1 2017-11-28 Jeff Law <law@redhat.com>
3 * gimple-ssa-evrp-analyze.c
4 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5 refine ranges if scev_initialized_p returns true.
6 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
8 2017-11-28 Julia Koval <julia.koval@intel.com>
10 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
11 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
12 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
13 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
14 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
15 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
16 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
17 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
18 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
19 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
20 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
21 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
22 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
23 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
24 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
25 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
26 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
27 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
28 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
29 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
30 __builtin_ia32_vpshrd_v2di_mask): New builtins.
31 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
33 2017-11-28 Julia Koval <julia.koval@intel.com>
35 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
36 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
37 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
38 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
39 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
40 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
41 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
42 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
43 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
44 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
45 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
46 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
47 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
48 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
49 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
50 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
51 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
52 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
53 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
54 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
55 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
56 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
57 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
58 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
59 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
60 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
61 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
62 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
63 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
64 __builtin_ia32_vpshld_v2di_mask): New builtins.
65 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
66 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
68 2017-11-28 Richard Biener <rguenther@suse.de>
70 PR tree-optimization/80776
71 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
73 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
75 (evrp_range_analyzer::record_ranges_from_incoming_edges):
76 If the incoming edge is an effective fallthru because the other
77 edge only reaches a __builtin_unreachable () then record ranges
78 derived from the controlling condition in SSA info.
79 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
80 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
82 2017-11-28 Olivier Hainque <hainque@adacore.com>
84 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
87 2017-11-28 Jakub Jelinek <jakub@redhat.com>
90 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
92 2017-11-28 Richard Biener <rguenther@suse.de>
95 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
96 copies generated from memcpy use a character array as reference
99 2017-11-28 Julia Koval <julia.koval@intel.com>
100 Sebastian Peryt <sebastian.peryt@intel.com>
102 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
103 c-family/c-cilkplus.o, c-family/array-notation-common.o,
104 cilk-common.o, cilk.h, cilk-common.c): Remove.
106 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
107 * builtins.c (is_builtin_name): Remove cilkplus condition.
108 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
109 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
110 cilk-builtins.def, cilkplus.def): Remove.
111 * cif-code.def (CILK_SPAWN): Remove.
112 * cilk-builtins.def: Delete.
113 * cilk-common.c: Ditto.
115 * cilkplus.def: Ditto.
116 * config/darwin.h (fcilkplus): Delete.
117 * cppbuiltin.c: Ditto.
118 * doc/extend.texi: Remove cilkplus doc.
119 * doc/generic.texi: Ditto.
120 * doc/invoke.texi: Ditto.
121 * doc/passes.texi: Ditto.
122 * gcc.c (fcilkplus): Remove.
123 * gengtype.c (cilk.h): Remove.
124 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
126 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
128 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
130 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
131 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
133 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
134 inline_read_section): Ditto.
135 * ipa-inline-analysis.c (cilk.h): Remove.
136 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
137 * lto-wrapper.c (merge_and_complain, append_compiler_options,
138 append_linker_options): Remove condition for fcilkplus.
139 * lto/lto-lang.c (cilk.h): Remove.
140 (lto_init): Remove condition for fcilkplus.
141 * omp-expand.c (expand_cilk_for_call): Delete.
142 (expand_omp_taskreg, expand_omp_for_static_chunk,
143 expand_omp_for): Remove cilkplus
145 (expand_cilk_for): Delete.
146 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
147 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
148 execute_lower_omp, diagnose_sb_0): Ditto.
149 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
150 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
151 * tree-nested.c: Ditto.
152 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
153 (dump_generic_node): Ditto.
154 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
155 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
157 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
159 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
161 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
162 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
163 for vector mode and !TARGET_SIMD.
165 2017-11-28 Jakub Jelinek <jakub@redhat.com>
167 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
169 * tree.h (SWITCH_LABELS): Remove.
170 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
171 assert SWITCH_BODY is non-NULL.
172 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
174 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
176 PR tree-optimization/80788
177 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
178 has TREE_OVERFLOW set, call drop_tree_overflow.
180 2017-11-28 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/83158
183 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
186 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
189 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
190 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
192 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
194 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
195 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
197 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
200 * varasm.c (compare_constant): Take the mode of the constants into
201 account when comparing floating point constants.
203 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
205 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
206 from explicit instantiation of debug_helper.
207 * vec.h (DEFINE_DEBUG_VEC): Ditto.
209 2017-11-27 Richard Biener <rguenther@suse.de>
211 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
214 2017-11-27 Richard Biener <rguenther@suse.de>
216 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
217 (type_hash_canon): Also clear int_cst_hash_table entry for
219 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
221 2017-11-27 Tamar Christina <tamar.christina@arm.com>
223 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
224 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
225 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
227 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
229 * hash-map.h (gt_cleare_cache): Avoid UB.
231 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
233 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
234 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
235 (decide_unroll_runtime_iterations): Likewise.
236 (decide_unroll_stupid): Likewise.
238 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
241 * config/i386/i386.md: Add a loop with incssp.
243 2017-11-27 Martin Jambor <mjambor@suse.cz>
245 PR tree-optimization/81248
246 * tree-sra.c (splice_param_accesses): Remove size check.
247 (decide_one_param_reduction): Fix size check.
248 * gimple-pretty-print.c (dump_profile): Silence warning.
249 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
251 2017-11-27 Jakub Jelinek <jakub@redhat.com>
254 * dbxout.c (lastlineno): New variable.
255 (dbx_debug_hooks): Use dbxout_switch_text_section as
256 switch_text_section debug hook.
257 (dbxout_function_end): Switch to current_function_section
258 rather than function_section. If crtl->has_bb_partition,
259 output just one N_FUN, depending on in_cold_section_p.
260 (dbxout_source_line): Remember last lineno in lastlineno.
261 (dbxout_switch_text_section): New function.
262 (dbxout_function_decl): Adjust dbxout_block caller.
263 (dbx_block_with_cold_children): New function.
264 (dbxout_block): Return true if any LBRAC/RBRAC have been
265 emitted. Use dbx_block_with_cold_children at depth == 0
266 in second partition. Add PARENT_BLOCKNUM argument, pass
267 it optionally adjusted to children. Output LBRAC/RBRAC
268 around recursive call only if the block is in the current
269 partition, if not and anything was output, emit empty
271 * final.c (final_scan_insn): Compute cold_function_name
272 before calling switch_text_section debug hook. Call
273 that hook even if dwarf2out_do_frame if not emitting
277 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
280 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
282 PR rtl-optimization/82488
283 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
285 2017-11-26 Julia Koval <julia.koval@intel.com>
287 * config/i386/i386.c (processor_target_table): Add skylake_cost for
289 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
292 2017-11-26 Julia Koval <julia.koval@intel.com>
294 * config/i386/driver-i386.c (host_detect_local_cpu):
295 Detect skylake-avx512.
297 2017-11-26 Julia Koval <julia.koval@intel.com>
299 * config.gcc: Add -march=cannonlake.
300 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
301 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
302 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
303 (PTA_CANNONLAKE): New.
304 (processor_target_table): Add cannonlake.
305 (ix86_option_override_internal): Ditto.
306 (fold_builtin_cpu): Ditto.
307 (get_builtin_code_for_version): Handle cannonlake.
308 (M_INTEL_COREI7_CANNONLAKE): New.
309 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
310 * doc/invoke.texi: Add -march=cannonlake.
312 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
314 * plugin.c (add_new_plugin): Use platform-specific library extensions.
315 (try_init_one_plugin): Alternative implementation for MinGW.
316 * Makefile.in (plugin_implib): New.
317 (gengtype-lex.c): Fix broken AIX workaround.
318 * configure: Regenerate.
319 * doc/plugins.texi: Document support for MinGW.
321 2017-11-25 Jakub Jelinek <jakub@redhat.com>
323 PR rtl-optimization/81553
324 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
325 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
326 is a shift where C1 has different mode than the whole shift, use C1's
327 mode for MULT rather than the shift's mode.
330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
331 builtins not enabled in the currently selected ISA.
333 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
335 PR tree-optimization/71026
336 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
337 (insert_reciprocals): Change to insert reciprocals before a division
338 by a square and to insert the square of a reciprocal.
339 (execute_cse_reciprocals_1): Change to consider division by a square.
340 (register_division_in): Add importance parameter.
342 2017-11-24 Richard Biener <rguenther@suse.de>
344 PR tree-optimization/82402
345 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
346 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
348 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
350 * match.pd (0-ptr): New transformation.
352 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
355 * ipa-inline.c (inline_small_functions): Set current badnes correctly
356 when skipping checking.
358 2017-11-24 Richard Biener <rguenther@suse.de>
360 PR tree-optimization/83128
361 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
362 (vn_reference_lookup_3): Likewise.
364 2017-11-24 Jakub Jelinek <jakub@redhat.com>
367 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
368 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
371 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
372 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
373 check for the latter with gimple_call_builtin_p. Do not handle
374 BUILT_IN_STPNCPY_CHK which is not a pass through call.
376 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
378 * config/arm/arm_neon.h: Fix pragma GCC push_options before
381 2017-11-23 Julia Koval <julia.koval@intel.com>
383 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
384 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
385 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
386 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
387 _mm512_maskz_expandloadu_epi16): New intrinsics.
388 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
389 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
390 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
391 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
392 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
393 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
394 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
395 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
396 _mm256_maskz_expandloadu_epi8): New intrinsics.
397 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
398 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
399 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
400 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
401 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
402 * config/i386/sse.md (VI248_VLBW): New iterator.
403 (expand<mode>_mask, expand<mode>_maskz): New patterns.
405 2017-11-23 Julia Koval <julia.koval@intel.com>
407 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
408 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
409 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
410 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
411 _mm512_mask_compressstoreu_epi16): New.
412 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
413 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
414 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
415 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
416 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
417 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
418 _mm256_mask_compressstoreu_epi8): New.
419 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
420 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
421 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
422 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
423 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
424 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
425 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
426 __builtin_ia32_compresshi128_mask,
427 __builtin_ia32_compressstoreuqi512_mask,
428 __builtin_ia32_compressstoreuhi512_mask,
429 __builtin_ia32_compressstoreuqi256_mask,
430 __builtin_ia32_compressstoreuqi128_mask,
431 __builtin_ia32_compressstoreuhi256_mask,
432 __builtin_ia32_compressstoreuhi128_mask): New builtins.
433 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
435 (ix86_expand_special_args_builtin): Handle new types.
436 (s4fma_expand): Handle new builtin array.
437 * config/i386/immintrin.h: Include new headers.
438 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
439 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
441 2017-11-23 Jakub Jelinek <jakub@redhat.com>
443 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
444 fixes. Declare temp and g variables at the top in order to avoid
445 {} in most of the cases.
447 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
449 * match.pd (ptr-0): New transformation.
451 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
453 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
454 (struct addr_mode_cost_table): New.
455 (struct tune_params): Add field addr_mode_costs.
456 * config/arm/arm.c (generic_addr_mode_costs): New.
457 (arm_slowmul_tune): Initialise addr_mode_costs field.
458 (arm_fastmul_tune): Likewise.
459 (arm_strongarm_tune): Likewise.
460 (arm_xscale_tune): Likewise.
461 (arm_9e_tune): Likewise.
462 (arm_marvell_pj4_tune): Likewise.
463 (arm_v6t2_tune): Likewise.
464 (arm_cortex_tune): Likewise.
465 (arm_cortex_a8_tune): Likewise.
466 (arm_cortex_a7_tune): Likewise.
467 (arm_cortex_a15_tune): Likewise.
468 (arm_cortex_a35_tune): Likewise.
469 (arm_cortex_a53_tune): Likewise.
470 (arm_cortex_a57_tune): Likewise.
471 (arm_exynosm1_tune): Likewise.
472 (arm_xgene1_tune): Likewise.
473 (arm_cortex_a5_tune): Likewise.
474 (arm_cortex_a9_tune): Likewise.
475 (arm_cortex_a12_tune): Likewise.
476 (arm_cortex_a73_tune): Likewise.
477 (arm_v7m_tune): Likewise.
478 (arm_cortex_m7_tune): Likewise.
479 (arm_v6m_tune): Likewise.
480 (arm_fa726te_tune): Likewise.
481 (arm_mem_costs): Use table lookup to calculate cost of addressing
484 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
486 * config/arm/arm.c (arm_mem_costs): New function.
487 (arm_rtx_costs_internal): Use arm_mem_costs.
489 2017-11-23 Mark Wielaard <mark@klomp.org>
491 * dwarf2out.c (init_sections_and_labels): Use generation to create
492 unique ranges_section_label and ranges_base_label. Return generation.
493 (output_rnglists): Add generation argument. Use generation to create
494 unique ranges labels.
495 (dwarf2out_finish): Get generation from init_sections_and_labels
496 and pass generation to output_rnglists.
498 2017-11-23 Mike Stump <mikestump@comcast.net>
499 Eric Botcazou <ebotcazou@adacore.com>
501 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
502 * cfgloop.h (struct loop): Add unroll field.
503 * function.h (struct function): Add has_unroll bitfield.
504 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
505 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
506 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
507 (pass_rtl_unroll_loops::gate): Likewise.
508 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
509 for which loop->unroll==1.
510 (decide_unroll_constant_iterations): Use note for consistency and
511 take loop->unroll into account. Return early if loop->unroll is set.
512 Fix thinko in existing test.
513 (decide_unroll_runtime_iterations): Use note for consistency and
514 take loop->unroll into account.
515 (decide_unroll_stupid): Likewise.
516 * lto-streamer-in.c (input_cfg): Read loop->unroll.
517 * lto-streamer-out.c (output_cfg): Write loop->unroll.
518 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
520 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
521 (print_loop): Print loop->unroll if set.
522 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
523 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
524 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
526 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
527 loop->unroll is set and smaller than the trip count. Otherwise bypass
528 entirely the heuristics if loop->unroll is set. Remove dead note.
529 Fix off-by-one bug in other note.
530 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
531 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
533 (tree_unroll_loops_completely): Make static.
534 (pass_complete_unroll::execute): Use correct type for variable.
535 (pass_complete_unrolli::execute): Fix formatting.
536 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
538 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
540 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
541 enable when TARGET_PREFER_AVX128 is set.
543 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
545 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
547 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
548 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
550 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
552 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
553 frequencies. Fix estimation of aggregate parameters.
555 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
557 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
560 2017-11-23 Tom de Vries <tom@codesourcery.com>
562 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
564 2017-11-23 Tom de Vries <tom@codesourcery.com>
566 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
569 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
572 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
573 sibcall_value_pcrel_fdpic): Use local variable instead of
575 (calli_tbr_rel): Add missing operand 2.
576 (call_valuei_tbr_rel): Add missing operand 3.
578 2017-11-23 Jakub Jelinek <jakub@redhat.com>
581 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
582 bitpos/bitsize covering the whole destination, use store_expr only if
583 the complex mode is the same. Otherwise, use expand_normal and if
584 it returns CONCAT, subreg each part separately instead of trying to
585 subreg the whole result.
587 2017-11-23 Richard Biener <rguenther@suse.de>
589 PR tree-optimization/23094
590 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
591 come here from walking over backedges in the first iteration.
592 (vn_reference_lookup_3): Skip clobbers that store the same value.
594 2017-11-23 Richard Biener <rguenther@suse.de>
596 PR tree-optimization/81403
597 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
598 a block we need a leader relative to.
599 (phi_translate_1): For nary processing require a leader from
600 get_representative_for given we run expression simplification
601 using match-and-simplify. Remove previous fix.
603 2017-11-22 Jeff Law <law@redhat.com>
605 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
606 Use new method allocate_value_range rather than accessing the
607 vrp_value_range_pool data member directly.
608 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
609 to use extract_range_from_stmt method to avoid need for
610 extract_range_from_assignment method.
611 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
612 method rather than setting values_propgated data member directly.
613 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
614 and values propagated data members and extract_range_from_assignment
615 method. Reorder private data members to conform to standards.
616 Add new methods set_lattice_propagation_complete and
617 allocate_value_range.
619 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
621 PR rtl-optimization/83030
622 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
623 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
624 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
625 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
626 CROSSING_JUMP_P flag is already set before setting it.
627 * cfgrtl.c (fixup_partition_crossing): Likewise.
628 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
631 2017-11-22 Jakub Jelinek <jakub@redhat.com>
633 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
634 Handle the case where both arguments are using gen_const_vec_series.
636 2017-11-22 David Malcolm <dmalcolm@redhat.com>
639 * pretty-print.c (pp_format): Move quoting implementation to
640 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
641 to pass address of "quote" local.
642 (pp_begin_quote): New function.
643 (pp_end_quote): New function.
644 * pretty-print.h (printer_fn): Convert penultimate param from bool
646 (pp_begin_quote): New decl.
647 (pp_end_quote): New decl.
648 * tree-diagnostic.c (default_tree_printer): Convert penultimate
649 param from bool to bool *.
650 * tree-diagnostic.h (default_tree_printer): Likewise.
652 2017-11-22 Jeff Law <law@redhat.com>
654 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
655 Initialize vr_values.
656 (evrp_range_analyzer::try_find_new_range): Call methods attached to
657 vr_values via vr_values class instance rather than delegators.
658 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
659 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
660 (evrp_range_analyzer::push_value_range): Likewise.
661 (evrp_range_analyzer::pop_value_range): Likewise.
662 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
663 most delegators. Those remaining are exposed as public interfaces.
664 Make vr_values a pointer and private.
665 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
667 (evrp_range_analyzer::get_vr_value): New method.
668 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
669 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
670 (class evrp_dom_walker): Attach evrp_folder class, initialize
671 it in the ctor. Remove temporary delegators.
672 (evrp_dom_walker::before_dom_children): Call methods in attached
673 evrp_range_analyzer class via class instance pointer. Use
674 free value_range_constant_singleton to remove need for
675 op_with_constant_singleton_value delegator method. Do not
676 create a vrp_prop class instance for every call! Narrow
677 scope of a couple variables.
678 (evrp_dom_walker::cleanup): Call methods in attached
679 evrp_range_analyzer class via class instance pointer.
680 * vr-values.h (class vr_values): Privatize many methods and
683 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
685 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
686 * cfgexpand.c (expand_debug_expr): Remove handling for them.
687 * expr.c (expand_expr_real_2): Likewise.
688 * fold-const.c (const_unop): Likewise.
689 * optabs-tree.c (optab_for_tree_code): Likewise.
690 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
691 * tree-inline.c (estimate_operator_cost): Likewise.
692 * tree-pretty-print.c (dump_generic_node): Likewise.
693 (op_code_prio): Likewise.
694 (op_symbol_code): Likewise.
695 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
696 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
697 * internal-fn.c (direct_internal_fn_optab): New function.
698 (direct_internal_fn_array, direct_internal_fn_supported_p
699 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
700 * fold-const-call.c (fold_const_reduction): New function.
701 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
703 * tree-vect-loop.c: Include internal-fn.h.
704 (reduction_code_for_scalar_code): Rename to...
705 (reduction_fn_for_scalar_code): ...this and return an internal
707 (vect_model_reduction_cost): Take an internal_fn rather than
709 (vect_create_epilog_for_reduction): Likewise. Build calls rather
711 (vectorizable_reduction): Use internal functions rather than tree
712 codes for the reduction operation. Update calls to the functions
714 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
715 Use calls to internal functions rather than REDUC tree codes.
716 * config/aarch64/aarch64-simd.md: Update comment accordingly.
718 2017-11-22 Olivier Hainque <hainque@adacore.com>
720 * config/vxworks.c (vxworks_override_options): Pick default
721 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
722 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
723 DWARF_GNAT_ENCODINGS_DEFAULT.
724 * config/vxworksae.h: Likewise.
726 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
728 PR tree-optimization/83104
729 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
732 2017-11-22 Marek Polacek <polacek@redhat.com>
733 H.J. Lu <hongjiu.lu@intel.com>
734 Jason Merrill <jason@redhat.com>
739 * calls.c (initialize_argument_information): Call
740 warn_parameter_passing_abi target hook.
741 (store_one_arg): Use 0 for empty record size. Don't push 0 size
743 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
744 * common.opt: Update -fabi-version description.
745 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
746 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
748 (ix86_is_empty_record): New function.
749 (ix86_warn_parameter_passing_abi): New function.
750 (TARGET_EMPTY_RECORD_P): Redefine.
751 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
752 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
753 * doc/tm.texi: Regenerated.
754 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
755 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
756 * dwarf2out.c (get_ultimate_context): Move to tree.c.
757 * explow.c (hard_function_value): Call arg_int_size_in_bytes
758 instead of int_size_in_bytes.
759 * expr.c (copy_blkmode_to_reg): Likewise.
760 * function.c (aggregate_value_p): Return 0 for empty types.
761 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
762 (locate_and_pad_parm): Call arg size_in_bytes instead
764 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
765 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
766 * target.def (empty_record_p, warn_parameter_passing_abi): New target
768 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
769 (std_gimplify_va_arg_expr): Skip empty records. Call
770 arg_size_in_bytes instead size_in_bytes.
771 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
772 * tree-core.h (tree_type_common): Add empty_flag.
773 (tree_decl_common): Update comments.
774 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
776 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
777 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
779 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
780 * tree.c (default_is_empty_type): New function.
781 (default_is_empty_record): New function.
782 (arg_int_size_in_bytes): New function.
783 (arg_size_in_bytes): New function.
784 (get_ultimate_context): New function.
785 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
786 TRANSLATION_UNIT_WARN_EMPTY_P.
787 (default_is_empty_record, arg_int_size_in_bytes,
788 arg_size_in_bytes, get_ultimate_context): Declare.
790 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
792 * config/arm/arm.c (cmse_clear_registers): New function.
793 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
794 code by call to cmse_clear_registers.
795 (cmse_nonsecure_entry_clear_before_return): Likewise.
797 2017-11-22 Tamar Christina <tamar.christina@arm.com>
799 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
800 (vdot_s32, vdotq_s32): New.
801 (vdot_lane_u32, vdotq_lane_u32): New.
802 (vdot_lane_s32, vdotq_lane_s32): New.
