intermediate_filename): Remove.
29318 (output_gcov_file): Use both for normal and intermediate format.
29319 (generate_results): Do not initialize special file for
29320 intermediate format.
29322 2017-04-10 Richard Biener <rguenther@suse.de>
29324 PR tree-optimization/80304
29325 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29328 2017-04-10 Nathan Sidwell <nathan@acm.org>
29331 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29332 (rs6000_init_builtins): Use it.
29334 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29336 * config/arm/arm.md (<mrc>): Add mode to SET source.
29337 (<mrrc>): Likewise.
29339 2017-04-10 Richard Biener <rguenther@suse.de>
29341 PR middle-end/80344
29342 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29344 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29347 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29348 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29349 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29350 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29351 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29352 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29353 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29354 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29355 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29356 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29357 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29358 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29359 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29360 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29361 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29362 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29363 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29364 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29365 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29366 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29367 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29368 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29369 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29371 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29373 PR rtl-optimization/70478
29374 * lra-constraints.c: Reverse the last patch.
29376 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29378 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29379 Add comment for WCHAR_T.
29381 2017-04-08 Martin Liska <mliska@suse.cz>
29384 2017-04-07 Martin Liska <mliska@suse.cz>
29387 * ipa-split.c (split_function): Add function part to a same comdat
29390 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29393 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29395 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29397 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29398 (vec_pairing): ...to this.
29399 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29400 (rs6000_sched_init): Adjust for name change.
29401 (struct rs6000_sched_context): Likewise.
29402 (rs6000_init_sched_context): Likewise.
29403 (rs6000_set_sched_context): Likewise.
29405 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29411 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29413 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29414 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29415 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29417 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29419 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29421 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29423 PR rtl-optimization/70703
29424 * ira-color.c (update_conflict_hard_regno_costs): Use
29425 int64_t instead of HOST_WIDE_INT.
29427 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29429 PR rtl-optimization/70478
29430 * lra-constraints.c (process_alt_operands): Disfavor alternative
29431 insn memory operands.
29433 2017-04-07 Jeff Law <law@redhat.com>
29435 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29436 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29438 2017-04-07 Martin Liska <mliska@suse.cz>
29441 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29442 Show error message instead of an ICE.
29444 2017-04-07 Martin Liska <mliska@suse.cz>
29447 * ipa-split.c (split_function): Add function part to a same comdat
29450 2017-04-07 Richard Biener <rguenther@suse.de>
29452 PR middle-end/80341
29453 * tree.c (get_unwidened): Also handle ! for_type case for
29455 * convert.c (do_narrow): Split out from ...
29456 (convert_to_integer_1): ... here. Do not pass final truncation
29457 type to get_unwidened for TRUNC_DIV_EXPR.
29459 2017-04-07 Richard Biener <rguenther@suse.de>
29461 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29463 (aff_combination_const): Adjust.
29464 (aff_combination_scale): Likewise.
29465 (aff_combination_add_elt): Likewise.
29466 (aff_combination_add_cst): Likewise.
29467 (aff_combination_convert): Likewise.
29468 (add_elt_to_tree): Likewise. Remove unused argument.
29469 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29471 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29473 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29475 * config/arm/arm.c (arm_default_short_enums): Use
29476 ARM_DEFAULT_SHORT_ENUMS.
29477 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29479 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29482 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29483 members with redundant out-of-class redeclaration.
29485 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29488 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29489 * config/i386/i386.md (*zero_extendsidi2):
29490 Add (?*x,*x) and (?*v,*v) alternatives.
29492 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29495 * config/i386/i386.c (ix86_expand_builtin)
29496 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29497 mode from insn data. Convert operands to insn operand mode.
29498 Copy operands that don't satisfy insn predicate to a register.
29500 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29502 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29505 2017-04-06 Richard Biener <rguenther@suse.de>
29507 PR tree-optimization/80334
29508 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29509 preserve alignment of accesses.
29511 2017-04-06 Richard Biener <rguenther@suse.de>
29513 PR tree-optimization/80262
29514 * tree-sra.c (build_ref_for_offset): Preserve address-space
29516 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29517 Drop useless address-space information on MEM_REF offsets.
29519 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29521 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29523 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29525 PR rtl-optimization/70703
29526 * ira-color.c (update_conflict_hard_regno_costs): Use
29527 HOST_WIDE_INT instead of long.
29529 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29532 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29533 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29535 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29536 for x86_64 target. Handle -m3dnowa option.
29538 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29540 PR rtl-optimization/70703
29541 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29542 (update_conflict_hard_regno_costs): Use long instead of unsigned
29543 arithmetic for cost calculation.
29545 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29546 Bernd Edlinger <bernd.edlinger@hotmail.de>
29549 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29552 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29555 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29556 ptr_mode with Pmode throughout.
29557 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29558 into probe_stack_range and use DImode.
29560 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29563 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29564 call_eh_return is true.
29566 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29568 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29569 Initialize last_match_fntype_index.
29571 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29574 * tree-nvr.c: Include internal-fn.h.
29575 (pass_return_slot::execute): Ignore internal calls without
29578 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29579 Richard Biener <rguenther@suse.de>
29582 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29583 captures used multiple times, except for the last use.
29584 * generic-match-head.c: Include gimplify.h.
29586 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29588 PR tree-optimization/79390
29589 * target.h (struct noce_if_info): Declare.
29590 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29591 * target.def (noce_conversion_profitable_p): New target hook.
29592 * ifcvt.h (struct noce_if_info): New type, moved from ...
29593 * ifcvt.c (struct noce_if_info): ... here.
29594 (noce_conversion_profitable_p): Renamed to ...
29595 (default_noce_conversion_profitable_p): ... this. No longer
29597 (noce_try_store_flag_constants, noce_try_addcc,
29598 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29599 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29600 instead of noce_conversion_profitable_p.
29601 * config/i386/i386.c: Include ifcvt.h.
29602 (ix86_option_override_internal): Don't override
29603 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29604 (ix86_noce_conversion_profitable_p): New function.
29605 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29606 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29607 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29608 * doc/tm.texi: Regenerated.
29610 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29612 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29615 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29618 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29619 instructions for small multiply cores.
29621 2017-04-04 Jeff Law <law@redhat.com>
29623 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29625 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29626 that are not set by the loop over the elements.
29628 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29631 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29632 int mode, convert_modes it to mode as unsigned, otherwise use
29633 lowpart_subreg to mode rather than SImode.
29634 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29635 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29636 Use DImode instead of SImode for the shift count operand.
29637 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29640 2017-04-04 Richard Biener <rguenther@suse.de>
29642 PR middle-end/80281
29643 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29644 arithmetic done for the negate or the plus. Simplify.
29645 (A - (-B) -> A + B): Likewise.
29646 * fold-const.c (split_tree): Make sure to not negate pointers.
29648 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29650 PR rtl-optimization/60818
29651 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29652 a compare of comparisons with the thing compared if this results
29653 in a different machine mode.
29655 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29657 * alias.c (base_alias_check): Fix typo in comment.
29658 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29659 * cgraphunit.c (symbol_table::compile): Likewise.
29660 * collect2.c (maybe_run_lto_and_relink): Likewise.
29661 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29662 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29663 * config/avr/avr.c (avr_map_op_t): Likewise.
29664 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29665 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29666 * config/epiphany/epiphany.md (movcc): Likewise.
29667 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29668 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29670 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29671 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29672 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29673 * config/sh/sh.c (sh_rtx_costs): Likewise.
29674 * fold-const.c (fold_truth_andor): Likewise.
29675 * genautomata.c (collapse_flag): Likewise.
29676 * gengtype.h (struct type::u::s): Likewise.
29677 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29678 * input.c (FORMAT_AMOUNT): Likewise.
29679 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29680 (known_aggs_to_agg_replacement_list): Likewise.
29681 * ipa-inline-analysis.c: Likewise.
29682 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29683 * ipa-polymorphic-call.c
29684 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29685 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29686 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29688 * modulo-sched.c (apply_reg_moves): Likewise.
29689 * omp-expand.c (build_omp_regions_1): Likewise.
29690 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29691 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29692 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29693 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29694 * value-prof.c: Likewise.
29695 * var-tracking.c (val_reset): Likewise.
29697 2017-04-03 Richard Biener <rguenther@suse.de>
29699 PR tree-optimization/80275
29700 * fold-const.c (split_address_to_core_and_offset): Handle
29703 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29705 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29706 descriptors is at least equal to that of functions.
29708 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29710 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29712 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29715 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29716 (mov<IMOD4:mode>): New expander.
29717 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29719 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29721 PR rtl-optimization/79405
29722 * fwprop.c (propagations_left): New variable.
29723 (forward_propagate_into): Decrement it.
29724 (fwprop_init): Initialize it.
29725 (fw_prop): If the variable has reached zero, stop propagating.
29726 (fwprop_addr): Ditto.
29728 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29731 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29732 a FUNCTION_DECL, pass it as decl instead of origin to
29735 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29737 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29740 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29743 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29744 TARGET_VSX_SMALL_INTEGER.
29746 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29748 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29749 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29751 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29753 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29754 extraction from odd-numbered MSA register.
29756 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29758 PR middle-end/80173
29759 * expmed.c (store_bit_field_1): Don't attempt to create
29760 a word subreg out of hard registers wider than word if they
29761 have HARD_REGNO_NREGS of 1 for their mode.
29763 PR middle-end/80163
29764 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29765 conversions to integer types wider than word and pointer.
29768 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29769 (rtx_equal_for_cselib_p): Pass 0 to it.
29770 * cselib.c (cselib_hasher::equal): Likewise.
29771 (rtx_equal_for_cselib_1): Add depth argument. If depth
29772 is 128, don't look up VALUE locs and punt. Increment
29773 depth in recursive calls when walking VALUE locs.
29775 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29777 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29778 (make_gcov_file_name): Use the canonical path name for generating
29780 (read_line): Fix handling of files with ascii null bytes.
29782 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29784 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29785 to initialise a vector register instead
29786 of using a const_int.
29788 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29790 PR translation/80189
29791 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29792 diagnostic messages.
29794 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29797 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29798 (dfp_diex_<mode>): Update mode of operand 1.
29799 * doc/extend.texi (dxex, dxexq): Document change to return type.
29800 (diex, diexq): Document change to argument type.
29802 2017-03-30 Martin Jambor <mjambor@suse.cz>
29805 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29806 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29807 it reflects the signature changes performed at the callee side.
29808 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29809 to cgraph_build_function_type_skip_args.
29810 (build_function_decl_skip_args): Adjust call to the above function.
29812 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29815 * config/i386/sse.md
29816 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29817 register as dest whenever it is a MEM not rtx_equal_p to the
29818 corresponding dup operand, and when forcing into reg move the
29819 reg into the memory afterwards.
29820 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29821 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29822 for the force_reg mode.
29823 (avx512vl_vextractf128<mode>): Use register as dest either
29824 always when a MEM, or when it is a MEM not rtx_equal_p to the
29825 corresponding dup operand, or even not when it is a CONST_VECTOR
29826 depending on the mode and lo vs. hi.
29827 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29829 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29830 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29831 Likewise. Require that operands[2] is even.
29832 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29833 Remove extraneous parens. Require that operands[2] is a multiple
29835 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29836 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29837 disallow memory then.
29839 2017-03-30 Richard Biener <rguenther@suse.de>
29841 PR tree-optimization/77498
29842 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29843 to non-constants over backedges.
29845 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29847 PR rtl-optimization/80233
29848 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29849 as last_combined_insn. Do not test for BARRIER_P separately.
29851 2017-03-29 Andreas Schwab <schwab@suse.de>
29854 * calls.c (prepare_call_address): Convert funexp to Pmode before
29855 copying to temp reg.
29857 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29859 PR tree-optimization/80158
29860 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29861 Handle possible future case of more than one alternate
29863 (replace_rhs_if_not_dup): Likewise.
29864 (replace_one_candidate): Likewise.
29866 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29868 PR rtl-optimization/80193
29869 * ira.c (ira): Do not check allocation for LRA.
29871 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29873 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29874 (nvptx_output_simt_exit): Declare.
29875 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29876 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29877 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29878 (nvptx_declare_function_name): ...here. Emit declaration of local
29879 memory space buffer for omp_simt_enter insn.
29880 (nvptx_output_unisimt_switch): New.
29881 (nvptx_output_softstack_switch): New.
29882 (nvptx_output_simt_enter): New.
29883 (nvptx_output_simt_exit): New.
29884 * config/nvptx/nvptx.h (struct machine_function): New fields
29885 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29886 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29887 (UNSPECV_SIMT_EXIT): Ditto.
29888 (omp_simt_enter_insn): New insn.
29889 (omp_simt_enter): New expansion.
29890 (omp_simt_exit): New insn.
29891 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29893 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29894 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29895 (expand_GOMP_SIMT_EXIT): New.
29896 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29897 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29898 (GOMP_SIMT_EXIT): Ditto.
29899 * target-insns.def (omp_simt_enter): New insn.
29900 (omp_simt_exit): Ditto.
29901 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29903 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29904 (lower_rec_input_clauses): Likewise.
29905 (lower_lastprivate_clauses): Handle SIMT privatization.
29907 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29908 (ompdevlow_adjust_simt_enter): New.
29909 (find_simtpriv_var_op): New.
29910 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29911 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29913 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29914 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29915 (copy_decl_for_dup_finish): Ditto.
29917 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29919 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29922 * config/i386/i386.c (ix86_option_override_internal): Always
29923 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29925 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29927 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29929 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29931 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29932 mark new edge's irreducible flag accordign to it.
29933 (vect_do_peeling): Check loop preheader edge's irreducible flag
29934 and pass it to function slpeel_add_loop_guard.
29936 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29938 PR tree-optimization/80218
29939 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29940 Update block frequencies and counts.
29942 2017-03-28 Richard Biener <rguenther@suse.de>
29944 PR tree-optimization/78644
29945 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29946 of a simplification result we may not use it at all.
29948 2017-03-28 Richard Biener <rguenther@suse.de>
29951 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29952 without arguments, generate default definition of a SSA name.
29954 2017-03-28 Richard Biener <rguenther@suse.de>
29956 PR middle-end/80222
29957 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29958 TYPE_REF_CAN_ALIAS_ALL references.
29959 * fold-const.c (fold_indirect_ref_1): Likewise.
29961 2017-03-28 Martin Liska <mliska@suse.cz>
29964 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29965 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29967 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29968 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29970 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29971 (EXTRA_SPECS): Define.
29972 (SUBTARGET_EXTRA_SPECS): Likewise.
29973 (SUBTARGET_CPP_SPEC): Likewise.
29974 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29975 SUBTARGET_EXTRA_SPECS.
29976 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29978 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29980 * config/arc/simdext.md (vst64_insn): Update pattern.
29981 (vld32wh_insn): Likewise.
29982 (vld32wl_insn): Likewise.
29983 (vld64_insn): Likewise.
29984 (vld32_insn): Likewise.
29986 2017-03-28 Marek Polacek <polacek@redhat.com>
29989 * fold-const.c (fold_comparison): Use protected_set_expr_location
29990 instead of SET_EXPR_LOCATION.
29992 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29994 * tree.c (add_expr): Avoid name lookup warning.
29996 2017-03-27 Jeff Law <law@redhat.com>
29998 PR tree-optimization/80216
29999 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30000 function name. Limit recursion depth.
30001 (record_temporary_equivalences): Corresponding changes.
30003 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30005 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30008 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30011 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30013 * cfgcleanup.c (reg_note_cfa_p): New array.
30014 (insns_have_identical_cfa_notes): New function.
30015 (old_insns_match_p): Don't cross-jump in between /f
30016 and non-/f instructions. If both i1 and i2 are frame related,
30017 verify all CFA notes, their order and content.
30019 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30022 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30023 HImode and SImode with zero extend to DImode to one insn.
30024 (bswap<mode>2_extenddi): Likewise.
30025 (bswapsi2_extenddi): Likewise.
30026 (bswaphi2_extendsi): Likewise.
30027 (bswaphi2): Combine bswap HImode and SImode into one insn.
30028 Separate memory insns from swapping register.
30029 (bswapsi2): Likewise.
30030 (bswap<mode>2): Likewise.
30031 (bswaphi2_internal): Delete, no longer used.
30032 (bswapsi2_internal): Likewise.
30033 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30034 store, and gpr<-gpr swap insns.
30035 (bswap<mode>2_store): Likewise.
30036 (bswaphi2_reg): Register only splitter, combine with the splitter.
30037 (bswaphi2 splitter): Likewise.
30038 (bswapsi2_reg): Likewise.
30039 (bswapsi2 splitter): Likewise.
30040 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30041 the insns into load, store, and register/register insns.
30042 (bswapdi2_ldbrx): Likewise.
30043 (bswapdi2_load): Likewise.
30044 (bswapdi2_store): Likewise.
30045 (bswapdi2_reg): Likewise.
30047 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30049 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30050 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30052 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30055 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30058 special handling for target option conflicts between dform
30059 options (-mpower9-dform, -mpower9-dform-vector,
30060 -mpower9-dform-scalar) and -mno-direct-move.
30062 2017-03-27 Richard Biener <rguenther@suse.de>
30064 PR tree-optimization/80181
30065 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30067 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30069 * config/arc/predicates.md (move_double_src_operand): Replace the
30070 call to move_double_src_operand with a call to address_operand.
30072 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30074 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30075 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30076 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30078 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30080 * config/arc/predicates.md (long_immediate_loadstore_operand):
30081 Consider scaled addresses cases.
30083 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30085 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30086 restored when in interrupt.
30087 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30088 doesn't have delay slot.
30090 2017-03-27 Richard Biener <rguenther@suse.de>
30093 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30094 inlined thunk clones.
30096 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30099 * asan.c (instrument_derefs): Copy over last operand from
30100 original COMPONENT_REF to the new COMPONENT_REF with
30101 DECL_BIT_FIELD_REPRESENTATIVE.
30102 * ubsan.c (instrument_object_size): Likewise.
30104 2017-03-27 Richard Biener <rguenther@suse.de>
30106 PR tree-optimization/80170
30107 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30108 sure DR/SCEV didnt fold in constants we do not see when looking
30109 at the reference base alignment.
30111 2017-03-27 Richard Biener <rguenther@suse.de>
30113 PR middle-end/80171
30114 * gimple-fold.c (fold_ctor_reference): Properly guard against
30115 NULL return value from canonicalize_constructor_val.
30117 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30120 * config/i386/i386.c (ix86_expand_builtin)
30121 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30122 flags reg setting and flags reg using instructions.
30123 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30124 clobbering instructions to zero extend op2.
30126 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30128 * doc/install.texi (Configuration) <--with-aix-soname>:
30129 Update link to AIX ld.
30131 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30133 PR rtl-optimization/80160
30134 PR rtl-optimization/80159
30135 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30136 reg_alternate_class into account.
30138 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30141 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30142 to consider in curr_insn_transform.
30144 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30146 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30147 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30148 and emit_mode_inner.
30150 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30152 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30153 argument to the overloaded builtin variants. Use the new flag to
30154 deprecate certain builtin variants.
30155 * config/s390/s390-builtin-types.def: Add new builtin types.
30156 * config/s390/s390-builtins.h: Support new flags field for
30157 overloaded builtins.
30158 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30159 (s390_macro_to_expand): Enable vector float data type.
30160 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30161 builtins by incrementing the __VEC__ version number.
30162 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30164 (s390_resolve_overloaded_builtin): Emit error messages depending
30165 on the builtin flags.
30166 * config/s390/s390.c (s390_expand_builtin): Support additional
30167 flags argument. Change error message to match the messages
30168 emitted in s390-c.c.
30169 * config/s390/s390.md: New UNSPEC_* constants.
30170 (op_type): Add new instruction types.
30171 * config/s390/vecintrin.h: Add new builtins and test data class
30173 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30174 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30175 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30176 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30177 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30178 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30180 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30181 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30182 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30183 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30185 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30186 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30187 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30188 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30190 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30191 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30192 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30193 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30194 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30195 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30196 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30198 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30199 ("vec_scatter_element<V_HW_4:mode>_DI")
30200 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30201 ("vec_fpint<mode>", "vflls")
30202 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30203 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30204 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30205 ("*vec_cmphe<mode>_cc"): ... these.
30207 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30208 mode constant instead of magic value.
30210 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30212 * config/s390/s390.c (s390_expand_vec_compare): Support other
30213 vector floating point modes than just V2DF.
30214 (s390_expand_vcond): Likewise.
30215 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30216 (s390_cannot_change_mode_class): Prevent mode changes between TF
30217 and V1TF in vector registers.
30218 * config/s390/s390.md (DF, SF): New mode attributes.
30219 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30220 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30221 SFmode support for VRs.
30222 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30224 (VFT, VF_HW): New mode iterators.
30225 (vw, sdx): New mode attributes.
30226 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30227 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30228 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30229 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30230 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30231 also the new vector floating point modes. Renaming to ...
30233 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30234 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30235 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30236 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30237 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30238 ("vec_unordered<mode>"): ... these.
30240 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30241 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30242 ("*vec_extendv2df"): New insn definitions.
30244 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30246 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30247 ("mulditi3_2", "*muldi3_sign"): New patterns.
30248 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30249 rename the pattern definition.
30251 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30253 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30255 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30257 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30259 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30260 instruction if possible.
30261 * config/s390/vector.md (vec_halfnumelts): New mode
30263 ("*vec_vllezlf<mode>"): New pattern.
30265 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30267 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30268 ("popcountv4si2", "popcountv2di2"): Rename to ...
30269 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30270 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30272 ("popcount<mode>2_vxe"): New pattern.
30274 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30276 * common/config/s390/s390-common.c (processor_flags_table): Add
30278 * config.gcc: Add arch12.
30279 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30280 Default to arch12 for unknown CPU model numbers.
30281 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30282 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30283 PROCESSOR_max sanity check.
30284 * config/s390/s390-opts.h (enum processor_type): Add
30286 * config/s390/s390.c (processor_table): Add arch12.
30287 (s390_expand_builtin): Add check for B_VXE flag.
30288 (s390_issue_rate): Add PROCESSOR_ARCH12.
30289 (s390_get_sched_attrmask): Likewise.
30290 (s390_get_unit_mask): Likewise.
30291 (s390_sched_score): Enable z13 scheduling for arch12.
30292 (s390_sched_reorder): Likewise.
30293 (s390_sched_variable_issue): Likewise.
30294 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30296 (s390_tune_attr): Use z13 scheduling also for arch12.
30297 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30298 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30299 (TARGET_VXE_P): New macros.
30300 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30301 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30302 * config/s390/s390.opt: Add arch12 as processor_type.
30304 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30306 * config/s390/s390.md
30307 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30308 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30309 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30311 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30312 Rename expanders to ...
30314 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30315 ("fixuns_truncdddi2_emu"): ... these.
30317 ("fixuns_trunc<mode>si2_emu"): New expander.
30319 ("*fixuns_truncdfdi2_z13"): Rename to ...
30320 ("*fixuns_truncdfdi2_vx"): ... this.
30322 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30324 * config/s390/2964.md: Remove the single element vector compare
30325 instructions which are no longer used.
30326 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30328 (s390_canonicalize_comparison): Remove handling of DFmode
30330 (s390_expand_vec_compare_scalar): Remove function.
30331 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30332 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30334 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30338 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30340 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30341 will anyway by matched by mov<mode>_64dfp.
30343 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30345 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30346 vlef/vstef. Add missing operand to vleif.
30348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30350 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30351 pair for all vector types with 64 bit elements.
30352 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30353 * config/s390/vector.md (V_HW_64): ... here.
30354 (V_128_NOSINGLE): New mode iterator.
30355 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30356 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30357 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30358 ("*vec_load_pairv2di"): Change to ...
30359 ("*vec_load_pair<mode>"): ... this one.
30361 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30363 * config/s390/constraints.md: Add comments.
30364 (jKK): Reject element sizes > 8 bytes.
30365 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30367 * config/s390/s390.md: Add the s_operand checks formerly in
30368 s390_split_ok_p to various splitters where they are still
30370 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30371 for 128 bit vectors. Plus two splitters.
30373 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30375 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30381 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30382 error if the boundary argument is not constant.
30384 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30386 PR rtl-optimization/80112
30387 * loop-doloop.c (doloop_condition_get): Don't check condition
30388 if cmp isn't SET with IF_THEN_ELSE src.