805 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
808 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
809 for unsigned values with fewer HWIs than the precision.
810 (test_overflow): New function.
811 (wide_int_cc_tests): Call it.
813 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
815 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
816 has the same mode class as Pmode.
818 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
820 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
821 padding_bits_to_clear_ptr.
822 (cmse_nonsecure_entry_clear_before_return): Likewise.
824 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
826 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
827 auto_sbitap instead of integer bitfield to control register needing
830 2017-11-22 Jakub Jelinek <jakub@redhat.com>
832 PR tree-optimization/83044
833 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
834 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
835 that need the upper bound. Subtract offset from
836 get_addr_base_and_unit_offset only if positive and subtract it
837 before division by eltsize rather than after it.
840 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
841 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
842 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
846 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
847 expand_binop with *mul_widen_optab, make sure at least one of the
848 operands doesn't have VOIDmode.
851 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
853 PR rtl-optimization/82044
854 PR tree-optimization/82042
855 * dse.c (record_store): Check for overflow.
856 (check_mem_read_rtx): Properly check for overflow if width == -1, call
857 add_wild_read instead of clear_rhs_from_active_local_stores on
858 overflow and log it into dump_file.
860 2017-11-22 Richard Biener <rguenther@suse.de>
862 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
863 fake edges to exit when looking for a place to insert.
864 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
866 (insert_into_preds_of_block): Commit edge insertion immediately,
867 assert that doesn't require new BBs.
868 (fini_pre): Release expressions.
869 (pass_pre::execute): Shuffle things around a bit, if the fn
870 is too large do not compute AVAIL either as this is really the
873 2017-11-22 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/83089
876 * tree-if-conv.c (pass_if_conversion::execute): If anything
877 changed reset SCEV and free the number of iteration estimates.
879 2017-11-21 Martin Sebor <msebor@redhat.com>
881 PR tree-optimization/82945
882 * calls.h (warn_nonstring_bound): Remove unused function.
884 2017-11-21 Martin Sebor <msebor@redhat.com>
886 PR tree-optimization/82945
887 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
888 * calls.h (maybe_warn_nonstring_arg): Declare new function.
889 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
891 (initialize_argument_information): Call maybe_warn_nonstring_arg.
892 * calls.h (get_attr_nonstring_decl): Declare new function.
893 * doc/extend.texi (attribute nonstring): Update.
894 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
895 get_attr_nonstring_decl and handle it.
896 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
897 detection of nul-termination.
898 (strlen_to_stridx): Change to a pointer.
899 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
900 (pass_strlen::execute): Same.
902 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
904 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
905 for the new option -mprefer-vector-width=[none|128|256|512].
906 * config/i386/i386.c (ix86_target_string): remove old style options
907 -mprefer-avx256 and make -mprefer-avx128 as alias.
908 (ix86_option_override_internal): Apply defaults for the
909 -mprefer-vector-width=[128|256] option.
910 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
911 Implement macros to work with -mprefer-vector-width=.
912 * config/i386/i386.opt: Implemented option
913 -mprefer-vector-width=[none|128|256|512].
914 * doc/invoke.texi: Documentation for
915 -mprefer-vector-width=[none|128|256|512].
917 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
919 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
921 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
923 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
925 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
927 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
929 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
930 * expr.c (expand_expr_real_2): Likewise.
931 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
932 fold_binary_loc): Likewise.
933 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
934 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
935 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
936 MINUS_EXPR transformations.
937 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
938 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
939 * tree-inline.c (estimate_operator_cost): Likewise.
940 * tree-pretty-print.c (dump_generic_node, op_code_prio,
941 op_symbol_code): Likewise.
942 * tree-vect-stmts.c (vectorizable_operation): Likewise.
943 * vr-values.c (extract_range_from_binary_expr): Likewise.
944 * varasm.c (initializer_constant_valid_p_1): Likewise.
945 * tree.def: New tree code POINTER_DIFF_EXPR.
947 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
949 * config/i386/i386.md (*bswap<mode>2_movbe): Add
950 integer suffix to movbe mnemonic.
951 (*bswaphi2_movbe): Ditto.
952 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
954 2017-11-21 Jakub Jelinek <jakub@redhat.com>
957 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
958 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
959 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
961 2017-11-21 Martin Liska <mliska@suse.cz>
963 * tree-inline.c (expand_call_inline): Remove not needed
966 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
967 Jakub Jelinek <jakub@redhat.com>
970 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
971 Remove static keyword from f variable.
973 2017-11-21 Jakub Jelinek <jakub@redhat.com>
975 PR tree-optimization/83086
976 * gimple-ssa-store-merging.c
977 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
978 rather than n.base_addr.
980 2017-11-21 Martin Liska <mliska@suse.cz>
982 PR rtl-optimization/82044
983 PR tree-optimization/82042
984 * dse.c (check_mem_read_rtx): Check for overflow.
986 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
988 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
991 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
993 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
994 and its corresponding call arg location note.
996 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
997 Andrew Burgess <andrew.burgess@embecosm.com>
999 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1001 (arc_return_slot_offset): Likewise.
1002 (arc_eh_return_address_location): New declaration.
1003 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1004 (MUST_SAVE_REGISTER): Add exception handler case.
1005 (MUST_SAVE_RETURN_ADDR): Likewise.
1006 (arc_frame_pointer_required): Likewise.
1007 (arc_frame_pointer_needed): New function.
1008 (arc_compute_frame_size): Changed.
1009 (arc_expand_prologue): Likewise.
1010 (arc_expand_epilogue): Likewise.
1011 (arc_initial_elimination_offset): Likewise.
1012 (arc_return_slot_offset): Delete.
1013 (arc_eh_return_address_location): New function.
1014 (arc_builtin_setjmp_frame_value): Likewise.
1015 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1016 (EH_RETURN_STACKADJ_RTX): Define.
1017 (EH_RETURN_HANDLER_RTX): Likewise.
1018 * config/arc/arc.md (eh_return): Delete.
1020 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1022 * print-tree.h (debug_vec_tree): Remove prototype.
1023 * gdbinit.in (pvt): Remove macro.
1025 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1027 PR tree-optimization/83047
1028 * gimple-ssa-store-merging.c
1029 (imm_store_chain_info::output_merged_store): If the loads with the
1030 same vuse are in different basic blocks, for load_gsi pick a load
1031 location that is dominated by the other loads.
1034 * config/i386/i386.c (ix86_memmodel_check): Start
1035 -Winvalid-memory-model diagnostics with lowercase letter.
1038 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1039 set in_cold_section_p to the partition containing loc_list->first.
1040 When seeing loc_list->last_before_switch node, update secname and
1041 perform range_across_switch second partition handling only after that.
1044 * run-rtl-passes.c: Include debug.h.
1045 (run_rtl_passes): Call debug_hooks->assembly_start.
1046 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1050 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1051 OPTAB_DIRECT in calls to expand_simple_binop.
1053 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1056 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1058 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1061 * spellcheck.h (best_match::blithely_get_best_candidate): New
1064 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1066 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1067 with lower case letter.
1069 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1071 * config/i386/i386.md (bswaphi2): New expander.
1072 (*bswaphi2_movbe): New insn pattern.
1073 (bswaphi -> rorhi pepehole2): New peephole pattern.
1075 2017-11-20 Jeff Law <law@redhat.com>
1077 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1078 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1079 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1080 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1081 gimple-ssa-evrp-analyze.h.
1083 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1084 set BB_VISITED here.
1085 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1087 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1089 * tree-predcom.c: Add general comment on Store-Store chains.
1090 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1092 (get_chain_last_ref_at): Rename into...
1093 (get_chain_last_write_at): ...this.
1094 (get_chain_last_write_before_load): New function.
1095 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1096 CT_STORE_STORE when write reference is added.
1097 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1098 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1099 (initialize_root_vars_store_elim_1): Ditto.
1100 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1101 definition is created.
1102 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1103 chain by replacing it with dominant stored value.
1105 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1107 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1109 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1111 * vr-values.c (extract_range_from_binary_expr): Use a full range
1114 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1116 * config/arm/arm.md (R4_REGNUM): Define constant.
1117 (nonsecure_call_internal): Remove r4 clobber.
1118 (nonsecure_call_value_internal): Likewise.
1119 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1120 clobber and resequence match_operands.
1121 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1122 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1123 (nonsecure_call_value_reg_thumb2): Likewise.
1125 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1127 PR tree-optimization/78821
1128 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1129 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1130 to the address of the base rather than the base itself.
1131 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1132 (find_bswap_or_nop_finalize): New function.
1133 (find_bswap_or_nop): Use it.
1134 (bswap_replace): Return a tree rather than bool, change first
1135 argument from gimple * to gimple_stmt_iterator, allow inserting
1136 into an empty sequence, allow ins_stmt to be NULL - then emit
1137 all stmts into gsi. Fix up MEM_REF address gimplification.
1138 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1139 (struct store_immediate_info): Add N and INS_STMT non-static
1141 (store_immediate_info::store_immediate_info): Initialize them
1142 from newly added ctor args.
1143 (merged_store_group::apply_stores): Formatting fixes. Sort by
1145 (stmts_may_clobber_ref_p): For stores call also
1146 refs_anti_dependent_p.
1147 (gather_bswap_load_refs): New function.
1148 (imm_store_chain_info::try_coalesce_bswap): New method.
1149 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1150 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1151 (imm_store_chain_info::output_merged_store): Fail if number of
1152 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1153 and NOP_EXPR rhs_code.
1154 (pass_store_merging::process_store): Compute n and ins_stmt, if
1155 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1156 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1158 (pass_store_merging::execute): Calculate dominators.
1160 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1161 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1162 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1163 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1164 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1165 class pass_optimize_bswap, bswap_replace,
1166 pass_optimize_bswap::execute): Moved to ...
1167 * gimple-ssa-store-merging.c: ... this file.
1168 Include optabs-tree.h.
1169 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1170 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1171 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1172 anonymous namespace, remove static keywords.
1173 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1174 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1176 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1179 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1181 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1183 * vec.h (debug_helper): New function.
1184 (DEFINE_DEBUG_VEC): New macro.
1185 * hash-set.h (debug_helper): New function.
1186 (DEFINE_DEBUG_HASH_SET): New macro.
1187 * cfg.c (debug_slim (edge)): New function.
1188 Call DEFINE_DEBUG_VEC for edges.
1189 Call DEFINE_DEBUG_HASH_SET for edges.
1190 * cfghooks.c (debug_slim (basic_block)): New function.
1191 Call DEFINE_DEBUG_VEC for basic blocks.
1192 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1193 * print-tree.c (debug_slim): New function to handle trees.
1194 Call DEFINE_DEBUG_VEC for trees.
1195 Call DEFINE_DEBUG_HASH_SET for trees.
1196 (debug (vec<tree, va_gc>) &): Remove.
1197 (debug (<vec<tree, va_gc>) *): Remove.
1198 * print-rtl.c (debug_slim): New function to handle const_rtx.
1199 Call DEFINE_DEBUG_VEC for rtx_def.
1200 Call DEFINE_DEBUG_VEC for rtx_insn.
1201 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1202 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1203 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1204 (debug (vec<rtx_insn *> *ptr): Remove.
1205 (debug_insn_vector): Remove.
1206 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1208 2017-11-20 Tom de Vries <tom@codesourcery.com>
1210 PR rtl-optimization/82020
1211 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1212 IF_THEN_ELSE condition.
1214 2017-11-19 Jeff Law <law@redhat.com>
1216 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1217 of degenerates resulting from ignoring an edge.
1219 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1222 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1224 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1227 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1228 for uninitialied counts.
1230 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1233 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1235 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1238 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1240 2017-11-19 Tom de Vries <tom@codesourcery.com>
1242 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1245 2017-11-19 Tom de Vries <tom@codesourcery.com>
1247 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1250 2017-11-19 Tom de Vries <tom@codesourcery.com>
1252 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1255 2017-11-19 Tom de Vries <tom@codesourcery.com>
1257 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1259 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1261 2017-11-19 Tom de Vries <tom@codesourcery.com>
1263 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1265 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1266 semicolon after MCORE_EXPORT_NAME call.
1268 2017-11-19 Tom de Vries <tom@codesourcery.com>
1271 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1272 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1274 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1276 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1277 (cgraph_edge::clone): Cleanup updating of profile.
1278 * ipa-cp.c (update_profiling_info): Likewise.
1279 * ipa-inline-transform.c (inline_transform): Likewise.
1280 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1281 * ipa-split.c (execute_split_functions): Do not split when function
1283 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1284 * profile-count.c (profile_count::dump): Add global0.
1285 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1287 (profile_count::to_sreal_scale): Likewise.
1288 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1289 (profile_count::combine_with_ipa_count): New function.
1290 * profile-count.h (profile_guessed_global0adjusted): New.
1291 (profile_count::adjusted_zero): New.
1292 (profile_count::global0adjusted): New.
1293 (profile_count::combine_with_ipa_count): New.
1294 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1295 correct profile of return block of split functions.
1296 (copy_cfg_body): Remove unused profile_count.
1297 (copy_body): Likewise.
1298 (expand_call_inline): Update.
1299 (tree_function_versioning): Update.
1301 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1303 * hash-set.h (hash_set::empty): New.
1304 * tree-ssa-threadbackward.h: Delete.
1305 * tree-ssa-threadbackward.c (class thread_jumps): New.
1306 Move max_threaded_paths into class.
1307 (fsm_find_thread_path): Remove arguments that are now in class.
1308 (profitable_jump_thread_path): Rename to...
1309 (thread_jumps::profitable_jump_thread_path): ...this.
1310 (convert_and_register_jump_thread_path): Rename to...
1311 (thread_jumps::convert_and_register_current_path): ...this.
1312 (check_subpath_and_update_thread_path): Rename to...
1313 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1314 (register_jump_thread_path_if_profitable): Rename to...
1315 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1316 (handle_phi): Rename to...
1317 (thread_jumps::handle_phi): ...this.
1318 (handle_assignment): Rename to...
1319 (thread_jumps::handle_assignment): ...this.
1320 (fsm_find_control_statement_thread_paths): Rename to...
1321 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1322 (find_jump_threads_backwards): Rename to...
1323 (thread_jumps::find_jump_threads_backwards): ...this.
1324 Initialize path local data.
1325 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1326 from within thread_jumps class.
1327 (pass_early_thread_jumps::execute): Same.
1329 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1331 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1333 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1335 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1336 to canonical location.
1338 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1341 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1343 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1345 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1347 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1348 rcrt1.o%s/grcrt1.o%s for -static-pie.
1350 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1352 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1353 ix86_shift_rotate_cost): Break out from ...
1354 (ix86_rtx_costs): ... here.
1355 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1358 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1360 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1361 when entry block was promoted unlikely.
1362 (estimate_bb_frequencies): Increase frequency scale.
1363 * profile-count.h (profile_count): Export precision info.
1365 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1367 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1368 disturbin profile of entry block.
1370 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1372 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1375 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1377 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1378 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1380 2017-11-17 Jeff Law <law@redhat.com>
1382 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1383 from evrp_dom_walker class. Various methods moved into new class.
1384 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1385 (evrp_range_analyzer::enter): New method.
1386 (evrp_range_analyzer::leave): New method.
1387 (evrp_dom_walker): Remove delegators no longer needed by this class.
1388 Replace vr_values data member with evrp_range_analyzer
1390 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1391 method extracted from evrp_dom_walker::before_dom_children.
1392 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1393 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1395 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1396 Add private copy constructor and move assignment operators.
1397 Privatize methods and class data where trivially possible.
1398 (evrp_dom_walker::cleanup): New function, extracted from
1399 execute_early_vrp. Simplify access to class data.
1401 * vr-values.h (get_output_for_vrp): Prototype.
1402 * vr-values.c (get_output_for_vrp): New function extracted from
1403 vrp_visit_assignment_or_call and extract_range_from_stmt.
1404 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1406 2017-11-17 Luis Machado <luis.machado@linaro.org>
1408 * config/aarch64/aarch64.c
1409 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1410 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1411 tune_params::AUTOPREFETCHER_WEAK.
1413 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1416 * config/arm/arm.c (arm_valid_target_attribute_rec):
1417 Parse "arch=" and "+<ext>".
1418 (arm_valid_target_attribute_tree): Re-init global options.
1419 (arm_option_override): Make non-static.
1420 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1421 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1422 (__ARM_FEATURE_CRC32): Support undef.
1423 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1424 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1426 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1428 * gdbinit.in (break-on-diagnostic): New command.
1430 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1432 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1433 return to indirect jump for EH return if control-flow
1434 protection is enabled. Change explicit 'false' argument in
1435 pro_epilogue_adjust_stack with a value of
1437 * config/i386/i386.md (simple_return_indirect_internal):
1438 Remove SImode restriction to support 64-bit.
1440 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1442 * combine.c (added_notes_insn): New.
1443 (try_combine): Handle added_notes_insn like added_links_insn.
1444 Rewrite return value code.
1445 (distribute_notes): Set added_notes_insn to the earliest insn we added
1448 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1450 PR rtl-optimization/82621
1451 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1452 dest of one of those SETs is unused.
1454 2017-11-17 Richard Biener <rguenther@suse.de>
1457 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1458 * tree-pretty-print.c (dump_generic_node): Handle
1459 annot_expr_parallel_kind.
1460 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1461 * gimplify.c (gimple_boolify): Likewise.
1463 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1465 * config.gcc (extra_headers): Add cet.h for x86 targets.
1466 * config/i386/cet.h: New file.
1467 * doc/install.texi: Add --enable-cet/--disable-cet.
1469 2017-11-17 Richard Biener <rguenther@suse.de>
1471 PR tree-optimization/83017
1472 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1473 (gen_parallel_loop): Properly count iterations.
1474 (parallelize_loops): Handle loop->can_be_parallel independent
1475 of flag_loop_parallelize_all. Make static profitability test match
1477 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1478 * invoke.texi (parloops-min-per-thread): Document.
1480 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1482 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1483 upstreaming review comments.
1485 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1487 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1488 with fast unaligned access.
1489 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1491 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1493 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1494 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1496 2017-11-17 Richard Biener <rguenther@suse.de>
1498 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1499 folding of references.
1501 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1504 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1505 of replacing call to strncmp with corresponding call to strcmp when
1508 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1510 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1511 option prefer-avx256 for skylake-avx512 configuration.
1512 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1513 (get_builtin_code_for_version): Ditto.
1515 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1516 Monk Chiang <sh.chiang04@gmail.com>
1518 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1519 (FIXED_REGISTERS): Reserve more register numbers.
1520 (CALL_USED_REGISTERS): Likewise.
1521 (REG_ALLOC_ORDER): Likewise.
1522 (REG_CLASS_CONTENTS): Likewise.
1523 (REGISTER_NAMES): Likewise.
1525 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1526 Kito Cheng <kito.cheng@gmail.com>
1528 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1530 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1532 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1534 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1535 parameter list for vec_splats.
1537 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1539 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1541 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1543 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1546 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1547 with the required version of automake 1.11.6.
1549 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1551 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1553 (DU_C2_3_power9): Correct reservation combinations.
1554 (FP_DIV_power9, VEC_DIV_power9): New.
1555 (power9-alu): Split out rotate/shift...
1556 (power9-rot): ...to here, correct dispatch resource.
1557 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1559 (power9-fp): Correct latency.
1560 (power9-sdiv): Add div/sqrt resource.
1561 (power9-ddiv): Correct latency, add div/sqrt resource.
1562 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1563 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1565 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1567 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1569 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1570 switch statement mapping KF built-ins to TF built-ins if we don't
1571 have the proper ISA 3.0 assembler support.
1573 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1575 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1576 (lower_emutls_function_body): Do not compute it.
1578 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1580 * ipa-split.c (split_bb_info): Turn time to sreal.
1581 (split_point): Likewise.
1582 (dump_split_point): Likewise.
1583 (fine_split_points): Likewise.
1584 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1586 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1588 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1591 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1593 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1595 (expensive_function_p): Remove useless assert.
1596 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1598 2017-11-16 Martin Sebor <msebor@redhat.com>
1600 PR tree-optimization/82588
1601 PR tree-optimization/82583
1602 * tree-vrp.c (check_array_ref): Handle flexible array members,
1603 string literals, and inner indices.
1604 (search_for_addr_array): Add detail to diagnostics.
1606 2017-11-16 Nathan Sidwell <nathan@acm.org>
1610 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1611 (SET_DECL_ASSEMBLER_NAME): Forward to
1612 overwrite_decl_assembler_name.
1613 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1614 (overwrite_decl_assembler_name): Declare.
1615 * tree.c (overwrite_decl_assembler_name): New.
1616 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1617 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1618 (LANG_HOOKS_INITIALIZER): Add it.
1619 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1620 * langhooks.c (lhd_set_decl_assembler_name): Use
1621 SET_DECL_ASSEMBLER_NAME.
1622 (lhd_overwrite_decl_assembler_name): Default implementation.
1624 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1625 Jackson Woodruff <jackson.woodruff@arm.com>
1627 PR tree-optimization/71026
1628 * match.pd: Canonicalize constant multiplies in division.
1630 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1632 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1633 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1634 Remove OPT_fomit_frame_pointer entry.
1635 * common/config/arc/arc-common.c: Likewise.
1636 * common/config/arm/arm-common.c: Likewise.
1637 * common/config/avr/avr-common.c: Likewise.
1638 * common/config/c6x/c6x-common.c: Likewise.
1639 * common/config/cr16/cr16-common.c: Likewise.
1640 * common/config/cris/cris-common.c: Likewise.
1641 * common/config/epiphany/epiphany-common.c: Likewise.
1642 * common/config/fr30/fr30-common.c: Likewise.
1643 * common/config/frv/frv-common.c: Likewise.
1644 * common/config/ia64/ia64-common.c: Likewise.
1645 * common/config/iq2000/iq2000-common.c: Likewise.
1646 * common/config/lm32/lm32-common.c: Likewise.
1647 * common/config/m32r/m32r-common.c: Likewise.
1648 * common/config/mcore/mcore-common.c: Likewise.