30390 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30392 PR tree-optimization/80158
30393 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30394 replacing a candidate statement, also replace it for the
30395 candidate's alternate interpretation.
30396 (replace_rhs_if_not_dup): Likewise.
30397 (replace_one_candidate): Likewise.
30399 2017-03-24 Richard Biener <rguenther@suse.de>
30401 PR tree-optimization/80167
30402 * graphite-isl-ast-to-gimple.c
30403 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30405 (translate_isl_ast_to_gimple::get_rename): Likewise.
30407 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30409 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30410 handling of certain combinations of target options, including the
30411 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30412 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30414 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30417 * config/arm/arm.md (*load_multiple): Add reload_completed to
30418 matching condition.
30420 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30421 Richard Biener <rguenth@suse.de>
30423 PR tree-optimization/79908
30424 PR tree-optimization/80136
30425 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30426 been cast away, gimplify_and_add suffices.
30428 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30430 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30432 2017-03-23 Richard Biener <rguenther@suse.de>
30434 PR tree-optimization/80032
30435 * gimplify.c (gimple_push_cleanup): Forced unconditional
30436 cleanups still have to go to the conditional_cleanups
30439 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30441 PR tree-optimization/80072
30442 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30444 (next_operand_entry_id): Change type to unsigned int.
30445 (sort_by_operand_rank): Make sure to return the right return value
30446 even if unsigned fields are bigger than INT_MAX.
30447 (struct oecount): Change cnt and id type to unsigned int.
30448 (oecount_hasher::equal): Formatting fix.
30449 (oecount_cmp): Make sure to return the right return value
30450 even if unsigned fields are bigger than INT_MAX.
30451 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30454 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30455 TREE_READONLY on result if writing it more than once.
30458 * doc/invoke.texi (-fsanitize=thread): Document that with
30459 -fnon-call-exceptions atomics are not able to throw
30463 * tsan.c: Include tree-eh.h.
30464 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30465 maybe_clean_or_replace_eh_stmt where needed.
30466 (instrument_memory_accesses): Add cfg_changed argument.
30467 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30468 if it returned true.
30469 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30471 PR rtl-optimization/63191
30472 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30473 wrapper function, moved the whole old content into ...
30474 (ix86_delegitimize_address_1): ... this. New inline function.
30475 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30476 true as last argument instead of ix86_delegitimize_address.
30478 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30480 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30481 cortexa57_branch_cost.
30483 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30485 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30487 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30490 * doc/md.texi (Constraints): Document wA constraint.
30491 * config/rs6000/constraints.md (wA): New.
30492 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30493 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30494 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30495 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30497 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30500 * gimplify.c (is_oacc_declared): New function.
30501 (oacc_default_clause): Use it to set default flags for acc declared
30502 variables inside parallel regions.
30503 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30504 declared variables.
30505 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30506 declare attribute to any decl as necessary.
30508 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30511 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30512 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30513 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30514 (arm_arch_lpae): This.
30515 * config/arm/arm.c (arm_arch7ve): Rename into ...
30516 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30517 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30520 2017-03-22 Martin Liska <mliska@suse.cz>
30523 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30524 error message instead of an ICE.
30526 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30528 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30530 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30532 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30534 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30537 2017-03-21 Martin Sebor <msebor@redhat.com>
30539 * doc/extend.texi: Use "cannot" instead of "can't."
30540 * doc/hostconfig.texi: Same.
30541 * doc/install.texi: Same.
30542 * doc/invoke.texi: Same.
30543 * doc/loop.texi: Same.
30544 * doc/md.texi: Same.
30545 * doc/objc.texi: Same.
30546 * doc/rtl.texi: Same.
30547 * doc/tm.texi: Same.
30548 * doc/tm.texi.in: Same.
30549 * doc/trouble.texi: Same.
30551 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30554 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30555 (collect_checksum_attributes): Set it.
30556 (die_checksum_ordered): Use it.
30558 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30560 PR tree-optimization/79908
30561 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30562 change: For a VA_ARG whose LHS has been cast away, use
30563 force_gimple_operand to construct the side effects.
30565 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30567 PR translation/80001
30568 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30569 more amenable to translation.
30570 (oacc_loop_auto_partitions): Likewise.
30572 2017-03-21 Marek Polacek <polacek@redhat.com>
30573 Martin Sebor <msebor@redhat.com>
30575 PR tree-optimization/80109
30576 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30577 on INTEGRAL_TYPE_P.
30579 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30580 Segher Boessenkool <segher@kernel.crashing.org>
30583 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30584 check reg_used_between_p between insn and one of succ or succ2
30585 depending on if succ is artificial insn not inserted into insn
30588 2017-03-21 Martin Liska <mliska@suse.cz>
30590 PR gcov-profile/80081
30591 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30592 * doc/gcc.texi: Include gcov-dump stuff.
30593 * doc/gcov-dump.texi: New file.
30595 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30597 PR rtl-optimization/79150
30598 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30599 conditional jump, if the jump is the last insn of the loop.
30601 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30602 Richard Biener <rguenth@suse.de>
30604 PR tree-optimization/79908
30605 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30606 been cast away, use force_gimple_operand to construct the side
30609 2017-03-21 Martin Liska <mliska@suse.cz>
30611 PR libfortran/79956
30612 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30615 2017-03-21 Brad Spengler <spender@grsecurity.net>
30618 * plugin.c (htab_hash_plugin): New function.
30619 (add_new_plugin): Use it and adjust.
30620 (parse_plugin_arg_opt): Adjust.
30621 (init_one_plugin): Likewise.
30623 2017-03-21 Richard Biener <rguenther@suse.de>
30625 PR tree-optimization/80032
30626 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30627 if set force the cleanup to happen unconditionally.
30628 (gimplify_target_expr): Push inserted clobbers with force_uncond
30629 to avoid them being removed by control-dependent DCE.
30631 2017-03-21 Richard Biener <rguenther@suse.de>
30633 PR tree-optimization/80122
30634 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30635 va_arg_pack_len when the inlined call stmt requires pack
30637 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30639 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30642 * tsan.c (instrument_builtin_call): If the memory model argument
30643 is not a constant, assume it is valid.
30646 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30649 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30651 PR rtl-optimization/79910
30652 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30653 if its dest is used by an insn before I2 (other than the combined
30654 insns themselves, which are properly handled already).
30656 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30659 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30661 * combine.c (record_used_regs): New static function.
30662 (try_combine): Handle situations where there is an additional
30663 instruction between I2 and I3 which needs to have a LOG_LINK
30667 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30669 * combine.c (try_combine): Delete redundant i1 test. Call
30670 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30672 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30675 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30676 alternatives 13/14.
30678 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30680 PR tree-optimization/80054
30681 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30682 the optimization if a PHI or any of its arguments is not dominated
30683 by the candidate's basis. Use gphi* rather than gimple* as
30685 (replace_profitable_candidates): Clean up a gimple* variable that
30686 should be a gphi* variable.
30688 2017-03-20 Martin Sebor <msebor@redhat.com>
30691 * doc/extend.texi (attribute constructor): Document present limitation.
30693 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30696 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30697 __POWER9_VECTOR__ #ifdef control, change template definition to
30698 use Power9-specific built-in function.
30699 (vec_any_eq): Likewise.
30700 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30701 to control outcomes from this test.
30702 (vector_ae_<mode>p): For VEC_F modes, likewise.
30704 2017-03-20 Ian Lance Taylor <iant@google.com>
30706 * config/i386/i386.c (ix86_function_regparm): Save an extra
30707 register for -fsplit-stack with DECL_STATIC_CHAIN.
30709 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30712 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30713 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30715 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30717 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30719 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30722 2017-03-20 Marek Polacek <polacek@redhat.com>
30725 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30727 2017-03-20 Richard Biener <rguenther@suse.de>
30729 PR tree-optimization/80113
30730 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30731 allocate extra SSA name for PHI def.
30732 (add_close_phis_to_outer_loops): Likewise.
30733 (add_close_phis_to_merge_points): Likewise.
30734 (copy_loop_close_phi_args): Likewise.
30735 (copy_cond_phi_nodes): Likewise.
30737 2017-03-20 Martin Liska <mliska@suse.cz>
30739 PR middle-end/79753
30740 * tree-chkp.c (chkp_build_returned_bound): Do not build
30741 returned bounds for a LHS that's not a BOUNDED_P type.
30743 2017-03-20 Martin Liska <mliska@suse.cz>
30747 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30748 COMPLEX_CST and VECTOR_CST.
30750 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30753 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30754 target operand. A new splitter adds the clobber statement in case
30755 the target operand is dead anyway.
30757 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30759 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30760 to age-old versions of binutils and glibc.
30762 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30764 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30766 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30768 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30770 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30772 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30773 requirement for binutils 2.13.
30775 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30777 * combine.c (try_combine): Delete redundant i1 test. Call
30778 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30780 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30782 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30783 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30785 <riscv64-*-elf>: Re-arrange section
30786 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30787 <riscv32-*-linux>: Likewise.
30788 <riscv64-*-elf>: Likewise
30789 <riscv64-*-linux>: Likewise.
30791 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30794 * aarch64.opt(verbose-cost-dump): Fix typo.
30796 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30799 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30800 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30802 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30804 * reload.c (find_reloads): When reloading a nonoffsettable address,
30805 use RELOAD_OTHER for it and its address reloads.
30807 PR rtl-optimization/79910
30808 * combine.c (record_used_regs): New static function.
30809 (try_combine): Handle situations where there is an additional
30810 instruction between I2 and I3 which needs to have a LOG_LINK
30813 2017-03-17 Jeff Law <law@redhat.com>
30815 PR tree-optimization/71437
30816 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30817 conditional in the hash table first.
30818 (vrp_dom_walker::before_dom_children): Extract condition from
30819 ASSERT_EXPR. Record condition, its inverion and any implied
30820 conditions as well.
30822 2017-03-17 Marek Polacek <polacek@redhat.com>
30823 Markus Trippelsdorf <markus@trippelsdorf.de>
30825 PR tree-optimization/80079
30826 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30829 2017-03-17 Richard Biener <rguenther@suse.de>
30831 PR middle-end/80075
30832 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30833 Properly verify the LHS before the RHS possibly claims to be
30835 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30838 2017-03-17 Martin Jambor <mjambor@suse.cz>
30840 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30841 (List of -O2 options): Likewise.
30842 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30845 2017-03-17 Tom de Vries <tom@codesourcery.com>
30847 * gcov-dump.c (print_usage): Print bug_report_url.
30849 2017-03-17 Richard Biener <rguenther@suse.de>
30851 PR middle-end/80050
30852 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30853 (parser::peek): Likewise.
30855 2017-03-17 Richard Biener <rguenther@suse.de>
30857 PR tree-optimization/80048
30858 * sese.c (free_sese_info): Properly release rename_map and
30859 copied_bb_map elements.
30861 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30863 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30864 Add linked-list forward and backlinks. Insert on
30865 construction, remove on destruction.
30866 (class pass_store_merging): Add m_stores_head field.
30867 (pass_store_merging::terminate_and_process_all_chains):
30868 Iterate over m_stores_head list.
30869 (pass_store_merging::terminate_all_aliasing_chains):
30871 (pass_store_merging::execute): Check for debug stmts first.
30872 Push new chains onto the m_stores_head stack.
30874 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30877 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30878 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30879 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30881 2017-03-16 Jeff Law <law@redhat.com>
30883 PR tree-optimization/71437
30884 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30885 member function. Implementation moved into after_dom_children
30886 member function and into the threader's thread_outgoing_edges
30888 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30889 some code into new thread_outgoing_edges.
30890 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30891 definition. Simplify marker handling (do it here). Assume we always
30892 have the available expression and the const/copies tables.
30893 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30895 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30896 * tree-vrp.c (equiv_stack): No longer file scoped.
30897 (vrp_dom_walker): New class.
30898 (vrp_dom_walker::before_dom_children): New member function.
30899 (vrp_dom_walker::after_dom_children): Likewise.
30900 (identify_jump_threads): Setup domwalker. Use it rather than
30901 walking edges in a random order by hand. Simplify setup/finalization.
30902 (finalize_jump_threads): Remove.
30903 (vrp_finalize): Do not call identify_jump_threads here.
30904 (execute_vrp): Do it here instead and call thread_through_all_blocks
30907 PR tree-optimization/71437
30908 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30910 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30912 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30913 (dom_opt_dom_walker::thread_across_edge): Remove
30914 handle_dominating_asserts argument. All callers changed.
30915 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30916 changes. Remove calls to lhs_of_dominating_assert. Other
30917 uses of handle_dominating_asserts turn into unconditional code
30918 (simplify_control_stmt_condition_1): Likewise.
30919 (simplify_control_stmt_condition): Likewise.
30920 (thread_through_normal_block, thread_across_edge): Likewise.
30921 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30922 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30923 object if it is not an SSA_NAME.
30924 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30925 before calling into the VRP specific simplifiers.
30926 (identify_jump_threads): Remove handle_dominating_asserts
30929 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30932 * tree-diagnostic.c (default_tree_printer): No longer static.
30933 * tree-diagnostic.h (default_tree_printer): New prototype.
30935 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30937 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30938 Change ins into fmov.
30940 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30942 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30943 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30944 Use h_con constraint for operand 1.
30945 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30946 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30948 2017-03-15 Jeff Law <law@redhat.com>
30950 PR tree-optimization/71437
30951 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30952 (record_temporary_equivalences): Use it.
30954 PR tree-optimization/71437
30955 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30956 tree-ssa-scopedtables.
30957 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30958 (record_conditions, record_cond, vuse_eq): Likewise.
30959 (record_edge_info): Adjust to API tweak of record_conditions.
30960 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30961 (record_temporary_equivalences, optimize_stmt): Likewise.
30962 (eliminate_redundant_computations): Likewise.
30963 (record_equivalences_from_stmt): Likewise.
30964 * tree-ssa-scopedtables.c: Include options.h and params.h.
30965 (vuse_eq): New function, moved from tree-ssa-dom.c
30966 (build_and_record_new_cond): Likewise.
30967 (record_conditions): Likewise. Accept vector of conditions rather
30968 than edge_equivalence structure for first argument.
30969 for the first argument.
30970 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30971 from tree-ssa-dom.c.
30972 (avail_exprs_stack::record_cond): Likewise.
30973 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30974 from tree-ssa-dom.c.
30975 (avail_exprs_stack): Add new member functions lookup_avail_expr
30977 (record_conditions): Declare.
30979 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30982 * lra-constraints.c (process_alt_operands): Increase reject for
30983 reloading an input/output operand.
30985 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30988 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30989 insns to convert from signed/unsigned char/short to IEEE 128-bit
30991 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30993 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30996 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30997 subreg of inner mode for values already in registers.
30999 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31001 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31002 iteration reg is used after the loop.
31004 2017-03-14 Martin Sebor <msebor@redhat.com>
31006 PR tree-optimization/79800
31007 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31008 precision in negative-positive range.
31009 (format_floating): Call non-const overload with adjusted precision.
31011 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31014 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31017 2017-03-14 Martin Sebor <msebor@redhat.com>
31019 PR middle-end/80020
31020 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31021 * builtins.def (aligned_alloc): Use it.
31024 * Makefile.in (GTFILES): Add calls.c.
31025 * calls.c: Include "gt-calls.h".
31027 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31029 PR rtl-optimization/79728
31030 * regs.h (struct target_regs): New field
31031 x_contains_allocatable_regs_of_mode.
31032 (contains_allocatable_regs_of_mode): New macro.
31033 * reginfo.c (init_reg_sets_1): Initialize it, and change
31034 contains_reg_of_mode so it includes global regs as well.
31035 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31036 rather than contains_regs_of_mode.
31038 2017-03-14 Martin Liska <mliska@suse.cz>
31040 * doc/invoke.texi: Document options that can't be combined with
31041 -fcheck-pointer-bounds.
31043 2017-03-14 Martin Liska <mliska@suse.cz>
31045 PR middle-end/79831
31046 * doc/invoke.texi (-Wchkp): Document the option.
31048 2017-03-14 Martin Liska <mliska@suse.cz>
31050 * Makefile.in: Install gcov-dump.
31052 2017-03-14 Martin Liska <mliska@suse.cz>
31054 * multiple_target.c (expand_target_clones): Bail out for
31055 an invalid attribute.
31057 2017-03-14 Richard Biener <rguenther@suse.de>
31059 * alias.c (struct alias_set_entry): Pack properly.
31060 * cfgloop.h (struct loop): Likewise.
31061 * cse.c (struct set): Likewise.
31062 * ipa-utils.c (struct searchc_env): Likewise.
31063 * loop-invariant.c (struct invariant): Likewise.
31064 * lra-remat.c (struct cand): Likewise.
31065 * recog.c (struct change_t): Likewise.
31066 * rtl.h (struct address_info): Likewise.
31067 * symbol-summary.h (function_summary): Likewise.
31068 * tree-loop-distribution.c (struct partition): Likewise.
31069 * tree-object-size.c (struct object_size_info): Likewise.
31070 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31071 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31072 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31073 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31074 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31075 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31076 (struct _stmt_vec_info): Likewise.
31078 2017-03-14 Martin Liska <mliska@suse.cz>
31081 * multiple_target.c (create_dispatcher_calls): Check that
31082 a target can create a function dispatcher.
31084 2017-03-14 Martin Liska <mliska@suse.cz>
31087 * multiple_target.c (expand_target_clones): Drop local.local
31088 flag for default implementation.
31090 2017-03-14 Richard Biener <rguenther@suse.de>
31092 PR tree-optimization/80030
31093 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31095 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31097 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31098 gcc_fallthrough() instead of __attribute__((fallthrough));
31100 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31102 * doc/gcc.texi: Remove "up" link to (DIR).
31103 * doc/gccint.texi: Ditto.
31105 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31107 * doc/install.texi (Specific) <avr>: Remove reference to
31110 2017-03-13 Jeff Law <law@redhat.com>
31112 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31113 attribute rather than comments.
31115 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31116 match_scratch operand is highest.
31118 2017-03-13 Martin Liska <mliska@suse.cz>
31120 PR middle-end/78339
31121 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31122 is a CHKP clone, use original declaration.
31124 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31126 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31127 (arc_conditional_register_usage): Use a different allocation order
31128 when optimizing for size.
31129 * common/config/arc/arc-common.c (arc_option_optimization_table):
31130 Section anchors default on when optimizing for size.
31132 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31134 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31136 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31138 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31139 * config/arc/arc.md (cpu_facility): Add cd variant.
31140 (*movqi_insn): Add code density variant.
31141 (*movhi_insn): Likewise.
31142 (*movqi_insn): Likewise.
31143 (*addsi3_mixed): Likewise.
31144 (subsi3_insn): Likewise.
31146 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31148 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31150 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31152 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31153 expressions with MINUS and UNARY ops.
31155 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31158 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31160 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31161 between vec_select and vector argument.
31162 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31163 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31164 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31165 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31166 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31167 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31169 2017-03-13 Richard Biener <rguenther@suse.de>
31172 * params.def (vect-max-peeling-for-alignment): Fix typo.
31174 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31176 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31177 issue that only occurred with binutils below 2.18.
31179 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31181 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31182 refer to binutils 2.11/2.12 minimum.
31184 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31186 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31187 ftp.kernel.org and simplify binutils requirement.
31189 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31191 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31193 (Optimize Options): Ditto. Also remove redundancy.
31195 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31197 PR translation/79848
31198 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31200 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31201 to G_ to avoid double translation.
31203 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31205 PR translation/79923
31206 * auto-profile.c (get_combined_location): Convert leading
31207 character of diagnostics to lower case and remove trailing period.
31208 (read_profile): Likewise for various diagnostics.
31209 * config/arm/arm.c (arm_option_override): Remove trailing period
31210 from various diagnostics.
31211 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31212 (msp430_expand_delay_cycles): Likewise.
31214 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31217 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31218 full command-line argument, rather than just "str".
31219 (aarch64_validate_march): Likewise.
31220 (aarch64_validate_mtune): Likewise.
31222 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31224 PR rtl-optimization/78911
31225 * lra-assigns.c (must_not_spill_p): New function.
31226 (spill_for): Use it.
31228 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31230 PR tree-optimization/79981
31231 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31232 ATOMIC_COMPARE_EXCHANGE ifn result.
31233 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31234 IFN_ATOMIC_COMPARE_EXCHANGE.
31236 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31239 * opts.c (parse_sanitizer_options): Add missing question mark to
31240 "did you mean" message.
31242 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31244 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31246 (VMULEUH_UNS): Likewise.
31247 (VMULOUB_UNS): Likewise.
31248 (VMULOUH_UNS): Likewise.
31249 * config/rs6000/rs6000.c (builtin_function_type): Remove
31250 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31252 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31255 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31256 x with result of extra_parsing_for_operand_code_0.
31257 (function_reader::extra_parsing_for_operand_code_0): Convert
31258 return type from void to rtx, returning x. When reading
31259 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31260 larger size containing struct block_symbol.
31262 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31264 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31265 -mfloat128-hardware without -m64.
31267 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31270 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31271 entries to the case statement that marks unsigned arguments to
31272 overloaded functions.
31274 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31276 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31277 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31279 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31282 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31283 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31285 2017-03-10 Martin Liska <mliska@suse.cz>
31289 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31290 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31293 2017-03-10 Olivier Hainque <hainque@adacore.com>
31295 * tree-switch-conversion (array_value_type): Start by resetting
31296 candidate type to it's main variant.
31298 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31300 PR rtl-optimization/79909
31301 * combine.c (try_combine): Use simplify_replace_rtx on individual
31302 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31303 of the whole CALL_INSN_FUNCTION_USAGE.
31305 PR tree-optimization/79972
31306 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31307 get_range_info on SSA_NAMEs. Formatting fixes.
31309 2017-03-10 Richard Biener <rguenther@suse.de>
31310 Jakub Jelinek <jakub@redhat.com>
31312 PR tree-optimization/77975
31313 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31314 edge to be constant.
31315 (get_val_for): For constant x return it. Formatting fix.
31316 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31317 would use the same bases as the current one.
31319 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31321 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31322 instead of vec_select for V1TImode.
31323 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31325 (VSX_LE_128): Add V1TI to this mode iterator.
31326 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31327 (*vsx_le_perm_store_<mode>): Likewise.
31328 (pre-reload splitter for VSX stores): Likewise.
31329 (post-reload splitter for VSX stores): Likewise.
31330 (*vsx_xxpermdi2_le_<mode>): Likewise.
31331 (*vsx_lxvd2x2_le_<mode>): Likewise.
31332 (*vsx_stxvd2x2_le_<mode>): Likewise.
31334 2017-03-09 Michael Eager <eager@eagercon.com>
31336 Correct failures with --enable-checking=yes,rtl.
31338 * config/microblaze/microblaze.c (microblaze_expand_shift):
31339 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31340 test for const0_rtx.
31341 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31342 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31344 2017-03-09 Richard Biener <rguenther@suse.de>
31346 PR tree-optimization/79977
31347 * graphite-scop-detection.c (scop_detection::merge_sese):
31348 Handle the case of extra exits to blocks dominating the entry.
31350 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31352 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31355 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31357 PR rtl-optimization/79949
31358 * lra-constraints.c (process_alt_operands): Check memory when
31359 trying to predict a cycle. Print about the overall increase.
31361 2017-03-09 Richard Biener <rguenther@suse.de>
31363 PR middle-end/79971
31364 * gimple-expr.c (useless_type_conversion_p): Preserve
31365 TYPE_SATURATING for fixed-point types.
31367 2017-03-09 Richard Biener <rguenther@suse.de>
31370 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31371 alignment of BLKmode params.
31373 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31376 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31377 (VALL_NO_V2Q): Likewise.
31379 * config/aarch64/aarch64-simd.md
31380 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31382 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31383 VALL_NO_V2Q mode iterator.
31384 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31386 2017-03-09 Martin Liska <mliska@suse.cz>
31388 PR tree-optimization/79631
31389 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31390 tree_int_cst_sign_bit just for INTEGER constants.
31392 2017-03-09 Martin Liska <mliska@suse.cz>
31396 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31399 2017-03-09 Marek Polacek <polacek@redhat.com>
31402 * tree.c (inchash::add_expr): Handle TREE_VEC.
31404 2017-03-09 Martin Liska <mliska@suse.cz>
31407 (chkp_narrow_size_and_offset): New function.
31408 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31409 (void chkp_parse_bit_field_ref): New function.
31410 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31411 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31413 2017-03-09 Martin Liska <mliska@suse.cz>
31416 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31417 (chkp_find_bounds_1): Remove gcc_unreachable.
31419 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31422 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31423 BUILT_IN_SYNC*, determine the access type from the size suffix and
31424 always build a MEM_REF with that type. Handle forgotten
31425 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31428 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31429 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31430 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31431 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31432 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31433 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31434 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31435 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31436 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31437 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31438 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31439 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31440 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31441 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31442 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31443 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31444 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31445 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31446 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31447 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31448 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31449 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31450 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31451 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31452 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31453 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31454 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31455 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31456 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31457 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31458 definitions outside of __OPTIMIZE__ guarded section.
31461 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31462 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31463 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31464 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31467 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31469 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31470 ("vfenez<mode>"): Add missing constraints.
31472 2017-03-08 Martin Sebor <msebor@redhat.com>
31475 * config/nds32/nds32.c (nds32_option_override):
31476 Fix misspelled diagnostic.
31478 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31481 * gimplify.c (gimplify_omp_for): Replace index var in outer
31482 taskloop statement with an artificial variable and add
31483 OMP_CLAUSE_PRIVATE clause for it.
31485 2017-03-08 Richard Biener <rguenther@suse.de>
31487 PR tree-optimization/79955
31488 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31489 for accesses that are completely outside of the variable.
31491 2017-03-08 Andrew Haley <aph@redhat.com>
31493 PR tree-optimization/79943
31494 * tree-ssa-loop-split.c (compute_new_first_bound): When
31495 calculating the new upper bound, (END-BEG) should be added, not
31498 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31500 * config/avr/avr.md (setmemhi): Make sure match_dup
31501 operand number comes before match_scratch.
31503 2017-03-08 Richard Biener <rguenther@suse.de>
31505 PR tree-optimization/79920
31506 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31507 with ncopies == 1 to ...
31508 (vect_transform_slp_perm_load): ... here. Properly compute
31509 all element loads by iterating VF times over the group. Do
31510 not handle ncopies (computed in a broken way) in
31511 vect_create_mask_and_perm.
31513 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31516 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31517 is a uniform vector, use uniform_vector_p return value instead of
31518 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31520 2017-03-07 Marek Polacek <polacek@redhat.com>
31522 PR middle-end/79809
31523 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31524 (alloca_call_type): Likewise.
31526 2017-03-07 Martin Liska <mliska@suse.cz>
31528 * gcov.c (process_args): Put comment to correct location.
31530 2017-03-07 Martin Liska <mliska@suse.cz>
31532 PR middle-end/68270
31533 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31534 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31535 (chkp_narrow_bounds_for_field): Likewise.
31536 (chkp_parse_array_and_component_ref): Pass one more argument to
31539 2017-03-07 Richard Biener <rguenther@suse.de>
31541 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31544 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31546 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31547 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31549 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31552 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31553 to end of description.
31554 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31556 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31558 PR rtl-optimization/79901
31559 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31561 (*avx512f_<code><mode>3<mask_name>): ... this.
31562 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31563 iterator instead of VI8_AVX2_AVX512BW.
31565 PR rtl-optimization/79901
31566 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31567 min/max expander, expand it using expand_vec_cond_expr.
31570 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31573 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31576 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31577 to void * for PCH reasons.
31578 * dwarf2out.c (output_loc_operands, output_die): Cast
31579 v.val_vec.array to unsigned char *.
31581 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31584 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31587 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31589 PR rtl-optimization/79571
31590 * lra-constraints.c (process_alt_operands): Calculate static
31591 reject and subtract it from overall when only addresses will be
31594 2017-03-06 Julia Koval <julia.koval@intel.com>
31597 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31598 incoming stack boundary to 128 for 64-bit targets.
31600 2017-03-06 Richard Biener <rguenther@suse.de>
31602 PR tree-optimization/79894
31603 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31604 to NULL after folding it.
31606 2017-03-06 Richard Biener <rguenther@suse.de>
31608 PR tree-optimization/79824
31609 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31610 check disabling peeling for gaps.
31612 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31614 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31615 attributes): Document gettimeofday.
31617 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31619 * config/s390/s390.c (s390_option_override_internal): Set
31620 PARAM_MIN_VECT_LOOP_BOUND
31622 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31624 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31625 * config/s390/s390.md: Likewise.
31627 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31630 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31631 (<avx2_avx512>_perm<mode>): Rename to ...
31632 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31634 (<avx512>_perm<mode>_mask): Rename to ...
31635 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31637 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31638 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31639 instead of VI8F_256_512.
31640 (avx512f_perm<mode>): New define_expand.
31641 (avx512f_perm<mode>_mask): Likewise.
31642 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31643 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31645 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31647 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31648 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31650 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31652 2017-03-06 Martin Liska <mliska@suse.cz>
31655 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31656 when having a SSA NAME w/o VAR_DECL assigned to it.
31658 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31660 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31661 msa_dpsub_<su>_d): Fix MODE for vec_select.
31663 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31665 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31667 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31669 2017-03-06 Richard Biener <rguenther@suse.de>
31671 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31672 * plugin.c (register_plugin_info): Likewise.
31673 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31675 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31677 * config/i386/sse.md (sse_storehps, sse_storelps,
31678 avx_<castmode><avxsizesuffix>_<castmode>,
31679 avx512f_<castmode><avxsizesuffix>_<castmode>,
31680 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31681 in condition that at least one operand is not a MEM.
31683 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31685 PR middle-end/79805
31686 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31687 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31689 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31690 gimple_call_nothrow_p flag based on whether original builtin can throw.
31691 If it can, emit following stmts on the fallthrough edge.
31692 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31693 don't create new bb if inserting just debug stmts on the edge, try to
31694 insert them on the fallthru bb or just reset debug stmts.
31696 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31699 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31700 restore recog_data (including the operand rtxes inside it) around
31701 the call to get_insn_template.
31703 2017-03-03 Martin Sebor <msebor@redhat.com>
31705 PR tree-optimization/79699
31706 * context.c (context::~context): Free MPFR caches to avoid
31707 a memory leak on program exit.
31709 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31711 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31712 Use wide_int::ulow () instead of .elt (0).
31714 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31716 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31717 (*pushxf): Limit oF constraint to 32bit targets and add oC
31718 constraint for 64bit targets.
31719 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31720 (*pushdf): Change rmF constraint to rmC.
31722 2017-03-03 Martin Liska <mliska@suse.cz>
31724 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31725 Remove unused variable.
31727 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31730 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31731 is a memory operand, increase num_memory.
31732 (ix86_expand_args_builtin): Likewise.
31734 2017-03-03 Jan Hubicka <jh@suse.cz>
31737 * ipa-devirt.c (maybe_record_node): Properly handle
31738 __cxa_pure_virtual visibility.
31740 2017-03-03 Martin Liska <mliska@suse.cz>
31742 PR tree-optimization/79803
31743 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31745 (pass_loop_prefetch::execute): Disabled optimization if an
31746 assumption about L1 cache size is not met.
31748 2017-03-03 Martin Liska <mliska@suse.cz>
31750 PR rtl-optimization/79574
31751 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31752 (hash_scan_set): Likewise.
31753 (dump_hash_table): Likewise.
31754 (hoist_code): Likewise.
31756 2017-03-03 Richard Biener <rguenther@suse.de>
31758 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31759 in place of uhigh ().
31760 (fixed_convert_from_real): Likewise.
31762 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31765 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31767 2017-03-03 Richard Biener <rguenther@suse.de>
31769 PR middle-end/79818
31770 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31771 TYPE_OVERFLOW_UNDEFINED check.
31773 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31775 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31777 (vector_ae_<mode>_p): Likewise.
31778 (vector_nez_<mode>_p): Likewise.
31779 (vector_ne_v2di_p): Likewise.
31780 (vector_ae_v2di_p): Likewise.
31781 (vector_ne_<mode>_p): Likewise.
31782 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31784 (vsx_tsqrt<mode>2_fe): Likewise.
31786 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31789 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31791 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31793 PR rtl-optimization/79780
31794 * cprop.c (one_cprop_pass): When second and further conditional trap
31795 in a single basic block is turned into an unconditional trap, turn it
31796 into a deleted note to avoid RTL verification failures.
31798 2017-03-02 Richard Biener <rguenther@suse.de>
31800 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31802 2017-03-02 Richard Biener <rguenther@suse.de>
31804 PR tree-optimization/79345
31806 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31807 param and abort the walk, returning -1 if it is hit.
31808 (walk_aliased_vdefs): Take a limit param and pass it on.
31809 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31810 defaulting to 0 and return a signed int.
31811 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31812 (check_defs): New helper.
31813 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31814 about uninitialized memory.
31815 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31816 bogus uninitialized warning.
31817 (fixed_convert_from_real): Likewise.
31819 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31821 PR tree-optimization/66768
31822 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31823 iv_use if base object can't be determined.
31825 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31827 PR tree-optimization/79345
31828 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31829 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31830 (get_pattern_stats): Initialize it.
31831 * genemit.c (gen_expand): Verify match_scratch numbers come after
31832 match_operand/match_dup numbers.
31833 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31834 match_scratch numbers.
31835 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31837 * config/s390/s390.md (trunctdsd2): Likewise.
31839 2017-03-02 Richard Biener <rguenther@suse.de>
31841 * wide-int.h (wide_int_storage::operator=): Implement in terms
31844 2017-03-02 Richard Biener <rguenther@suse.de>
31846 PR tree-optimization/79777
31847 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31848 the to insert expression to sth existing.
31850 2017-03-01 Martin Sebor <msebor@redhat.com>
31852 PR middle-end/79692
31853 * gimple-ssa-sprintf.c
31854 (directive::known_width_and_precision): New function.
31855 (format_integer): Use it.
31856 (get_mpfr_format_length): Consider the full range of precision
31857 when computing %g output with the # flag. Set the likely byte
31858 count to 3 rather than 1 when precision is indeterminate.
31859 (format_floating): Correct the lower bound of precision.
31861 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31863 * doc/invoke.texi: Document default code model for 64-bit Linux.
31865 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31868 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31869 udiv rather than div since input pattern is unsigned.
31871 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31873 * config/i386/i386.c (print_reg): Warn for values of
31874 unsupported size in integer register.
31876 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31879 * config/rs6000/predicates.md (current_file_function_operand): Do
31880 not allow self calls to be local if the function is replaceable.
31882 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31885 * config/rs6000/altivec.h (vec_ctz and others): Change the
31886 preprocessor macro that controls conditional compilation from
31887 _ARCH_PWR9 to __POWER9_VECTOR__.
31888 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31889 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31890 control (instead of _ARCH_PWR9 control) so that template
31891 definition uses power9-specific function.
31892 (vec_any_eq): Likewise.
31893 (vec_all_ne): Change macro definition to use a power9-specific
31894 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31895 _ARCH_PWR9 control).
31896 (vec_any_eq) Likewise.
31897 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31898 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31899 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31900 support for xvcmpnedp instruction.
31901 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31902 macro expansion so that Power9 implementation of vec_all_ne does
31903 not use the AltiVec predicate framework.
31904 (VCMPNEH_P): Likewise.
31905 (VCMPNEW_P): Likewise.
31906 (VCMPNED_P): Likewise.
31907 (VCMPNEFP_P): Likewise.
31908 (VCMPNEDP_P): Likewise.
31909 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31910 implementation of vec_any_eq to not use AltiVec predicate
31912 (VCMPAEH_P): Likewise.
31913 (VCMPAEW_P): Likewise.
31914 (VCMPAED_P): Likewise.
31915 (VCMPAEFP_P): Likewise.
31916 (VCMPAEDP_P): Likewise.
31917 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31918 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31919 not use the AltiVec predicate framework.
31920 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31921 of vec_any_eq to not use AltiVec predicate framework.
31922 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31923 support for predefined __POWER9_VECTOR__ macro to indicate that
31924 Power9 instruction selection is enabled.
31925 (altivec_overloaded_builtins): Remove extraneous
31926 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31927 function argument types RS6000_BTI_bool_V16QI and
31928 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31929 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31930 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31931 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31932 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31933 Power9 for implementations of vec_cmpne. Change the signature for
31934 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31935 (representing vec_all_ne) to remove the previously described first
31936 argument of type RS6000_BTI_INTSI, as this was an artifact of
31937 reliance on the AltiVec predicate framework, which is no longer
31938 used in the implementation of these functions. Add
31939 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31940 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31941 since, unlike the AltiVec predicate framework implementation, we
31942 do not share function descriptors between vec_alle and vec_anyeq.
31943 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31944 set of modes that receive special treatment even when
31945 TARGET_P9_VECTOR is true. The special treatment emits code that
31946 does not depend on Power9 instructions.
31947 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31948 define_expand to not rely on AltiVec predicate framework.
31949 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31951 (vector_ne_v2di_p): Change this define_expand to not rely on
31952 AltiVec predicate framework.
31953 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31955 (vector_ne_<mode>_p): Change this define_expand to not rely on
31956 AltiVec predicate framework.
31957 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31959 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31960 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31961 define_insn pattern.
31962 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31963 define_insn pattern because the xvcmpne<VSs>. instruction is not
31965 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31966 instruction is not supported.
31968 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31970 * config/nvptx/nvptx.c: Include intl.h.
31972 2017-03-01 Martin Jambor <mjambor@suse.cz>
31975 * ipa-prop.h (ipa_bits): Removed field known.
31976 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31977 to pointers. Adjusted their comments to warn about their sharing.
31978 (ipcp_transformation_summary): Change bits to a vector of pointers.
31979 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31980 (ipa_get_ipa_bits_for_value): Declare.
31981 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31982 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31983 (ipa_bits_hash_table): Likewise.
31984 (ipa_vr_ggc_hash_traits): Likewise.
31985 (ipa_vr_hash_table): Likewise.
31986 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31987 being pointers and vr_known being removed.
31988 (ipa_set_jf_unknown): Likewise.
31989 (ipa_get_ipa_bits_for_value): New function.
31990 (ipa_set_jfunc_bits): Likewise.
31991 (ipa_get_value_range): New overloaded functions.
31992 (ipa_set_jfunc_vr): Likewise.
31993 (ipa_compute_jump_functions_for_edge): Use the above functions to
31994 construct bits and vr parts of jump functions.
31995 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31996 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31998 (ipcp_grow_transformations_if_necessary): Also allocate
31999 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32001 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32002 them. Fix too long lines.
32003 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32004 vr_known being removed.
32005 (ipa_read_jump_function): Use new setter functions to construct bits
32006 and vr parts of jump functions or set them to NULL.
32007 (write_ipcp_transformation_info): Adjust for bits being pointers.
32008 (read_ipcp_transformation_info): Likewise.
32009 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32011 Include gt-ipa-prop.h.
32012 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32014 (ipcp_store_bits_results): Likewise.
32015 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32016 Do not write to existing jump functions but use a temporary instead.
32018 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32021 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32022 attempt to use its first operand as BIT_FIELD_REF base.
32024 2017-03-01 Richard Biener <rguenther@suse.de>
32026 PR middle-end/79721
32027 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32028 interpolating formula in wrapping arithmetic.
32029 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32031 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32033 PR tree-optimization/79734
32034 * tree-vect-generic.c (expand_vector_condition): Optimize
32035 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32036 Handle VEC_COND_EXPR where comparison has different inner width from
32037 type's inner width.
32039 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32041 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32042 markup, and similar issues. Remove @opindex entries for things
32043 that aren't options. Add missing -mmpy-option entries.
32045 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32047 PR tree-optimization/79737
32048 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32049 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32050 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32051 instead of byte_size. Formatting fix.
32052 (shift_bytes_in_array_right): Formatting fix.
32054 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32057 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32058 condition on optimize for the leaf function test.
32060 2017-02-28 Martin Liska <mliska@suse.cz>
32063 * read-rtl-function.c (function_reader::handle_unknown_directive):
32064 Bail out when one uses -flto.
32066 2017-02-28 Martin Liska <mliska@suse.cz>
32068 * common.opt: Replace space with tabular for options of <number>
32070 * config/i386/i386.opt: Show <number> value for
32071 -mlarge-data-threshold.
32072 * opts.c (print_filtered_help): Do not display number in hexadecimal
32075 2017-02-28 Martin Liska <mliska@suse.cz>
32077 * common.opt: Fix --help=option -Q for options which are of
32080 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32082 * config/i386/i386.c (print_reg): Error out for values
32083 of 8-bit size in invalid integer register.
32085 2017-02-28 Martin Sebor <msebor@redhat.com>
32087 PR tree-optimization/79691
32088 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32090 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32093 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32094 gcc_unreachable with output_operand_lossage.
32096 2017-02-28 Richard Biener <rguenther@suse.de>
32098 PR tree-optimization/79740
32099 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32101 (visit_nary_op): Insert the nary into the hashtable if we
32102 pattern-matched sth.
32103 * tree-ssa-pre.c (eliminate_insert): Robustify.
32105 2017-02-28 Richard Biener <rguenther@suse.de>
32107 PR middle-end/79731
32108 * fold-const.c (decode_field_reference): Reject out-of-bound
32111 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32113 * config/i386/i386.c: Include intl.h.
32114 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32115 instead of just cond ? "..." : "...".
32116 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32117 * coverage.c (read_counts_file): Likewise.
32118 * omp-offload.c: Include intl.h.
32119 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32120 of just cond ? "..." : "...".
32121 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32122 of just cond ? "..." : "...".
32124 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32127 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32129 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32130 'tune for' CPU name.
32131 * config/arm/arm-cpu-data.h: Regenerated.
32133 2017-02-28 Richard Biener <rguenther@suse.de>
32135 PR tree-optimization/79732
32136 * tree-inline.c (expand_call_inline): Do not shadow var.
32138 2017-02-28 Richard Biener <rguenther@suse.de>
32140 PR tree-optimization/79723
32141 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32142 address-space properly.
32144 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32146 * doc/optinfo.texi (Optimization groups): Fix option used for
32148 * doc/invoke.texi (-fopt-info): Document "omp".
32149 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32150 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32151 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32152 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32153 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32155 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32157 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32158 OPTGROUP_OMP with "omp".
32160 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32163 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32164 for arithmetic shift of unsigned V2DI.
32166 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32168 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32169 arc/linux.h headers.
32170 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32171 (LINK_SPEC): Likewise.
32172 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32173 (EXTRA_SPECS): Likewise.
32174 (STARTFILE_SPEC): Likewise.
32175 (ENDFILE_SPEC): Likewise.
32176 (LIB_SPEC): Likewise.
32177 (TARGET_SDATA_DEFAULT): Likewise.
32178 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32179 (MULTILIB_DEFAULTS): Likewise.
32180 (DWARF2_UNWIND_INFO): Likewise.
32181 * config/arc/big.h: New file.
32182 * config/arc/elf.h: Likewise.
32183 * config/arc/linux.h: Likewise.
32184 * config/arc/t-uClibc: Remove.
32186 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32188 PR tree-optimization/77536
32189 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32190 (tree_transform_and_unroll_loop): Use above function to compute the
32191 estimated niter of unrolled loop and use it when scaling profile.
32192 Also use count info rather than frequency if it's non-zero.
32193 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32194 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32195 (vect_transform_loop): Call above function.
32197 2017-02-27 Richard Biener <rguenther@suse.de>
32199 PR tree-optimization/45397
32200 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32201 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32202 (visit_nary_op): Add pattern matching for CSEing sign-changed
32203 or truncated operations with wider ones.
32205 2017-02-27 Richard Biener <rguenther@suse.de>
32207 PR tree-optimization/79690
32208 * tree-vect-stmts.c (vectorizable_store): Use vector type
32209 built from the DR with address-space.
32211 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32213 * doc/invoke.texi (Optimize Options): Refine the description
32214 of asan-use-after-return.
32216 2017-02-25 Alan Modra <amodra@gmail.com>
32218 PR rtl-optimization/79584
32219 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32220 base, not ad->base_term, the reg within base. Remove assertion
32221 that ad->base == ad->base_term. Replace gen_int_mode using
32222 bogus mode with const0_rtx.
32224 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32226 PR middle-end/79396
32227 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32228 FMA_EXPR like tcc_binary or tcc_unary.
32230 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32233 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32235 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32236 (output_loc_operands): Handle DW_OP_call_ref and
32237 DW_OP_GNU_variable_value.
32238 (struct variable_value_struct): New type.
32239 (struct variable_value_hasher): Likewise.
32240 (variable_value_hash): New variable.
32241 (string_types): Remove.
32242 (copy_loc_descr): New function.
32243 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32244 (prepend_loc_descr_to_each): New function.
32245 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32246 instead of add_loc_descr_to_each if the first argument is single
32247 location list and the second has multiple.
32248 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32249 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32250 when looking for variable value which doesn't have other location info.
32251 (loc_list_from_tree): Formatting fix.
32252 (gen_array_type_die): Simplify DW_AT_string_length handling.
32253 (adjust_string_types): Remove.
32254 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32255 string_types. Call resolve_variable_values.
32256 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32257 (resolve_addr_in_expr): Likewise. Add A argument.
32258 (copy_deref_exprloc): Remove deref argument. Adjust for the
32259 original expression being DW_OP_GNU_variable_value with optionally
32260 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32261 optionally after it.
32262 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32263 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32264 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32265 (variable_value_hasher::hash, variable_value_hasher::equal): New
32267 (resolve_variable_value_in_expr, resolve_variable_value,
32268 resolve_variable_values, note_variable_value_in_expr,
32269 note_variable_value): New functions.
32270 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32273 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32276 * opts.h (handle_generated_option): Add GENERATED_P argument.
32277 * opts-common.c (handle_option): Adjust function comment.
32278 (handle_generated_option): Add GENERATED_P argument, pass it to
32280 (control_warning_option): Pass false to handle_generated_option
32282 * opts.c (maybe_default_option): Pass true to handle_generated_option
32284 * optc-gen.awk: Likewise.
32286 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32288 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32289 a REG, look at the REG it is a SUBREG of.
32290 (splitter for cmpeqsi_t): Ditto.
32292 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32294 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32295 the special USEs with the pattern of the insn, not the insn itself.
32297 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32300 * doc/invoke.texi: Document -mload-store-pairs.
32302 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32303 Sandra Loosemore <sandra@codesourcery.com>
32305 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32306 argument isn't a CONST_INT.
32307 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32308 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32309 (nios2_validate_compare): Bypass alternate compare logic if *op2
32310 is not a CONST_INT.
32311 (ldstwm_operation_p): Return false if first_base is not a REG or
32312 if first_offset is not a CONST_INT.
32314 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32316 * config/cris/cris.md: Use correct operand in a define_peephole2.
32318 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32320 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32322 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32324 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32325 this_insn if it is an INSN or JUMP_INSN.
32326 (force_offsettable): Look at base, not at addr.
32327 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32328 on things that aren't necessarily CONST_INTs.
32330 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32332 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32333 -mfpmath=sse is the default also for x86-32 targets with SSE2
32334 instruction set when @option{-ffast-math} is enabled
32336 2017-02-24 Jeff Law <law@redhat.com>
32338 PR rtl-optimizatoin/79286
32339 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32342 2017-02-24 Richard Biener <rguenther@suse.de>
32344 PR tree-optimization/79389
32345 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32348 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32350 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32351 function comment to reflect reality.
32352 (loop_exits_before_overflow): Fix typo in function description.
32354 2017-02-24 Richard Biener <rguenther@suse.de>
32356 PR tree-optimization/79389
32357 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32358 properly that a threading opportunity exists. Detect conditional
32359 copy/constant propagation opportunities.
32361 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32363 * config/visium/visium.md (type): Add trap.
32364 (b): New mode attribute.
32365 (*btst): Rename into...
32366 (*btst<mode>): ...this and adjust.
32367 (*cbranchsi4_btst_insn): Rename into...
32368 (*cbranch<mode>4_btst_insn): ...this and adjust.
32369 (trap): New define_insn.
32371 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32373 PR tree-optimization/79389
32374 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32375 (noce_reversed_cond_code): New function.
32376 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32377 reversed_comparison_code. Formatting fix.