1649 * common/config/microblaze/microblaze-common.c: Likewise.
1650 * common/config/mips/mips-common.c: Likewise.
1651 * common/config/mmix/mmix-common.c: Likewise.
1652 * common/config/mn10300/mn10300-common.c: Likewise.
1653 * common/config/nios2/nios2-common.c: Likewise.
1654 * common/config/pa/pa-common.c: Likewise.
1655 * common/config/pdp11/pdp11-common.c: Likewise.
1656 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1657 * common/config/riscv/riscv-common.c: Likewise.
1658 * common/config/rs6000/rs6000-common.c: Likewise.
1659 * common/config/rx/rx-common.c: Likewise.
1660 * common/config/s390/s390-common.c: Likewise.
1661 * common/config/sh/sh-common.c: Likewise.
1662 * common/config/sparc/sparc-common.c: Likewise.
1663 * common/config/tilegx/tilegx-common.c: Likewise.
1664 * common/config/tilepro/tilepro-common.c: Likewise.
1665 * common/config/v850/v850-common.c: Likewise.
1666 * common/config/visium/visium-common.c: Likewise.
1667 * common/config/xstormy16/xstormy16-common.c: Likewise.
1668 * common/config/xtensa/xtensa-common.c: Likewise.
1669 * invoke.texi (-fomit-frame-pointer): Update documentation.
1671 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1673 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1675 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1677 * cfg.c (scale_bbs_frequencies_int,
1678 cale_bbs_frequencies_gcov_type): Remove.
1679 * cfg.h (scale_bbs_frequencies_int,
1680 cale_bbs_frequencies_gcov_type): Remove.
1682 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1684 * tree-ssa-loop-manip.c
1685 (scale_dominated_blocks_in_loop): Update to profile counts.
1686 (tree_transform_and_unroll_loop): Likewise.
1688 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1690 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1691 scale_bbs_frequencies_int.
1693 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1695 * final.c (compute_alignments): Use counts rather than frequencies.
1697 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1699 * cfgloopanal.c: Include sreal.h
1700 (average_num_loop_insns): Use counts and sreal for accounting.
1702 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1704 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1707 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1708 Kito Cheng <kito.cheng@gmail.com>
1710 * config/nds32/constraints.md: Provide more constraints.
1711 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1712 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1713 support constraints usage.
1715 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1717 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1719 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1720 Kito Cheng <kito.cheng@gmail.com>
1722 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1723 * config/nds32/nds32.opt: Refine the layout.
1724 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1725 TARGET_EXT_STRING): Support new options.
1726 * config/nds32/nds32.h: Likewise.
1727 * config/nds32/nds32.md: Likewise.
1728 * config/nds32/nds32-predicates.c: Likewise.
1729 * config/nds32/constraints.md: Likewise.
1730 * common/config/nds32/nds32-common.c: Likewise.
1732 2017-11-16 Julia Koval <julia.koval@intel.com>
1735 * config/i386/gfniintrin.h: Add sse check.
1736 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1738 2017-11-16 Julia Koval <julia.koval@intel.com>
1740 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1741 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1742 (ix86_handle_option): Handle -mavx512vbmi2.
1743 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1744 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1745 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1746 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1747 (ix86_valid_target_attribute_inner_p): Ditto.
1748 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1749 * config/i386/i386.opt (mavx512vbmi2): New option.
1750 * doc/invoke.texi: Add new option.
1752 2017-11-16 Julia Koval <julia.koval@intel.com>
1754 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1755 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1756 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1757 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1758 _mm512_gf2p8mul_epi8): New intrinsics.
1759 * config/i386/i386-builtin-types.def
1760 (V64QI_FTYPE_V64QI_V64QI): New type.
1761 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1762 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1763 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1764 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1765 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1766 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1768 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1770 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1771 explicitly as a stream of bytes.
1773 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1775 * config/rs6000/altivec.h (vec_xst_be): New #define.
1776 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1777 and externalize from *altivec_vperm_<mode>_internal.
1778 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1780 (XL_BE_V8HI): Likewise.
1781 (XL_BE_V4SI): Likewise.
1782 (XL_BE_V4SI): Likewise.
1783 (XL_BE_V2DI): Likewise.
1784 (XL_BE_V4SF): Likewise.
1785 (XL_BE_V2DF): Likewise.
1786 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1787 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1788 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1789 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1790 VSX_BUILTIN_VEC_XST_BE.
1791 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1792 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1794 (altivec_init_builtins): Replace conditional calls to def_builtin
1795 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1796 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1797 with unconditional calls. Remove calls to def_builtin for
1798 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1799 __builtin_vec_xst_be.
1800 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1801 to define_expand, and add alternate RTL generation for P8.
1802 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1803 vsx_ld_elemrev_v8hi.
1804 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1805 add alternate RTL generation for P8.
1806 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1807 vsx_ld_elemrev_v16qi.
1808 (vsx_st_elemrev_v8hi): Convert define_insn
1809 to define_expand, and add alternate RTL generation for P8.
1810 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1811 vsx_st_elemrev_v8hi.
1812 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1813 add alternate RTL generation for P8.
1814 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1815 vsx_st_elemrev_v16qi.
1817 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1820 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1821 TARGET_AVX512ER check.
1822 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1823 neither -mzeroupper nor -mno-zeroupper is used and
1824 TARGET_EMIT_VZEROUPPER is set.
1825 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1826 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1828 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1831 folding of vector compares.
1832 (fold_build_vec_cmp): New helper function.
1833 (fold_compare_helper): New helper function.
1834 (builtin_function_type): Add compare builtins to the list of functions
1835 having unsigned arguments. Cosmetic updates to comment indentation.
1836 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1837 the not+eq combination.
1839 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1841 PR tree-optimization/82726
1842 PR tree-optimization/70754
1843 * tree-predcom.c (order_drefs_by_pos): New function.
1844 (combine_chains): Move code setting has_max_use_after to...
1845 (try_combine_chains): ...here. New parameter. Sort combined chains
1846 according to position information.
1847 (tree_predictive_commoning_loop): Update call to above function.
1848 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1850 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1852 PR tree-optimization/82726
1854 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1856 PR tree-optimization/70754
1857 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1858 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1859 combined stmt before it if not NULL.
1860 (combine_chains): Process refs reversely and compute dominance point
1864 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1866 PR tree-optimization/79663
1867 * tree-predcom.c (combine_chains): Process refs in reverse order
1868 only for ZERO length chains, and add explaining comment.
1870 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1872 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1873 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1874 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1876 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1878 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1880 2017-11-15 Martin Liska <mliska@suse.cz>
1882 * tree-cfg.c (pass_warn_function_return::execute):
1883 Compare warn_return_type for greater than zero.
1885 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1889 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1890 to return true on Xeon and not on Xeon Phi.
1891 (ix86_check_avx256_register): Changed to ...
1892 (ix86_check_avx_upper_register): ... this. Add extra check for
1893 VALID_AVX512F_REG_OR_XI_MODE.
1894 (ix86_avx_u128_mode_needed): Changed
1895 ix86_check_avx256_register to ix86_check_avx_upper_register.
1896 (ix86_check_avx256_stores): Changed to ...
1897 (ix86_check_avx_upper_stores): ... this. Changed
1898 ix86_check_avx256_register to ix86_check_avx_upper_register.
1899 (ix86_avx_u128_mode_after): Changed
1900 avx_reg256_found to avx_upper_reg_found. Changed
1901 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1902 (ix86_avx_u128_mode_entry): Changed
1903 ix86_check_avx256_register to ix86_check_avx_upper_register.
1904 (ix86_avx_u128_mode_exit): Ditto.
1905 * config/i386/i386.h: (host_detect_local_cpu): New define.
1907 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1909 * config/arm/xgene1.md (xgene1): Split into automatons
1910 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1911 (xgene1_f_load): Adjust reservations and/or types.
1912 (xgene1_f_store): Likewise.
1913 (xgene1_load_pair): Likewise.
1914 (xgene1_store_pair): Likewise.
1915 (xgene1_fp_load1): Likewise.
1916 (xgene1_load1): Likewise.
1917 (xgene1_store1): Likewise.
1918 (xgene1_move): Likewise.
1919 (xgene1_alu): Likewise.
1920 (xgene1_simd): Likewise.
1921 (xgene1_bfm): Likewise.
1922 (xgene1_neon_load1): Likewise.
1923 (xgene1_neon_store1): Likewise.
1924 (xgene1_neon_logic): Likewise.
1925 (xgene1_neon_st1): Likewise.
1926 (xgene1_neon_ld1r): Likewise.
1927 (xgene1_alu_cond): Added.
1928 (xgene1_shift_reg): Likwise.
1929 (xgene1_bfx): Likewise.
1930 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1932 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1935 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1937 (can_widen_mult_without_libcall): New function.
1938 (expand_mul_overflow): If only checking unsigned mul overflow,
1939 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1940 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1941 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1942 (expand_DIVMOD): Formatting fix.
1943 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1944 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1945 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1947 PR tree-optimization/82977
1948 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1949 constructed temporary to strlen_to_stridx.put.
1951 2017-11-15 Martin Liska <mliska@suse.cz>
1953 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1954 * configure: Regenerate.
1956 2017-11-15 Martin Liska <mliska@suse.cz>
1959 * config/sh/sh-mem.cc: Use proper probability for
1962 2017-11-14 Jeff Law <law@redhat.com>
1964 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1965 the red zone for stack_clash_protection_final_dynamic_probe targets
1966 when the total dynamic stack size is zero bytes.
1968 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1969 blocks is post order.
1971 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1973 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1974 * final.c (rest_of_clean_state): Set it for the
1975 -fcompare-debug dump.
1976 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1977 class when TDF_COMPARE_DEBUG is set.
1979 * dwarf2out.c (gen_producer_string): Discard
1982 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1985 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1986 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1987 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1988 (__TGMATH_REAL_2_3): Remove macros.
1989 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1990 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1992 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1994 (remquo): Define using __TGMATH_REAL_3.
1996 2017-11-14 Jeff Law <law@redhat.com>
1998 * vr-values.c: New file with contents extracted from tree-vrp.c.
1999 * Makefile.in (OBJS): Add vr-values.o
2000 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2001 (set_value_range, set_and_canonicalize_value_range): Likewise.
2002 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2003 (value_range_constant_singleton, symbolic_range_p): Likewise.
2004 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2005 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2006 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2007 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2008 (find_case_label_range, find_case_label_index): Likewise.
2009 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2010 (range_int_cst_singleton_p, value_inside_range): Likewise.
2011 (get_single_symbol): Likewise.
2012 (switch_update): Move structure definition here.
2013 (to_remove_edges, to_update_switch_stmts): Provide externs.
2014 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2015 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2016 (vrp_val_is_min, set_value_range): Likewise.
2017 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2018 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2019 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2020 (range_is_nonnull, range_int_cst_p): Likewwise.
2021 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2022 (get_single_symbol, operand_less_p): Likewise
2023 (compare_values_warnv, compare_values): Likewise.
2024 (value_inside_range, value_range_constant_singleton): Likewise.
2025 (zero_nonzero_bitgs_from_vr): Likewise.
2026 (extract_range_from_binary_expr_1): Likewise.
2027 (overflow_comparison_p): Likewise.
2028 (to_remove_edges, to_update_switch_stmts): Likewise.
2029 (find_case_label-index, find_case_label_range): Likewise.
2030 (switch_update, set_value_range_to_nonnegative): Remove.
2031 (set_value_range_to_truthvalue): Likewise.
2032 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2033 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2034 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2035 (find_case_label_ranges, test_for_singularity): Likewise.
2036 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2037 (x_vr_values): Move to its remaining use site.
2039 2017-11-10 Jeff Law <law@redhat.com>
2041 * vr-values.h (VR_INITIALIZER): Move #define here.
2042 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2043 * Makefile.in (OBJS): Add tree-evrp.o
2044 * tree-vrp.h (assert_info): Move structure definition here.
2045 (set_value_range_to_varying): Prototype.
2046 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2047 (infer_value_range, register_edge_assert_for): Likewise.
2048 (stmt_interesting_for_vrp): Likewise.
2049 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2050 (set_value_range_to_varying): No longer static.
2051 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2052 (infer_value_range, register_edge_assert_for): Likewise.
2054 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2056 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2057 generate the XXBRD instruction.
2059 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2060 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2061 as being compatible if -mabi=ieeelongdouble.
2062 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2063 to setup float128 built-ins with hardware support.
2064 (BU_FLOAT128_HW_2): Likewise.
2065 (BU_FLOAT128_HW_3): Likewise.
2066 (BU_FLOAT128_HW_VSX_1): Likewise.
2067 (BU_FLOAT128_HW_VSX_2): Likewise.
2068 (scalar_extract_expq): Change float128 built-in functions to
2069 accommodate having both KFmode and TFmode functions. Use the
2070 KFmode variant as the default.
2071 (scalar_extract_sigq): Likewise.
2072 (scalar_test_neg_qp): Likewise.
2073 (scalar_insert_exp_q): Likewise.
2074 (scalar_insert_exp_qp): Likewise.
2075 (scalar_test_data_class_qp): Likewise.
2076 (sqrtf128_round_to_odd): Delete processing the round to odd
2077 built-in functions as special built-in functions, and define them
2078 as float128 built-ins. Use the KFmode variant as the default.
2079 (truncf128_round_to_odd): Likewise.
2080 (addf128_round_to_odd): Likewise.
2081 (subf128_round_to_odd): Likewise.
2082 (mulf128_round_to_odd): Likewise.
2083 (divf128_round_to_odd): Likewise.
2084 (fmaf128_round_to_odd): Likewise.
2085 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2086 support for KFmode and TFmode xststdcqp calls.
2087 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2088 point, switch the built-in handlers for the get/set float128
2089 exponent, get float128 mantissa, float128 test built-ins, and the
2090 float128 round to odd built-in functions. Eliminate creating the
2091 float128 round to odd built-in functions as special built-ins.
2092 (rs6000_init_builtins): Eliminate special creation of the float128
2093 round to odd built-in functions.
2094 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2095 function insns to support both TFmode and KFmode variants.
2096 (xsxsigqp_<mode>): Likewise.
2097 (xsiexpqpf_<mode>): Likewise.
2098 (xsiexpqp_<mode>): Likewise.
2099 (xststdcqp_<mode>): Likewise.
2100 (xststdcnegqp_<mode>): Likewise.
2101 (xststdcqp_<mode>): Likewise.
2103 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2105 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2106 unused path_in_freq_ptr parameter.
2107 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2109 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2111 * ipa-inline.c (edge_badness): Dump sreal frequency.
2112 (compute_inlined_call_time): Match natural implementaiton ...
2113 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2114 forgotten division by CGRAPH_FREQ_BASE.
2116 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2118 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2119 Solaris 11. Update comment.
2120 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2122 * config/sol2.h (STARTFILE_SPEC): Likewise.
2123 * configure: Regenerate.
2125 2017-11-14 Carl Love <cel@us.ibm.com>
2127 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2128 le_ and be_ prefixes to swap* variables. Remove
2129 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2132 2017-11-14 Jason Merrill <jason@redhat.com>
2134 Support GTY((cache)) on hash_map.
2135 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2136 (ggc_cache_remove): Override it instead of ggc_mx.
2137 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2138 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2139 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2140 (simple_cache_map_traits): Override maybe_mx.
2141 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2142 (hash_map): Friend gt_cleare_cache.
2143 (gt_cleare_cache): New.
2144 * tree.h (tree_cache_traits): New hash_map traits class.
2145 (tree_cache_map): New typedef.
2147 2017-11-14 Richard Biener <rguenther@suse.de>
2149 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2150 paramter and handling.
2151 (cleanup_control_flow_bb): Likewise.
2152 (cleanup_control_flow_pre): New helper performing a DFS walk
2153 to call cleanup_control_flow_bb in PRE order.
2154 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2155 via cleanup_control_flow_pre.
2157 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2159 * config/aarch64/aarch64-simd.md
2160 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2161 (*aarch64_simd_bsl<mode>_alt): Likewise.
2162 (aarch64_simd_bsldi_internal): New.
2163 (aarch64_simd_bsldi_alt): Likewise.
2165 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2167 * tracer.c (better_p): Do not compare frequencies.
2168 * reg-stack.c (better_edge): Likewise.
2169 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2172 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2174 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2175 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2176 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2177 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2178 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2179 * predict.c (maybe_hot_frequency_p): Inline to ...
2180 (maybe_hot_count_p): ... here; rewrite to counts.
2181 (counts_to_freqs): Rename to ...
2182 (update_max_bb_count): ... this one.
2183 (expensive_function_p): Use counts.
2184 (estimate_bb_frequencies): Update.
2185 (rebuild_frequencies): Update.
2186 * predict.h (counts_to_freqs): Rename to ...
2187 (update_max_bb_count): ... this one.
2188 * profile.c (compute_branch_probabilities): Add debug info
2189 * tree-inline.c (expand_call_inline): Update debug info.
2190 (optimize_inline_calls): Use update_max_bb_count..
2191 (tree_function_versioning): Use update_max_bb_count..
2192 * value-prof.c (gimple_value_profile_transformations):
2193 Do not use update_max_bb_count.
2195 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2197 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2198 always use frequencies.
2200 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2202 * bb-reorder.c: Remove frequencies from comments.
2203 (better_edge_p): Use profile counts.
2204 (find_traces): Dump profile counts.
2205 (rotate_loop): Use profile counts.
2206 (find_traces_1_round): Likewise.
2207 (connect_better_edge_p): Use counts instead of probabilities for
2209 (copy_bb_p): Drop early check for non-0 frequency.
2210 (sanitize_hot_paths): Update comments.
2212 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2214 * ipa-split.c (struct split_point): Add count.
2215 (consider_split): Do not compute incoming frequency; compute incoming
2216 count and store it to split_point.
2217 (split_function): Set count of the call to split part correctly.
2219 2017-11-13 Carl Love <cel@us.ibm.com>
2221 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2223 2017-11-13 Tom Tromey <tom@tromey.com>
2225 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2227 2017-11-13 Carl Love <cel@us.ibm.com>
2229 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2230 Add support for builtins:
2231 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2232 vector {un,}signed {char,int,short},
2233 vector {un,}signed {char,int,short}) arguments.
2234 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2235 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2236 Add BU_P9V_AV_2 expansions for the builtins.
2237 * config/rs6000/altivec.h (vec_first_match_index,
2238 vec_first_mismatch_index, vec_first_match_or_eos_index,
2239 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2240 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2241 new extern declaration.
2242 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2243 * config/rs6000/vsx.md (first_match_index_<mode>,
2244 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2245 first_mismatch_or_eos_index_<mode>): Add define expand.
2246 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2247 * doc/extend.texi: Update the built-in documenation file for the new
2250 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2252 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2253 into the min/max operations for _Float<N> and _Float<N>X types.
2255 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2258 * dwarf2out.c (do_eh_frame): New static variable.
2259 (dwarf2out_begin_prologue): Set it.
2260 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2262 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2264 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2266 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2268 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2271 * debug.h (dwarf2out_do_eh_frame): Declare.
2272 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2273 (dwarf2out_do_frame): Use it.
2274 (dwarf2out_do_cfi_asm): Likewise.
2275 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2276 (dwarf2out_assembly_start): Likewise.
2277 (dwarf2out_begin_prologue): Fix comment.
2278 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2279 if the target needs either debug or unwind DWARF2 info.
2280 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2281 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2282 -ftrapping-math, -ftrapv and -fwrapv.
2284 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2286 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2287 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2288 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2289 (estimate_edge_size_and_time): Likewise.
2290 (ipa_merge_fn_summary_after_inlining): Likewise.
2291 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2292 (compute_uninlined_call_time): Use sreal_frequency.
2293 (compute_inlined_call_time): Likewise.
2294 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2295 * profile-count.c: Include sreal.h.
2296 (profile_count::to_sreal_scale): New.
2297 * profile-count.h: Forward declare sreal.
2298 (profile_count::to_sreal_scale): Declare.
2300 2017-11-13 Nathan Sidwell <nathan@acm.org>
2302 * diagnostic.c (maybe_line_and_column): New.
2303 (diagnostic_get_location_text): Use it.
2304 (diagnostic_report_current_module): Likewise.
2305 (test_diagnostic_get_location_text): Add tests.
2307 2017-11-13 Luis Machado <luis.machado@linaro.org>
2309 * doc/md.texi (Specifying processor pipeline description): Fix
2310 incorrect latency for the div instruction example.
2312 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2314 PR tree-optimization/78821
2315 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2316 that bit_not_p is the same.
2317 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2318 (split_group): Count precisely bit_not_p bits in each statement.
2319 (invert_op): New function.
2320 (imm_store_chain_info::output_merged_store): Use invert_op to
2321 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2322 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2324 2017-11-13 Martin Liska <mliska@suse.cz>
2326 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2327 (struct source_info): Likewise.
2328 (add_branch_counts): Likewise.
2329 (add_line_counts): Likewise.
2330 (function_summary): Likewise.
2331 (output_intermediate_line): Likewise.
2332 (generate_results): Likewise.
2334 2017-11-13 Martin Liska <mliska@suse.cz>
2336 * gcov.c (struct block_info): Remove typedef for block_t.
2337 (struct line_info): Likewise.
2338 (line_info::has_block): Likewise.
2339 (EXIT_BLOCK): Likewise.
2340 (unblock): Likewise.
2341 (circuit): Likewise.
2342 (get_cycles_count): Likewise.
2343 (process_file): Likewise.
2344 (read_graph_file): Likewise.
2345 (solve_flow_graph): Likewise.
2346 (find_exception_blocks): Likewise.
2347 (add_line_counts): Likewise.
2348 (accumulate_line_info): Likewise.
2349 (output_line_details): Likewise.
2351 2017-11-13 Martin Liska <mliska@suse.cz>
2353 * gcov.c (struct arc_info): Remove typedef for arc_t.
2354 (struct line_info): Likewise.
2355 (add_branch_counts): Likewise.
2356 (output_branch_count): Likewise.
2357 (function_info::~function_info): Likewise.