32378 (noce_try_store_flag): Test rev_cond != NULL in addition to
32379 reversed_comparison_code.
32380 (noce_try_store_flag_constants): Likewise.
32381 (noce_try_store_flag_mask): Likewise.
32382 (noce_try_addcc): Use rev_cond if non-NULL instead of
32383 reversed_comparison_code.
32384 (noce_try_cmove_arith): Likewise. Formatting fixes.
32385 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32386 (noce_find_if_block): Initialize rev_cond.
32387 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32388 instead of false as last argument never attempt to reverse it
32391 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32393 PR tree-optimization/79663
32394 * tree-predcom.c (combine_chains): Process refs in reverse order
32395 only for ZERO length chains, and add explaining comment.
32397 2017-02-23 Jeff Law <law@redhat.com>
32399 PR tree-optimization/79578
32400 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32401 in call to operand_equal_p.
32403 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32406 * config/i386/cpuid.h: Fix another undefined behavior.
32408 2017-02-23 Richard Biener <rguenther@suse.de>
32410 PR tree-optimization/79683
32411 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32412 vector types for data-refs.
32414 2017-02-23 Martin Liska <mliska@suse.cz>
32416 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32418 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32420 PR middle-end/79665
32421 * internal-fn.c (get_range_pos_neg): Moved to ...
32422 * tree.c (get_range_pos_neg): ... here. No longer static.
32423 * tree.h (get_range_pos_neg): New prototype.
32424 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32425 are known to be in between 0 and signed maximum inclusive, try to
32426 expand both unsigned and signed divmod and use the cheaper one from
32429 2017-02-22 Jeff Law <law@redhat.com>
32431 PR tree-optimization/79578
32432 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32433 to compare base operands.
32435 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32438 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32439 gpc_reg_operand instead of fpr_reg_operand.
32441 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32443 * config/mips/mips.c (mips_return_in_memory): Force FP
32444 vector types to be returned in memory for o32 ABI.
32446 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32448 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32449 instead of DW_TAG_member for static data member declarations and don't
32450 set no_linkage_name for static inline data members.
32451 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32454 2017-02-22 Martin Liska <mliska@suse.cz>
32456 * doc/invoke.texi: Replace inequality signs with square brackets
32459 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32462 * lra-constraints.c (simplify_operand_subreg): Handle
32463 WORD_REGISTER_OPERATIONS targets.
32465 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32468 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32469 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32470 elimination by swapping fld*.
32472 2017-02-22 Richard Biener <rguenther@suse.de>
32474 PR tree-optimization/79673
32475 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32476 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32477 irrelevant address-space qualifiers and avoiding a
32478 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32480 2017-02-22 Richard Biener <rguenther@suse.de>
32482 PR tree-optimization/79666
32483 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32484 to not symbolically negate if that may introduce undefined
32487 2017-02-22 Martin Liska <mliska@suse.cz>
32490 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32491 * data-streamer-out.c (streamer_write_gcov_count_stream):
32493 * value-prof.c (stream_out_histogram_value): Make assert more
32494 precise based on type of counter.
32496 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32499 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32500 Use nonimmediate_operand instead of memory_operand for operand 1.
32501 (float-extend standard_x87sse_constant_load splitter): Ditto.
32503 2017-02-21 Jeff Law <law@redhat.com>
32505 PR tree-optimization/79621
32506 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32507 blocks with edges to themselves.
32509 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32512 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32513 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32514 Use gimple_call_builtin_p.
32517 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32518 on temporarily removed DEBUG_INSNs.
32520 PR tree-optimization/79649
32521 * tree-loop-distribution.c (classify_partition): Give up on
32522 non-generic address space loads/stores.
32524 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32526 * doc/loop.texi (Loop manipulation): Remove nonexistent
32527 tree_ssa_loop_version from the documentation.
32528 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32530 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32533 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32534 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32535 * config/rs6000/rs6000.c: Include except.h.
32536 (rs6000_expand_split_stack_prologue): Call
32537 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32539 2017-02-21 Martin Jambor <mjambor@suse.cz>
32542 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32543 have been analyzed.
32545 2017-02-21 Martin Jambor <mjambor@suse.cz>
32547 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32548 for backward compatibility only.
32549 * doc/invoke.texi (Option Summary): Remove all references to
32550 -fipa-cp-alignment.
32552 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32556 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32558 * lra-constraints.c (curr_insn_transform): Handle
32559 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32561 2017-02-21 Martin Liska <mliska@suse.cz>
32563 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32565 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32567 PR translation/79638
32568 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32570 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32573 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32574 (arm_function_ok_for_sibcall): Return false for an indirect call by
32575 descriptor if all the argument registers are used.
32576 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32577 alignment of the function.
32579 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32581 PR tree-optimization/61441
32582 * simplify-rtx.c (simplify_const_unary_operation): For
32583 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32584 the sNaN unmodified.
32586 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32588 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32589 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32590 instead of SYSTEM_HEADER_DIR.
32592 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32593 Martin Liška <mliska@suse.cz>
32595 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32596 Fix typos and grammar, use active voice, and clarify.
32598 2017-02-20 Marek Polacek <polacek@redhat.com>
32600 PR middle-end/79537
32601 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32604 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32606 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32609 * config/i386/i386.c (ix86_expand_builtin): Handle
32610 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32611 ix86_builtins_isa[fcode].isa as a requirement of those
32612 flags and any other flag in the bitmask.
32613 (ix86_init_mmx_sse_builtins): Use 0 instead of
32614 ~OPTION_MASK_ISA_64BIT as mask.
32615 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32616 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32617 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32618 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32620 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32623 * lra-constraints.c (split_reg): Check requested split mode
32624 is supported by the register.
32626 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32628 * lra-constraints.c (simplify_operand_subreg): Remove early
32631 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32634 * lra-constraints.c (curr_insn_transform): Tighten condition
32635 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32637 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32640 * lra-constraints.c (curr_insn_transform): Handle
32641 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32643 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32646 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32648 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32650 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32653 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32656 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32658 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32661 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32663 * final.c (last_columnnum, override_columnnum): New variables.
32664 (final_start_function): Set last_columnnum, pass it to begin_prologue
32665 hook and pass 0 to dwarf2out_begin_prologue.
32666 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32667 to source_line debug hook.
32668 (notice_source_line): Compute last_columnnum and for debug_column_info
32669 return true on column changes.
32670 * debug.h (struct gcc_debug_hooks): Add column argument to
32671 source_line and begin_prologue hooks.
32672 (debug_nothing_int_charstar_int_bool): Remove prototype.
32673 (debug_nothing_int_int_charstar,
32674 debug_nothing_int_int_charstar_int_bool): New prototypes.
32675 (dwarf2out_begin_prologue): Add column argument.
32676 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32677 begin_prologue hooks.
32678 (debug_nothing_int_charstar_int_bool): Remove.
32679 (debug_nothing_int_int_charstar,
32680 debug_nothing_int_int_charstar_int_bool): New functions.
32681 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32682 through to dwarf2out_source_line.
32683 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32684 (dwarf2out_source_line): Add column argument, emit it if requested.
32685 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32687 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32688 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32689 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32690 through to dwarf2out_begin_prologue.
32691 (vmsdbgout_source_line): Add column argument, pass it through to
32692 dwarf2out_source_line.
32693 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32694 dbxout_source_line caller.
32695 (dbxout_source_line): Add column argument.
32697 * common.opt (gno-column-info, gcolumn-info): New options.
32698 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32699 (check_die): Also test for multiple DW_AT_decl_column attributes.
32700 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32701 DW_AT_decl_column if requested.
32702 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32704 (gen_variable_die): Likewise.
32705 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32706 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32709 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32710 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32711 (ix86_handle_option): Handle OPT_m3dnowa.
32712 * doc/invoke.texi (-m3dnowa): Document.
32713 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32714 -m3dnowa instead of -m3dnow -march=athlon.
32717 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32718 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32720 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32723 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32724 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32725 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32726 generator for vsx_xxpermdi_<mode>_be.
32727 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32728 force big-endian semantics.
32729 (vsx_xxpermdi_<mode>_be): New define_expand with same
32730 implementation as previous version of vsx_xxpermdi_<mode>.
32732 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32734 PR tree-optimization/79327
32735 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32736 variable, its initialization and use.
32738 2017-02-17 Julia Koval <julia.koval@intel.com>
32740 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32741 (OPTION_MASK_ISA_PKU_UNSET): New.
32742 (ix86_handle_option): Handle -mrdpid.
32743 * config/i386/cpuid.h (bit_RDPID): New.
32744 * config/i386/driver-i386.c (host_detect_local_cpu):
32745 Detect RDPID feature.
32746 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32747 * config/i386/i386-c.c (ix86_target_macros_internal):
32749 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32750 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32751 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32752 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32753 * config/i386/i386.md (define_insn "rdpid"): New.
32754 * config/i386/i386.opt Add -mrdpid.
32755 * config/i386/immintrin.h (_rdpid_u32): New.
32757 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32759 PR rtl-optimization/79541
32760 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32761 instead of transforming it into USE.
32763 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32765 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32766 If HONOR_SNANS (SFmode) force the input to a register.
32767 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32768 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32769 an frsp or similar insn.
32771 2017-02-17 Martin Liska <mliska@suse.cz>
32773 PR rtl-optimization/79577
32774 * params.def (selsched-max-sched-times): Increase minimum to 1.
32776 2017-02-17 Martin Liska <mliska@suse.cz>
32778 PR rtl-optimization/79574
32779 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32781 2017-02-17 Martin Liska <mliska@suse.cz>
32783 PR tree-optimization/79529
32784 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32785 ssa_defined_default_def_p to handle cases which are implicitly
32787 * tree-ssa.c (ssa_defined_default_def_p): New function.
32788 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32789 which are implicitly defined.
32790 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32792 2017-02-17 Richard Biener <rguenther@suse.de>
32794 PR middle-end/79576
32795 * params.def (max-ssa-name-query-depth): Limit to 10.
32797 2017-02-17 Richard Biener <rguenther@suse.de>
32799 PR tree-optimization/79552
32800 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32803 2017-02-17 Richard Biener <rguenther@suse.de>
32806 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32808 2017-02-17 Marek Polacek <polacek@redhat.com>
32810 PR middle-end/79536
32811 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32812 (fold_negate_expr): New wrapper.
32814 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32816 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32817 Correct terminology and de-emphasize pre-standard behavior.
32819 2017-02-16 Alan Modra <amodra@gmail.com>
32821 PR rtl-optimization/79286
32822 * ira.c (def_dominates_uses): New function.
32823 (update_equiv_regs): Don't create an equivalence for insns that
32824 may trap where the register def does not dominate the use.
32826 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32828 PR rtl-optimization/78127
32829 * lra.c (lra): Call lra_eliminate before finish the loop after
32832 2017-02-16 Richard Biener <rguenther@suse.de>
32834 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32836 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32837 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32838 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32839 (isl_val_int_from_wi): New function.
32840 (extract_affine_gmp): Rename to ...
32841 (extract_affine_wi): ... this, take a widest_int.
32842 (extract_affine_int): Just wrap extract_affine_wi.
32843 (add_param_constraints): Use isl_val_int_from_wi.
32844 (add_loop_constraints): Likewise, and extract_affine_wi.
32846 2017-02-15 Jeff Law <law@redhat.com>
32848 PR middle-end/79521
32849 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32850 ira_init_register_move_cost_if_necessary.
32852 2017-02-15 Martin Sebor <msebor@redhat.com>
32854 PR middle-end/32003
32855 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32856 removed in a prior commit.
32858 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32860 PR tree-optimization/79347
32861 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32862 counters during peeling.
32864 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32866 * Makefile.in (site.exp): Remove "set ISLVER".
32868 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32871 * real.c (real_from_integer): Call real_convert even for decimal.
32873 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32876 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32878 2017-02-14 Andrew Pinski <apinski@cavium.com>
32880 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32881 cores and change the partno/implementer to be correct.
32882 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32883 the 'B" as the implementer.
32884 * config/aarch64/aarch64-tune.md: Regenerate.
32886 2017-02-14 Carl Love <cel@us.ibm.com>
32888 * config/rs6000/rs6000.c: Add case statement entry to make the
32889 xvcvuxdsp built-in argument unsigned.
32890 * config/rs6000/vsx.md: Fix the source and return operand types so they
32891 match the instruction definitions from the ISA document. Fix typo
32892 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32895 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32898 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32899 member early_clobber_alts.
32900 * lra-lives.c (reg_early_clobber_p): New.
32901 (process_bb_lives): Use it.
32902 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32903 (debug_operand_data): Initialize early_clobber_alts.
32904 (setup_operand_alternative): Set up early_clobber_alts.
32905 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32906 alternatives to new_insn_reg.
32907 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32909 (lra_update_insn_regno_info): Pass the new arg.
32911 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32913 PR middle-end/79505
32914 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32915 (new_oacc_loop_raw): Don't clear already cleared fields.
32918 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32919 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32920 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32922 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32925 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32927 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32930 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32931 the extra instruction to the right place to store 128-bit constant
32934 2017-02-14 Martin Sebor <msebor@redhat.com>
32936 PR middle-end/79448
32937 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32938 warning for strings of unknown length.
32940 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32942 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32944 2017-02-14 Jeff Law <law@redhat.com>
32947 * ira-costs.c (scan_one_insn): Initialize register move costs
32948 for pseudos seen in USE/CLOBBER insns.
32950 PR tree-optimization/79095
32951 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32952 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32953 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32954 if the operands are known to be not equal, then the resulting range
32956 (intersect_ranges): If the new range is ~[0,0] and the old range is
32957 wide, then prefer ~[0,0].
32958 * tree-vrp.c (overflow_comparison_p_1): New function.
32959 (overflow_comparison_p): New function.
32960 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32961 if NAME is used in an overflow test.
32962 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32963 overflow check that can be expressed as an equality test, then adjust
32964 ops to be that equality test.
32966 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32968 * config/s390/s390-builtin-types.def: Remove flags argument.
32969 * config/s390/s390.c (s390_init_builtins): Likewise.
32971 2017-02-14 Martin Liska <mliska@suse.cz>
32973 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32974 vector. Fix trailing white spaces.
32976 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32978 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32981 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32983 PR rtl-optimization/68664
32984 * config/arm/arm.c (arm_sched_can_speculate_insn):
32985 New function. Declare prototype.
32986 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32988 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32990 PR rtl-optimization/68664
32991 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32993 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32995 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32997 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32998 max skip bytes for function, loop and jump.
33000 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33002 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33003 ABS_EXPR for gimple dump.
33005 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33008 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33010 PR tree-optimization/79408
33011 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33012 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33013 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33014 also if rhs1 is INTEGER_CST.
33016 2017-02-14 Richard Biener <rguenther@suse.de>
33018 PR middle-end/79432
33019 * tree-into-ssa.c (insert_phi_nodes): When the function can
33020 have abnormal edges rewrite SSA names with broken use-def
33021 dominance out of SSA and register them for PHI insertion.
33023 2017-02-13 Martin Sebor <msebor@redhat.com>
33025 PR middle-end/79496
33026 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33027 clearing info.nowrite flag when snprintf size argument is a range.
33029 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33031 * cprop.c (cprop_jump): Add missing space in string literal.
33032 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33033 (get_constraint_for_component_ref): Likewise.
33034 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33035 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33036 * lra-constraints.c (process_alt_operands): Likewise.
33037 * ipa-inline.c (inline_small_functions): Likewise.
33038 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33039 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33040 * trans-mem.c (diagnose_tm_1_op): Likewise.
33041 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33042 (grid_parallel_clauses_gridifiable): Likewise.
33044 * config/nvptx/mkoffload.c (process): Add space in between
33047 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33048 "MOD4_SSE_REGS" and "ALL_REGS".
33050 * spellcheck.c (test_data): Add , in between "foo" and "food".
33052 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33055 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33056 boundary crossing check and subsequent code generation agree.
33058 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33060 * config/aarch64/aarch64.c (has_memory_op): Delete.
33061 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33064 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33066 PR rtl-optimization/79388
33067 PR rtl-optimization/79450
33068 * combine.c (distribute_notes): When removing TEM_INSN for which
33069 corresponding dest has last value recorded, invalidate that last
33072 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33074 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33075 of explicit '@'. Add missing assembly comment marker on branch costs
33078 2017-02-13 Nathan Sidwell <nathan@acm.org>
33080 * gengtype-lex.l (<in_struct>): Add '/'.
33082 2017-02-13 Martin Liska <mliska@suse.cz>
33085 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33087 2017-02-13 Richard Biener <rguenther@suse.de>
33089 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33091 * configure: Re-generate.
33092 * config.in: Likewise.
33093 * graphite-dependences.c: Simplify as if
33094 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33095 * graphite-isl-ast-to-gimple.c: Likewise.
33096 * graphite-optimize-isl.c: Likewise.
33097 * graphite-poly.c: Likewise.
33098 * graphite-sese-to-poly.c: Likewise.
33099 * graphite.h: Likewise.
33100 * toplev.c: Include isl/version.h and use isl_version () for
33101 printing the ISL version.
33102 * doc/install.texi: Update ISL requirement.
33104 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33106 * doc/standards.texi (Standards): Update reference to
33109 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33111 * doc/extend.texi (Named Address Spaces): sourceware.org now
33113 * doc/install.texi (Binaries): Ditto.
33116 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33118 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33119 standard terminology "stringize"/"stringizing" throughout.
33120 * doc/cppinternals.texi: Likewise.
33122 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33124 * doc/extend.texi: Fix some spelling mistakes and typos.
33125 * doc/invoke.texi: Likewise.
33127 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33130 * params.def (inline-min-speedup) Change from 10 to 8.
33132 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33134 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33137 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33140 * ipa-inline-analysis.c (get_minimal_bb): New function.
33141 (record_modified): Use it.
33142 (remap_edge_change_prob): Handle also ancestor functions.
33144 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33146 * doc/contrib.texi (Contributors): Remove broken link into
33147 the Mauve CVS repository.
33149 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33151 PR middle-end/79454
33152 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33153 result computation whenever lhs doesn't have vector mode, not
33154 just when it has BLKmode.
33156 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33158 * doc/makefile.texi (profiledbootstrap): Refer to the
33159 installation instructions only in textual form.
33161 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33164 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33166 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33168 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33169 (Specific): Update mingw-w64 reference.
33171 (Specific): Remove broken link to Renesas RX processor.
33173 2017-02-10 Richard Biener <rguenther@suse.de>
33175 * toplev.c (process_options): Do not mention obsolete graphite
33176 options when printing sorry message about missing graphite support.
33177 Mention -floop-nest-optimize.
33179 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33181 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33182 (vtst_p16): Likewise.
33183 (vtstq_p8): Likewise.
33184 (vtstq_p16): Likewise.
33186 (vtstq_p64): Likewise.
33187 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33188 (vset_lane_p64): New.
33189 (vsetq_lane_p64): New.
33191 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33193 PR tree-optimization/79411
33194 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33195 stmt operands are SSA_NAMEs used in abnormal phis.
33196 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33199 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33202 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33205 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33207 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33209 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33211 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33215 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33216 not arbitrary TREE_CONSTANT.
33219 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33220 "omp declare target link" attribute unless is_global_var.
33221 * omp-offload.c (find_link_var_op): Likewise.
33223 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33224 Chung-Lin Tang <cltang@codesourcery.com>
33226 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33228 (gimplify_adjust_omp_clauses): Don't delete TILE.
33229 (gimplify_omp_for): Deal with TILE.
33230 * internal-fn.c (expand_GOACC_TILE): New function.
33231 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33233 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33234 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33236 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33237 avoid DIV for outermost collapse var.
33238 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33239 Remove out of date comments, fix whitespace.
33240 * omp-general.c (omp_extract_for_data): Deal with tiling.
33241 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33242 adjust OLF_DIM_BASE value.
33243 (struct omp_for_data): Add tiling field.
33244 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33245 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33246 for auto loops. Remove default auto determining, moved to
33247 oacc_loop_fixed_partitions.
33248 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33249 stmts, add e_mask field.
33250 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33251 (oacc_thread_numbers): Use oacc_dim_call.
33252 (oacc_xform_tile): New.
33253 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33254 (finish_oacc_loop): Adjust for ifns vector.
33255 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33256 add case for GOACC_TILE fns.
33257 (oacc_loop_xform_loop): Delete.
33258 (oacc_loop_process): Iterate over call list directly, and add
33259 handling for GOACC_TILE fns.
33260 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33262 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33263 vector partitioning to outer loops. Assign 2 partitions to loops
33264 when available. Add TILE handling.
33265 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33266 (execite_oacc_device_lower): Process GOACC_TILE fns,
33267 ignore unknown specs.
33268 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33269 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33270 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33272 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33274 * configure.ac (ACX_BUGURL): Update.
33275 * configure: Regenerate.
33277 2017-02-09 Richard Biener <rguenther@suse.de>
33279 PR tree-optimization/69823
33280 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33281 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33283 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33285 * config/arc/arc-c.def: Add __NPS400__ definition.
33286 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33287 (TARGET_NPS400): Define.
33289 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33291 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33293 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33294 pointer, arch_info.
33295 (arc_cpu_types): Fill the arch_info field with a pointer into the
33296 arc_arch_types table.
33297 (arc_selected_cpu): Declare.
33298 * config/arc/arc.c (arc_selected_cpu): Make global.
33299 (arc_selected_arch): Delete.
33300 (arc_base_cpu): Delete.
33301 (arc_override_options): Remove references to deleted variables,
33302 update access to arch information.
33303 (ARC_OPT): Update access to arch information.
33304 (ARC_OPTX): Likewise.
33305 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33306 (TARGET_ARC600): Update access to arch information.
33307 (TARGET_ARC601): Likewise.
33308 (TARGET_ARC700): Likewise.
33309 (TARGET_EM): Likewise.
33310 (TARGET_HS): Likewise.
33311 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33314 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33317 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33318 condition/operands for integer GE/LE/GEU/LEU operations.
33320 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33322 PR translation/79397
33323 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33326 2017-02-08 Martin Jambor <mjambor@suse.cz>
33329 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33330 whether allocation happened.
33331 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33332 nothing was allocated.
33334 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33336 PR tree-optimization/79408
33337 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33338 constant, but SSA_NAME with a known integer range, use the minimum
33339 of that range instead of op1 to determine if modulo can be replaced
33340 with its first operand.
33342 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33344 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33346 2017-02-08 Richard Biener <rguenther@suse.de>
33348 PR tree-optimization/71824
33349 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33350 Check all loops contained in the merged region.
33352 2017-02-07 Andrew Pinski <apinski@cavium.com>
33354 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33356 2017-02-07 Andrew Pinski <apinski@cavium.com>
33358 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33359 (thunderxt88): Likewise.
33360 (thunderxt81): Disable LSE and change v8.1 to v8.
33361 (thunderxt83): Likewise.
33363 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33364 Richard Biener <rguenther@suse.de>
33366 PR middle-end/79399
33367 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33368 type from int to size_t.
33369 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33371 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33373 PR rtl-optimization/79386
33374 * cprop.c (bypass_conditional_jumps): Initialize
33375 bypass_last_basic_block already before splitting bbs after
33376 unconditional traps...
33377 (bypass_conditional_jumps): ... rather than here.
33380 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33381 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33382 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33383 fix -masm=intel patterns.
33385 2017-02-07 Richard Biener <rguenther@suse.de>
33387 PR tree-optimization/79256
33388 PR middle-end/79278
33389 * builtins.c (get_object_alignment_2): Use min_align_of_type
33390 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33391 and ADJUST_FIELD_ALIGN.
33393 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33395 * doc/tm.texi: Regenerate.
33396 * stor-layout.c (layout_decl): Adjust.
33397 (update_alignment_for_field): Likewise.
33398 (place_field): Likewise.
33399 (min_align_of_type): Likewise.
33400 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33401 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33402 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33403 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33404 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33405 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33406 * config/i386/i386.c (x86_field_alignment): Likewise.
33407 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33408 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33409 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33410 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33411 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33412 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33416 2017-01-30 Richard Biener <rguenther@suse.de>
33418 PR tree-optimization/79256
33419 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33420 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33423 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33425 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33426 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33427 builtins to SImode and emit a zero-extend, if necessary.
33429 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33431 * docs/invoke.texi (RISC-V Options): Alphabetize.