2358 (circuit): Likewise.
2359 (output_intermediate_line): Likewise.
2360 (read_graph_file): Likewise.
2361 (solve_flow_graph): Likewise.
2362 (find_exception_blocks): Likewise.
2363 (add_line_counts): Likewise.
2364 (accumulate_line_info): Likewise.
2365 (output_line_details): Likewise.
2366 (output_function_details): Likewise.
2368 2017-11-13 Martin Liska <mliska@suse.cz>
2370 * gcov.c (struct function_info): Remove typedef for function_t.
2371 (struct source_info): Likewise.
2372 (source_info::get_functions_at_location): Likewise.
2373 (solve_flow_graph): Likewise.
2374 (find_exception_blocks): Likewise.
2375 (add_line_counts): Likewise.
2376 (output_intermediate_file): Likewise.
2377 (process_file): Likewise.
2378 (generate_results): Likewise.
2379 (release_structures): Likewise.
2380 (read_graph_file): Likewise.
2381 (read_count_file): Likewise.
2382 (accumulate_line_counts): Likewise.
2383 (output_lines): Likewise.
2385 2017-11-13 Martin Liska <mliska@suse.cz>
2387 * gcov.c (function_info::function_info): Remove num_counts
2388 and add vector<gcov_type>.
2389 (function_info::~function_info): Use the vector.
2390 (process_file): Likewise.
2391 (read_graph_file): Likewise.
2392 (read_count_file): Likewise.
2393 (solve_flow_graph): Likewise.
2395 2017-11-13 Martin Liska <mliska@suse.cz>
2397 * gcov.c (function_info::is_artificial): New function.
2398 (process_file): Erase all artificial early.
2399 (generate_results): Skip as all artificial are already
2402 2017-11-13 Martin Liska <mliska@suse.cz>
2404 * gcov.c (read_graph_file): Store to global vector of functions.
2405 (read_count_file): Iterate the vector.
2406 (process_file): Likewise.
2407 (generate_results): Likewise.
2408 (release_structures): Likewise.
2410 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2412 PR tree-optimization/82954
2413 * gimple-ssa-store-merging.c
2414 (imm_store_chain_info::coalesce_immediate_stores): If
2415 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2417 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2419 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2420 Upddate call to ENDIAN_LANE_N.
2421 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2422 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2423 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2424 and use aarch64_endian_lane_rtx.
2425 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2427 2017-11-12 Tom de Vries <tom@codesourcery.com>
2429 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2431 2017-11-12 Tom de Vries <tom@codesourcery.com>
2433 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2435 2017-11-12 Tom de Vries <tom@codesourcery.com>
2437 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2439 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2440 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2441 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2443 2017-11-11 Martin Sebor <msebor@redhat.com>
2446 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2449 * prefic.c (translate_name): Replace strncpy with memcpy to
2450 avoid -Wstringop-truncation.
2452 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2454 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2456 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2458 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2460 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2462 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2465 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2467 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2468 enable generating XXBRH if the value is in a vector register.
2469 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2470 value is in a vector register.
2471 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2472 register to register bswap64's instead of doing the GPR sequence
2473 used on previous machines.
2474 (bswapdi2_xxbrd): New insn.
2475 (bswapdi2_reg): Disallow on ISA 3.0.
2476 (register to register bswap64 splitter): Do not split the insn on
2477 ISA 3.0 systems that use XXBRD.
2479 2017-11-10 Martin Sebor <msebor@redhat.com>
2482 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2484 (find_subframework_file): Same.
2486 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2488 * auto-profile.c (afdo_indirect_call): Drop frequency.
2489 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2490 (cgraph_node::create_edge): Drop frequency argument.
2491 (cgraph_node::create_indirect_edge): Drop frequency argument.
2492 (cgraph_edge::make_speculative): Drop frequency arguments.
2493 (cgraph_edge::resolve_speculation): Do not update frequencies
2494 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2495 (cgraph_node::dump): Check consistency in IPA mode.
2496 (cgraph_edge::maybe_hot_p): Use IPA counter.
2497 (cgraph_edge::verify_count_and_frequency): Rename to ...
2498 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2499 (cgraph_node::verify_node): Update.
2500 * cgraph.h (struct cgraph_edge): Drop frequency.
2501 (cgraph_edge::frequency): New function.
2502 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2504 (cgraph_edge::rebuild_edges): Likewise.
2505 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2506 (duplicate_thunk_for_node): Do not pass frequency.
2507 (cgraph_node::create_clone): Scale only counts.
2508 (cgraph_node::create_virtual_clone): Do not pass frequency.
2509 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2510 (cgraph_node::create_version_clone): Do not pass frequency.
2511 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2512 (cgraph_node::expand_thunk): Do not pass frequency.
2513 (cgraph_node::create_wrapper): Do not pass frequency.
2514 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2516 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2517 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2518 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2519 (gather_caller_stats): Use frequency function.
2520 (ipcp_cloning_candidate_p): Use frequency function.
2521 (ipcp_propagate_stage): Use frequency function.
2522 (get_info_about_necessary_edges): Use frequency function.
2523 (update_profiling_info): Update only IPA profile.
2524 (update_specialized_profile): Use frequency functoin.
2525 (perhaps_add_new_callers): Update only IPA profile.
2526 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2527 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2528 (dump_ipa_call_summary): Use frequency function.
2529 (estimate_edge_size_and_time): Use frequency function.
2530 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2531 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2532 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2533 (update_noncloned_counts): ... ths one; scale counts only.
2534 (clone_inlined_nodes): Do not scale frequency.
2535 (inline_call): Do not pass frequency.
2536 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2537 (compute_inlined_call_time): Use IPA profile.
2538 (want_inline_small_function_p): Use IPA profile.
2539 (want_inline_self_recursive_call_p): Use IPA profile.
2540 (edge_badness): Use IPA profile.
2541 (lookup_recursive_calls): Use IPA profile.
2542 (recursive_inlining): Do not pass frequency.
2543 (resolve_noninline_speculation): Do not update frequency.
2544 (inline_small_functions): Collect max of IPA profile.
2545 (dump_overall_stats): Dump IPA porfile.
2546 (dump_inline_stats): Dump IPA porfile.
2547 (ipa_inline): Collect IPA stats.
2548 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2549 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2550 (ipa_propagate_frequency): Use frequency function.
2551 (ipa_profile): Cleanup.
2552 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2553 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2554 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2555 (input_node): Do not stream frequency.
2556 (input_edge): Do not stream frequency.
2557 (merge_profile_summaries): Scale only IPA profiles.
2558 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2559 * predict.c (drop_profile): Do not recompute frequency.
2560 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2561 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2562 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2563 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2564 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2565 * tree-inline.c (copy_bb): Do not scale frequency.
2566 (expand_call_inline): Do not scale frequency.
2567 (tree_function_versioning): Do not scale frequency.
2568 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2570 2017-11-10 Julia Koval <julia.koval@intel.com>
2572 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2573 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2574 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2575 (_mm256_maskz_gf2p8affine_epi64_epi8)
2576 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2577 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2578 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2579 (__builtin_ia32_vgf2p8affineqb_v32qi)
2580 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2581 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2583 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2587 (arm_option_override): Refactor.
2588 (arm_option_reconfigure_globals): New.
2589 (arm_options_perform_arch_sanity_checks): New.
2590 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2592 (arm_options_perform_arch_sanity_checks): Likewise
2594 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2596 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2597 (power9-qpmul): New.
2598 * rs6000/rs6000.md ("type" attr): Add qmul.
2599 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2600 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2601 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2603 2017-11-10 Martin Sebor <msebor@redhat.com>
2606 * builtins.c (compute_objsize): Handle arrays that
2607 compute_builtin_object_size likes to fail for. Make extern.
2608 * builtins.h (compute_objsize): Declare.
2609 (check_strncpy_sizes): New function.
2610 (expand_builtin_strncpy): Call check_strncpy_sizes.
2611 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2612 -Wstringop-truncation.
2613 (gimple_fold_builtin_strncat): Same.
2614 * gimple.c (gimple_build_call_from_tree): Set call location.
2615 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2616 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2617 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2618 (handle_builtin_strlen): Use strlen_to_stridx.
2619 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2621 Use strlen_to_stridx.
2622 (pass_strlen::execute): Release strlen_to_stridx.
2623 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2624 (-Wstringop-truncation): Document new option.
2626 2017-11-10 Martin Liska <mliska@suse.cz>
2628 PR gcov-profile/82702
2629 * gcov.c (main): Handle intermediate files in a different
2631 (get_gcov_intermediate_filename): New function.
2632 (output_gcov_file): Remove support of intermediate files.
2633 (generate_results): Allocate intermediate file.
2634 (release_structures): Clean-up properly fn_end.
2635 (output_intermediate_file): Start iterating with line 1.
2637 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2639 PR tree-optimization/82929
2640 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2641 ops_swapped_p non-static data member.
2642 (store_immediate_info::store_immediate_info): Clear it.
2643 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2644 ops set ops_swapped_p.
2645 (count_multiple_uses): Handle ops_swapped_p.
2647 2017-11-10 Martin Liska <mliska@suse.cz>
2649 * coverage.c (coverage_init): Stream information about
2650 support of has_unexecuted_blocks.
2651 * doc/gcov.texi: Document that.
2652 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2653 * gcov.c (read_graph_file): Likewise.
2654 (output_line_beginning): Fix a small issue with
2657 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2659 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2660 reference of trivial component.
2662 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2665 * gimple-ssa-store-merging.c
2666 (pass_store_merging::terminate_all_aliasing_chains): For
2667 gimple_store_p stmts also call refs_output_dependent_p.
2669 PR rtl-optimization/82913
2670 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2673 2017-11-09 Jeff Law <law@redhat.com>
2675 * vr-values.h: New file with vr_values class.
2676 * tree-vrp.c: Include vr-values.h
2677 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2678 data objects into the vr_values class.
2679 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2680 (get_value_range): Make it a member function within vr_values class.
2681 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2682 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2683 (op_with_constant_singleton_value_range): Likewise.
2684 (extract_range_for_var_from_comparison_expr): Likewise.
2685 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2686 (extract_range_from_binary_expr): Likewise.
2687 (extract_range_from_unary_expr): Likewise.
2688 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2689 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2690 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2691 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2692 (compare_name_with_value, compare_names): Likewise.
2693 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2694 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2695 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2696 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2697 (extract_range_from_phi_node): Likewise.
2698 (simplify_truth_ops_using_ranges): Likewise.
2699 (simplify_div_or_mod_using_ranges): Likewise.
2700 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2701 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2702 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2703 (simplify_float_conversion_using_ranges): Likewise.
2704 (simplify_internal_call_using_ranges): Likewise.
2705 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2706 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2707 poitner via x_vr_values for calls into gimple folder.
2708 (vrp_initialize_lattice): Make this the vr_values ctor.
2709 (vrp_free_lattice): Make this the vr_values dtor.
2710 (set_vr_value): New function.
2711 (class vrp_prop): Add vr_values data member. Add various member
2712 functions as well as member functions that delegate to vr_values.
2713 (check_array_ref): Make a member function within vrp_prop class.
2714 (search_for_addr_array, vrp_initialize): Likewise.
2715 (vrp_finalize): Likewise. Revamp to avoid direct access to
2716 vr_value, values_propagated, etc.
2717 (check_array_bounds): Extract vrp_prop class instance pointer from
2718 walk info structure. Use it to call member functions.
2719 (check_all_array_refs): Make a member function within vrp_prop class.
2720 Smuggle class instance pointer via walk info structure.
2721 (x_vr_values): New local static.
2722 (vrp_valueize): Use x_vr_values to get class instance.
2723 (vr_valueize_1): Likewise.
2724 (class vrp_folder): Add vr_values data member. Add various member
2725 functions as well as member functions that delegate to vr_values.
2726 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2727 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2728 class instance from vr_values. Use it to call member functions.
2729 (vrp_dom_walker): Add vr_values data member.
2730 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2731 instance via x_vr_values.
2732 (identify_jump_threads): Accept vr_values as argument. Store
2733 it into the walker structure.
2734 (evrp_dom_walker): Add vr_values class data member. Add various
2736 (evrp_dom_walker::try_find_new_range): Use vr_values data
2737 member to access the memory allocator.
2738 (evrp_dom_walker::before_dom_children): Store vr_values class
2739 instance into the vrp_folder class.
2740 (evrp_dom_walker::push_value_range): Rework to avoid direct
2741 access to num_vr_values and vr_value.
2742 (evrp_dom_walker::pop_value_range): Likewise.
2743 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2744 Use vr_values to get to dump_all_value_ranges member function.
2745 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2746 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2747 Pass vr_values class instance down to identify_jump_threads.
2748 Remove call to vrp_free_lattice.
2749 (debug_all_value_ranges): Remove.
2751 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2752 (vrp_folder): Likewise.
2754 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2755 Get it from the existing bitmap instead.
2756 (vrp_intersect_ranges_1): Likewise.
2758 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2760 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2762 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2763 set bit_not_p to it.
2764 (imm_store_chain_info::coalesce_immediate_stores): Break group
2765 if bit_not_p is different.
2766 (count_multiple_uses, split_group,
2767 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2768 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2769 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2770 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2771 to store_immediate_info ctor.
2773 2017-11-09 Jim Wilson <jimw@sifive.com>
2775 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2776 (scan_prog_file): Likewise.
2778 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2780 * bb-reorder.c (max_entry_frequency): Remove.
2781 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2782 connect_traces, push_to_next_round_p): Remove prototypes.
2783 (find_traces_1_round): Use counts only.
2784 (push_to_next_round_p): Likewise.
2785 (find_traces): Likewise.
2786 (rotate_loop): Likewise.
2787 (find_traces_1_round): Likewise.
2788 (connect_traces): Likewise.
2789 (edge_order): Likewise.
2791 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2793 * config/arm/arm.c (output_return_instruction): Add comments to
2794 indicate requirement for cmse_nonsecure_entry return to account
2795 for the size of clearing instruction output here.
2796 (thumb_exit): Likewise.
2797 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2798 return in hardfloat mode.
2800 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2802 * config/rs6000/rs6000.c (machine_function): Add a bool,
2803 "toc_is_wrapped_separately".
2804 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2805 if it wasn't explicitly set or unset, we are optimizing for speed, and
2806 doing separate shrink-wrapping.
2807 (rs6000_get_separate_components): Enable the TOC component if
2808 saving the TOC register in the prologue.
2809 (rs6000_components_for_bb): Handle the TOC component.
2810 (rs6000_emit_prologue_components): Store the TOC register where needed.
2811 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2812 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2814 2017-11-09 Martin Jambor <mjambor@suse.cz>
2816 * ipa-param-manipulation.c: New file.
2817 * ipa-param-manipulation.h: Likewise.
2818 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2819 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2820 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2821 (ipa_parm_adjustment): Likewise.
2822 (ipa_parm_adjustment_vec): Likewise.
2823 (ipa_get_vector_of_formal_parms): Moved declaration to
2824 ipa-param-manipulation.h.
2825 (ipa_get_vector_of_formal_parm_types): Likewise.
2826 (ipa_modify_formal_parameters): Likewise.
2827 (ipa_modify_call_arguments): Likewise.
2828 (ipa_combine_adjustments): Likewise.
2829 (ipa_dump_param_adjustments): Likewise.
2830 (ipa_modify_expr): Likewise.
2831 (ipa_get_adjustment_candidate): Likewise.
2832 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2833 ipa-param-manipulation.c.
2834 (ipa_get_vector_of_formal_parm_types): Likewise.
2835 (ipa_modify_formal_parameters): Likewise.
2836 (ipa_modify_call_arguments): Likewise.
2837 (ipa_modify_expr): Likewise.
2838 (get_ssa_base_param): Likewise.
2839 (ipa_get_adjustment_candidate): Likewise.
2840 (index_in_adjustments_multiple_times_p): Likewise.
2841 (ipa_combine_adjustments): Likewise.
2842 (ipa_dump_param_adjustments): Likewise.
2843 * tree-sra.c: Also include ipa-param-manipulation.h
2844 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2847 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2848 Alan Hayward <alan.hayward@arm.com>
2849 David Sherwood <david.sherwood@arm.com>
2851 * doc/sourcebuild.texi (vect_masked_store): Document.
2853 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2854 Alan Hayward <alan.hayward@arm.com>
2855 David Sherwood <david.sherwood@arm.com>
2857 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2859 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2860 Alan Hayward <alan.hayward@arm.com>
2861 David Sherwood <david.sherwood@arm.com>
2863 * doc/sourcebuild.texi (vect_variable_length): Document.
2865 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2866 Alan Hayward <alan.hayward@arm.com>
2867 David Sherwood <david.sherwood@arm.com>
2869 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2871 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2872 Alan Hayward <alan.hayward@arm.com>
2873 David Sherwood <david.sherwood@arm.com>
2875 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2877 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2878 Alan Hayward <alan.hayward@arm.com>
2879 David Sherwood <david.sherwood@arm.com>
2881 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2882 previously undocumented selectors.
2883 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2885 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2887 * doc/rtl.texi (const_vector): Say that elements can be
2888 const_wide_ints too.
2889 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2890 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2891 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2892 * optabs.c (expand_vector_broadcast): Likewise.
2894 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2895 Alan Hayward <alan.hayward@arm.com>
2896 David Sherwood <david.sherwood@arm.com>
2898 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2899 scaled index even if the unscaled address was invalid.
2900 Don't increase the complexity of using a scale in that case.
2902 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2903 Alan Hayward <alan.hayward@arm.com>
2904 David Sherwood <david.sherwood@arm.com>
2906 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2907 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2908 UNITS_PER_WORD bytes.
2909 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2910 based on REGMODE_NATURAL_SIZE of the inner mode.
2911 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2912 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2913 * expmed.c (lowpart_bit_field_p): Divide the value up into
2914 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2915 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2916 whether something is likely to occupy more than one register.
2918 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2921 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2922 profile_count::adjust_for_ipa_scaling.
2923 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2924 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2926 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2928 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2930 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2931 (split_group): Add total_orig and total_new arguments, estimate the
2932 number of statements related to the store group without store merging
2933 and with store merging.
2934 (imm_store_chain_info::output_merged_store): Adjust split_group
2935 callers, punt if estimated number of statements with store merging
2936 is not smaller than estimated number of statements without it.
2938 (handled_load): Remove has_single_use checks.
2939 (pass_store_merging::process_store): Likewise.
2941 2017-11-09 Richard Biener <rguenther@suse.de>
2943 PR tree-optimization/82902
2944 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2946 2017-11-09 Martin Liska <mliska@suse.cz>
2949 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2952 2017-11-09 Martin Liska <mliska@suse.cz>
2954 PR tree-optimization/82669
2955 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2957 2017-11-09 Martin Liska <mliska@suse.cz>
2959 PR gcov-profile/48463
2960 * coverage.c (coverage_begin_function): Output also end locus
2961 of a function and information whether the function is
2963 * gcov-dump.c (tag_function): Parse and print the information.
2964 * gcov.c (INCLUDE_MAP): Add include.
2965 (INCLUDE_SET): Likewise.
2966 (struct line_info): Move earlier in the source file because
2967 of vector<line_info> in function_info structure.
2968 (line_info::line_info): Likewise.
2969 (line_info::has_block): Likewise.
2970 (struct source_info): Add new member index.
2971 (source_info::get_functions_at_location): New function.
2972 (function_info::group_line_p): New function.
2973 (output_intermediate_line): New function.
2974 (output_intermediate_file): Use the mentioned function.
2975 (struct function_start): New.
2976 (struct function_start_pair_hash): Likewise.
2977 (process_file): Add code that identifies group functions.
2978 Assign lines either to global or function scope.
2979 (generate_results): Skip artificial functions.
2980 (find_source): Assign index for each source file.
2981 (read_graph_file): Read new flag artificial and end_line.
2982 (add_line_counts): Assign it either to global of function scope.
2983 (accumulate_line_counts): Isolate core of the function to
2984 accumulate_line_info and call it for both function and global
2986 (accumulate_line_info): New function.
2987 (output_line_beginning): Fix GNU coding style.
2988 (print_source_line): New function.
2989 (output_line_details): Likewise.
2990 (output_function_details): Likewise.
2991 (output_lines): Iterate both source (global) scope and function
2993 (struct function_line_start_cmp): New class.
2994 * doc/gcov.texi: Reflect changes in documentation.
2996 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2999 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3000 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3001 and similarly for not instead of neg.
3003 2017-11-08 Andi Kleen <ak@linux.intel.com>
3005 * config/i386/i386.opt: Add -mforce-indirect-call.
3006 * config/i386/predicates.md: Check for flag_force_indirect_call.
3007 * doc/invoke.texi: Document -mforce-indirect-call
3009 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3011 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3013 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3014 config/riscv/riscv.c (predict.h): New include.
3015 (riscv_slow_unaligned_access_p): No longer static.
3016 (riscv_block_move_straight): Add require.
3017 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3019 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3022 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3023 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3024 nonimmediate_operand predicate for operand 1 instead of
3027 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3029 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3031 * config/aarch64/constraints.md (Uml): New constraint.
3032 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3035 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3037 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3038 of two vec_duplicates into a vec_concat.
3040 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3042 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3043 Simplify vec_merge of vec_duplicate and vec_concat.
3044 * config/aarch64/constraints.md (Utq): New constraint.
3045 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3048 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3050 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3051 Simplify vec_merge of vec_duplicate and const_vector.
3052 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3054 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3055 mode iterator. Update predicate on operand 1 to
3056 handle non-const_vec constants. Delete constraints.
3057 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3059 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3061 PR tree-optimization/78821
3062 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3064 (store_operand_info::store_operand_info): Initialize it to false.
3065 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3066 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3067 store in the group, and if chain_info is non-NULL, to ignore altogether
3069 (compatible_load_p): Fail if bit_not_p does not match.
3070 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3071 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3072 (pass_store_merging::process_store): Adjust
3073 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3074 call terminate_all_aliasing_chains newly when adding a store into
3075 a chain with non-NULL chain_info.