33433 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33435 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33438 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33440 * config/riscv/riscv.c: New file.
33441 * common/config/riscv/riscv-common.c: Likewise.
33442 * config.gcc: Likewise.
33443 * config/riscv/constraints.md: Likewise.
33444 * config/riscv/elf.h: Likewise.
33445 * config/riscv/generic.md: Likewise.
33446 * config/riscv/linux.h: Likewise.
33447 * config/riscv/multilib-generator: Likewise.
33448 * config/riscv/peephole.md: Likewise.
33449 * config/riscv/pic.md: Likewise.
33450 * config/riscv/predicates.md: Likewise.
33451 * config/riscv/riscv-builtins.c: Likewise.
33452 * config/riscv/riscv-c.c: Likewise.
33453 * config/riscv/riscv-ftypes.def: Likewise.
33454 * config/riscv/riscv-modes.def: Likewise.
33455 * config/riscv/riscv-opts.h: Likewise.
33456 * config/riscv/riscv-protos.h: Likewise.
33457 * config/riscv/riscv.h: Likewise.
33458 * config/riscv/riscv.md: Likewise.
33459 * config/riscv/riscv.opt: Likewise.
33460 * config/riscv/sync.md: Likewise.
33461 * config/riscv/t-elf-multilib: Likewise.
33462 * config/riscv/t-linux: Likewise.
33463 * config/riscv/t-linux-multilib: Likewise.
33464 * config/riscv/t-riscv: Likewise.
33465 * configure.ac: Likewise.
33466 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33467 Waterman as RISC-V maintainers.
33468 * doc/install.texi: Add RISC-V entries.
33469 * doc/invoke.texi: Add RISC-V options section.
33470 * doc/md.texi: Add RISC-V constraints section.
33471 * configure: Regenerated.
33473 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33476 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33477 false values to be constant vectors with all 0 or all 1 bits set.
33478 (vcondu<mode><mode>): Likewise.
33479 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33481 (fpmask_comparison_operator): Update comment.
33482 (vecint_comparison_operator): New predicate.
33483 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33484 vector conditionals when the true and false values are constant
33485 vectors with all 0 bits or all 1 bits set.
33487 2017-02-06 Martin Sebor <msebor@redhat.com>
33489 PR tree-optimization/79376
33490 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33492 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33494 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33495 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33496 to simplify split condition.
33498 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33500 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33501 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33504 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33506 PR rtl-optimization/68664
33507 * target.def (can_speculate_insn): New hook.
33508 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33509 * doc/tm.texi: Regenerate.
33510 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33511 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33512 (rs6000_sched_can_speculate_insn): New function.
33514 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33516 PR tree-optimization/79284
33517 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33518 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33519 vectorizable_mask_load_store, vectorizable_operation,
33520 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33521 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33522 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33523 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33524 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33525 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33526 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33527 is_gimple_assign (stmt). Replace another such test with
33528 is_gimple_assign (stmt).
33530 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33533 * config/avr/avr.c (rtl-iter.h): Include it.
33534 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33535 (avr_legitimate_combined_insn): ...and implementation.
33537 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33539 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33540 * config/s390/s390.c (s390_const_operand_ok)
33541 (s390_canonicalize_comparison, s390_extract_part)
33542 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33543 (s390_contiguous_bitmask_p, s390_rtx_costs)
33544 (legitimize_pic_address): Likewise.
33545 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33546 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33547 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33548 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33549 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33551 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33553 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33554 REGNO($0) == REGNO($1).
33556 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33558 * config/s390/linux.h(SIZE_TYPE): Add comment.
33560 2017-02-06 Julian Brown <julian@codesourcery.com>
33561 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33562 Virendra Pathak <virendra.pathak@broadcom.com>
33564 * config/aarch64/aarch64-cores.def: Change the scheduler
33566 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33567 * config/aarch64/thunderx2t99.md: New file.
33569 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33571 * doc/standards.texi (Go Language): Update link to language
33574 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33576 * tree-eh.c (lower_resx): Sanitize profile.
33577 (cleanup_empty_eh_move_lp): Likewise.
33579 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33582 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33584 * cfgloopmanip.h (loop_version): Update prototype.
33585 * modulo-sched.c (sms_schedule): Update call of loop_version.
33586 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33587 * tree-parloops.c (gen_parallel_loop): Likewise.
33588 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33589 * tree-ssa-loop-split.c (split_loop): Likewise.
33590 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33591 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33593 2017-02-05 Martin Liska <mliska@suse.cz>
33596 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33598 (print_operand_address): Initialize a struct to zero.
33600 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33602 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33603 garbage collector only in textual form.
33605 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33607 * doc/extend.texi (x86 specific memory model extensions for
33608 transactional memory): Simplify a phrase.
33610 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33613 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33614 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33615 (atomic_storedi_1): Likewise.
33617 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33619 PR tree-optimization/79338
33620 * tree-parloops.c (gather_scalar_reductions): Don't call
33621 vect_analyze_loop_form for loop->inner before destroying loop's
33624 2017-02-03 Martin Sebor <msebor@redhat.com>
33626 PR tree-optimization/79327
33627 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33628 when precision has resulted in leading zeros.
33629 (format_integer): Adjust the likely counter to assume an unknown
33630 argument that may be zero is non-zero.
33632 2017-02-03 Jason Merrill <jason@redhat.com>
33635 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33636 avoid copying non-taken branch.
33638 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33640 PR tree-optimization/79340
33641 * tree-vect-loop.c (vectorizable_reduction): Release
33642 vec_defs elements after safe_splicing them into other vectors.
33645 PR tree-optimization/79327
33646 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33647 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33649 (format_integer): Use wide_int_to_tree instead of build_int_cst
33650 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33651 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33652 of shortest and longest sequence.
33654 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33656 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33657 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33659 2017-02-03 Walter Lee <walt@tilera.com>
33662 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33663 after initial stackframe link reg save.
33664 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33666 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33669 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33670 wu for stxssp alternative.
33672 2017-02-03 Martin Sebor <msebor@redhat.com>
33674 PR tree-optimization/79352
33675 * gimple-fold.c (get_range_strlen): Add argument.
33676 (get_range_strlen): Change return type to bool.
33677 (get_maxval_strlen): Pass in a dummy argument.
33678 * gimple-fold.h (get_range_strlen): Change return type to bool.
33679 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33680 * tree.h (array_at_struct_end_p): Add argument.
33681 * tree.c (array_at_struct_end_p): Handle it.
33683 2017-02-03 Martin Liska <mliska@suse.cz>
33686 * multiple_target.c (create_dispatcher_calls): Redirect edge
33687 from a caller of a dispatcher.
33688 (expand_target_clones): Make the clones local.
33689 (ipa_target_clone): Do both target clones and resolvers.
33690 (ipa_dispatcher_calls): Remove the pass.
33691 (pass_dispatcher_calls::gate): Likewise.
33692 (make_pass_dispatcher_calls): Likewise.
33693 * passes.def (pass_target_clone): Put as very first IPA early
33696 2017-02-03 Martin Liska <mliska@suse.cz>
33698 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33699 in case of a function with ifunc attribute.
33701 2017-02-03 Martin Liska <mliska@suse.cz>
33703 * cgraph.c (cgraph_node::dump): Dump function version info.
33704 * symtab.c (symtab_node::dump_base): Add missing new line.
33706 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33708 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33709 (ifcombine_ifandif): Use it.
33711 2017-02-03 Martin Liska <mliska@suse.cz>
33713 * doc/invoke.texi: Document default value for
33714 use-after-scope-direct-emission-threshold.
33716 2017-02-03 Martin Liska <mliska@suse.cz>
33718 PR tree-optimization/79339
33719 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33720 (format_floating): Likewise.
33722 2017-02-03 Martin Liska <mliska@suse.cz>
33725 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33727 (ipa_node_params_t::remove): Likewise.
33728 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33729 initialization from removed ipa_node_params_t::insert.
33730 (ipa_node_params::~ipa_node_params): Move from removed
33731 ipa_node_params_t::release.
33732 * symbol-summary.h (symbol_summary::m_released): New member.
33733 Do not release a summary twice. Do not allow to call finalizer
33734 for types of a summary that live in GGC memory.
33736 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33738 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33741 2017-02-02 Martin Sebor <msebor@redhat.com>
33743 PR middle-end/79275
33744 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33745 (format_string): Tighten up the range of output for non-constant
33746 strings and correct the expected range for wide non-constant strings.
33748 2017-02-02 Martin Sebor <msebor@redhat.com>
33750 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33752 PR middle-end/32003
33753 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33755 (-fdump-tree-@var): Add to index and document how to come up
33756 with pass-specific option and dump file names.
33757 (-fdump-passes): Clarify where to look for output.
33759 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33761 PR middle-end/77445
33762 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33763 statistics of the analyzed path; allow threading for speed when
33764 any of BBs along the path are optimized for speed.
33766 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33768 PR middle-end/78468
33769 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33770 settings of the virtual registers.
33773 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33775 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33776 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33779 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33781 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33782 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33784 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33786 * config/s390/s390.md: Add missing comments with the expanded
33788 * config/s390/vector.md: Likewise.
33789 * config/s390/vx-builtins.md: Likewise.
33791 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33794 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33795 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33796 conditions on a single line.
33798 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33800 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33801 __S390_VX__ to __VX__.
33803 2017-02-01 Andrew Pinski <apinski@cavium.com>
33805 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33806 stmt_info to record_stmt_cost.
33807 (vect_get_known_peeling_cost): Pass stmt_info if known to
33809 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33810 cpu_vector_cost field into
33811 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33812 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33813 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33814 splitting of scalar_stmt_cost and vec_stmt_cost.
33815 (thunderx_vector_cost): Likewise.
33816 (cortexa57_vector_cost): LIkewise.
33817 (exynosm1_vector_cost): Likewise.
33818 (xgene1_vector_cost): Likewise.
33819 (thunderx2t99_vector_cost): Improve after the splitting of the two
33821 (aarch64_builtin_vectorization_cost): Update for the splitting of
33822 scalar_stmt_cost and vec_stmt_cost.
33824 2017-02-01 Torvald Riegel <triegel@redhat.com>
33825 Richard Henderson <rth@redhat.com>
33827 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33828 conditional on existance of a fast atomic load.
33829 * optabs-query.c (can_atomic_load_p): New function.
33830 * optabs-query.h (can_atomic_load_p): Declare it.
33831 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33832 no fast atomic load is available for the particular size of access.
33833 (expand_atomic_compare_and_swap): Likewise.
33834 (expand_atomic_load): Likewise.
33835 (expand_atomic_store): Likewise.
33836 (expand_atomic_fetch_op): Likewise.
33837 * testsuite/lib/target-supports.exp
33838 (check_effective_target_sync_int_128): Remove x86 because it provides
33839 no fast atomic load.
33840 (check_effective_target_sync_int_128_runtime): Likewise.
33842 2017-02-01 Richard Biener <rguenther@suse.de>
33844 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33845 (graphite_transform_loops): Provide opt-info for optimized nests.
33846 * tree-parloop.c (parallelize_loops): Provide opt-info for
33847 parallelized loops.
33849 2017-02-01 Richard Biener <rguenther@suse.de>
33851 PR middle-end/79315
33852 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33853 was not set before.
33855 2017-02-01 Richard Biener <rguenther@suse.de>
33857 PR tree-optimization/71824
33858 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33859 Verify the loops are valid in the merged SESE region.
33860 (scop_detection::can_represent_loop_1): Check analyzing the
33861 evolution of the number of iterations in the region succeeds.
33863 2017-01-31 Ian Lance Taylor <iant@golang.org>
33865 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33866 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33868 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33870 PR preprocessor/79210
33871 * input.c (get_substring_ranges_for_loc): Replace line_width
33872 assertion with error-handling.
33874 2017-01-31 Richard Biener <rguenther@suse.de>
33876 PR tree-optimization/77318
33877 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33878 (create_pw_aff_from_tree): Take loop parameter.
33879 (add_condition_to_pbb): Pass loop of the condition to
33880 create_pw_aff_from_tree.
33882 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33884 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33885 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33887 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33891 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33893 (convert_int_to_float128): Likewise.
33894 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33895 (convert_int_to_float128): Likewise.
33896 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33897 (UNSPEC_IEEE128_CONVERT): Likewise.
33898 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33899 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33900 Use local variables for IBM extended format.
33901 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33902 (fix_trunc<mode>si2_fprs): Likewise.
33903 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33904 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33905 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33906 to know that we can now have integers of all sizes in vector
33908 (fix<uns>_<mode>di2_hw): Likewise.
33909 (float<uns>_<mode>si2_hw): Likewise.
33910 (fix_<mode>si2_hw): Likewise.
33911 (fixuns_<mode>si2_hw): Likewise.
33912 (float<uns>_<mode>di2_hw): Likewise.
33913 (float_<mode>di2_hw): Likewise.
33914 (float_<mode>si2_hw): Likewise.
33915 (floatuns_<mode>di2_hw): Likewise.
33916 (floatuns_<mode>si2_hw): Likewise.
33917 (xscvqp<su>wz_<mode>): Delete, no longer used.
33918 (xscvqp<su>dz_<mode>): Likewise.
33919 (xscv<su>dqp_<mode>): Likewise.
33920 (ieee128_mfvsrd_64bit): Likewise.
33921 (ieee128_mfvsrd_32bit): Likewise.
33922 (ieee128_mfvsrwz): Likewise.
33923 (ieee128_mtvsrw): Likewise.
33924 (ieee128_mtvsrd_64bit): Likewise.
33925 (ieee128_mtvsrd_32bit): Likewise.
33927 2017-01-31 Martin Liska <mliska@suse.cz>
33930 * ipa-prop.c (ipa_free_all_node_params): Call release method
33931 instead of ~sumbol_summary to not to trigger double times
33934 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33936 PR tree-optimization/71691
33937 * bitmap.h (class auto_bitmap): New.
33938 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33939 is_maybe_undefined instead of ssa_undefined_value_p.
33941 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33943 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33944 __S390_ARCH_LEVEL__ to __ARCH__.
33946 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33948 PR tree-optimization/79267
33949 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33950 if should_remove_lhs_p is true.
33952 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33955 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33956 (add_alignment_attribute): New.
33957 (base_type_die): Add alignment attribute.
33958 (subrange_type_die): Likewise.
33959 (modified_type_die): Likewise.
33960 (gen_array_type_die): Likewise.
33961 (gen_descr_array_type_die: Likewise.
33962 (gen_enumeration_type_die): Likewise.
33963 (gen_subprogram_die): Likewise.
33964 (gen_variable_die): Likewise.
33965 (gen_field_die): Likewise.
33966 (gen_ptr_to_mbr_type_die): Likewise.
33967 (gen_struct_or_union_type_die): Likewise.
33968 (gen_subroutine_type_die): Likewise.
33969 (gen_typedef_die): Likewise.
33970 (base_type_cmp): Compare alignment attribute.
33972 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33975 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33976 (setb_unsigned) New pattern for setb with CCUNS.
33977 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33978 subfc./subfe sequence to avoid overflow problems. Generate a
33979 shorter sequence with cmpld/setb for power9.
33980 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33981 for generating subfc. instruction.
33982 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33983 now uses this instruction.
33985 2017-01-30 Ian Lance Taylor <iant@google.com>
33988 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33989 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33991 2017-01-30 Martin Sebor <msebor@redhat.com>
33993 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33994 Move constant to the right of a relational operator.
33995 (get_mpfr_format_length, format_character, format_string): Ditto.
33996 (should_warn_p, maybe_warn): Same.
33998 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34000 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34003 * asan.c (get_translation_unit_decl): Remove function.
34004 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34006 2017-01-30 Martin Liska <mliska@suse.cz>
34008 PR gcov-profile/79259
34009 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34010 -fprofile-generate.
34012 2017-01-30 Martin Liska <mliska@suse.cz>
34015 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34016 Initialize variables with NULL value.
34018 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34021 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34023 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34025 2017-01-30 Richard Biener <rguenther@suse.de>
34027 PR tree-optimization/79276
34028 * tree-vrp.c (process_assert_insertions): Properly adjust common
34029 when removing a duplicate.
34031 2017-01-30 Richard Biener <rguenther@suse.de>
34033 PR tree-optimization/79256
34034 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34035 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34037 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34039 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34042 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34043 ("*r<noxa>sbg_<mode>_sll_bitmask")
34044 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34045 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34046 Use contiguous_bitmask_nowrap_operand.
34048 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34051 * config/rs6000/altivec.h (vec_xl): Revise #define.
34052 (vec_xst): Likewise.
34054 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34056 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34058 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34060 PR rtl-optimization/79194
34061 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34062 traps before call to bypass_conditional_jumps.
34064 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34066 PR tree-optimization/71374
34067 * lra-constraints.c (check_conflict_input_operands): New.
34068 (match_reload): Use it.
34070 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34073 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34074 account to calculate conflict_set.
34076 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34078 PR rtl-optimization/78559
34079 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34080 other_insn in combine.
34082 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34084 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34085 uint16_type_node for BT_UINT16.
34087 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34089 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34090 "RTL Tests" to menu.
34091 (GIMPLE Tests): New node.
34092 (RTL Tests): New node.
34094 2017-01-27 Richard Biener <rguenther@suse.de>
34096 PR tree-optimization/79245
34097 * tree-loop-distribution.c (distribute_loop): Apply cost
34098 modeling also to detected patterns.
34100 2017-01-27 Richard Biener <rguenther@suse.de>
34102 PR tree-optimization/71433
34103 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34104 (compare_assert_loc): New function.
34105 (process_assert_insertions): Sort and optimize assert locations
34106 to remove duplicates and push down identical assertions on
34107 edges to their destination block.
34109 2017-01-27 Richard Biener <rguenther@suse.de>
34111 PR tree-optimization/79244
34112 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34113 out SSA names even if abnormal.
34115 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34117 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34118 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34119 instead of MPFR_RNDN.
34121 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34124 * arm.c (arm_option_override): Don't call build_target_option_node
34125 until after doing all option overrides.
34126 (arm_valid_target_attribute_tree): Likewise.
34128 2017-01-27 Martin Liska <mliska@suse.cz>
34130 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34131 for {cd}tors and C++ {cd}tors.
34133 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34135 * config/s390/s390.md ("*setmem_long_and")
34136 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34138 2017-01-26 Martin Sebor <msebor@redhat.com>
34140 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34143 2017-01-26 Martin Sebor <msebor@redhat.com>
34145 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34146 HAVE_DFmode before using XFmode or DFmode.
34147 (parse_directive): Avoid using the z length modifier to avoid
34148 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34150 PR middle-end/78703
34151 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34152 to accept adjustment as an array.
34153 (get_int_range): New function.
34154 (struct directive): Make width and prec arrays.
34155 (directive::set_width, directive::set_precision): Call get_int_range.
34156 (format_integer, format_floating): Handle width and precision ranges.
34157 (format_string, parse_directive): Same.
34159 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34162 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34163 comdat_type_p set, just clone them, but keep the children in the
34167 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34168 which have direct callers with -fvar-tracking-assignments enabled
34170 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34171 inside of type units.
34173 2017-01-26 Martin Sebor <msebor@redhat.com>
34175 PR middle-end/78703
34176 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34178 (struct format_result): Replace number_chars, number_chars_min,
34179 and number_chars_max with a single member of struct result_range.
34181 (format_result::operator+=): Adjust.
34182 (struct fmtresult): Remove bounded. Handle likely and unlikely
34184 (fmtresult::adjust_for_width_or_precision): New function.
34185 (fmtresult:type_max_digits): New function.
34186 (bytes_remaining): Handle likely and unlikely counters.
34187 (min_bytes_remaining): Remove.
34188 (format_percent): Simplify.
34189 (format_integer, format_floating): Set likely and unlikely counters.
34190 (get_string_length, format_character, format_string): Same.
34191 (format_plain, should_warn_p): New function.
34192 (maybe_warn): Call should_warn_p. Update diagnostic messages
34193 and handle those for all directives, including plain strings.
34194 (format_directive): Handle likely and unlikely counters.
34195 Remove unnecessary quoting from diagnostics. Add an informational
34197 (add_bytes): Remove.
34198 (pass_sprintf_length::compute_format_length): Simplify.
34199 (try_substitute_return_value): Handle likely and unlikely counters.
34201 2017-01-26 Carl Love <cel@us.ibm.com>
34203 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34204 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34206 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34209 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34210 endianess for subregs into account.
34211 * lra-constraints.c (lra_constraints): Do risky transformations
34212 always on the first iteration.
34213 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34214 last_call_used_reg_set.
34215 (process_bb_lives): Define and use last_call_used_reg_set.
34216 * lra.c (lra): Always continue after lra_constraints on the first
34219 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34221 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34223 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34225 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34227 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34228 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34229 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34230 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34231 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34232 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34233 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34234 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34235 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34237 2017-01-26 Marek Polacek <polacek@redhat.com>
34240 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34241 for the third operand.
34243 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34245 PR middle-end/79236
34246 * omp-low.c (struct omp_context): Add simt_stmt field.
34247 (scan_omp_for): Return omp_context *.
34248 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34249 context to the _simt_ SIMD stmt.
34250 (lower_omp_for): For combined SIMD with sibling _simt_
34251 SIMD, make sure to use the same decls in _looptemp_
34252 clauses as in the sibling.
34254 2017-01-26 David Sherwood <david.sherwood@arm.com>
34256 PR middle-end/79212
34257 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34260 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34263 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34264 emit fld b; fld a; if possible.
34266 * brig-builtins.def: Update copyright years.
34267 * config/arm/arm_acle_builtins.def: Update copyright years.
34269 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34272 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34273 constraint instead of o for the stxsd instruction.
34275 2017-01-25 Carl Love <cel@us.ibm.com>
34277 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34278 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34280 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34282 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34284 2017-01-25 Richard Biener <rguenther@suse.de>
34286 PR tree-optimization/69264
34287 * target.def (vector_alignment_reachable): Improve documentation.
34288 * doc/tm.texi: Regenerate.
34289 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34291 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34292 earlier changes with respect to TYPE_USER_ALIGN.
34293 (vector_alignment_reachable_p): Likewise. Improve dumping.
34295 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34298 * config/arm/arm.md (xordi3): Force constant operand into a register
34301 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34303 * doc/invoke.texi (-fstore-merging): Correct default optimization
34304 levels at which it is enabled.
34305 (-O): Move -fstore-merging from list to...
34308 2017-01-25 Richard Biener <rguenther@suse.de>
34311 * omp-expand.c: Include debug.h.
34312 (expand_omp_taskreg): Make sure to generate early debug before
34313 outlining anything from a function.
34314 (expand_omp_target): Likewise.
34315 (grid_expand_target_grid_body): Likewise.
34317 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34320 * asan.c (get_translation_unit_decl): New function.
34321 (asan_add_global): Extract modules file name from globals
34322 TRANSLATION_UNIT_DECL name.
34324 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34327 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34328 for long calls with APCS frame and VFP.
34330 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34332 * cfg.c (original_copy_tables_initialized_p): New function.
34333 * cfg.h (original_copy_tables_initialized_p): New decl.
34334 * cfgrtl.c (relink_block_chain): Guard the call to
34335 free_original_copy_tables with a call to
34336 original_copy_tables_initialized_p.
34337 * cgraph.h (symtab_node::native_rtl_p): New decl.
34338 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34339 (symtab_node::needed_p): Don't assert for early assembly output
34340 for __RTL functions.
34341 (cgraph_node::finalize_function): Set "force_output" for __RTL
34343 (cgraph_node::analyze): Bail out early for __RTL functions.
34344 (analyze_functions): Update assertion to support __RTL functions.
34345 (cgraph_node::expand): Bail out early for __RTL functions.
34346 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34348 * function.h (struct function): Update comment for field
34350 * gimple-expr.c: Include "tree-pass.h".
34351 (gimple_has_body_p): Return false for __RTL functions.
34352 * Makefile.in (OBJS): Add run-rtl-passes.o.
34353 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34355 (gcc::pass_manager::get_clean_slate): New accessor.
34356 * passes.c: Include "insn-addr.h".
34357 (should_skip_pass_p): Add logging. Update logic for running
34358 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34359 property-provider override so it is only done for gimple passes.
34361 (skip_pass): New function.
34362 (execute_one_pass): Call skip_pass when skipping passes.
34363 * read-md.c (md_reader::read_char): Support filtering
34364 the input to a subset of line numbers.
34365 (md_reader::md_reader): Initialize fields
34366 m_first_line and m_last_line.
34367 (md_reader::read_file_fragment): New function.
34368 * read-md.h (md_reader::read_file_fragment): New decl.
34369 (md_reader::m_first_line): New field.
34370 (md_reader::m_last_line): New field.
34371 * read-rtl-function.c (function_reader::create_function): Only
34372 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34373 curr_properties. Set DECL_INITIAL to a dummy block.
34374 (read_rtl_function_body_from_file_range): New function.
34375 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34377 * run-rtl-passes.c: New file.
34378 * run-rtl-passes.h: New file.
34380 2017-01-24 Jeff Law <law@redhat.com>
34382 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34385 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34387 PR tree-optimization/79159
34388 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34389 (record_nonwrapping_iv): Improve boundary using above function if no
34390 value range information.
34392 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34393 Martin Jambor <mjambor@suse.cz>
34395 * brig-builtins.def: New file.
34396 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34397 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34398 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34399 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34400 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34401 * builtin-types.def (BT_INT8): New.
34402 (BT_INT16): Likewise.
34403 (BT_UINT8): Likewise.
34404 (BT_UINT16): Likewise.
34405 (BT_FN_ULONG): Likewise.
34406 (BT_FN_UINT_INT): Likewise.
34407 (BT_FN_UINT_ULONG): Likewise.
34408 (BT_FN_UINT_LONG): Likewise.
34409 (BT_FN_UINT_PTR): Likewise.
34410 (BT_FN_ULONG_PTR): Likewise.
34411 (BT_FN_INT8_FLOAT): Likewise.
34412 (BT_FN_INT16_FLOAT): Likewise.
34413 (BT_FN_UINT32_FLOAT): Likewise.
34414 (BT_FN_UINT16_FLOAT): Likewise.
34415 (BT_FN_UINT8_FLOAT): Likewise.
34416 (BT_FN_UINT64_FLOAT): Likewise.
34417 (BT_FN_UINT16_UINT32): Likewise.
34418 (BT_FN_UINT32_UINT16): Likewise.
34419 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34420 (BT_FN_INT_PTR_INT): Likewise.
34421 (BT_FN_UINT_PTR_UINT): Likewise.
34422 (BT_FN_LONG_PTR_LONG): Likewise.
34423 (BT_FN_ULONG_PTR_ULONG): Likewise.
34424 (BT_FN_VOID_UINT64_UINT64): Likewise.
34425 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34426 (BT_FN_INT8_INT8_INT8): Likewise.
34427 (BT_FN_INT16_INT16_INT16): Likewise.
34428 (BT_FN_INT_INT_INT): Likewise.
34429 (BT_FN_UINT_FLOAT_UINT): Likewise.
34430 (BT_FN_FLOAT_UINT_UINT): Likewise.
34431 (BT_FN_ULONG_UINT_UINT): Likewise.
34432 (BT_FN_ULONG_UINT_PTR): Likewise.
34433 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34434 (BT_FN_UINT_UINT_UINT): Likewise.
34435 (BT_FN_VOID_UINT_PTR): Likewise.
34436 (BT_FN_UINT_UINT_PTR: Likewise.
34437 (BT_FN_UINT32_UINT64_PTR): Likewise.
34438 (BT_FN_INT_INT_UINT_UINT): Likewise.
34439 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34440 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34441 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34442 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34443 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34444 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34445 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34446 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34447 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34448 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34449 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34450 * doc/frontends.texi: List BRIG FE.
34451 * doc/install.texi (Testing): Add BRIG tesring requirements.
34452 * doc/invoke.texi (Overall Options): Mention BRIG.
34453 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34455 2017-01-24 Richard Biener <rguenther@suse.de>
34457 PR translation/79208
34458 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34460 2017-01-24 Martin Jambor <mjambor@suse.cz>
34463 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34464 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34465 and known_contexts.
34467 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34469 PR middle-end/79123
34470 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34471 casts from signed to unsigned really don't have a range.
34473 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34475 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34476 GMP_RNDx for compatiblity.
34478 2017-01-24 Martin Liska <mliska@suse.cz>
34481 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34482 that would prevent us to call alloca with -1 as argument.
34484 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34486 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34487 Avoid -Wformat-security warning.
34489 2017-01-23 Andrew Pinski <apinski@cavium.com>
34491 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34494 2017-01-23 Martin Sebor <msebor@redhat.com>
34496 PR middle-end/78703
34497 * gimple-ssa-sprintf.c (warn_level): New global.
34498 (format_integer): Use it here and throughout the rest of the file.
34499 Use the same switch to compute sign as base.
34500 (maybe_warn): New function.
34501 (format_directive): Factor out warnings into maybe_warn.
34502 Add debugging output. Use warn_level.
34503 (add_bytes): Use warn_level.
34504 (pass_sprintf_length::compute_format_length): Add debugging output.
34505 (try_substitute_return_value): Same.
34506 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34508 PR middle-end/78703
34509 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34510 (struct fmtresult, format_integer, format_floating): Adjust.
34511 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34512 (get_string_length, format_string,format_directive): Same.
34513 (pass_sprintf_length::compute_format_length): Same.
34514 (try_substitute_return_value): Simplify slightly.
34516 PR middle-end/78703
34517 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34518 (fmtresult::operator+=): Outlined.
34519 (struct fmtresult): Add ctors.
34520 (struct conversion_spec): Rename...
34521 (struct directive): ...to this. Add and remove data members.
34522 (directive::set_width, directive::set_precision): New functions.
34523 (format_percent): Use fmtresult ctor.
34524 (get_width_and_precision): Remove.
34525 (format_integer): Make naming changes. Avoid computing width and
34527 (format_floating): Same. Adjust indentation.
34528 (format_character, format_none): New functions.
34529 (format_string): Moved character handling to format_character.
34530 (format_directive): Remove arguments, change return type.
34531 (parse_directive): New function.
34532 (pass_sprintf_length::compute_format_length): Move directive
34533 parsing to parse_directive.
34535 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34537 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34538 (assign_assembler_name_if_needed): ... this.
34539 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34540 (assign_assembler_name_if_needed): ... this.
34541 (free_lang_data_in_cgraph): Adjust callers.
34542 * cgraphunit.c (cgraph_node::analyze): Likewise.
34543 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34546 2017-01-23 Richard Biener <rguenther@suse.de>
34548 PR tree-optimization/79088
34549 PR tree-optimization/79188
34550 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34551 resetting loop bounds after last path deletion. Reset loop
34552 bounds of the target loop, make code match the comments.
34553 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34554 Make sure loops need no fixups.
34556 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34558 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34559 exponent support with double type for first argument.
34560 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34561 type returned by __builtin_vec_extract_sig,
34562 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34563 functions from "vector int" to "vector unsigned int" or from
34564 "vector long long int" to "vector unsigned long long int".
34565 Changed type returned by __builtin_vec_extract_exp,
34566 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34567 functions from "vector int" to "vector unsigned int" or from
34568 "vector long long int" to "vector unsigned long long int".
34569 Changed return type of __builtin_vec_test_data_class,
34570 __builtin_vec_test_data_class_sp, and
34571 __builtin_vec_test_data_class_dp from "vector int" to
34572 "vector bool int" or from "vector long long int" to "vector bool
34573 long long int" and changed second argument type from "unsigned
34574 int" to "int". Added new overloaded function forms "vector float
34575 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34576 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34577 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34578 double, vector unsigned long long int)" and "vector double
34579 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34580 long int)". Changed return type of
34581 __builtin_scalar_test_data_class and
34582 __builtin_scalar_test_data_class_sp and
34583 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34584 int" and changed second argument from "unsigned int" to "int".
34585 Changed type returned by __builtin_scalar_test_neg,
34586 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34587 from "int" to "bool int". Added new overloaded function form
34588 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34589 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34590 exponent double-precision with floating point first argument.
34591 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34592 documentation of scalar_test_data_class, scalar_test_neg,
34593 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34594 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34595 vec_test_data_class built-in functions to reflect refinements in
34596 their type signatures.
34598 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34600 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34602 (aarch64_elf_asm_destructor): Likewise.
34604 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34606 PR rtl-optimization/78634
34607 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34608 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34609 * ifcvt.c (noce_try_cmove): Add missing cost check.
34611 PR rtl-optimization/71724
34612 * combine.c (if_then_else_cond): Look for situations where it is
34613 beneficial to undo the work of one of the recursive calls.
34615 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34617 PR tree-optimization/70754
34618 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34619 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34620 combined stmt before it if not NULL.
34621 (combine_chains): Process refs reversely and compute dominance point
34624 2017-01-23 Martin Liska <mliska@suse.cz>
34626 PR tree-optimization/79196
34627 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34628 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34630 (strlen_optimize_stmt): Call the renamed function.
34632 2017-01-23 Michael Matz <matz@suse.de>
34634 PR tree-optimization/78384
34635 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34637 2017-01-23 Richard Biener <rguenther@suse.de>
34639 PR tree-optimization/79186
34640 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34641 both incoming edges before moving an assert.
34643 2017-01-23 Martin Jambor <mjambor@suse.cz>
34645 * ipa-prop.c (load_from_param_1): Removed.
34646 (load_from_unmodified_param): Bits from load_from_param_1 put back
34648 (load_from_param): Removed.
34649 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34650 with stmt. Reverted back to use of load_from_unmodified_param.
34652 2017-01-23 Martin Jambor <mjambor@suse.cz>
34655 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34656 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34657 field a pointer to garbage collected vector, mark lattices and
34658 ipcp_orig_node with GTY((skip)).
34659 (ipa_get_param_count): Adjust to descriptors being a pointer.
34660 (ipa_get_param): Likewise.
34661 (ipa_get_type): Likewise.
34662 (ipa_get_param_move_cost): Likewise.
34663 (ipa_set_param_used): Likewise.
34664 (ipa_get_controlled_uses): Likewise.
34665 (ipa_set_controlled_uses): Likewise.
34666 (ipa_is_param_used): Likewise.
34667 (ipa_node_params_t): Move into garbage collector. New methods insert
34669 (ipa_node_params_sum): Annotate wth GTY(()).
34670 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34672 (ipa_load_from_parm_agg): Adjust declaration.
34673 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34674 * ipa-profile.c (ipa_profile): Likewise.
34675 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34676 (ipa_populate_param_decls): Make descriptors parameter garbage
34678 (ipa_dump_param): Adjust to descriptors being a pointer.
34679 (ipa_alloc_node_params): Likewise.
34680 (ipa_initialize_node_params): Likewise.
34681 (load_from_param_1): Make descriptors parameter garbage collected.
34682 (load_from_unmodified_param): Likewise.
34683 (load_from_param): Likewise.
34684 (ipa_load_from_parm_agg): Likewise.
34685 (ipa_node_params::~ipa_node_params): Removed.
34686 (ipa_free_all_node_params): Remove call to delete operator.
34687 (ipa_node_params_t::insert): New.
34688 (ipa_node_params_t::remove): Likewise.
34689 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34690 copy known_csts and known_contexts vectors.
34691 (ipa_read_node_info): Adjust to descriptors being a pointer.
34692 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34694 (ipcp_transform_function): Make descriptors variable garbage
34697 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34699 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34700 * config/i386/avx512dqintrin.h: Ditto.
34701 * config/i386/avx512fintrin.h: Ditto.
34702 * config/i386/i386.c: Handle new builtins.
34703 * config/i386/i386-builtin.def: Add new builtins.
34704 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34705 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34707 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34708 Martin Liska <mliska@suse.cz>
34710 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34711 * asan.c (asan_expand_poison_ifn): Support stores and use
34712 appropriate ASAN report function.
34713 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34714 * internal-fn.def (ASAN_POISON_USE): Declare.
34715 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34716 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34717 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34718 ASAN_POISON calls w/o LHS.
34719 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34720 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34721 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34722 * gimplify.c (asan_poison_variables): Add attribute
34723 use_after_scope_memory to variables that really needs to live
34725 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34726 having the attribute.
34728 2017-01-23 Martin Liska <mliska@suse.cz>
34730 * asan.c (create_asan_shadow_var): New function.
34731 (asan_expand_poison_ifn): Likewise.
34732 * asan.h (asan_expand_poison_ifn): New declaration.
34733 * internal-fn.c (expand_ASAN_POISON): Likewise.
34734 * internal-fn.def (ASAN_POISON): New builtin.
34735 * sanopt.c (pass_sanopt::execute): Expand
34736 asan_expand_poison_ifn.
34737 * tree-inline.c (copy_decl_for_dup_finish): Make function
34739 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34740 * tree-ssa.c (is_asan_mark_p): New function.
34741 (execute_update_addresses_taken): Rewrite local variables
34742 (identified just by use-after-scope as addressable) into SSA.
34744 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34746 * doc/install.texi (Specific): opensource.apple.com uses https
34747 now. Remove trailing slash.
34749 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34751 * README.Portability: Remove note on an Irix compatibility issue.
34753 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34755 * gcov.c (INCLUDE_ALGORITHM): Define.
34756 (INCLUDE_VECTOR): Define.
34757 No longer include <vector> and <algorithm> directly.
34759 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34761 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34763 * doc/invoke.texi (Code Gen Options): Ditto.
34765 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34768 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34770 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34772 rtl-optimization/79125
34773 * cprop.c (local_cprop_pass): Handle cases where we make an
34774 unconditional trap.
34776 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34780 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34781 read from, for big endian.
34783 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34785 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34786 register pauth builtins for LP64 only.
34788 2017-01-20 Marek Polacek <polacek@redhat.com>
34791 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34794 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34796 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34798 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34799 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34800 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34802 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34805 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34806 in big-endian mode when they are not a single duplicated value.
34808 2017-01-20 Richard Biener <rguenther@suse.de>
34810 * BASE-VER: Bump to 7.0.1.
34812 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34814 * omp-low.c (omplow_simd_context): New struct. Use it...
34815 (lower_rec_simd_input_clauses): ...here and...
34816 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34817 references to idx, lane, max_vf, is_simt.
34819 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34821 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34824 2017-01-20 Martin Jambor <mjambor@suse.cz>
34826 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34827 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34829 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34830 (GTFILES): Rename hsa.c to hsa-common.c.
34831 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34832 * hsa-dump.c: Likewise.
34833 * hsa-gen.c: Likewise.
34834 * hsa-regalloc.c: Likewise.
34835 * ipa-hsa.c: Likewise.
34836 * omp-expand.c: Likewise.
34837 * omp-low.c: Likewise.
34838 * toplev.c: Likewise.
34840 2017-01-20 Marek Polacek <polacek@redhat.com>
34843 * doc/invoke.texi: Document -Wduplicated-branches.
34844 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34845 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34846 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34847 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34848 return 0 only when not OEP_LEXICOGRAPHIC.
34849 (fold_build_cleanup_point_expr): Use the expression
34850 location when building CLEANUP_POINT_EXPR.
34851 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34852 * tree.c (add_expr): Handle error_mark_node.
34854 2017-01-20 Martin Liska <mliska@suse.cz>
34857 * tree-profile.c (init_ic_make_global_vars): Do not call
34859 (gimple_init_gcov_profiler): Likewise.
34861 2017-01-20 Martin Liska <mliska@suse.cz>
34864 * cgraph.h (maybe_create_reference): Remove argument and
34866 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34868 * ipa-cp.c (create_specialized_node): Likewise.
34869 * symtab.c (symtab_node::maybe_create_reference): Handle
34870 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34872 2017-01-20 Martin Liska <mliska@suse.cz>
34874 * read-rtl-function.c (function_reader::create_function): Use
34875 build_decl instread of build_decl_stat.
34877 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34879 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34880 * config/i386/avx512dqintrin.h: Ditto.
34881 * config/i386/avx512fintrin.h: Ditto.
34882 * config/i386/i386-builtin-types.def: Add new types.
34883 * config/i386/i386.c: Handle new types.
34884 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34885 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34886 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34887 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34888 (__builtin_ia32_kshiftridi): New.
34889 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34891 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34895 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34896 define to rs6000_init_stack_protect_guard.
34897 (rs6000_init_stack_protect_guard): New function.
34899 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34900 Yunqiang Su <yunqiang.su@imgtec.com>
34902 * config.gcc (supported_defaults): Add madd4.
34903 (with_madd4): Add validation.
34904 (all_defaults): Add madd4.
34905 * config/mips/mips.opt (mmadd4): New option.
34906 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34908 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34910 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34911 (ISA_HAS_FUSED_MADD4): Likewise.
34912 * doc/invoke.texi (-mmadd4): Document the new option.
34913 * doc/install.texi (--with-madd4): Document the new option.
34915 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34917 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34918 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34919 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34920 (aarch64_init_pauth_hint_builtins): New.
34921 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34922 (aarch64_expand_builtin): Expand new builtins.
34924 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34926 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34927 * combine-stack-adj.c (no_unhandled_cfa): Handle
34928 REG_CFA_TOGGLE_RA_MANGLE.
34929 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34930 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34931 info for return address signing.
34932 (aarch64_expand_epilogue): Likewise.
34934 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34936 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34937 * config/aarch64/aarch64-protos.h
34938 (aarch64_return_address_signing_enabled): New declaration.
34939 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34941 (aarch64_expand_prologue): Sign return address before it's pushed onto
34943 (aarch64_expand_epilogue): Authenticate return address fetched from
34945 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34946 (aarch64_attributes): New function attributes for "sign-return-address".
34947 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34948 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34949 ("*do_return"): Generate combined instructions according to key index.
34950 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34951 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34953 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34954 * config/aarch64/aarch64.opt (msign-return-address=): New.
34955 * doc/extend.texi (AArch64 Function Attributes): Documents
34956 "sign-return-address=".
34957 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34959 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34961 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34962 overall option summary.
34964 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34966 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34967 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34968 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34969 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34971 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34973 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34974 -mpower9-minmax by default for -mcpu=power9.
34975 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34976 128-bit floating point.
34978 2017-01-20 Alan Modra <amodra@gmail.com>
34980 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34981 optimizing for size.
34983 2017-01-20 Alan Modra <amodra@gmail.com>
34986 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34987 for strcmp and strncmp from corresponding builtin decl.
34989 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34991 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34992 instead of i386/rtems-64.h.
34993 * config/i386/rtems-64.h: Remove.
34995 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34999 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35001 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35003 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35005 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35006 Change int to HOST_WIDE_INT.
35007 * config/aarch64/aarch64-protos.h
35008 (aarch64_simd_gen_const_vector_dup): Likewise.
35009 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35011 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35013 * langhooks-def.h (lhd_type_for_size): New decl.
35014 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35015 * langhooks.c (lhd_type_for_size): New function, taken from
35018 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35020 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35021 define_bypass for CR latency.
35022 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35023 (power9-alu2): Add define_bypass for CR latency.
35025 (power9-mul): Update insn latency.
35026 (power9-mul-compare): Update insn latency, bypass latency and remove
35029 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35031 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35033 * config/aarch64/aarch64.md
35034 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35035 aarch64_nopcrelative_literal_loads.
35036 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35038 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35040 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35041 TARGET_LOONGSON_3A.
35042 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35044 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35047 * config.gcc (supported_defaults): Add lxc1-sxc1.
35048 (with_lxc1_sxc1): Add validation.
35049 (all_defaults): Add lxc1-sxc1.
35050 * config/mips/mips.opt (mlxc1-sxc1): New option.
35051 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35053 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35054 __mips_no_lxc1_sxc1.
35055 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35056 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35057 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35059 2017-01-19 Richard Biener <rguenther@suse.de>
35061 PR tree-optimization/72488
35062 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35063 sure to restore SSA info.
35064 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35066 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35068 PR rtl-optimization/79121
35069 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35070 of the inner type when shifting an extended value.
35072 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35075 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35076 interposable aliases.
35078 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35081 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35082 Use the evmergelohi instruction.
35083 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35084 (mov_sitf_e500_subreg8_2_be): Likewise.
35085 (mov_sitf_e500_subreg12_2_le): Likewise.
35086 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35087 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35088 (mov_sitf_e500_subreg8_2_le): Likewise.
35089 (mov_sitf_e500_subreg12_2_be): Likewise.
35091 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35093 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35094 attribute from vecsimple to vecperm.
35095 (altivec_vbpermq2): Likewise.
35097 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35100 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35102 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35103 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35104 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35105 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35106 case where N arg is SIZE_MAX.
35107 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35108 (cmpstrsi): Add pattern.
35110 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35113 __builtin_vec_revb builtins.
35114 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35115 built-in functions to support generation of the ISA 3.0 XXBR<x>
35116 vector byte reverse instructions.
35117 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35118 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35119 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35120 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35121 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35122 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35123 (P9V_BUILTIN_VEC_REVB): Likewise.
35124 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35125 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35126 (p9_xxbrq_v16qi): Likewise.
35127 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35128 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35129 (p9_xxbrh_v8hi): Likewise.
35130 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35131 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35132 vec_revb built-in functions.
35134 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35136 PR rtl-optimization/78952
35137 * config/i386/i386.md (any_extract): New code iterator.
35138 (*insvqi_2): Use any_extract for source operand.
35139 (*insvqi_3): Use any_shiftrt for source operand.
35141 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35143 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35145 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35147 2017-01-18 Matthias Klose <doko@ubuntu.com>
35149 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35151 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35153 * config/rs6000/altivec.h (vec_bperm): Change #define.
35154 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35155 (altivec_vbpermq2): New define_insn.
35156 (altivec_vbpermd): Likewise.
35157 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35158 function interface.
35159 (VBPERMD): Likewise.
35160 (VBPERM): New polymorphic function interface.
35161 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35162 Add entries for P9V_BUILTIN_VEC_VBPERM.
35163 * doc/extend.texi: Add interfaces for vec_bperm.
35165 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35167 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35168 first letter of error messages.
35169 (s390_resolve_overloaded_builtin): Likewise.
35170 * config/s390/s390.c (s390_expand_builtin): Likewise.
35171 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35172 (s390_valid_target_attribute_inner_p): Likewise.
35173 * config/s390/s390.md ("tabort"): Likewise.
35175 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35177 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35178 (ISA_AVOID_DIV_HILO): New macro.
35179 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35180 (ISA_HAS_DDIV): Likewise.
35182 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35184 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35186 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35188 * doc/invoke.texi (fabi-version): Spelling fix.
35190 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35193 * doc/invoke.texi (fabi-version): Mention mangling fix for
35196 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35199 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35201 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35204 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35205 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35207 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35209 (stack_protect_test): Ditto.
35210 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35211 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35213 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35214 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35215 -mstack-protector-guard-offset=.
35216 (RS/6000 and PowerPC Options): Ditto.
35218 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35220 * config/i386/i386.h (MASK_CLASS_P): New define.
35221 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35222 there are no registers from different register sets also when
35223 mask registers are used. Update function comment.
35224 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35225 to (*k/*r) and (*k/*km) alternatives.
35227 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35229 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35230 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35231 (EH_RETURN_HANDLER_RTX): New define.
35232 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35233 Force frame pointer in EH return functions.
35234 (aarch64_expand_epilogue): Add barrier for eh_return.
35235 (aarch64_final_eh_return_addr): Remove.
35236 (aarch64_eh_return_handler_rtx): New function.
35237 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35239 (aarch64_eh_return_handler_rtx): New prototype.
35241 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35243 * config/rs6000/altivec.h (vec_rlmi): New #define.
35244 (vec_vrlnm): Likewise.
35245 (vec_rlnm): Likewise.
35246 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35247 (UNSPEC_VRLNM): Likewise.
35248 (VIlong): New mode iterator.
35249 (altivec_vrl<VI_char>mi): New define_insn.
35250 (altivec_vrl<VI_char>nm): Likewise.
35251 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35253 (VRLDNM): Likewise.
35254 (RLNM): New polymorphic function entry.
35255 (VRLWMI): New monomorphic function entry.
35256 (VRLDMI): Likewise.
35257 (RLMI): New polymorphic function entry.
35258 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35259 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35260 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35263 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35266 * dwarf2out.c (field_byte_offset): Restore the
35267 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35268 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35269 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35272 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35275 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35277 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35280 * dwarf2out.c (add_data_member_location_attribute): For constant
35281 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35282 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35283 DW_AT_byte_size attributes.