3077 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3079 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3081 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3083 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3085 (aarch64_layout_frame): Initialise emit_frame_chain.
3086 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3087 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3089 2017-11-08 Martin Liska <mliska@suse.cz>
3091 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3092 of gimple_call_internal_p.
3094 2017-11-07 Tom de Vries <tom@codesourcery.com>
3096 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3098 2017-11-07 Tom de Vries <tom@codesourcery.com>
3100 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3103 2017-11-08 Martin Liska <mliska@suse.cz>
3106 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3108 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3110 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3112 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3115 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3116 (*cmp<mode>_ccz_1): New insn with $k alternative.
3119 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3120 RTX_COMM_COMPARE as commutative as well.
3121 (ix86_binary_operator_ok): Formatting fix.
3122 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3123 *<code><mode>3<mask_name><round_saeonly_name>,
3124 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3125 *<s>mul<mode>3_highpart<mask_name>,
3126 *vec_widen_umult_even_v16si<mask_name>,
3127 *vec_widen_umult_even_v8si<mask_name>,
3128 *vec_widen_umult_even_v4si<mask_name>,
3129 *vec_widen_smult_even_v16si<mask_name>,
3130 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3131 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3132 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3133 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3134 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3135 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3136 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3137 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3138 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3139 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3140 ix86_binary_operator_ok. Formatting fixes.
3141 (*<plusminus_insn><mode>3<mask_name><round_name>,
3142 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3145 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3147 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3148 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3149 eq and ne if TARGET_ISEL.
3150 (cmp): New code_iterator.
3151 (UNS, UNSU_, UNSIK): New code_attrs.
3152 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3153 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3155 ("eq<mode>3"): ... this.
3156 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3158 ("ne<mode>3"): ... this.
3160 2017-11-07 Julia Koval <julia.koval@intel.com>
3163 * common/config/i386/i386-common.c
3164 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3165 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3166 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3167 * config/i386/i386.opt: Ditto.
3168 * config/i386/i386.c (ix86_target_string): Ditto.
3169 (ix86_option_override_internal): Ditto.
3170 (ix86_init_mpx_builtins): Move MPX to args2.
3171 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3172 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3173 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3174 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3175 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3176 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3177 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3179 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3182 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3183 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3184 (zero-extendsidi peephole2): Remove peephole.
3186 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3189 * stor-layout.c: Include attribs.h.
3190 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3191 explicit lookup of "aligned" attribute.
3193 2017-11-07 Andrew Waterman <andrew@sifive.com>
3195 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3196 (riscv_expand_block_move): Likewise.
3197 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3199 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3200 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3201 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3202 (riscv_adjust_block_mem): Likewise.
3203 (riscv_block_move_loop): Likewise.
3204 (riscv_expand_block_move): Likewise.
3205 gcc/config/riscv/riscv.md (movmemsi): New pattern.
3207 2017-11-07 Michael Clark <michaeljclark@mac.com>
3209 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3210 (MUSL_DYNAMIC_LINKER): Likewise.
3212 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3214 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3215 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3218 2017-11-07 Richard Biener <rguenther@suse.de>
3220 * match.pd: Fix build.
3222 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3223 Jackson Woodruff <jackson.woodruff@arm.com>
3225 PR tree-optimization/71026
3226 * match.pd: Canonicalize negate in division.
3228 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3231 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3233 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3235 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3236 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3238 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3240 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3241 non-scalar integral types.
3242 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3243 (-(A-B), -(~A)): New transformations.
3245 2017-11-07 Tom de Vries <tom@codesourcery.com>
3247 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3248 semicolon after "do {} while (0)".
3249 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3250 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3251 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3252 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3253 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3255 2017-11-07 Tom de Vries <tom@codesourcery.com>
3257 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3258 after "do {} while (0)".
3259 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3260 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3261 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3262 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3263 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3265 2017-11-07 Tom de Vries <tom@codesourcery.com>
3268 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3270 (arm_rtx_costs_internal): Add missing semicolon after
3271 HANDLE_NARROW_SHIFT_ARITH call.
3273 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3275 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3276 disable isel if it was not set explicitly.
3278 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3280 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3281 (add_type_attribute) likewise.
3283 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3285 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3286 of struct ix86_frame.
3287 (ix86_initial_elimination_offset): Likewise.
3288 (ix86_expand_split_stack_prologue): Likewise.
3290 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3292 * tree-vrp.h (enum value_range_type): Update stale comment.
3294 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3295 Alan Hayward <alan.hayward@arm.com>
3296 David Sherwood <david.sherwood@arm.com>
3298 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3299 (aarch64_expand_vec_perm_const): Take the number of units too.
3300 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3301 (aarch64_expand_vec_perm_const): Likewise.
3302 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3303 (vec_perm<mode>): Update accordingly.
3305 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3306 Alan Hayward <alan.hayward@arm.com>
3307 David Sherwood <david.sherwood@arm.com>
3309 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3310 Take the number of units too.
3311 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3312 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3313 but check for a vector mode before rather than after the call.
3314 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3315 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3316 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3317 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3318 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3319 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3320 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3321 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3322 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3323 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3324 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3325 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3326 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3327 (aarch64_sqdmull2_n<mode>): Update accordingly.
3329 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3330 Alan Hayward <alan.hayward@arm.com>
3331 David Sherwood <david.sherwood@arm.com>
3333 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3334 the number of units too.
3335 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3336 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3337 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3338 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3339 (vec_store_lanesxi<mode>): Update accordingly.
3341 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3342 Alan Hayward <alan.hayward@arm.com>
3343 David Sherwood <david.sherwood@arm.com>
3345 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3346 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3347 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3348 of units rather than the mode.
3349 * config/aarch64/iterators.md (nunits): New mode attribute.
3350 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3351 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3352 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3353 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3354 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3355 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3356 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3357 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3358 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3359 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3360 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3361 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3362 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3363 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3364 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3365 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3366 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3367 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3368 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3369 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3370 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3371 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3372 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3373 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3374 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3375 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3376 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3377 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3378 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3379 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3380 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3381 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3382 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3383 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3384 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3385 (aarch64_simd_vec_setv2di): Likewise.
3387 2017-11-06 Carl Love <cel@us.ibm.com>
3389 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3391 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3392 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3394 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3396 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3397 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3398 Add power 8 macro expansions.
3399 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3400 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3401 power 8 instructions. (VSX_XXBR): Add iterator.
3403 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3405 * config/arm/arm.md (predicable_short_it): Change default to "no",
3406 improve documentation, remove uses that are identical to the default.
3407 (enabled_for_depr_it): Rename to enabled_for_short_it.
3408 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3409 Remove default uses.
3410 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3411 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3412 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3413 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3415 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3418 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3419 float128 helper macros, which are no longer used after deleting
3420 the old 'q' built-in functions, and moving the round to odd
3421 built-in functions to being special built-in functions.
3422 (BU_FLOAT128_2): Likewise.
3423 (BU_FLOAT128_1_HW): Likewise.
3424 (BU_FLOAT128_2_HW): Likewise.
3425 (BU_FLOAT128_3_HW): Likewise.
3426 (FABSQ): Delete old 'q' built-in functions.
3427 (COPYSIGNQ): Likewise.
3428 (SQRTF128_ODD): Move round to odd built-in functions to be
3429 special built-in functions, so that we can handle
3430 -mabi=ieeelongdouble.
3431 (TRUNCF128_ODD): Likewise.
3432 (ADDF128_ODD): Likewise.
3433 (SUBF128_ODD): Likewise.
3434 (MULF128_ODD): Likewise.
3435 (DIVF128_ODD): Likewise.
3436 (FMAF128_ODD): Likewise.
3437 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3438 built-in names to 'f128'.
3439 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3440 old 'q' built-in functions, as the machine independent code for
3441 'f128' built-in functions handles this.
3442 (rs6000_expand_builtin): Add expansion for float128 round to odd
3443 functions, keying off on -mabi=ieeelongdouble of whether to use
3444 the KFmode or TFmode variant.
3445 (rs6000_init_builtins): Initialize the _Float128 round to odd
3447 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3448 _Float128 'q' built-in functions are now mapped into the new
3449 'f128' built-in functions.
3451 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3453 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3454 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3456 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3459 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3460 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3461 * toplev.c: Include "ipa-fnsummary.h".
3462 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3464 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3466 PR tree-optimization/82838
3467 * gimple-ssa-store-merging.c
3468 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3469 on a separate gimple_seq which is then appended to seq.
3471 2017-11-06 Jeff Law <law@redhat.com>
3474 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3475 (get_probe_interval): New functions.
3476 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3477 (ix86_adjust_stack_and_probe): Likewise.
3478 (output_adjust_stack_and_probe): Likewise.
3479 (ix86_emit_probe_stack_range): Likewise.
3480 (ix86_expand_prologue): Likewise.
3482 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3484 PR tree-optimization/82816
3485 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3486 if the modes of the two types are the same.
3487 (convert_plusminus_to_widen): Likewise.
3489 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3491 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3493 (usadv16qi): New define_expand.
3494 (usadv8hi): New define_expand.
3496 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3499 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3501 (inline_transform): Likewise.
3502 * predict.c (counts_to_freqs): Remove useless conditional.
3503 * profile-count.h (profile_count::apply_scale): Move sanity check.
3504 * tree-inline.c (copy_bb): Always scale.
3505 (copy_cfg_body): Likewise.
3507 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3510 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3512 (*cmp_ite0): Add enabled_for_depr_it attribute.
3513 (*cmp_ite1): Likewise.
3515 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3517 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3520 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3522 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3523 return the value by pointer.
3524 (extract_range_from_multiplicative_op_1): Update accordingly.
3525 Return as soon as an operation fails.
3527 2017-11-05 Tom de Vries <tom@codesourcery.com>
3530 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3531 (DEF_SANITIZER_BUILTIN): ... here.
3532 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3533 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3535 2017-11-05 Tom de Vries <tom@codesourcery.com>
3538 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3540 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3541 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3542 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3544 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3545 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3546 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3548 2017-11-05 Tom de Vries <tom@codesourcery.com>
3551 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3554 2017-11-04 Michael Clark <michaeljclark@mac.com>
3556 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3557 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3559 (*addsi3_extended): Likewise.
3560 (*addsi3_extended2): Likewise.
3561 (<optab>si3): Likewise.
3562 (<optab>di3): Likewise.
3563 (<optab><mode>3): Likewise.
3564 (<*optabe>si3_internal): Likewise.
3565 (zero_extendqi<SUPERQI:mode>2): Likewise.
3566 (*add<mode>hi3): Likewise.
3567 (*xor<mode>hi3): Likewise.
3568 (<optab>di3): Likewise.
3569 (*<optab>si3_extend): Likewise.
3570 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3571 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3572 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3574 2017-11-04 Andrew Waterman <andrew@sifive.com>
3576 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3577 TARGET_STRICT_ALIGN based upon -mtune argument.
3579 2017-11-04 Andrew Waterman <andrew@sifive.com>
3581 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3583 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3585 config/i386/i386.c (choose_basereg): Use optional scratch
3586 register and add assertion.
3587 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3588 needed, and don't allocate stack.
3589 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3590 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3591 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3593 2017-11-03 Jeff Law <law@redhat.com>
3595 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3596 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3597 to probe at the start of a noreturn function.
3599 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3601 PR tree-optimization/78821
3602 * gimple-ssa-store-merging.c: Update the file comment.
3603 (MAX_STORE_ALIAS_CHECKS): Define.
3604 (struct store_operand_info): New type.
3605 (store_operand_info::store_operand_info): New constructor.
3606 (struct store_immediate_info): Add rhs_code and ops data members.
3607 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3608 and op1r arguments to the ctor, initialize corresponding data members.
3609 (struct merged_store_group): Add load_align_base and load_align
3611 (merged_store_group::merged_store_group): Initialize them.
3612 (merged_store_group::do_merge): Update them.
3613 (merged_store_group::apply_stores): Pick the constant for
3614 encode_tree_to_bitpos from one of the two operands, or skip
3615 encode_tree_to_bitpos if neither operand is a constant.
3616 (class pass_store_merging): Add process_store method decl. Remove
3617 bool argument from terminate_all_aliasing_chains method decl.
3618 (pass_store_merging::terminate_all_aliasing_chains): Remove
3619 var_offset_p argument and corresponding handling.
3620 (stmts_may_clobber_ref_p): New function.
3621 (compatible_load_p): New function.
3622 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3623 if there is overlap and rhs_code is not INTEGER_CST. For
3624 non-overlapping stores terminate group if rhs is not mergeable.
3625 (get_alias_type_for_stmts): Change first argument from
3626 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3627 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3628 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3630 (get_location_for_stmts): Change first argument from
3631 auto_vec<gimple *> & to vec<gimple *> &.
3632 (struct split_store): Remove orig_stmts data member, add orig_stores.
3633 (split_store::split_store): Create orig_stores rather than orig_stmts.
3634 (find_constituent_stmts): Renamed to ...
3635 (find_constituent_stores): ... this. Change second argument from
3636 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3637 to info structures rather than the statements.
3638 (split_group): Rename ALLOW_UNALIGNED argument to
3639 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3640 it. Adjust find_constituent_stores caller.
3641 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3642 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3643 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3644 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3645 (mem_valid_for_store_merging): New function.
3646 (handled_load): New function.
3647 (pass_store_merging::process_store): New method.
3648 (pass_store_merging::execute): Use process_store method. Adjust
3649 terminate_all_aliasing_chains caller.
3651 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3653 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3654 Return true for more constants, symbols and label references.
3655 (aarch64_valid_floating_const): Remove unused function.
3657 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3660 * config/aarch64/aarch64.c (aarch64_layout_frame):
3661 Undo forcing of LR at bottom of frame.
3663 2017-11-03 Jeff Law <law@redhat.com>
3666 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3667 for int_registers_saved.
3669 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3670 extracted from tree-ssa-dom.c.
3671 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3672 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3673 (record_equivalences_from_incoming_edge): Add additional argument
3674 to single_pred_edge_ignoring_loop_edges call.
3675 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3676 (uncprop_dom_walker::before_dom_children): Add additional argument
3677 to single_pred_edge_ignoring_loop_edges call.
3678 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3679 single_pred_edge_ignoring_loop_edges rather than open coding.
3680 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3682 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3684 * match.pd (-(-A)): Rewrite.
3686 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3688 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3689 (rs6000_emit_int_cmove): New declaration.
3690 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3691 (rs6000_emit_sISEL): Delete.
3692 (rs6000_emit_int_cmove): Make non-static.
3693 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3694 instead of rs6000_emit_sISEL.
3696 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3698 * asan.c (create_cond_insert_point): Maintain profile.
3699 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3701 * basic-block.h (struct basic_block_def): Remove frequency.
3702 (EDGE_FREQUENCY): Use to_frequency
3703 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3705 (find_traces): Update to use to_frequency.
3706 (find_traces_1_round): Likewise; use only IPA counts.
3707 (bb_to_key): Likewise.
3708 (connect_traces): Use IPA counts only.
3709 (copy_bb_p): Update to use to_frequency.
3710 (fix_up_crossing_landing_pad): Likewise.
3711 (sanitize_hot_paths): Likewise.
3712 * bt-load.c (basic_block_freq): Likewise.
3713 * cfg.c (init_flow): Set count_max to uninitialized.
3714 (check_bb_profile): Remove frequencies; check counts.
3715 (dump_bb_info): Do not dump frequencies.
3716 (update_bb_profile_for_threading): Update counts only.
3717 (scale_bbs_frequencies_int): Likewise.
3718 (MAX_SAFE_MULTIPLIER): Remove.
3719 (scale_bbs_frequencies_gcov_type): Update counts only.
3720 (scale_bbs_frequencies_profile_count): Update counts only.
3721 (scale_bbs_frequencies): Update counts only.
3722 * cfg.h (struct control_flow_graph): Add count-max.
3723 (update_bb_profile_for_threading): Update prototype.
3724 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3725 (find_many_sub_basic_blocks): Likewise.
3726 * cfgcleanup.c (try_forward_edges): Likewise.
3727 (try_crossjump_to_edge): Likewise.
3728 * cfgexpand.c (expand_gimple_cond): Likewise.
3729 (expand_gimple_tailcall): Likewise.
3730 (construct_init_block): Likewise.
3731 (construct_exit_block): Likewise.
3732 * cfghooks.c (verify_flow_info): Check consistency of counts.
3733 (dump_bb_for_graph): Do not dump frequencies.
3734 (split_block_1): Do not update frequencies.
3735 (split_edge): Do not update frequencies.
3736 (make_forwarder_block): Do not update frequencies.
3737 (duplicate_block): Do not update frequencies.
3738 (account_profile_record): Do not update frequencies.
3739 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3740 for global heuristics.
3741 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3742 (expected_loop_iterations_unbounded): Use counts only.
3743 * cfgloopmanip.c (scale_loop_profile): Simplify.
3744 (create_empty_loop_on_edge): Simplify
3746 (duplicate_loop_to_header_edge): Simplify
3747 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3748 (update_br_prob_note): Take care of removing note when profile
3750 (relink_block_chain): Do not dump frequency.
3751 (rtl_account_profile_record): Use to_frequency.
3752 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3753 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3754 (cgraph_update_edges_for_call_stmt_node): Likewise.
3755 (cgraph_edge::verify_count_and_frequency): Update.
3756 (cgraph_node::verify_node): Temporarily disable frequency verification.
3757 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3758 to_cgraph_frequency.
3759 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3760 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3762 (cgraph_node::expand_thunk): Update profile.
3763 * except.c (dw2_build_landing_pads): Do not update frequency.
3764 * final.c (compute_alignments): Use to_frequency.
3765 (dump_basic_block_info): Do not dump frequency.
3766 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3767 (dump_gimple_bb_header): Do not dump frequency.
3768 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3770 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3771 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3772 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3773 (init_before_recovery): Do not update frequency.
3774 (sched_create_recovery_edges): Do not update frequency.
3775 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3776 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3777 (ipa_cp_c_finalize): Set max_count to uninitialized.
3778 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3779 (param_change_prob): Use counts.
3780 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3782 * ipa-split.c (consider_split): Use to_frequency.
3783 (split_function): Use to_frequency.
3784 * ira-build.c (loop_compare_func): Likewise.
3785 (mark_loops_for_removal): Likewise.
3786 (mark_all_loops_for_removal): Likewise.
3787 * loop-doloop.c (doloop_modify): Do not update frequency.
3788 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3790 * lto-streamer-in.c (input_function): Update count_max.
3791 * omp-expand.c (expand_omp_taskreg): Update count_max.
3792 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3793 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3794 (maybe_hot_count_p): Use ipa counts only.
3795 (maybe_hot_bb_p): Simplify.
3796 (maybe_hot_edge_p): Simplify.
3797 (probably_never_executed): Do not take frequency argument.
3798 (probably_never_executed_bb_p): Do not pass frequency.
3799 (probably_never_executed_edge_p): Likewise.
3800 (combine_predictions_for_bb): Check that profile is nonzero.
3801 (propagate_freq): Do not set frequency.
3802 (drop_profile): Simplify.
3803 (counts_to_freqs): Simplify.
3804 (expensive_function_p): Use to_frequency.
3805 (propagate_unlikely_bbs_forward): Simplify.
3806 (determine_unlikely_bbs): Simplify.
3807 (estimate_bb_frequencies): Add hack to silence graphite issues.
3808 (compute_function_frequency): Use ipa counts.
3809 (pass_profile::execute): Update.
3810 (rebuild_frequencies): Use counts only.
3811 (force_edge_cold): Use counts only.
3812 * profile-count.c (profile_count::dump): Dump new count types.
3813 (profile_count::differs_from_p): Check compatiblity.
3814 (profile_count::to_frequency): New function.
3815 (profile_count::to_cgraph_frequency): New function.
3816 * profile-count.h (struct function): Declare.
3817 (enum profile_quality): Add profile_guessed_local and
3818 profile_guessed_global0.
3819 (class profile_proability): Decrease number of bits to 29;
3820 update from_reg_br_prob_note and to_reg_br_prob_note.
3821 (class profile_count: Update comment; decrease number of bits
3822 to 61. Check compatibility.
3823 (profile_count::compatible_p): New private member function.
3824 (profile_count::ipa_p): New member function.
3825 (profile_count::operator<): Handle global zero correctly.
3826 (profile_count::operator>): Handle global zero correctly.
3827 (profile_count::operator<=): Handle global zero correctly.
3828 (profile_count::operator>=): Handle global zero correctly.
3829 (profile_count::nonzero_p): New member function.
3830 (profile_count::force_nonzero): New member function.
3831 (profile_count::max): New member function.
3832 (profile_count::apply_scale): Handle IPA scalling.
3833 (profile_count::guessed_local): New member function.
3834 (profile_count::global0): New member function.
3835 (profile_count::ipa): New member function.
3836 (profile_count::to_frequency): Declare.
3837 (profile_count::to_cgraph_frequency): Declare.
3838 * profile.c (OVERLAP_BASE): Delete.
3839 (compute_frequency_overlap): Delete.
3840 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3841 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3842 * sched-ebb.c (rank): Use counts only.
3843 * shrink-wrap.c (handle_simple_exit): Use counts only.
3844 (try_shrink_wrapping): Use counts only.
3845 (place_prologue_for_one_component): Use counts only.
3846 * tracer.c (find_best_predecessor): Use to_frequency.
3847 (find_trace): Use to_frequency.
3848 (tail_duplicate): Use to_frequency.
3849 * trans-mem.c (expand_transaction): Do not update frequency.
3850 * tree-call-cdce.c: Do not update frequency.
3851 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3852 (gimple_merge_blocks): Likewise.
3853 (gimple_split_edge): Likewise.
3854 (gimple_duplicate_sese_region): Likewise.
3855 (gimple_duplicate_sese_tail): Likewise.
3856 (move_sese_region_to_fn): Likewise.
3857 (gimple_account_profile_record): Likewise.
3858 (insert_cond_bb): Likewise.
3859 * tree-complex.c (expand_complex_div_wide): Likewise.
3860 * tree-eh.c (lower_resx): Update profile.
3861 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3863 (initialize_cfun): Do not initialize frequencies
3864 (freqs_to_counts): Delete.
3865 (copy_cfg_body): Ignore count parameter.
3866 (copy_body): Update.
3867 (expand_call_inline): Update count_max.
3868 (optimize_inline_calls): Update count_max.
3869 (tree_function_versioning): Update count_max.
3870 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3871 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3873 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3874 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3875 (try_peel_loop): Likewise.
3876 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3878 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3879 (tree_transform_and_unroll_loop): Do not use frequencies
3880 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3881 Use reliable prediction only.
3882 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3883 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3884 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3885 probability scaling.
3886 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3887 not update frequency
3888 (any_remaining_duplicated_blocks): Likewise.
3889 (update_profile): Likewise.
3890 (estimated_freqs_path): Delete.
3891 (freqs_to_counts_path): Delete.
3892 (clear_counts_path): Delete.
3893 (ssa_fix_duplicate_block_edges): Likewise.
3894 (duplicate_thread_path): Likewise.
3895 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3896 * tree-tailcall.c (decrease_profile): Do not update frequency.
3897 (eliminate_tail_call): Likewise.
3898 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3899 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3900 (optimize_mask_stores): Likewise.
3901 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3902 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3903 (ubsan_expand_ptr_ifn): Update profile.
3904 * value-prof.c (gimple_ic): Simplify.
3905 * value-prof.h (gimple_ic): Update prototype.
3906 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3907 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3909 (want_inline_self_recursive_call_p): Likewise.
3910 (resolve_noninline_speculation): Only cummulate defined counts.
3911 (inline_small_functions): Use nonzero_p.
3912 (ipa_inline): Do not access freed node.
3914 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3916 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3917 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3919 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3921 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3922 non-legitimate address.
3924 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3926 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3927 (*lt0_<mode>di, *lt0_<mode>si): New.
3929 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3931 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3932 TARGET_PAIRED_FLOAT.
3934 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3935 Jim Wilson <jim.wilson@linaro.org>
3937 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3938 * config/aarch64/aarch64-tune.md: Regenerated.
3939 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3940 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3942 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3944 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3945 (arc_expand_prologue): Restore blink for millicode.
3946 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3948 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3951 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3952 gen_vec_duplicate after forcing the scalar into a register.
3954 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3956 * combine (try_combine): Print the insns input to try_combine to the
3959 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3962 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3963 Remove second argument from aarch64_process_target_attr call.
3964 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3966 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3968 (aarch64_handle_attr_arch): Remove second argument.
3969 (aarch64_handle_attr_cpu): Ditto.
3970 (aarch64_handle_attr_tune): Ditto.
3971 (aarch64_handle_attr_isa_flags): Ditto.
3972 (aarch64_process_one_target_attr): Ditto.
3973 (aarch64_process_target_attr): Ditto.
3974 (aarch64_option_valid_attribute_p): Remove second argument.
3975 on aarch64_process_target_attr call.
3977 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3979 * diagnostic.c: Include "selftest-diagnostic.h".
3980 (selftest::assert_location_text): New function.
3981 (selftest::test_diagnostic_get_location_text): New function.
3982 (selftest::diagnostic_c_tests): Call it.
3984 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3986 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3987 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3988 (class selftest::test_diagnostic_context): Move to...
3989 * selftest-diagnostic.c: New file.
3990 * selftest-diagnostic.h: New file.
3992 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3994 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3995 offset range for FT32B.
3996 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3997 * config/ft32/ft32.md: Add TARGET_NOPM.
3998 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3999 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4001 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4003 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4005 2017-11-02 Jeff Law <law@redhat.com>
4007 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4008 virtual keyword on FINAL OVERRIDE members.
4010 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4011 virtuals together. Add virtual destructor.
4012 (substitute_and_fold_engine): Similarly.
4014 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4016 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4018 2017-11-02 Richard Biener <rguenther@suse.de>
4020 PR tree-optimization/82795
4021 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4023 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4025 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4027 (gcc_GAS_CHECK_FEATURE): Remove.
4028 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4029 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4030 * configure: Regenerate.
4032 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4034 * config/arc/arc.c (hwloop_optimize): Account for empty
4037 2017-11-02 Richard Biener <rguenther@suse.de>
4040 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4041 Truncate ARRAY_REF index and element size.
4043 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4045 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4047 2017-11-01 Jeff Law <law@redhat.com>
4049 * tree-ssa-ccp.c (ccp_folder): New class derived from
4050 substitute_and_fold_engine.
4051 (ccp_folder::get_value): New member function.
4052 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4053 (ccp_fold_stmt): Remove prototype.
4054 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4055 * tree-ssa-copy.c (copy_folder): New class derived from
4056 substitute_and_fold_engine.
4057 (copy_folder::get_value): Renamed from get_value.
4058 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4059 * tree-vrp.c (vrp_folder): New class derived from
4060 substitute_and_fold_engine.
4061 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4062 (vrp_folder::get_value): New member function.
4063 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4064 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4065 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4066 provide a class interface to folder/substitute routines.
4067 (ssa_prop_fold_stmt_fn): Remove typedef.
4068 (ssa_prop_get_value_fn): Likewise.
4069 (subsitute_and_fold): Remove prototype.
4070 (replace_uses_in): Likewise.
4071 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4072 Renamed from replace_uses_in. Call the virtual member function
4073 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4074 (substitute_and_fold_dom_walker): Remove initialization of
4075 data member entries for calbacks. Add substitute_and_fold_engine
4076 member and initialize it.
4077 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4078 member functions for get_value, replace_phi_args_in c
4079 replace_uses_in, and fold_stmt calls.
4080 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4081 substitute_and_fold. Remove assert. Update ctor call.
4083 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4084 (ssa_prop_visit_phi_fn): Likewise.
4085 (class ssa_propagation_engine): New class to provide an interface
4087 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4089 (ssa_prop_visit_phi): Likewise.
4090 (ssa_propagation_engine::simulate_stmt): Moved into class.
4091 Call visit_phi/visit_stmt from the class rather than via
4092 file scoped static variables.
4093 (ssa_propagation_engine::simulate_block): Moved into class.
4094 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4095 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4096 set file scoped statics for the visit_stmt/visit_phi callbacks.
4097 * tree-complex.c (complex_propagate): New class derived from
4098 ssa_propagation_engine.
4099 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4100 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4101 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4103 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4104 ssa_propagation_engine.
4105 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4106 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4107 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4108 * tree-ssa-copy.c (copy_prop): New class derived from
4109 ssa_propagation_engine.
4110 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4111 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4112 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4113 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4114 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4115 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4116 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4118 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4120 PR rtl-optimization/82778
4121 PR rtl-optimization/82597
4122 * compare-elim.c (struct comparison): Add in_a_setter field.
4123 (find_comparison_dom_walker::before_dom_children): Remove killed
4124 bitmap and df_simulate_find_defs call, instead walk the defs.
4125 Compute last_setter and initialize in_a_setter. Merge definitions
4126 with first initialization for a few variables.
4127 (try_validate_parallel): Use insn_invalid_p instead of
4128 recog_memoized. Return insn rather than just the pattern.
4129 (try_merge_compare): Fix up comment. Don't uselessly test if
4130 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4132 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4136 Alan Hayward <alan.hayward@arm.com>
4137 David Sherwood <david.sherwood@arm.com>
4139 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4140 aarch64_hard_regno_nregs to get the number of registers
4143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4144 Alan Hayward <alan.hayward@arm.com>
4145 David Sherwood <david.sherwood@arm.com>
4147 * config/aarch64/constraints.md (Upl): Rename to...
4149 * config/aarch64/aarch64.md
4150 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4153 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4154 Alan Hayward <alan.hayward@arm.com>
4155 David Sherwood <david.sherwood@arm.com>
4157 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4158 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4161 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4162 Alan Hayward <alan.hayward@arm.com>
4163 David Sherwood <david.sherwood@arm.com>
4165 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4166 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4167 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4169 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4170 of a single element.
4171 * config/aarch64/iterators.md: Add a comment above the permute
4172 unspecs to say that they are generated directly by
4173 aarch64_expand_vec_perm_const.
4174 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4176 2017-11-01 Nathan Sidwell <nathan@acm.org>
4178 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4180 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4182 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4183 and medany code models, and describe what they do.
4185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4187 Revert accidental duplicate:
4189 * combine.c (can_change_dest_mode): Reject changes in
4190 REGMODE_NATURAL_SIZE.
4192 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4194 PR rtl-optimization/64682
4195 PR rtl-optimization/69567
4196 PR rtl-optimization/69737
4197 PR rtl-optimization/82683
4198 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4199 register mentioned in the note, drop the note, unless it came from I3,
4200 in which case it should go to I3 again.
4202 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4204 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4205 and return false if not.
4206 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4210 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4211 range is known to be empty.
4213 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4214 Alan Hayward <alan.hayward@arm.com>
4215 David Sherwood <david.sherwood@arm.com>
4217 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4218 and CONST_VECTOR_NUNITS instead of computing the number of units from
4219 the byte sizes of the vector and element.
4220 (simplify_binary_operation_1): Likewise.
4221 (simplify_const_binary_operation): Likewise.
4222 (simplify_ternary_operation): Likewise.
4224 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4225 Alan Hayward <alan.hayward@arm.com>
4226 David Sherwood <david.sherwood@arm.com>
4228 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4229 (int_mem_offset): ...this new function.
4230 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4231 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4232 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4235 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4236 Alan Hayward <alan.hayward@arm.com>
4237 David Sherwood <david.sherwood@arm.com>
4239 * lower-subreg.c (interesting_mode_p): New function.
4240 (compute_costs, find_decomposable_subregs, decompose_register)
4241 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4242 (resolve_clobber, dump_choices): Use it.
4244 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4245 Alan Hayward <alan.hayward@arm.com>
4246 David Sherwood <david.sherwood@arm.com>
4248 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4250 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4251 Alan Hayward <alan.hayward@arm.com>
4252 David Sherwood <david.sherwood@arm.com>
4254 * alias.c (find_base_value, find_base_term): Only process integer
4255 truncations. Check the precision rather than the size.
4257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4258 Alan Hayward <alan.hayward@arm.com>
4259 David Sherwood <david.sherwood@arm.com>
4261 * machmode.h (is_narrower_int_mode): New function
4262 * optabs.c (expand_float, expand_fix): Use it.
4263 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4265 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4266 Alan Hayward <alan.hayward@arm.com>
4267 David Sherwood <david.sherwood@arm.com>
4269 * rtl.h (narrower_subreg_mode): New function.
4270 * ira-color.c (update_costs_from_allocno): 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 * optabs-query.h (convert_optab_p): New function, split out from...
4277 (convert_optab_handler): ...here.
4278 (widening_optab_handler): Delete.
4279 (find_widening_optab_handler): Remove permit_non_widening parameter.
4280 (find_widening_optab_handler_and_mode): Likewise. Provide an
4281 override that operates on mode class wrappers.
4282 * optabs-query.c (widening_optab_handler): Delete.
4283 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4284 parameter. Assert that the two modes are the same class and that
4285 the "from" mode is narrower than the "to" mode. Use
4286 convert_optab_handler instead of widening_optab_handler.
4287 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4288 instead of widening_optab_handler.
4289 * expr.c (expand_expr_real_2): Update calls to
4290 find_widening_optab_handler.
4291 * optabs.c (expand_widen_pattern_expr): Likewise.
4292 (expand_binop_directly): Take the insn_code as a parameter.
4293 (expand_binop): Only call find_widening_optab_handler for
4294 conversion optabs; use optab_handler otherwise. Update calls
4295 to find_widening_optab_handler and expand_binop_directly.
4296 Use convert_optab_handler instead of widening_optab_handler.
4297 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4298 find_widening_optab_handler and use scalar_mode rather than
4300 (convert_plusminus_to_widen): Likewise.
4302 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4303 Alan Hayward <alan.hayward@arm.com>
4304 David Sherwood <david.sherwood@arm.com>
4306 * machmode.h (fixed_size_mode): New class.
4307 * rtl.h (get_pool_mode): Return fixed_size_mode.
4308 * gengtype.c (main): Add fixed_size_mode.
4309 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4310 (get_raw_arg_mode): Likewise.
4311 * doc/tm.texi: Regenerate.
4312 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4313 * targhooks.c (default_get_reg_raw_mode): Likewise.
4314 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4315 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4316 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4317 (msp430_get_raw_result_mode): Likewise.
4318 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4319 * dbxout.c (dbxout_parms): Require fixed-size modes.
4320 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4321 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4322 * omp-low.c (lower_oacc_reductions): Likewise.
4323 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4324 (simplify_subreg): Update accordingly.
4325 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4326 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4327 that aren't fixed-size.
4328 (get_pool_mode): Return a fixed_size_mode.
4329 (output_constant_pool_2): Take a fixed_size_mode.
4331 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4332 Alan Hayward <alan.hayward@arm.com>
4333 David Sherwood <david.sherwood@arm.com>
4335 * doc/rtl.texi (vec_series): Document.
4336 (const): Say that the operand can be a vec_series.
4337 * rtl.def (VEC_SERIES): New rtx code.
4338 * rtl.h (const_vec_series_p_1): Declare.
4339 (const_vec_series_p): New function.
4340 * emit-rtl.h (gen_const_vec_series): Declare.
4341 (gen_vec_series): Likewise.
4342 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4343 (gen_vec_series): Likewise.
4344 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4345 * simplify-rtx.c (simplify_unary_operation): Handle negations
4347 (simplify_binary_operation_series): New function.
4348 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4349 (test_vector_ops_series): New function.
4350 (test_vector_ops): Call it.
4351 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4352 gen_const_vec_series.
4353 (altivec_lvsr): Likewise.
4354 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4356 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4357 Alan Hayward <alan.hayward@arm.com>
4358 David Sherwood <david.sherwood@arm.com>
4360 * doc/rtl.texi (const): Update description of address constants.
4361 Say that vector constants are allowed too.
4362 * common.md (E, F): Use CONSTANT_P instead of checking for
4364 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4365 checking for CONST_VECTOR.
4366 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4368 * expr.c (expand_expr_real_2): Check for vector modes instead
4369 of checking for CONST_VECTOR.
4370 * rtl.h (const_vec_p): New function.
4371 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4372 (unwrap_const_vec_duplicate): Handle them here too.
4374 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4375 David Malcolm <dmalcolm@redhat.com>
4376 Alan Hayward <alan.hayward@arm.com>
4377 David Sherwood <david.sherwood@arm.com>
4379 * rtl.h (vec_duplicate_p): New function.
4380 * selftest-rtl.c (assert_rtx_eq_at): New function.
4381 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4382 (assert_rtx_eq_at): Declare.
4383 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4384 * selftest-run-tests.c (selftest::run_tests): Call it.
4385 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4386 (simplify_unary_operation_1): Recursively handle vector duplicates.
4387 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4389 (simplify_subreg): Handle subregs of vector duplicates.
4390 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4391 (selftest::simplify_rtx_c_tests): New functions.
4393 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4394 Alan Hayward <alan.hayward@arm.com>
4395 David Sherwood <david.sherwood@arm.com>
4397 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4398 (gen_vec_duplicate): Likewise.
4399 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4401 (gen_const_vector): ...here.
4402 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4403 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4404 whose elements are all equal.
4405 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4406 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4407 (simplify_relational_operation): Likewise.
4408 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4410 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4411 (aarch64_expand_vector_init): Likewise.
4412 * config/arm/arm.c (neon_vdup_constant): Likewise.
4413 (neon_expand_vector_init): Likewise.
4414 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4415 (arm_block_set_unaligned_vect): Likewise.
4416 (arm_block_set_aligned_vect): Likewise.
4417 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4418 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4419 (expand_vec_perm_even_odd_pack): Likewise.
4420 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4421 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4422 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4423 gen_const_vec_duplicate.
4424 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4425 * config/mips/mips.c (mips_gen_const_int_vector): Use
4426 gen_const_vec_duplicate.
4427 (mips_expand_vector_init): Use CONST0_RTX.
4428 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4429 (define_split): Use gen_const_vec_duplicate.
4430 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4431 (define_split): Use gen_const_vec_duplicate.
4432 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4433 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4434 * config/spu/spu.c (spu_const): Likewise.
4436 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4437 Alan Hayward <alan.hayward@arm.com>
4438 David Sherwood <david.sherwood@arm.com>
4440 * combine.c (can_change_dest_mode): Reject changes in
4441 REGMODE_NATURAL_SIZE.
4443 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4445 * configure.ac (--enable-libssp): New.
4446 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4447 trying to determine target-specific default. Adjust indentation.
4448 * configure: Regenerated.
4449 * doc/install.texi (Configuration): Expand --disable-libssp
4452 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4454 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4457 2017-10-31 Martin Jambor <mjambor@suse.cz>
4460 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4462 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4464 * auto-profile.c (autofdo_source_profile::read): Use
4465 UNKNOWN_LOCATION rather than 0.
4466 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4467 (warning_at): ...this overload.
4468 (warning_at_rich_loc_n): Rename to...
4469 (warning_n): ...this overload.
4470 (error_at_rich_loc): Rename to...
4471 (error_at): ...this overload.
4472 (pedwarn_at_rich_loc): Rename to...
4473 (pedwarn): ...this overload.
4474 (permerror_at_rich_loc): Rename to...
4475 (permerror): ...this overload.
4476 (inform_at_rich_loc): Rename to...
4477 (inform): ...this overload.
4478 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4479 (diagnostic_n_impl_richloc): Rename to...
4480 (diagnostic_n_impl): ...this rich_location *-based decl.
4481 (inform_at_rich_loc): Rename to...
4482 (inform): ...this, and add an assertion.
4483 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4484 (warning_at_rich_loc): Rename to...
4485 (warning_at): ...this, and add an assertion.
4486 (warning_at_rich_loc_n): Rename to...
4487 (warning_n): ...this, and add an assertion.
4488 (warning_n): Update location_t-based implementation for removal of
4489 location_t-based diagnostic_n_impl.
4490 (pedwarn_at_rich_loc): Rename to...
4491 (pedwarn): ...this, and add an assertion.
4492 (permerror_at_rich_loc): Rename to...
4493 (permerror): ...this, and add an assertion.
4494 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4495 (error_at_rich_loc): Rename to...
4496 (error_at): ...this, and add an assertion.
4497 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4498 (driver::do_spec_on_infiles): Likewise.
4499 * substring-locations.c (format_warning_va): Update for renaming
4500 of inform_at_rich_loc.
4502 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4504 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4505 _Float<N>X built-in functions so that the variant without the
4506 "__builtin_" prefix is only enabled for the GNU C and Objective C
4507 languages when they are in non-strict ANSI/ISO mode.
4508 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4509 * target.def (floatn_builtin_p): Add a target hook to control
4510 whether _Float<N> and _Float<N>X built-in functions without the
4511 "__builtin_" prefix are enabled, and return true for C and
4512 Objective C in the default hook. Include langhooks.h in
4514 * targhooks.h (default_floatn_builtin_p): Likewise.
4515 * targhooks.c (default_floatn_builtin_p): Likewise.
4516 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4517 floatn_builtin_p target hook.
4518 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4520 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4521 Eric Botcazou <ebotcazou@adacore.com>
4523 PR rtl-optimization/81803
4524 * lra-constraints.c (curr_insn_transform): Also reload the whole
4525 register for a strict subreg no wider than a word if this is for
4526 a WORD_REGISTER_OPERATIONS target.
4528 2017-10-31 Jason Merrill <jason@redhat.com>
4530 * gdbinit.in: Skip over inlines from timevar.h.
4532 2017-10-31 Martin Liska <mliska@suse.cz>
4534 * doc/gcov.texi: Document new option.
4535 * gcov.c (print_usage): Likewise print it.
4536 (process_args): Support the argument.
4537 (format_count): New function.
4538 (format_gcov): Use the function.
4540 2017-10-31 Martin Liska <mliska@suse.cz>
4542 * gcov.c (struct name_map): do not use typedef.
4543 Define operator== and operator<.
4544 (name_search): Remove.
4545 (name_sort): Remove.
4546 (main): Do not allocate names.
4547 (process_file): Add vertical space.
4548 (generate_results): Use std::find.
4549 (release_structures): Do not release memory.
4550 (find_source): Use std::find.
4552 2017-10-31 Martin Liska <mliska@suse.cz>
4554 * gcov.c (struct line_info): Remove it's typedef.
4555 (line_info::line_info): Add proper ctor.
4556 (line_info::has_block): Do not use a typedef.
4557 (struct source_info): Do not use typedef.
4558 (circuit): Likewise.
4559 (get_cycles_count): Likewise.
4560 (output_intermediate_file): Iterate via vector iterator.
4561 (add_line_counts): Use std::vector methods.
4562 (accumulate_line_counts): Likewise.
4563 (output_lines): Likewise.
4565 2017-10-31 Martin Liska <mliska@suse.cz>
4567 * gcov.c (struct source_info): Remove typedef.
4568 (source_info::source_info): Add proper ctor.
4569 (accumulate_line_counts): Use struct, not it's typedef.
4570 (output_gcov_file): Likewise.
4571 (output_lines): Likewise.
4572 (main): Do not allocate an array.
4573 (output_intermediate_file): Use size of vector container.
4574 (process_file): Resize the vector.
4575 (generate_results): Do not preallocate, use newly added vector
4577 (release_structures): Do not release sources.
4578 (find_source): Use vector methods.
4579 (add_line_counts): Do not use typedef.
4581 2017-10-31 Martin Liska <mliska@suse.cz>
4583 * doc/gcov.texi: Document that.
4584 * gcov.c (add_line_counts): Mark lines with a non-executed
4586 (output_line_beginning): Handle such lines.
4587 (output_lines): Pass new argument.
4588 (output_intermediate_file): Print it in intermediate format.
4590 2017-10-31 Martin Liska <mliska@suse.cz>
4592 * color-macros.h: New file.
4593 * diagnostic-color.c: Factor out color related to macros to
4595 * doc/gcov.texi: Document -k option.