35285 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35287 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35288 after forcing to constant memory when the code model is medium.
35290 2017-01-17 Julia Koval <julia.koval@intel.com>
35293 * config/i386/avx512fintrin.h
35294 (_mm512_i32gather_ps): Change __addr type to void const*.
35295 (_mm512_mask_i32gather_ps): Ditto.
35296 (_mm512_i32gather_pd): Ditto.
35297 (_mm512_mask_i32gather_pd): Ditto.
35298 (_mm512_i64gather_ps): Ditto.
35299 (_mm512_mask_i64gather_ps): Ditto.
35300 (_mm512_i64gather_pd): Ditto.
35301 (_mm512_mask_i64gather_pd): Ditto.
35302 (_mm512_i32gather_epi32): Ditto.
35303 (_mm512_mask_i32gather_epi32): Ditto.
35304 (_mm512_i32gather_epi64): Ditto.
35305 (_mm512_mask_i32gather_epi64): Ditto.
35306 (_mm512_i64gather_epi32): Ditto.
35307 (_mm512_mask_i64gather_epi32): Ditto.
35308 (_mm512_i64gather_epi64): Ditto.
35309 (_mm512_mask_i64gather_epi64): Ditto.
35310 (_mm512_i32scatter_ps): Change __addr type to void*.
35311 (_mm512_mask_i32scatter_ps): Ditto.
35312 (_mm512_i32scatter_pd): Ditto.
35313 (_mm512_mask_i32scatter_pd): Ditto.
35314 (_mm512_i64scatter_ps): Ditto.
35315 (_mm512_mask_i64scatter_ps): Ditto.
35316 (_mm512_i64scatter_pd): Ditto.
35317 (_mm512_mask_i64scatter_pd): Ditto.
35318 (_mm512_i32scatter_epi32): Ditto.
35319 (_mm512_mask_i32scatter_epi32): Ditto.
35320 (_mm512_i32scatter_epi64): Ditto.
35321 (_mm512_mask_i32scatter_epi64): Ditto.
35322 (_mm512_i64scatter_epi32): Ditto.
35323 (_mm512_mask_i64scatter_epi32): Ditto.
35324 (_mm512_i64scatter_epi64): Ditto.
35325 (_mm512_mask_i64scatter_epi64): Ditto.
35326 * config/i386/avx512pfintrin.h
35327 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35328 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35329 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35330 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35331 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35332 (_mm512_prefetch_i32scatter_ps): Ditto.
35333 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35334 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35335 (_mm512_prefetch_i64scatter_pd): Ditto.
35336 (_mm512_prefetch_i64scatter_ps): Ditto.
35337 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35338 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35339 * config/i386/avx512vlintrin.h
35340 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35341 (_mm_mmask_i32gather_ps): Ditto.
35342 (_mm256_mmask_i32gather_pd): Ditto.
35343 (_mm_mmask_i32gather_pd): Ditto.
35344 (_mm256_mmask_i64gather_ps): Ditto.
35345 (_mm_mmask_i64gather_ps): Ditto.
35346 (_mm256_mmask_i64gather_pd): Ditto.
35347 (_mm_mmask_i64gather_pd): Ditto.
35348 (_mm256_mmask_i32gather_epi32): Ditto.
35349 (_mm_mmask_i32gather_epi32): Ditto.
35350 (_mm256_mmask_i32gather_epi64): Ditto.
35351 (_mm_mmask_i32gather_epi64): Ditto.
35352 (_mm256_mmask_i64gather_epi32): Ditto.
35353 (_mm_mmask_i64gather_epi32): Ditto.
35354 (_mm256_mmask_i64gather_epi64): Ditto.
35355 (_mm_mmask_i64gather_epi64): Ditto.
35356 (_mm256_i32scatter_ps): Change __addr type to void*.
35357 (_mm256_mask_i32scatter_ps): Ditto.
35358 (_mm_i32scatter_ps): Ditto.
35359 (_mm_mask_i32scatter_ps): Ditto.
35360 (_mm256_i32scatter_pd): Ditto.
35361 (_mm256_mask_i32scatter_pd): Ditto.
35362 (_mm_i32scatter_pd): Ditto.
35363 (_mm_mask_i32scatter_pd): Ditto.
35364 (_mm256_i64scatter_ps): Ditto.
35365 (_mm256_mask_i64scatter_ps): Ditto.
35366 (_mm_i64scatter_ps): Ditto.
35367 (_mm_mask_i64scatter_ps): Ditto.
35368 (_mm256_i64scatter_pd): Ditto.
35369 (_mm256_mask_i64scatter_pd): Ditto.
35370 (_mm_i64scatter_pd): Ditto.
35371 (_mm_mask_i64scatter_pd): Ditto.
35372 (_mm256_i32scatter_epi32): Ditto.
35373 (_mm256_mask_i32scatter_epi32): Ditto.
35374 (_mm_i32scatter_epi32): Ditto.
35375 (_mm_mask_i32scatter_epi32): Ditto.
35376 (_mm256_i32scatter_epi64): Ditto.
35377 (_mm256_mask_i32scatter_epi64): Ditto.
35378 (_mm_i32scatter_epi64): Ditto.
35379 (_mm_mask_i32scatter_epi64): Ditto.
35380 (_mm256_i64scatter_epi32): Ditto.
35381 (_mm256_mask_i64scatter_epi32): Ditto.
35382 (_mm_i64scatter_epi32): Ditto.
35383 (_mm_mask_i64scatter_epi32): Ditto.
35384 (_mm256_i64scatter_epi64): Ditto.
35385 (_mm256_mask_i64scatter_epi64): Ditto.
35386 (_mm_i64scatter_epi64): Ditto.
35387 (_mm_mask_i64scatter_epi64): Ditto.
35388 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35389 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35390 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35391 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35392 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35393 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35394 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35395 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35396 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35397 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35398 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35399 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35400 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35401 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35402 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35403 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35404 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35405 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35406 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35407 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35408 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35409 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35410 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35411 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35412 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35413 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35414 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35415 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35416 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35417 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35418 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35419 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35420 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35421 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35422 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35423 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35424 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35425 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35426 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35427 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35428 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35429 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35430 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35431 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35432 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35433 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35434 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35435 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35436 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35437 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35438 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35439 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35440 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35441 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35442 definitions accordingly.
35444 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35445 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35448 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35451 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35454 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35455 hard regs for inner regno.
35457 2017-01-17 Martin Liska <mliska@suse.cz>
35460 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35461 assumption and add comment.
35463 2017-01-17 Nathan Sidwell <nathan@acm.org>
35465 * ipa-visibility.c (localize_node): New function, broken out of ...
35466 (function_and_variable_visibility): ... here. Call it.
35468 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35470 PR middle-end/77445
35471 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35472 correctly set frequency of oudgoing edge.
35473 (duplicate_thread_path): Fix profile updating.
35475 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35478 * configure.ac: Add GCC_BASE_VER.
35479 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35480 version from BASE-VER file.
35481 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35482 (gcc.o): Depend on $(BASEVER).
35483 * common.opt (dumpfullversion): New option.
35484 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35485 * doc/invoke.texi: Document -dumpfullversion.
35486 * doc/install.texi: Document --with-gcc-major-version-only.
35487 * configure: Regenerated.
35489 2017-01-17 Richard Biener <rguenther@suse.de>
35491 PR tree-optimization/71433
35492 * tree-vrp.c (register_new_assert_for): Merge same asserts
35493 on all incoming edges.
35494 (process_assert_insertions_for): Handle insertions at the
35497 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35499 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35500 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35502 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35505 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35508 2017-01-17 Alan Modra <amodra@gmail.com>
35511 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35512 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35513 symbolic stack limit when pic.
35515 2017-01-16 Martin Sebor <msebor@redhat.com>
35517 PR tree-optimization/78608
35518 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35520 2017-01-16 Jeff Law <law@redhat.com>
35523 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35524 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35525 for several include directories that may be relative to sysroot.
35526 * config/i386/x-mingw32 (gplus_includedir): Define.
35527 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35528 (native_system_includedir): Likewise.
35529 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35530 override if TARGET_SYSTEM_ROOT is defined.
35531 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35533 PR tree-optimization/79090
35534 PR tree-optimization/33562
35535 PR tree-optimization/61912
35536 PR tree-optimization/77485
35537 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35538 and computed trims into the dump file.
35540 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35542 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35544 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35547 * gimplify.c (gimplify_init_constructor): If want_value and
35548 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35552 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35553 sequence. Formatting fixes.
35554 (doloop_optimize): Formatting fixes.
35557 * gcc.c (debug_level_greater_than_spec_func): New function.
35558 (static_spec_functions): Add debug-level-gt spec function.
35559 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35561 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35562 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35563 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35564 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35565 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35566 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35568 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35570 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35571 QImode fixups to general and mask registers only.
35573 2017-01-16 Carl Love <cel@us.ibm.com>
35575 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35576 for built-in functions
35577 vector signed char vec_nabs (vector signed char)
35578 vector signed short vec_nabs (vector signed short)
35579 vector signed int vec_nabs (vector signed int)
35580 vector signed long long vec_nabs (vector signed long long)
35581 vector float vec_nabs (vector float)
35582 vector double vec_nabs (vector double)
35583 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35585 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35586 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35587 * doc/extend.texi: Update the documentation file for the new built-in
35590 2017-01-16 Martin Sebor <msebor@redhat.com>
35592 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35595 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35597 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35598 UNSPEC_VSX__XXSPLTD to require special splat handling.
35600 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35603 * system.h: Poison strndup.
35605 2017-01-16 Alan Modra <amodra@gmail.com>
35608 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35611 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35613 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35615 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35617 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35618 call recog here. Assert that INSN_CODE (insn) is non-negative.
35620 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35623 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35625 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35626 in the currently scheduled RTL fragment.
35628 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35630 PR rtl-optimization/78751
35631 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35634 2017-01-14 Jeff Law <law@redhat.com>
35636 PR tree-optimization/79090
35637 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35638 variable length stores.
35639 (compute_trims): Delete dead assignment to *trim_tail.
35640 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35643 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35645 PR rtl-optimization/78626
35646 PR rtl-optimization/78727
35647 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35648 of a block, and split such blocks after everything else is finished.
35650 2017-01-14 Alan Modra <amodra@gmail.com>
35653 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35654 target legitimate_combined_insn.
35655 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35656 (rs6000_legitimate_combined_insn): New function.
35657 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35659 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35660 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35661 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35663 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35665 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35667 2017-01-13 Jeff Law <law@redhat.com>
35669 PR tree-optimization/33562
35670 PR tree-optimization/61912
35671 PR tree-optimization/77485
35672 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35674 (delete_dead_assignment): Likewise.
35675 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35676 statement to delete_dead_call and delete_dead_assignment.
35678 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35681 * substring-locations.c (format_warning_va): Strengthen case 1 so
35682 that both endpoints of the substring must be within the format
35683 range for just the substring to be printed.
35685 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35687 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35688 * config/i386/i386.c (ix86_target_string): Add missing options
35689 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35690 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35691 flags_other and ix86_target_other to flags2_other. Display unknown
35693 (ix86_valid_target_attribute_inner_p): Add missing options and
35694 reorder options by implied ISAs, as in ix86_target_string.
35696 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35698 * hash-table.h (hash_table::too_empty_p): New function.
35699 (hash_table::expand): Use it.
35700 (hash_table::traverse): Likewise.
35701 (hash_table::empty_slot): Use sizeof (value_type) instead of
35702 sizeof (PTR) to convert bytes to elements. Shrink the table
35703 if the current size is excessive for the current number of
35706 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35708 * ira-costs.c (record_reg_classes): Break from the inner loop
35709 early once alt_fail is known to be true. Update outer loop
35710 handling accordingly.
35712 2017-01-13 Jeff Law <law@redhat.com>
35714 * tree-ssa-dse.c (decrement_count): New function.
35715 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35716 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35717 when we know the partially dead statement is a mem* function.
35719 PR tree-optimization/61912
35720 PR tree-optimization/77485
35721 * tree-ssa-dse.c: Include expr.h.
35722 (maybe_trim_constructor_store): New function.
35723 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35725 PR tree-optimization/33562
35726 PR tree-optimization/61912
35727 PR tree-optimization/77485
35728 * doc/invoke.texi: Document new dse-max-object-size param.
35729 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35730 * tree-ssa-dse.c: Include params.h.
35731 (dse_store_status): New enum.
35732 (initialize_ao_ref_for_dse): New, partially extracted from
35734 (valid_ao_ref_for_dse, normalize_ref): New.
35735 (setup_live_bytes_from_ref, compute_trims): Likewise.
35736 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35737 (maybe_trim_partially_dead_store): Likewise.
35738 (maybe_trim_complex_store): Likewise.
35739 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35740 Track what bytes live from the original store. Return tri-state
35741 for dead, partially dead or live.
35742 (dse_dom_walker): Add constructor, destructor and new private members.
35743 (delete_dead_call, delete_dead_assignment): New extracted from
35745 (dse_optimize_stmt): Make a member of dse_dom_walker.
35746 Use initialize_ao_ref_for_dse.
35748 PR tree-optimization/33562
35749 PR tree-optimization/61912
35750 PR tree-optimization/77485
35751 * sbitmap.h (bitmap_count_bits): Prototype.
35752 (bitmap_clear_range, bitmap_set_range): Likewise.
35753 * sbitmap.c (bitmap_clear_range): New function.
35754 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35756 2017-01-13 Martin Liska <mliska@suse.cz>
35759 * function.c (set_cfun): Add new argument force.
35760 * function.h (set_cfun): Likewise.
35761 * ipa-inline-transform.c (inline_call): Use the function when
35762 strict alising from is dropped for function we inline to.
35764 2017-01-13 Richard Biener <rguenther@suse.de>
35766 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35767 for dumping GIMPLE INTEGER_CSTs.
35769 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35771 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35772 to 201112L since C++17.
35774 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35777 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35778 if -fsanitize=kernel-address is present.
35780 2017-01-13 Richard Biener <rguenther@suse.de>
35782 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35783 as _Literal ( type ) number in case usual suffixes do not
35784 preserve all information.
35786 2017-01-13 Richard Biener <rguenther@suse.de>
35788 PR tree-optimization/77283
35789 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35790 and ssa-iterators.h.
35791 (is_feasible_trace): Implement a cost model based on joiner
35794 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35797 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35798 char or short to __float128/_Float128 directly.
35800 2017-01-12 Martin Sebor <msebor@redhat.com>
35802 to -Wformat-overflow.
35803 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35804 (min_bytes_remaining): Same.
35805 (get_string_length): Same.
35806 (format_string): Same.
35807 (format_directive): Same.
35809 (pass_sprintf_length::handle_gimple_call): Same.
35811 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35813 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35814 info.nowrite calls with no lhs that can't throw. Return bool
35815 whether gsi_remove has been called or not.
35816 (pass_sprintf_length::handle_gimple_call): Return bool whether
35817 try_substitute_return_value called gsi_remove. Formatting fix.
35818 (pass_sprintf_length::execute): Don't use gsi_remove if
35819 handle_gimple_call returned true.
35822 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35823 be removed due to side-effects, don't remove following barrier nor
35824 turn the successor edge into fallthru edge.
35826 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35829 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35830 element-reversing loads and stores as not swappable.
35832 2017-01-12 Nathan Sidwell <nathan@acm.org>
35833 Nicolai Stange <nicstange@gmail.com>
35835 * combine.c (try_combine): Don't ignore result of overlap checking
35836 loop. Combine overlap & asm check into single loop.
35838 2017-01-12 Richard Biener <rguenther@suse.de>
35840 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35841 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35843 2017-01-12 Richard Biener <rguenther@suse.de>
35845 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35846 and TS_TARGET_OPTION directly derive from TS_BASE.
35847 * tree-core.h (tree_optimization_option): Derive from tree_base.
35848 (tree_target_option): Likewise.
35850 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35852 * config/i386/i386.c (memory_address_length): Increase len
35853 only when rip_relative_addr_p returns false.
35855 2017-01-11 Julia Koval <julia.koval@intel.com>
35857 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35858 (OPTION_MASK_ISA_SGX_SET): New.
35859 (ix86_handle_option): Handle OPT_msgx.
35860 * config.gcc: Added sgxintrin.h.
35861 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35862 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35863 * config/i386/i386.c (ix86_target_string): Add -msgx.
35865 (ix86_option_override_internal): Handle new options.
35866 (ix86_valid_target_attribute_inner_p): Add sgx.
35867 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35868 * config/i386/i386.opt: Add msgx.
35869 * config/i386/sgxintrin.h: New file.
35870 * config/i386/x86intrin.h: Add sgxintrin.h.
35872 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35875 * fold-const.c (maybe_nonzero_address): Return 1 for function
35877 (tree_single_nonzero_warnv_p): Don't handle function local objects
35881 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35884 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35887 * opts.c: Include "spellcheck.h"
35888 (struct string_fragment): New struct.
35889 (struct edit_distance_traits<const string_fragment &>): New
35891 (get_closest_sanitizer_option): New function.
35892 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35894 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35896 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35898 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35899 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35900 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35901 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35902 for initial die_offset if dwarf_split_debug_info.
35903 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35904 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35906 (output_skeleton_debug_sections): Formatting fix. Use
35907 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35908 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35910 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35912 * config/arm/cortex-a53.md: Add bypasses for
35913 cortex_a53_r2f_cvt.
35914 (cortex_a53_r2f): Only use for transfers.
35915 (cortex_a53_f2r): Likewise.
35916 (cortex_a53_r2f_cvt): Add reservation for conversions.
35917 (cortex_a53_f2r_cvt): Likewise.
35919 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35921 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35922 to all inlined functions, change static to extern.
35924 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35927 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35929 (arm_assemble_integer): Likewise.
35931 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35933 * config.gcc: Use new awk script to check CPU, FPU and architecture
35934 parameters for --with-... options.
35935 * config/arm/parsecpu.awk: New file
35936 * config/arm/arm-cpus.in: New file.
35937 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35939 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35941 * config/arm/t-arm: Update dependency rules.
35942 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35943 of processing .def files.
35944 * config/arm/genopt.sh: Deleted.
35945 * config/arm/gentune.sh: Deleted.
35946 * config/arm/arm-cores.def: Deleted.
35947 * config/arm/arm-arches.def: Deleted.
35948 * config/arm/arm-fpus.def: Deleted.
35949 * config/arm/arm-tune.md: Regenerated.
35950 * config/arm/arm-tables.opt: Regenerated.
35951 * config/arm/arm-cpu.h: New generated file.
35952 * config/arm/arm-cpu-data.h: New generated file.
35953 * config/arm/arm-cpu-cdata.h: New generated file.
35955 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35958 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35960 (input_varpool_node): Unpack dynamically_initialized bit.
35962 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35964 PR rtl-optimization/79032
35965 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35966 the alignment of the adjusted memory reference against that of MODE,
35967 instead of the alignment of the original memory reference.
35969 2017-01-11 Martin Jambor <mjambor@suse.cz>
35971 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35973 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35974 decorated functions.
35976 2017-01-11 Richard Biener <rguenther@suse.de>
35978 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35979 set range/nonnull info for PHI results. Do not set it on
35980 stmts marked for removal.
35982 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35984 * expr.c (store_field): In the bitfield case, fetch the return value
35985 from the registers before applying a single big-endian adjustment.
35986 Always do a final load for a BLKmode value not larger than a word.
35988 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35991 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35992 that we correctly handle column numbers greater than
35993 LINE_MAP_MAX_COLUMN_NUMBER.
35995 2017-01-10 Martin Sebor <msebor@redhat.com>
35997 PR middle-end/78245
35998 * gimple-ssa-sprintf.c (get_destination_size): Call
35999 {init,fini}object_sizes.
36000 * tree-object-size.c (addr_object_size): Adjust.
36001 (pass_through_call): Adjust.
36002 (pass_object_sizes::execute): Adjust.
36003 * tree-object-size.h (fini_object_sizes): Declare.
36005 2017-01-10 Martin Sebor <msebor@redhat.com>
36007 PR tree-optimization/78775
36008 * builtins.c (get_size_range): Move...
36009 * calls.c: ...to here.
36010 (alloc_max_size): Accept zero argument.
36011 (operand_signed_p): Remove.
36012 (maybe_warn_alloc_args_overflow): Call get_size_range.
36013 * calls.h (get_size_range): Declare.
36015 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36017 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36018 from TI's devices.csv file as of September 2016.
36019 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36021 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36023 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36024 * doc/invoke.texi: Likewise.
36025 * doc/md.texi: Likewise.
36026 * doc/objc.texi: Likewise.
36028 2017-01-10 Joshua Conner <joshconner@google.com>
36030 * config/arm/fuchsia-elf.h: New file.
36031 * config/fuchsia.h: New file.
36032 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36033 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36035 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36037 2016-01-10 Richard Biener <rguenther@suse.de>
36039 PR tree-optimization/79034
36040 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36041 Propagate out degenerate PHIs in the joiner.
36043 2017-01-10 Martin Liska <mliska@suse.cz>
36045 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36046 (sort_congruence_classes_by_decl_uid): Likewise.
36047 (sort_congruence_class_groups_by_decl_uid): Likewise.
36048 (sem_item_optimizer::merge_classes): Sort class, groups in these
36049 classes and members in the groups by DECL_UID of declarations.
36050 This would make merge operations stable.
36052 2017-01-10 Martin Liska <mliska@suse.cz>
36054 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36055 usage of m_classes_vec.
36056 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36057 (sem_item_optimizer::get_group_by_hash): Likewise.
36058 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36059 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36060 (sem_item_optimizer::verify_classes): Likewise.
36061 (sem_item_optimizer::process_cong_reduction): Likewise.
36062 (sem_item_optimizer::dump_cong_classes): Likewise.
36063 (sem_item_optimizer::merge_classes): Likewise.
36064 * ipa-icf.h (congruence_class_hash): Rename from
36065 congruence_class_group_hash. Remove declaration of m_classes_vec.
36067 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36069 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36070 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36071 * config.gcc: Add avx512vpopcntdqintrin.h.
36072 * config/i386/avx512vpopcntdqintrin.h: New.
36073 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36074 * config/i386/i386-builtin-types.def: Add new types.
36075 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36076 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36077 __builtin_ia32_vpopcountq_v8di_mask): New.
36078 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36079 __AVX512VPOPCNTDQ__.
36080 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36081 (PTA_AVX512VPOPCNTDQ): Define.
36082 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36083 TARGET_AVX512VPOPCNTDQ_P): Define.
36084 * config/i386/i386.opt: Add mavx512vpopcntdq.
36085 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36086 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36088 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36090 PR middle-end/77484
36091 * predict.def (PRED_CALL): Set to 67.
36093 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36095 * expr.c (store_field): In the bitfield case, if the value comes from
36096 a function call and is of an aggregate type returned in registers, do
36097 not modify the field mode; extract the value in all cases if the mode
36098 is BLKmode and the size is not larger than a word.
36100 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36103 * config/i386/cpuid.h: Fix undefined behavior.
36105 2017-01-04 Jeff Law <law@redhat.com>
36107 PR tree-optimization/79007
36108 PR tree-optimization/67955
36109 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36110 conservative for pt.null when flag_non_call_exceptions is on.
36112 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36114 PR translation/79019
36115 PR translation/79020
36116 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36117 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36118 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36120 * config/avr/avr.opt (maccumulate-args): Likewise.
36121 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36122 * common.opt (freport-bug): Likewise.
36123 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36124 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36125 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36126 translatable string.
36127 * config/i386/i386.c (function_value_32): Likewise.
36128 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36129 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36131 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36132 * common/config/msp430/msp430-common.c (msp430_handle_option):
36134 * symtab.c (symtab_node::verify_base): Likewise.
36135 * opts.c (set_debug_level): Likewise.
36136 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36137 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36138 missing whitespace to translatable strings.
36139 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36140 * config/sh/superh.h: Likewise.
36141 * config/i386/xopintrin.h: Likewise.
36142 * config/i386/znver1.md: Likewise.
36143 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36144 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36145 * double-int.h (struct double_int): Likewise.
36146 * double-int.c (div_and_round_double): Likewise.
36147 * wide-int.cc: Likewise.
36148 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36149 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36150 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36151 (crossjumps_occurred): ... this.
36152 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36155 PR tree-optimization/78899
36156 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36157 returning bool return struct loop *, NULL for failure and the new
36159 (versionable_outer_loop_p): Don't version outer loop if it has
36160 dont_vectorized bit set.
36161 (tree_if_conversion): When versioning outer loop, ensure
36162 tree_if_conversion is performed also on the inner loop of the
36163 non-vectorizable outer loop copy.
36164 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36165 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36166 prevent vectorization of it.
36167 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36168 the outer loop vectorization of the non-scalar version is attempted
36169 before vectorization of the inner loop in scalar version. If
36170 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36171 vectorization of its inner loop.
36172 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36173 has 2 inner loops, rename also on edges from bb whose single pred
36174 is outer_loop->header. Fix typo in function comment.
36176 2017-01-09 Martin Sebor <msebor@redhat.com>
36179 * asan.c (asan_emit_stack_protection): Increase local buffer size
36180 to avoid snprintf truncation warning.
36182 2017-01-09 Andrew Pinski <apinski@cavium.com>
36184 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36185 to reference thunderx2t99 for the tuning structure
36186 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36188 (thunderx2t99_extra_costs): This.
36189 * config/aarch64/aarch64-tune.md: Regenerate.
36190 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36191 (vulcan_addrcost_table): This.
36192 (vulcan_regmove_cost): Rename to ...
36193 (thunderx2t99_regmove_cost): This.
36194 (vulcan_vector_cost): Rename to ...
36195 (thunderx2t99_vector_cost): this.
36196 (vulcan_branch_cost): Rename to ...
36197 (thunderx2t99_branch_cost): This.
36198 (vulcan_tunings): Rename to ...
36199 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36200 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36202 2017-01-09 Martin Jambor <mjambor@suse.cz>
36206 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36207 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36208 (propagate_vr_accross_jump_function): Use the above function for all
36209 value range computations for pass-through jump functions and type
36210 converasion from explicit value range values.
36211 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36212 parameters from TYPE_ARG_TYPES.
36213 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36214 (ipa_write_node_info): Stream type of the actual argument.
36215 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36217 2017-01-09 Martin Liska <mliska@suse.cz>
36220 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36221 (lookup_compiler): Do not show error message with have_E.
36223 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36225 PR tree-optimization/78938
36226 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36227 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36228 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36229 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36232 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36234 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36237 2017-01-09 Richard Biener <rguenther@suse.de>
36239 PR tree-optimization/78997
36240 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36241 name condition properly.
36243 2017-01-09 Richard Biener <rguenther@suse.de>
36246 * dwarf2out.c (is_cxx): New overload with context.
36247 (is_naming_typedef_decl): Use it.
36249 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36251 * invoke.texi (Option Summary): Correct spacing in option lists
36252 and add line breaks to fix over-long lines.
36254 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36256 PR middle-end/17660
36258 * extend.texi (Common Variable Attributes): Add xref to GCC
36259 Internals manual to explain mode attribute keywords.
36261 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36264 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36265 and Preprocessor Options.
36266 (Options for Linking): Document -pthread here....
36267 (RS/6000 and PowerPC Options): ...not here.
36268 (Solaris 2 Options): ...or here.
36269 * doc/cppopts.texi: Document -pthread.
36271 2017-01-08 Martin Sebor <msebor@redhat.com>
36273 PR middle-end/77708
36274 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36275 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36276 New member functions.
36277 (format_directive): Used them.
36279 (pass_sprintf_length::handle_gimple_call): Same.
36280 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36281 to avoid truncation for any argument.
36282 (extract_affine_mul): Same.
36283 * tree.c (get_file_function_name): Same.
36285 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36287 PR middle-end/77484
36288 * predict.def (PRED_INDIR_CALL): Set to 86.
36290 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36292 PR preprocessor/54124
36293 * doc/cppopts.texi: Reformat -d subtable to list the full name
36294 of the options. Add cross-reference to the docs for the general
36295 compiler -d options.
36296 * doc/invoke.texi (Developer Options): Add cross-reference to the
36297 preprocessor-specific -d option documentation.
36299 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36301 PR preprocessor/13498
36302 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36303 redudant material, and reflect new command-line options.
36304 (System Headers): Likewise.
36306 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36308 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36309 -isystem, and -idirafter. Copy-edit.
36310 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36311 default for -ftrack-macro-expansion. Delete obsolete and
36312 badly-formatted implementation details about -fdebug-cpp output.
36313 * doc/cppwarnopts.texi: Copy-edit.
36315 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36318 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36319 that the transition from a max line width >= 1<<10 to narrower
36320 lines works correctly.
36322 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36324 * doc/options.texi (PerFunction): New.
36325 * opt-functions.awk (switch_flags): Map both Optimization and
36326 PerFunction to CL_OPTIMIZATION.
36327 * opth-gen.awk: Test for PerFunction flag along with
36329 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36330 it only when the latter is present. Skip those that don't in
36331 the hash function generator.
36332 * common.opt (fvar-tracking): Mark as PerFunction instead of
36334 (fvar-tracking-assignments): Likewise.
36335 (fvar-tracking-assignments-toggle): Likewise.
36336 (fvar-tracking-uninit): Likewise.
36338 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36340 PR translation/79018
36341 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36344 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36347 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36348 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36349 TARGET_LONG_JUMP_TABLE_OFFSETS.
36350 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36351 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36352 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36353 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36354 * config/m68k/m68k.md (tablejump expander): Likewise.
36355 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36356 TARGET_LONG_JUMP_TABLE_OFFSETS.
36357 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36358 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36360 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36361 David Holsgrove <david.holsgrove@xilinx.com>
36363 * common/config/microblaze/microblaze-common.c
36364 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36365 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36367 * config/microblaze/microblaze.c (microblaze_must_save_register)
36368 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36370 (microblaze_eh_return): New function.
36371 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36372 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36373 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36374 * config/microblaze/microblaze.md (eh_return): New pattern.
36376 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36378 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36379 GCC_DIAGNOSTIC_STRINGIFY): Define.
36381 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36383 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36385 * config/arm/arm.md (<mcrr>): New.
36387 * config/arm/arm.c (arm_arch5te): New.
36388 (arm_option_override): Set arm_arch5te.
36389 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36391 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36392 (arm_mcrr_qualifiers): ... this. New.
36393 (MRRC_QUALIFIERS): Define to...
36394 (arm_mrrc_qualifiers): ... this. New.
36395 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36397 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36398 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36399 (MRRCI, mrrc, MRRC): New.
36400 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36401 VUNSPEC_MRRC2): New.
36403 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36405 * config/arm/arm.md (<mcr>): New.
36407 * config/arm/arm.c (arm_coproc_builtin_available): Add
36408 support for mcr, mrc, mcr2 and mrc2.
36409 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36410 (arm_mcr_qualifiers): ... this. New.
36411 (MRC_QUALIFIERS): Define to ...
36412 (arm_mrc_qualifiers): ... this. New.
36413 (MCR_QUALIFIERS): Define to ...
36414 (arm_mcr_qualifiers): ... this. New.
36415 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36417 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36418 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36419 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36420 VUNSPEC_MRC2): New.
36422 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36424 * config/arm/arm.md (*ldc): New.
36428 * config/arm/arm.c (arm_coproc_builtin_available): Add
36429 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36430 (arm_coproc_ldc_stc_legitimate_address): New.
36431 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36432 'qualifier_const_pointer'.
36433 (LDC_QUALIFIERS): Define to...
36434 (arm_ldc_qualifiers): ... this. New.
36435 (STC_QUALIFIERS): Define to...
36436 (arm_stc_qualifiers): ... this. New.
36437 * config/arm/arm-protos.h
36438 (arm_coproc_ldc_stc_legitimate_address): New.
36439 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36440 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36441 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36442 stc2, stcl, stc2l): New.
36443 * config/arm/constraints.md (Uz): New.
36444 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36445 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36446 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36447 VUNSPEC_STC2L): New.
36449 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36451 * config/arm/arm.md (<cdp>): New.
36452 * config/arm/arm.c (neon_const_bounds): Rename this ...
36453 (arm_const_bounds): ... this.
36454 (arm_coproc_builtin_available): New.
36455 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36456 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36457 (CDP_QUALIFIERS): Define to...
36458 (arm_cdp_qualifiers): ... this. New.
36460 (arm_expand_builtin_args): Add case for 6 arguments.
36461 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36462 (arm_const_bounds): ... this.
36463 (arm_coproc_builtin_available): New.
36464 * config/arm/arm_acle.h (__arm_cdp): New.
36466 * config/arm/arm_acle_builtins.def (cdp): New.
36468 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36469 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36470 'arm_const_bounds'.
36471 * config/arm/types.md (coproc): New.
36472 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36473 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36474 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36475 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36477 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36479 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36480 (UBINOP_QUALIFIERS): New.
36482 (acle_builtin_data): New. Change comment.
36483 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36484 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36485 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36486 arm_acle_builtins.def.
36487 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36488 (arm_init_acle_builtins): New.
36489 (CRC32_BUILTIN): Remove.
36490 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36491 crc32cb, crc32ch and crc32cw.
36492 (arm_init_crc32_builtins): Remove.
36493 (arm_init_builtins): Use arm_init_acle_builtins rather
36494 than arm_init_crc32_builtins.
36495 (arm_expand_acle_builtin): New.
36496 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36497 * config/arm/arm_acle_builtins.def: New.
36499 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36501 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36502 (arm_builtin_datum): ... this.
36503 (arm_init_neon_builtin): Rename to ...
36504 (arm_init_builtin): ... this. Add a new parameters PREFIX
36505 and USE_SIG_IN_NAME.
36506 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36507 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36508 'arm_builtin_datum'.
36509 (arm_init_vfp_builtins): Likewise.
36510 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36511 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36512 (arm_expand_neon_args): Rename to ...
36513 (arm_expand_builtin_args): ... this. Rename builtin_arg
36514 enum values and differentiate between ARG_BUILTIN_MEMORY
36515 and ARG_BUILTIN_NEON_MEMORY.
36516 (arm_expand_neon_builtin_1): Rename to ...
36517 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36518 values, arm_expand_builtin_args and add bool parameter NEON.
36519 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36520 (arm_expand_vfp_builtin): Likewise.
36521 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36523 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36525 PR middle-end/77484
36526 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36527 * predict.c (tree_estimate_probability_bb): Reverse direction of
36528 polymorphic call predictor.
36530 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36532 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36533 (determine_pass_name_match): ...this new function and...
36534 (should_skip_pass_p): ...this new function.
36536 2017-01-06 Nathan Sidwell <nathan@acm.org>
36538 * ipa-visibility.c (function_and_variable_visibility): Reformat
36539 comments and long lines. Remove extrneous if.
36540 * symtab.c (symtab_node::make_decl_local): Fix code format.
36541 (symtab_node::set_section_for_node): Fix comment typo.
36543 2017-01-06 Martin Liska <mliska@suse.cz>
36546 * lra-constraints.c: Rename invariant to lra_invariant.
36547 * predict.c (set_even_probabilities): Initialize e to NULL.
36549 2017-01-05 Martin Sebor <msebor@redhat.com>
36551 PR tree-optimization/78910
36552 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36553 (format_integer): Correct off-by-one error in the handling
36554 of precision with negative numbers in signed conversions..
36556 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36558 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36560 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36562 PR tree-optimization/71016
36563 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36564 factor_out_conditional_conversion. Formatting fix.
36565 (factor_out_conditional_conversion): Add cond_stmt argument.
36566 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36567 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36570 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36572 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36573 read-rtl-function.o, and selftest-rtl.o.
36574 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36575 (selftest::aarch64_test_loading_full_dump): New function.
36576 (selftest::aarch64_run_selftests): New function.
36577 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36578 selftest::aarch64_run_selftests.
36579 * config/i386/i386.c
36580 (selftest::ix86_test_loading_dump_fragment_1): New function.
36581 (selftest::ix86_test_loading_call_insn): New function.
36582 (selftest::ix86_test_loading_full_dump): New function.
36583 (selftest::ix86_test_loading_unspec): New function.
36584 (selftest::ix86_run_selftests): Call the new functions.
36585 * emit-rtl.c (maybe_set_max_label_num): New function.
36586 * emit-rtl.h (maybe_set_max_label_num): New decl.
36587 * function.c (instantiate_decls): Guard call to
36588 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36589 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36591 * gensupport.c (gen_reader::gen_reader): Pass "false"
36592 for new "compact" param of rtx_reader.
36593 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36594 rather than an empty string for NULL strings.
36595 * read-md.c: Potentially include config.h rather than bconfig.h.
36596 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36597 (have_error): New global, copied from errors.c.
36598 (md_reader::read_name): Rename to...
36599 (md_reader::read_name_1): ...this, adding "out_loc" param,
36600 and converting "missing name or number" to returning false, rather
36602 (md_reader::read_name): Reimplement in terms of read_name_1.
36603 (md_reader::read_name_or_nil): New function.
36604 (md_reader::read_string): Handle "(nil)" by returning NULL.
36605 (md_reader::md_reader): Add new param "compact".
36606 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36607 (md_reader::read_file): New method.
36608 * read-md.h (md_reader::md_reader): Add new param "compact".
36609 (md_reader::read_file): New method.
36610 (md_reader::is_compact): New accessor.
36611 (md_reader::read_name): Convert return type from void to file_location.
36612 (md_reader::read_name_or_nil): New decl.
36613 (md_reader::read_name_1): New decl.
36614 (md_reader::m_compact): New field.
36615 (noop_reader::noop_reader): Pass "false" for new "compact" param
36617 (rtx_reader::rtx_reader): Add new "compact" param.
36618 (rtx_reader::read_rtx_operand): Make virtual and convert return
36619 type from void to rtx.
36620 (rtx_reader::read_until): New decl.
36621 (rtx_reader::handle_any_trailing_information): New virtual function.
36622 (rtx_reader::postprocess): New virtual function.
36623 (rtx_reader::finalize_string): New virtual function.
36624 (rtx_reader::m_in_call_function_usage): New field.
36625 (rtx_reader::m_reuse_rtx_by_id): New field.
36626 * read-rtl-function.c: New file.
36627 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36628 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36629 (selftest::verify_three_block_rtl_cfg): New decl.
36630 * read-rtl-function.h: New file.
36631 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36632 For host, include function.h, memmodel.h, and emit-rtl.h.
36633 (one_time_initialization): New function.
36634 (struct compact_insn_name): New struct.
36635 (compact_insn_names): New array.
36636 (find_code): Handle insn codes in compact dumps.
36637 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36638 (bind_subst_iter_and_attr): Likewise.
36639 (add_condition_to_string): Likewise.
36640 (add_condition_to_rtx): Likewise.
36641 (apply_attribute_uses): Likewise.
36642 (add_current_iterators): Likewise.
36643 (apply_iterators): Likewise.
36644 (initialize_iterators): Guard usage of apply_subst_iterator with
36645 #ifdef GENERATOR_FILE.
36646 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36647 (md_reader::read_mapping): Likewise.
36648 (add_define_attr_for_define_subst): Likewise.
36649 (add_define_subst_attr): Likewise.
36650 (read_subst_mapping): Likewise.
36651 (check_code_iterator): Likewise.
36652 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36654 (one_time_initialization): New function.
36655 (rtx_reader::read_until): New method.
36656 (read_flags): New function.
36657 (parse_reg_note_name): New function.
36658 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36659 Handle reuse_rtx ids.
36660 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36661 Add parsing support for RTL dumps, mirroring the special-cases in
36662 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36663 values, and calling handle_any_trailing_information.
36664 (rtx_reader::read_rtx_operand): Convert return type from void
36665 to rtx, returning return_rtx. Handle case 'e'. Call
36666 finalize_string on XSTR and XTMPL fields.
36667 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36668 "(nil)" values were omitted. Call the postprocess vfunc on the
36670 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36671 class ctor. Initialize m_in_call_function_usage. Call
36672 one_time_initialization.
36673 * rtl-tests.c (selftest::test_uncond_jump): Call
36674 set_new_first_and_last_insn.
36675 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36676 * selftest-rtl.c: New file.
36677 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36678 (selftest::get_insn_by_uid): New decl.
36679 * selftest-run-tests.c (selftest::run_tests): Call
36680 read_rtl_function_c_tests.
36681 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36682 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36685 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36687 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36688 operands in a special way. Assert that pos+len <= mode precision.
36690 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36692 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36693 3 argument Alias with unlimited for the negative form.
36694 (fno-vect-cost-model): Removed.
36696 2017-01-05 Martin Liska <mliska@suse.cz>
36698 * hsa-gen.c (gen_hsa_divmod): New function.
36699 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36701 2017-01-05 Martin Liska <mliska@suse.cz>
36704 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36707 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36709 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36710 small constant length operands.
36712 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36714 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36715 between loop iterations.
36717 2017-01-05 Martin Liska <mliska@suse.cz>
36720 * gimplify.c (gimplify_decl_expr): Compare to
36721 asan_poisoned_variables instread of checking flags.
36722 (gimplify_target_expr): Likewise.
36723 (gimplify_expr): Likewise.
36724 (gimplify_function_tree): Conditionally initialize
36725 asan_poisoned_variables.
36727 2017-01-04 Jeff Law <law@redhat.com>
36729 PR tree-optimizatin/78812
36730 * rtl.h (contains_mem_rtx_p): Prototype.
36731 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36732 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36733 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36734 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36735 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36737 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36739 * input.c (assert_char_at_range): Default-initialize actual_range.
36741 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36743 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36744 to match the caller.
36746 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36748 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36749 insns after final jump in test to emit dummy move.
36751 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36753 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36754 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36756 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36758 * multiple_target.c (create_dispatcher_calls): Init e_next.
36759 * tree-ssa-loop-split.c (split_loop): Init border.
36760 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36763 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36768 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36769 (altivec_register_operand): Do not return true if the operand
36770 contains a SUBREG mixing SImode and SFmode.
36771 (vsx_register_operand): Likewise.
36772 (vsx_reg_sfsubreg_ok): New predicate.
36773 (vfloat_operand): Do not return true if the operand contains a
36774 SUBREG mixing SImode and SFmode.
36775 (vint_operand): Likewise.
36776 (vlogical_operand): Likewise.
36777 (gpc_reg_operand): Likewise.
36778 (int_reg_operand): Likewise.
36779 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36780 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36781 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36783 (rs6000_emit_move_si_sf_subreg): New helper function.
36784 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36785 fixup SUBREGs involving SImode and SFmode.
36786 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36787 numbers for the new peephole2 optimization.
36788 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36789 the GLIBC math library that do AND/IOR/XOR operations on single
36790 precision floating point.
36791 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36792 target macros to say whether we need to avoid SUBREGs mixing
36794 (TARGET_ALLOW_SF_SUBREG): Likewise.
36795 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36796 (UNSPEC_SI_FROM_SF): Likewise.
36797 (iorxor): Change spacing.
36798 (and_ior_xor): New iterator for AND, IOR, and XOR.
36799 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36800 (movdi_from_sf_zero_ext): Likewise.
36801 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36802 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36803 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36804 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36805 (fms<mode>4): Likewise.
36806 (fnma<mode>4): Likewise.
36807 (fnms<mode>4): Likewise.
36808 (nfma<mode>4): Likewise.
36809 (nfms<mode>4): Likewise.
36811 2017-01-04 Marek Polacek <polacek@redhat.com>
36814 * doc/invoke.texi: Document -Wpointer-compare.
36816 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36818 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36821 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36822 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36825 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36828 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36829 documentation of the powerpc_popcntb_ok attribute.
36830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36831 code to issue warning messages if a requested CPU configuration is
36832 not supported by the binary (assembler and loader) toolchain.
36833 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36834 made to define a built-in function that has been disabled.
36835 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36836 made to define a built-in function that has been disabled.
36837 (altivec_init_builtins): Add comment explaining why definition
36838 of the DST built-in functions is not preceded by an assertion
36839 check. Add assertions to prevent ICE if attempts are made to
36840 define an altivec predicate or an abs* built-in function that has
36842 (htm_init_builtins): Add comment explaining why definition of the
36843 htm built-in functions is not preceded by an assertion check.
36845 2017-01-04 Jeff Law <law@redhat.com>
36847 PR tree-optimizatin/67955
36848 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36849 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36850 the points-to solution does not include pt_null. Use DECL_PT_UID
36853 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36855 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36856 Use gen_int_mode instead of gen_lopwart for const_int operands.
36858 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36860 PR tree-optimization/71563
36861 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36862 out of range value - has low bits known to be zero.
36864 2017-01-04 Alan Modra <amodra@gmail.com>
36866 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36867 * configure: Regenerate.
36868 * config.in: Regenerate.
36870 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36873 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36874 a substring of the message, but strcmp with the whole message. Ifdef
36875 ENABLE_NLS, translate the message first using dgettext.
36877 2017-01-03 Jeff Law <law@redhat.com>
36879 PR tree-optimizatin/78856
36880 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36881 (mark_threaded_blocks): Remove code to truncate thread paths that
36882 cross multiple loop headers. Instead invalidate the cached loop
36883 iteration information and handle case of a thread path walking
36884 into an irreducible region.
36886 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36889 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36890 assertions. Add support for doing the signbit if the IEEE 128-bit
36891 floating point value is in a GPR.
36892 * config/rs6000/rs6000.md (Fsignbit): Delete.
36893 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36894 Update the length attribute if the value is in a GPR.
36895 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36896 the sign or zero extension instruction, since the value is always 0/1.
36897 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36900 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36901 extracting SImode to a GPR register so that we can generate a
36902 store, limit the vector to be in a traditional Altivec register
36903 for the vextuwrx instruction.
36905 2017-01-03 Ian Lance Taylor <iant@google.com>
36907 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36909 2017-01-03 Martin Sebor <msebor@redhat.com>
36911 PR tree-optimization/78696
36912 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36913 precision. Use MPFR for %f for greater fidelity. Correct handling
36915 (pass_sprintf_length::compute_format_length): Set width and precision
36916 specified by asrerisk to void_node for vararg functions.
36917 (try_substitute_return_value): Adjust dump output.
36919 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36921 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36923 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36925 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36926 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36927 -mlra/-mno-lra was passed to the compiler.
36929 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36931 PR rtl-optimization/65618
36932 * emit-rtl.c (try_split): Move initialization of "before" and
36933 "after" to just before the call to emit_insn_after_setloc.
36935 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36937 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36939 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36941 * dwarf2out.c (gen_enumeration_type_die): When
36942 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36944 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36946 PR tree-optimization/78965
36947 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36948 Change first argument from const call_info & to call_info &. For %n
36949 set info.nowrite to false.
36951 PR middle-end/78901
36952 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36953 possibly throwing calls.
36955 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36956 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36957 and fns handling, rather than in a separate case SSA_NAME.
36959 2017-01-02 Jeff Law <law@redhat.com>
36961 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36962 fixes for first_period and second_period variables.
36964 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36967 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36968 (*insvqi_1): New insn pattern.
36969 (*insvqi_1_mem_rex64): Ditto.
36970 (*insvqi_2): Ditto.
36971 (*insvqi_3): Rename from *insvqi.
36973 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36975 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36977 * doc/cfg.texi (Edges): Remove reference to Java.
36978 (Maintaining the CFG): Ditto.
36980 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36982 PR middle-end/77674
36983 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36984 transparent aliases.
36986 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36988 PR middle-end/77484
36989 * predict.def (PRED_CALL): Update hitrate.
36990 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36991 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36992 into direct/indirect/polymorphic variants.
36994 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36996 Update copyright years.
36998 * gcc.c (process_command): Update copyright notice dates.
36999 * gcov-dump.c (print_version): Ditto.
37000 * gcov.c (print_version): Ditto.
37001 * gcov-tool.c (print_version): Ditto.
37002 * gengtype.c (create_file): Ditto.
37003 * doc/cpp.texi: Bump @copying's copyright year.
37004 * doc/cppinternals.texi: Ditto.
37005 * doc/gcc.texi: Ditto.
37006 * doc/gccint.texi: Ditto.
37007 * doc/gcov.texi: Ditto.
37008 * doc/install.texi: Ditto.
37009 * doc/invoke.texi: Ditto.
37011 Copyright (C) 2017 Free Software Foundation, Inc.
37013 Copying and distribution of this file, with or without modification,
37014 are permitted in any medium without royalty provided the copyright
37015 notice and this notice are preserved.