4596 * gcov.c (INCLUDE_STRING): Include string.h.
4597 (print_usage): Add -k option.
4598 (process_args): Parse it.
4599 (pad_count_string): New function.
4600 (output_line_beginning): Likewise.
4601 (DEFAULT_LINE_START): New macro.
4602 (output_lines): Support color output.
4604 2017-10-31 Martin Liska <mliska@suse.cz>
4606 PR gcov-profile/82633
4607 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4608 their interaction with GCOV infrastructure.
4609 * configure.ac: Add -fkeep-{inline,static}-functions to
4611 * configure: Regenerate.
4613 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4616 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4618 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4621 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4622 into a register if it does not fit into an immediate offset field.
4624 2017-10-31 Olivier Hainque <hainque@adacore.com>
4626 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4628 2017-10-31 Julia Koval <julia.koval@intel.com>
4630 * config.gcc: Add gfniintrin.h.
4631 * config/i386/gfniintrin.h: New.
4632 * config/i386/i386-builtin-types.def
4633 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4634 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4635 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4636 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4637 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4638 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4639 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4640 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4641 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4642 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4643 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4644 * config/i386/immintrin.h: Include gfniintrin.h.
4645 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4647 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4649 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4651 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4653 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4654 (arm_ashldi3_1bit): Remove pattern.
4655 (ashrdi3): Remove shift by 1 expansion.
4656 (arm_ashrdi3_1bit): Remove pattern.
4657 (lshrdi3): Remove shift by 1 expansion.
4658 (arm_lshrdi3_1bit): Remove pattern.
4659 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4660 cost of ashldi3 by 1.
4661 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4662 (<shift>di3_neon): Likewise.
4664 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4666 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4667 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4668 and (*aarch64_simd_mov<VQ:mode>).
4669 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4670 pattern alternative.
4671 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4672 attributes to match pattern alternative.
4674 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4676 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4677 * config/rs6000/emmintrin.h: New file.
4678 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4680 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4682 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4683 * (movdi_vfp_cortexa8): Remove pattern.
4685 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4687 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4689 (Specific, alpha*-dec-osf5.1): Remove.
4690 (Specific, mips-sgi-irix5): Remove.
4691 (Specific, mips-sgi-irix6): Remove.
4693 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4696 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4697 arguments to clear_bit_region_be.
4699 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4701 * gimplify.c: Include memmodel.h.
4703 2017-10-30 Martin Jambor <mjambor@suse.cz>
4705 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4706 condition whether loop should be executed at all.
4708 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4710 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4711 gimple folding of vec_madd() intrinsics.
4712 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4713 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4714 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4716 2017-10-30 Richard Biener <rguenther@suse.de>
4718 PR tree-optimization/82762
4720 2017-10-23 Richard Biener <rguenther@suse.de>
4722 PR tree-optimization/82129
4724 2017-08-01 Richard Biener <rguenther@suse.de>
4726 PR tree-optimization/81181
4727 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4728 (compute_antic): ... end of iteration here.
4730 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4732 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4733 -std=iso9899:2017 and -std=gnu17.
4734 * doc/standards.texi (C Language): Document C17 support.
4735 * doc/cpp.texi (Overview): Mention -std=c17.
4736 (Standard Predefined Macros): Document C11 and C17 values of
4737 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4738 * doc/extend.texi (Inline): Do not list individual options for
4739 standards newer than C99.
4740 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4742 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4745 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4747 * asan.c (asan_finish_file): Align asan globals array by shadow
4750 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4753 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4754 (struct store_immediate_info): Add bitregion_start and bitregion_end
4756 (store_immediate_info::store_immediate_info): Add brs and bre
4757 arguments and initialize bitregion_{start,end} from those.
4758 (struct merged_store_group): Add bitregion_start, bitregion_end,
4759 align_base and mask fields. Drop unnecessary struct keyword from
4760 struct store_immediate_info. Add do_merge method.
4761 (clear_bit_region_be): Use memset instead of loop storing zeros.
4762 (merged_store_group::do_merge): New method.
4763 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4764 stores as long as the surrounding bitregions have no gaps.
4765 (merged_store_group::merge_overlapping): Use do_merge.
4766 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4767 is byte aligned, rather than requiring that start and width are
4768 byte aligned. Drop unnecessary struct keyword from
4769 struct store_immediate_info. Allocate and populate also mask array.
4770 Make start of the arrays relative to bitregion_start rather than
4771 start and size them according to bitregion_{end,start} difference.
4772 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4773 struct store_immediate_info.
4774 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4775 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4776 struct keyword from struct store_immediate_info.
4777 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4778 between stores as long as the surrounding bitregions have no gaps.
4780 (struct split_store): Add orig non-static data member.
4781 (split_store::split_store): Initialize orig to false.
4782 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4783 if there is exactly a single original stmt. Change stmts argument
4784 to pointer from reference, if NULL, don't push anything to it. Add
4785 first argument, use it to optimize skipping over orig stmts that
4786 are known to be before bitpos already. Simplify.
4787 (split_group): Return unsigned int count how many stores are or
4788 would be needed rather than a bool. Add allow_unaligned argument.
4789 Change split_stores argument from reference to pointer, if NULL,
4790 only do a dry run computing how many stores would be produced.
4791 Rewritten algorithm to use both alignment and misalign if
4792 !allow_unaligned and handle bitfield stores with gaps.
4793 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4794 from bitregion_start instead of start. Compute allow_unaligned
4795 here, if true, do 2 split_group dry runs to compute which one
4796 produces fewer stores and prefer aligned if equal. Punt if
4797 new count is bigger or equal than original before emitting any
4798 statements, rather than during that. Remove no longer needed
4799 new_ssa_names tracking. Replace num_stmts with
4800 split_stores.length (). Use 32-bit stack allocated entries
4801 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4802 if possible. Handle bitfields with gaps.
4803 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4804 Compute bitregion_{start,end} for the stores and construct
4805 store_immediate_info with that. Formatting fixes.
4807 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4810 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4811 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4813 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4815 * gimplify.c: Include tm_p.h.
4817 * common.opt (gcoff): Re-add as ignored option.
4818 (gcoff1, gcoff2, gcoff3): Likewise.
4820 * Makefile.in (OBJS): Delete sdbout.o.
4821 (GTFILES): Delete $(srcdir)/sdbout.c.
4822 * debug.h: Delete sdb_debug_hooks.
4823 * final.c: Delete sdbout.h include.
4824 (final_scan_insn): Delete SDB_DEBUG check.
4825 (rest_of_clean_state): Likewise.
4826 * output.h: Delete sdb_begin_function_line.
4829 * toplev.c: Delete sdbout.h include.
4830 (process_options): Delete SDB_DEBUG check.
4831 * tree-core.h (tree_type_common): Delete pointer field of
4833 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4834 TYPE_SYMTAB_POINTER.
4835 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4836 (TYPE_SYMTAB_IS_POINTER): Delete.
4837 (TYPE_SYMTAB_IS_DIE): Renumber.
4838 * xcoffout.c: Refer to former sdbout.c file.
4839 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4841 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4842 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4843 (Debugging Options): Delete -gcoff.
4844 (-gstabs): Delete SDB reference.
4846 (-gcoff@var{level}): Delete.
4847 * doc/passes.texi (Debugging information output): Delete SDB and
4848 sdbout.c references.
4849 * doc/tm.texi: Regenerate.
4850 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4851 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4853 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4854 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4855 (SDB_DEBUGGING_INFO): Delete.
4856 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4857 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4858 * target.def (output_source_filename): Delete COFF reference.
4860 * common.opt (gcoff): Delete.
4861 (gxcoff+): Update Negative chain.
4862 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4864 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4865 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4866 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4868 (expand_function_start): Change sdb reference to past tense.
4869 (expand_function_end): Change sdb reference to past tense.
4870 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4871 * opts.c (debug_type_names): Delete coff entry.
4872 (common_handle_option): Delete OPT_gcoff case.
4873 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4875 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4876 * config/cris/cris.h: Delete SDB reference in comment.
4877 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4878 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4879 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4880 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4882 (ix86_expand_prologue): Likewise.
4883 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4884 * config/ia64/ia64.h: Likewise.
4885 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4886 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4888 * config/mmix/mmix.h: Likewise.
4889 * config/nds32/nds32.c: Likewise.
4890 * config/stormy/storym16.h: Likewise.
4891 * config/visium/visium.h: Likewise.
4892 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4894 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4896 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4897 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4898 FRAME_POINTER_REGNUM point at high end of local var area.
4900 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4902 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4903 Move comment around. Do not reset best_edge for a copiable
4904 destination if the copy would cause a partition change.
4905 (better_edge_p): Remove redundant check.
4907 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4909 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4911 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4913 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4914 for math functions that have _Float<N> and _Float<N>X variants.
4915 (mathfn_built_in_2): Add support for math functions that have
4916 _Float<N> and _Float<N>X variants.
4917 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4918 (expand_builtin_mathfn_ternary): Add support for fma with
4919 _Float<N> and _Float<N>X variants.
4920 (expand_builtin): Likewise.
4921 (fold_builtin_3): Likewise.
4922 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4923 create math function _Float<N> and _Float<N>X variants as external
4925 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4926 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4927 the __builtin_ prefix and if not strict ansi, without the prefix.
4928 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4929 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4930 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4931 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4932 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4933 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4934 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4935 function signatures for fma _Float<N> and _Float<N>X variants.
4936 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4937 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4938 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4939 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4940 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4941 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4942 * gencfn-macros.c (print_case_cfn): Add support for math functions
4943 that have _Float<N> and _Float<N>X variants.
4944 (print_define_operator_list): Likewise.
4945 (fltfn_suffixes): Likewise.
4947 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4948 for math functions that have _Float<N> and _Float<N>X variants.
4949 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4950 and _Float<N>X variants.
4951 (COPYSIGN): Likewise.
4954 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4955 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4957 (integer_valued_read_call_p): Likewise.
4958 * fold-const-call.c (fold_const_call_ss): Likewise.
4959 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4960 _Float<N> and _Float<N>X variants.
4961 (fold_const_call_ssss): Add support for fma _Float<N> and
4962 _Float<N>X variants.
4963 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4964 support for copysign and fma _Float<N> and _Float<N>X variants.
4965 (backprop::process_builtin_call_use): Likewise.
4966 * tree-call-cdce.c (can_test_argument_range); Add support for
4967 sqrt _Float<N> and _Float<N>X variants.
4968 (edom_only_function): Likewise.
4969 (get_no_error_domain): Likewise.
4970 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4971 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4972 copysign _Float<N> and _Float<N>X variants.
4973 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4974 handled by machine independent code.
4975 (FMAF128): Likewise.
4976 * doc/cpp.texi (Common Predefined Macros): Document defining
4977 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4978 fma _Float<N> and _Float<N>X variants.
4980 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4983 * config/i386/i386-modes.def (CCFPU): Remove definition.
4984 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4985 (ix86_cc_modes_compatible): Ditto.
4986 (ix86_expand_carry_flag_compare): Ditto.
4987 (ix86_expand_int_movcc): Ditto.
4988 (ix86_expand_int_addcc): Ditto.
4989 (ix86_reverse_condition): Ditto.
4990 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4991 Return true/false for unordered/ordered fp comparisons.
4992 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4993 (ix86_prepare_fp_compare_args): Update for rename.
4994 (ix86_expand_fp_compare): Update for rename. Generate unordered
4995 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4996 (ix86_expand_sse_compare_and_jump): Ditto.
4997 * config/i386/predicates.md (fcmov_comparison_operator):
4998 Remove CCFPU mode handling.
4999 (ix86_comparison_operator): Ditto.
5000 (ix86_carry_flag_operator): Ditto.
5001 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5002 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5003 (*cmpu<mode>_cc_i387): Ditto.
5004 (FPCMP): Remove mode iterator.
5005 (unord): Remove mode attribute.
5006 (unord_subst): New define_subst transformation
5007 (unord): New define_subst attribute.
5009 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5010 (*cmpi<unord>xf_i387): Ditto.
5011 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5012 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5013 using unord_subst transformation.
5014 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5015 (round_saeonly): Also handle CCFP mode.
5016 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5017 Remove UNSPEC_SAHF unspec handling.
5019 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5021 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5023 2017-10-27 Jeff Law <law@redhat.com>
5025 * gimple-ssa-sprintf.c: Include domwalk.h.
5026 (class sprintf_dom_walker): New class, derived from dom_walker.
5027 (sprintf_dom_walker::before_dom_children): New function.
5028 (struct call_info): Moved into sprintf_dom_walker class
5029 (compute_formath_length, handle_gimple_call): Likewise.
5030 (sprintf_length::execute): Call the dominator walker rather
5031 than walking the statements.
5033 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5034 gimple statement locations.
5035 (check_array_bounds): Corresponding changes. Get the statement's
5036 location directly from wi->stmt.
5038 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5041 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5043 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5045 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5046 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5048 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5051 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5052 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5053 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5054 maybe_get_pool_constant.
5055 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5058 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5060 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5061 2.26 caveat. Update gas and gld versions.
5062 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5065 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5067 * cgraph.h (set_malloc_flag): Declare.
5068 * cgraph.c (set_malloc_flag_1): New function.
5069 (set_malloc_flag): Likewise.
5070 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5071 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5073 (read_ipa_call_summary): Add support for reading is_return_callee.
5074 (write_ipa_call_summary): Stream is_return_callee.
5075 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5076 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5077 ipa-prop.h, ipa-fnsummary.h.
5078 (pure_const_names): Change to static.
5079 (malloc_state_e): Define.
5080 (malloc_state_names): Define.
5081 (funct_state_d): Add field malloc_state.
5082 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5083 (check_retval_uses): New function.
5084 (malloc_candidate_p): Likewise.
5085 (analyze_function): Add support for malloc attribute.
5086 (pure_const_write_summary): Stream malloc_state.
5087 (pure_const_read_summary): Add support for reading malloc_state.
5088 (dump_malloc_lattice): New function.
5089 (propagate_malloc): New function.
5090 (warn_function_malloc): New function.
5091 (ipa_pure_const::execute): Call propagate_malloc and
5092 ipa_free_fn_summary.
5093 (pass_local_pure_const::execute): Add support for malloc attribute.
5094 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5095 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5097 2017-10-27 Martin Liska <mliska@suse.cz>
5099 PR gcov-profile/82457
5100 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5101 for fork-like functions to be properly instrumented.
5103 2017-10-27 Richard Biener <rguenther@suse.de>
5106 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5107 info when we redirected EH.
5109 2017-10-26 Michael Collison <michael.collison@arm.com>
5111 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5113 (<optab>_trunchf<GPI:mode>2: New pattern.
5114 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5115 * config/aarch64/iterators.md (wv): New mode attribute.
5116 (vf, VF): New mode attributes.
5117 (vgp, VGP): New mode attributes.
5118 (s): Update attribute with SImode and DImode prefixes.
5120 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5122 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5123 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5124 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5125 (nios2_option_overide): Initialize it. Don't allow R0-relative
5126 addressing with PIC.
5127 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5128 (nios2_symbolic_constant_p): Likewise.
5129 (nios2_legitimate_address_p): Likewise.
5130 (nios2_r0rel_section_name_p): New.
5131 (nios2_symbol_ref_in_r0rel_data_p): New.
5132 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5133 (r0rel_constant_p): New.
5134 (nios2_print_operand_address): Handle r0rel_constant_p.
5135 (nios2_cdx_narrow_form_p): Likewise.
5136 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5137 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5138 (Nios II Options): Document -mr0rel-sec.
5140 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5142 * config/nios2/nios2.c: Include xregex.h.
5143 (nios2_gprel_sec_regex): New.
5144 (nios2_option_overide): Initialize it. Don't allow GP-relative
5145 addressing with PIC.
5146 (nios2_small_section_name_p): Check for regex match.
5147 * config/nios2/nios2.opt (mgprel-sec=): New option.
5148 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5149 (Nios II Options): Document -mgprel-sec.
5151 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5153 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5155 2017-10-26 Tom de Vries <tom@codesourcery.com>
5157 PR tree-optimization/82707
5158 * gimple.c (gimple_copy): Fix unsharing of
5159 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5161 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5163 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5164 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5165 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5166 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5167 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5168 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5169 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5170 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5171 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5172 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5173 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5174 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5175 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5176 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5177 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5178 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5179 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5181 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5183 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5185 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5186 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5187 warning to rs6000.c. Remove the Undocumented flag, since it has
5189 (-mabi=ibmlongdouble): Likewise.
5190 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5191 already set, set the default format for long double.
5192 (rs6000_debug_reg_global): Print whether long double is IBM or
5194 (rs6000_option_override_internal): Rework setting long double
5195 format. Only warn if the user is changing the long double default
5196 and they did not use -Wno-psabi.
5197 * doc/invoke.texi (PowerPC options): Update the documentation for
5198 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5200 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5201 Alan Hayward <alan.hayward@arm.com>
5202 David Sherwood <david.sherwood@arm.com>
5204 * rtl.h (wider_subreg_mode): New function.
5205 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5206 rather than an unsigned int *.
5207 * ira-color.c (regno_max_ref_width): Replace with...
5208 (regno_max_ref_mode): ...this new variable.
5209 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5210 Use wider_subreg_mode.
5211 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5212 rather than an unsigned int *.
5213 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5214 (process_alt_operands): Likewise.
5215 (invariant_p): Likewise.
5216 * lra-spills.c (assign_mem_slot): Likewise.
5217 (add_pseudo_to_slot): Likewise.
5218 * lra.c (collect_non_operand_hard_regs): Likewise.
5219 (add_regs_to_insn_regno_info): Likewise.
5220 * reload1.c (regno_max_ref_width): Replace with...
5221 (regno_max_ref_mode): ...this new variable.
5222 (reload): Update accordingly. Update call to
5223 ira_sort_regnos_for_alter_reg.
5224 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5225 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5226 (scan_paradoxical_subregs): Likewise.
5228 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5230 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5231 (aarch64_frame): Add emit_frame_chain boolean.
5232 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5233 Move eh_return case to aarch64_layout_frame.
5234 (aarch64_layout_frame): Initialize emit_frame_chain.
5235 (aarch64_expand_prologue): Use emit_frame_chain.
5237 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5239 * config/aarch64/aarch64.c (aarch64_layout_frame):
5240 Ensure LR is always stored at the bottom of the callee-saves.
5241 Remove rarely used frame layout which saves callee-saves at top of
5242 frame, so the store of LR can be used as a valid probe in all cases.
5244 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5246 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5247 Improve unaligned TImode/TFmode base/offset split.
5249 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5250 Alan Hayward <alan.hayward@arm.com>
5251 David Sherwood <david.sherwood@arm.com>
5253 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5254 * combine.c (find_single_use_1): Likewise.
5255 (expand_field_assignment): Likewise.
5256 (move_deaths): Likewise.
5257 * lra-constraints.c (simplify_operand_subreg): Likewise.
5258 (curr_insn_transform): Likewise.
5259 * lra.c (collect_non_operand_hard_regs): Likewise.
5260 (add_regs_to_insn_regno_info): Likewise.
5261 * rtlanal.c (reg_referenced_p): Likewise.
5262 (covers_regno_no_parallel_p): Likewise.
5264 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5266 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5267 Don't print any bits outside the precision of the value.
5268 * wide-int.cc (test_printing): Add some new tests.
5270 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5272 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5273 supports -xbrace_comment option.
5274 * configure: Regenerate.
5275 * config.in: Regenerate.
5276 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5277 (ASM_CPU_SPEC): Use it.
5279 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5281 * target.def (static_rtx_alignment): New hook.
5282 * targhooks.h (default_static_rtx_alignment): Declare.
5283 * targhooks.c (default_static_rtx_alignment): New function.
5284 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5285 * doc/tm.texi: Regenerate.
5286 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5287 instead of targetm.constant_alignment. Remove call to
5289 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5290 (cris_preferred_mininum_alignment): New function, split out from...
5291 (cris_constant_alignment): ...here.
5292 (cris_static_rtx_alignment): New function.
5293 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5295 (ix86_constant_alignment): ...here.
5296 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5297 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5298 (mmix_static_rtx_alignment): New function.
5299 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5300 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5302 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5305 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5306 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5308 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5310 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5311 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5312 mark as a sign-extending load.
5313 (local_pic_load_u): Define.
5315 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5318 * fold-const.c (operand_equal_for_comparison_p): Also return true
5319 if ARG0 is a simple variant of ARG1 with narrower precision.
5320 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5322 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5324 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5326 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5327 scatter_static, scatter_per_elt.
5328 * x86-tune-costs.h: Add new cost entries.
5330 2017-10-25 Richard Biener <rguenther@suse.de>
5332 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5333 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5334 class pass_fre): Move to ...
5335 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5336 class pass_fre): ... here and adjust for statistics.
5338 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5341 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5342 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5343 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5344 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5347 2017-10-25 Richard Biener <rguenther@suse.de>
5349 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5350 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5351 eliminate_push_avail, eliminate_insert): Move inside...
5352 (class eliminate_dom_walker): ... this class in preparation
5354 (fini_eliminate): Remove by merging with ...
5355 (eliminate): ... this function. Adjust for class changes.
5356 (pass_pre::execute): Remove fini_eliminate call.
5357 (pass_fre::execute): Likewise.
5359 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5362 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5363 (VPERMI2, VPERMI2I): New mode iterators.
5364 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5365 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5367 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5368 mode iterator. Remove 3 old define_insn patterns.
5369 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5370 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5371 VPERMI2 mode iterator, remove the other two expanders.
5372 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5373 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5374 instructions, remove the other two patterns.
5375 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5376 mode iterator, remove the other two patterns.
5377 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5378 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5379 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5380 and adjust argument order accordingly.
5381 (ix86_expand_vec_perm): Adjust caller.
5382 (expand_vec_perm_1): Likewise.
5383 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5384 (expand_vec_perm_vpermt2_vpshub2): ... this.
5385 (ix86_expand_vec_perm_const_1): Adjust caller.
5386 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5389 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5390 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5391 (vec_shl_<mode>): Remove unused expander.
5392 (avx512bw_<shift_insn><mode>3): New define_insn.
5393 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5394 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5396 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5399 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5402 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5404 PR rtl-optimization/82396
5405 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5406 (autopref_multipass_init): Simplify initialization.
5407 (autopref_rank_data): Simplify sort order.
5408 * gcc/sched-int.h (autopref_multipass_data_): Remove
5409 multi_mem_insn_p, min_offset and max_offset.
5411 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5414 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5415 Check special value of flag_omit_frame_pointer.
5416 (aarch64_can_eliminate): Likewise.
5417 (aarch64_override_options_after_change_1): Simplify handling of
5418 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5420 2017-10-24 Richard Biener <rguenther@suse.de>
5422 PR tree-optimization/82697
5423 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5424 zero for conditional load and unconditional store.
5426 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5428 * doc/install.texi: Document bootstrap-cet.
5430 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5433 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5434 ENDBR instruction at function entrance if function is only
5437 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5440 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5441 patterns to better describe from which operation the CF is computed.
5442 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5443 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5444 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5445 is 0, use _0 suffixed expanders instead of emitting a comparison
5448 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5450 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5451 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5453 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5456 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5457 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5458 * loop-iv.c (iv_get_reaching_def): Likewise.
5459 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5460 variable is promoted and the partition contains undefined values.
5462 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5464 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5466 (nios2_address_cost): Define.
5467 (nios2_legitimize_address): Recognize (exp + constant) directly.
5468 (TARGET_ADDRESS_COST): Define.
5470 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5472 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5473 (nios2_symbolic_memory_operand_p): Declare.
5474 (nios2_split_large_constant): Declare.
5475 (nios2_split_symbolic_memory_operand): Declare.
5476 * config/nios2/nios2.c: Adjust includes.
5477 (nios2_symbolic_constant_allowed): New.
5478 (nios2_symbolic_constant_p): New.
5479 (nios2_plus_symbolic_constant_p): New.
5480 (nios2_valid_addr_expr_p): Recognize addresses involving
5482 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5483 (nios2_symbolic_memory_operand_p): New.
5484 (nios2_large_constant_p): New.
5485 (nios2_split_large_constant): New.
5486 (nios2_split_plus_large_constant): New.
5487 (nios2_split_symbolic_memory_operand): New.
5488 (nios2_legitimize_address): Code refactoring. Handle addresses
5489 involving symbolic constants.
5490 (nios2_emit_move_sequence): Likewise.
5491 (nios2_print_operand): Improve error output.
5492 (nios2_print_operand_address): Handle LO_SUM.
5493 (nios2_cdx_narrow_form_p): Likewise.
5494 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5495 operands involving symbolic constants.
5496 (movhi_internal, movsi_internal): Likewise.
5497 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5498 (extendhisi2, extendqi<mode>2): Likewise.
5500 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5502 * tree-pass.h (PROP_rtl_split_insns): Define.
5503 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5505 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5507 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5509 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5512 * target.def (const_not_ok_for_debug_p): Default to
5513 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5514 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5515 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5516 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5517 which targetm.const_not_ok_for_debug_p returned true.
5518 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5520 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5522 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5523 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5524 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5526 (ix86_const_not_ok_for_debug_p): New function.
5527 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5528 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5530 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5533 * system.h: Conditionally include "unique-ptr.h" if
5534 INCLUDE_UNIQUE_PTR is defined.
5535 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5536 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5538 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5540 * config/rl78/rl78.md: New define_expand "subdi3".
5542 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5545 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5546 DF_REF_INSN if DF_REF_INSN_INFO is false.
5548 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5550 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5551 xmm_move instead of sse_move.
5552 (sse_store_index): New function.
5553 (ix86_register_move_cost): Be more sensible about mismatch stall;
5554 model AVX moves correctly; make difference between sse->integer and
5556 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5557 moves; make difference between SSE and AVX.
5558 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5559 and zmm_move. Increase size of sse load and store tables;
5560 add unaligned load and store tables; add ssemmx_to_integer.
5561 * x86-tune-costs.h: Update all entries according to real
5562 move latencies from Agner Fog's manual and chip documentation.
5564 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5567 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5568 * config/i386/constraints.md (Wf): New constraint.
5569 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5570 (cmp<dwi>_doubleword): Removed.
5571 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5572 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5573 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5574 expand with cmp<dwi>_doubleword. For LTU and GEU use
5575 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5577 * common.opt (gcolumn-info): Enable by default.
5578 * doc/invoke.texi (gcolumn-info): Document new default.
5580 2017-10-23 Richard Biener <rguenther@suse.de>
5582 PR tree-optimization/82672
5583 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5584 Fold the stmt if we propagated into it.
5586 2017-10-23 Richard Biener <rguenther@suse.de>
5588 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5589 (bitmap_remove_expr_from_set): ... this. All callers call this
5590 for non-constant values.
5591 (bitmap_set_subtract): Rename to...
5592 (bitmap_set_subtract_expressions): ... this. Adjust and
5594 (bitmap_set_contains_value): Remove superfluous check.
5595 (bitmap_set_replace_value): Inline into single caller ...
5596 (bitmap_value_replace_in_set): ... here and simplify.
5597 (dependent_clean): Merge into ...
5598 (clean): ... this using an overload. Adjust.
5599 (prune_clobbered_mems): Adjust.
5600 (compute_antic_aux): Likewise.
5601 (compute_partial_antic_aux): Likewise.
5603 2017-10-23 Richard Biener <rguenther@suse.de>
5605 PR tree-optimization/82129
5607 2017-08-01 Richard Biener <rguenther@suse.de>
5609 PR tree-optimization/81181
5610 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5611 (compute_antic): ... end of iteration here.
5613 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5615 * target.def (starting_frame_offset): New hook.
5616 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5617 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5618 * doc/tm.texi.in: Regenerate.
5619 * hooks.h (hook_hwi_void_0): Declare.
5620 * hooks.c (hook_hwi_void_0): New function.
5621 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5622 STARTING_FRAME_OFFSET.
5623 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5624 * reload1.c (reload): Likewise.
5625 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5626 instead of STARTING_FRAME_OFFSET.
5627 * function.c (try_fit_stack_local): Likewise.
5628 (assign_stack_local_1): Likewise
5629 (instantiate_virtual_regs): Likewise.
5630 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5631 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5632 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5633 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5634 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5635 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5636 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5637 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5638 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5639 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5640 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5641 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5642 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5643 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5644 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5645 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5646 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5647 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5648 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5649 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5650 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5651 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5652 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5653 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5654 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5655 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5656 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5657 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5658 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5659 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5660 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5661 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5662 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5663 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5664 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5665 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5666 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5667 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5668 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5669 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5670 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5671 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5672 return a HOST_WIDE_INT.
5673 (avr_builtin_setjmp_frame_value): Use it instead of
5674 STARTING_FRAME_OFFSET.
5675 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5676 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5677 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5679 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5680 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5681 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5682 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5683 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5684 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5685 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5686 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5687 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5688 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5689 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5690 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5692 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5693 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5694 * config/mips/mips.c (mips_compute_frame_info): Refer to
5695 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5696 (mips_starting_frame_offset): New function.
5697 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5698 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5699 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5700 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5701 and return a HOST_WIDE_INT.
5702 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5703 (mmix_initial_elimination_offset): Refer to
5704 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5705 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5706 * config/pa/pa.c (pa_starting_frame_offset): New function.
5707 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5708 (pa_expand_prologue): Likewise.
5709 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5710 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5711 !FRAME_GROWS_DOWNWARD handling to...
5712 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5713 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5714 !FRAME_GROWS_DOWNWARD handling to...
5715 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5716 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5717 !FRAME_GROWS_DOWNWARD handling to...
5718 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5719 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5721 (rs6000_starting_frame_offset): New function.
5722 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5723 !FRAME_GROWS_DOWNWARD handling to...
5724 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5725 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5726 !FRAME_GROWS_DOWNWARD handling to...
5727 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5728 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5729 !FRAME_GROWS_DOWNWARD handling to...
5730 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5731 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5732 (rs6000_starting_frame_offset): New function.
5733 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5734 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5735 * config/vax/vax.c (vax_starting_frame_offset): New function.
5736 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5737 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5738 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5739 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5740 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5741 * system.h (STARTING_FRAME_OFFSET): Poison.
5743 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5745 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5746 SCALAR_TYPE_MODE instead of TYPE_MODE.
5748 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5749 Alan Hayward <alan.hayward@arm.com>
5750 David Sherwood <david.sherwood@arm.com>
5752 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5754 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5755 Alan Hayward <alan.hayward@arm.com>
5756 David Sherwood <david.sherwood@arm.com>
5758 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5760 2017-10-23 Richard Biener <rguenther@suse.de>
5762 PR tree-optimization/82129
5763 * tree-ssa-pre.c (bitmap_set_and): Remove.
5764 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5765 canonicalizing expressions in the set to those with lowest
5766 ID rather than taking that from the first edge.
5768 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5770 * combine.c (rtx_equal_for_field_assignment_p): Use
5771 byte_lowpart_offset.
5773 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5774 Alan Hayward <alan.hayward@arm.com>
5775 David Sherwood <david.sherwood@arm.com>
5777 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5778 to a SUBREG_PROMOTED_VAR.
5780 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5781 Alan Hayward <alan.hayward@arm.com>
5782 David Sherwood <david.sherwood@arm.com>
5784 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5785 (expand_debug_source_expr): Likewise.
5786 * combine.c (combine_simplify_rtx): Likewise.
5787 * cse.c (fold_rtx): Likewise.
5788 * optabs.c (expand_float): Likewise.
5789 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5790 (simplify_binary_operation_1): Likewise.
5792 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5793 Alan Hayward <alan.hayward@arm.com>
5794 David Sherwood <david.sherwood@arm.com>
5796 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5797 (record_promoted_value): Likewise.
5798 * expr.c (expand_expr_real_2): Likewise.
5799 * ree.c (update_reg_equal_equiv_notes): Likewise.
5800 (combine_set_extension): Likewise.
5801 * rtlanal.c (low_bitmask_len): Likewise.
5802 * simplify-rtx.c (neg_const_int): Likewise.
5803 (simplify_binary_operation_1): Likewise.
5805 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5806 Alan Hayward <alan.hayward@arm.com>
5807 David Sherwood <david.sherwood@arm.com>
5809 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5810 * regcprop.c (maybe_mode_change): Likewise.
5811 * reload1.c (alter_reg): Likewise.
5813 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5815 * inchash.h (inchash::hash::add_wide_int): New function.
5816 * lto-streamer-out.c (hash_tree): Use it.
5818 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5820 * inchash.h (inchash::hash::add_wide_int): Rename to...
5821 (inchash::hash::add_hwi): ...this.
5822 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5823 (polymorphic_call_target_hasher::hash): Likewise.
5824 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5825 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5826 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5827 * lto-streamer-out.c (hash_tree): Likewise.
5828 * optc-save-gen.awk: Likewise.
5829 * tree.c (add_expr): Likewise.
5831 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5834 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5835 for ordered inequality comparisons even with TARGET_IEEE_FP.
5837 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5840 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5841 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5842 Expand with cmp<dwi>_doubleword.
5844 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5846 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5847 List CET intrinsics.
5848 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5849 specific to -fcf-protection option.
5851 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5853 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5854 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5855 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5856 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5857 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5858 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5859 (extra_objs): Add cet.o for Linux/x86 targets.
5860 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5861 * config/i386/cet.c: New file.
5862 * config/i386/cetintrin.h: Likewise.
5863 * config/i386/t-cet: Likewise.
5864 * config/i386/cpuid.h (bit_SHSTK): New.
5865 (bit_IBT): Likewise.
5866 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5867 pass IBT and SHSTK bits.
5868 * config/i386/i386-builtin-types.def
5869 (VOID_FTYPE_UNSIGNED_PVOID): New.
5870 (VOID_FTYPE_UINT64_PVOID): Likewise.
5871 * config/i386/i386-builtin.def: Add CET intrinsics.
5872 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5873 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5874 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5875 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5877 * config/i386/i386.c (rest_of_insert_endbranch): New.
5878 (pass_data_insert_endbranch): Likewise.
5879 (pass_insert_endbranch): Likewise.
5880 (make_pass_insert_endbranch): Likewise.
5881 (ix86_notrack_prefixed_insn_p): Likewise.
5882 (ix86_target_string): Add -mibt, -mshstk flags.
5883 (ix86_option_override_internal): Add flag_cf_protection
5885 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5886 (ix86_print_operand): Add 'notrack' prefix output.
5887 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5888 (ix86_expand_builtin): Expand CET intrinsics.
5889 (x86_output_mi_thunk): Add 'endbranch' instruction.
5890 * config/i386/i386.h (TARGET_IBT): New.
5891 (TARGET_IBT_P): Likewise.
5892 (TARGET_SHSTK): Likewise.
5893 (TARGET_SHSTK_P): Likewise.
5894 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5895 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5896 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5897 (builtin_setjmp_setup): New pattern.
5898 (builtin_longjmp): Likewise.
5899 (rdssp<mode>): Likewise.
5900 (incssp<mode>): Likewise.
5901 (saveprevssp): Likewise.
5902 (rstorssp): Likewise.
5903 (wrss<mode>): Likewise.
5904 (wruss<mode>): Likewise.
5905 (setssbsy): Likewise.
5906 (clrssbsy): Likewise.
5907 (nop_endbr): Likewise.
5908 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5910 * config/i386/immintrin.h: Include <cetintrin.h>.
5911 * config/i386/linux-common.h
5912 (file_end_indicate_exec_stack_and_cet): New prototype.
5913 (TARGET_ASM_FILE_END): New.
5915 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5917 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5918 latencies instead of having separate table; make difference between
5919 integer and float costs.
5920 * i386.h (processor_costs): Remove scalar_stmt_cost,
5921 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5922 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5924 * x86-tune-costs.h: Remove entries which has been removed in
5925 procesor_costs from all tables; make cond_taken_branch_cost
5926 and cond_not_taken_branch_cost COST_N_INSNS based.
5928 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5930 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5932 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5934 * config/i386/i386.md (isa): Remove fma_avx512f.
5935 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5936 <avx512>_fmadd_<mode>_mask3<round_name>,
5937 <avx512>_fmsub_<mode>_mask<round_name>,
5938 <avx512>_fmsub_<mode>_mask3<round_name>,
5939 <avx512>_fnmadd_<mode>_mask<round_name>,
5940 <avx512>_fnmadd_<mode>_mask3<round_name>,
5941 <avx512>_fnmsub_<mode>_mask<round_name>,
5942 <avx512>_fnmsub_<mode>_mask3<round_name>,
5943 <avx512>_fmaddsub_<mode>_mask<round_name>,
5944 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5945 <avx512>_fmsubadd_<mode>_mask<round_name>,
5946 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5947 (*vec_widen_umult_even_v16si<mask_name>,
5948 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5949 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5951 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5953 * extend.texi: Add 'nocf_check' documentation.
5954 * gimple.texi: Add second parameter to
5955 gimple_build_call_from_tree.
5956 * invoke.texi: Add -fcf-protection documentation.
5957 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5959 2017-10-20 Richard Biener <rguenther@suse.de>
5961 PR tree-optimization/82473
5962 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5963 the largest input type.
5965 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5967 * c-attribs.c (handle_nocf_check_attribute): New function.
5968 (c_common_attribute_table): Add 'nocf_check' handling.
5969 * gimple-parser.c: Add second argument NULL to
5970 gimple_build_call_from_tree.
5971 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5972 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5974 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5975 * common.opt: Add fcf-protection flag.
5976 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5977 * flag-types.h: Add enum cf_protection_level.
5978 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5979 Add 'nocf_check' attribute propagation to gimple call.
5980 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5981 (gimple_build_call_from_tree): Update prototype.
5982 (gimple_call_nocf_check_p): New function.
5983 (gimple_call_set_nocf_check): Likewise.
5984 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5985 * ipa-icf.c: Add nocf_check attribute in statement hash.
5986 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5987 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5988 * toplev.c (process_options): Add flag_cf_protection handling.
5990 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5992 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5994 2017-10-20 Richard Biener <rguenther@suse.de>
5996 PR tree-optimization/82603
5997 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5998 remove false predicated stores.
6000 2017-10-20 Richard Biener <rguenther@suse.de>
6002 * graphite-isl-ast-to-gimple.c
6003 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6004 Remove return value and simplify, dump copied stmt after lhs
6006 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6007 Reduce dump verbosity.
6008 (gsi_insert_earliest): Likewise.
6009 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6010 * graphite.c (print_global_statistics): Adjust dumping.
6011 (print_graphite_scop_statistics): Likewise.
6012 (print_graphite_statistics): Do not dump loops here.
6013 (graphite_transform_loops): But here.
6015 2017-10-20 Nicolas Roche <roche@adacore.com>
6017 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6018 * configure: Regenerate.
6020 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6023 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6024 functions when optimizing replace GIMPLE_RETURN stmts with
6025 calls to __builtin_unreachable ().
6028 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6029 for -fsanitize=thread link of executables.
6030 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6031 link of executables.
6034 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6036 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6037 define_insns for logical vector shifts to use VI248_AVX512BW
6038 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6039 condition, useless isa and prefix attributes. Change the first
6040 2 of these define_insns to ...
6041 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6042 define_insn for avx512vl.
6043 (<shift_insn><mode>3): ... and this, new define_insn without
6044 masking for non-avx512vl.
6047 * config/i386/sse.md (*andnot<mode>3,
6048 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6049 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6050 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6051 not applied use empty suffix even for TARGET_AVX512VL.
6052 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6053 is applied, supply evex,evex or evex,evex,evex instead of just
6056 2017-10-20 Julia Koval <julia.koval@intel.com>
6058 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6059 (OPTION_MASK_ISA_GFNI_UNSET): New.
6060 (ix86_handle_option): Handle OPT_mgfni.
6061 * config/i386/cpuid.h (bit_GFNI): New.
6062 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6063 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6064 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6065 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6066 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6067 * config/i386/i386.opt: Add mgfni.
6069 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6071 * config/msp430/msp430.c (msp430_option_override): Disable
6072 -fdelete-null-pointer-checks.
6073 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6075 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6077 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6078 of x87 and SSE instructions.
6080 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6082 * asan.c (create_cond_insert_point): Do not update edge count.
6083 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6084 (afdo_propagate_circuit): Likewise.
6085 (afdo_calculate_branch_prob): Likewise.
6086 (afdo_annotate_cfg): Likewise.
6087 * basic-block.h (struct edge_def): Remove count.
6088 (edge_def::count): New accessor.
6089 * bb-reorder.c (rotate_loop): Update.
6090 (find_traces_1_round): Update.
6091 (connect_traces): Update.
6092 (sanitize_hot_paths): Update.
6093 * cfg.c (unchecked_make_edge): Update.
6094 (make_single_succ_edge): Update.
6095 (check_bb_profile): Update.
6096 (dump_edge_info): Update.
6097 (update_bb_profile_for_threading): Update.
6098 (scale_bbs_frequencies_int): Update.
6099 (scale_bbs_frequencies_gcov_type): Update.
6100 (scale_bbs_frequencies_profile_count): Update.
6101 (scale_bbs_frequencies): Update.
6102 * cfganal.c (connect_infinite_loops_to_exit): Update.
6103 * cfgbuild.c (compute_outgoing_frequencies): Update.
6104 (find_many_sub_basic_blocks): Update.
6105 * cfgcleanup.c (try_forward_edges): Update.
6106 (try_crossjump_to_edge): Update
6107 * cfgexpand.c (expand_gimple_cond): Update
6108 (expand_gimple_tailcall): Update
6109 (construct_exit_block): Update
6110 * cfghooks.c (verify_flow_info): Update
6111 (redirect_edge_succ_nodup): Update
6112 (split_edge): Update
6113 (make_forwarder_block): Update
6114 (duplicate_block): Update
6115 (account_profile_record): Update
6116 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6117 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6118 * cfgloopmanip.c (scale_loop_profile): Update.
6120 (lv_adjust_loop_entry_edge): Update.
6121 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6122 (force_nonfallthru_and_redirect): Update.
6123 (purge_dead_edges): Update.
6124 (rtl_flow_call_edges_add): Update.
6125 * cgraphunit.c (init_lowered_empty_function): Update.
6126 (cgraph_node::expand_thunk): Update.
6127 * gimple-pretty-print.c (dump_probability): Update.
6128 (dump_edge_probability): Update.
6129 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6130 * haifa-sched.c (sched_create_recovery_edges): Update.
6131 * hsa-gen.c (convert_switch_statements): Update.
6132 * ifcvt.c (dead_or_predicable): Update.
6133 * ipa-inline-transform.c (inline_transform): Update.
6134 * ipa-split.c (split_function): Update.
6135 * ipa-utils.c (ipa_merge_profiles): Update.
6136 * loop-doloop.c (add_test): Update.
6137 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6138 * lto-streamer-in.c (input_cfg): Update.
6139 (input_function): Update.
6140 * lto-streamer-out.c (output_cfg): Update.
6141 * modulo-sched.c (sms_schedule): Update.
6142 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6143 * predict.c (maybe_hot_edge_p): Update.
6144 (unlikely_executed_edge_p): Update.
6145 (probably_never_executed_edge_p): Update.
6146 (dump_prediction): Update.
6147 (drop_profile): Update.
6148 (propagate_unlikely_bbs_forward): Update.
6149 (determine_unlikely_bbs): Update.
6150 (force_edge_cold): Update.
6151 * profile.c (compute_branch_probabilities): Update.
6152 * reg-stack.c (better_edge): Update.
6153 * shrink-wrap.c (handle_simple_exit): Update.
6154 * tracer.c (better_p): Update.
6155 * trans-mem.c (expand_transaction): Update.
6156 (split_bb_make_tm_edge): Update.
6157 * tree-call-cdce.c: Update.
6158 * tree-cfg.c (gimple_find_sub_bbs): Update.
6159 (gimple_split_edge): Update.
6160 (gimple_duplicate_sese_region): Update.
6161 (gimple_duplicate_sese_tail): Update.
6162 (gimple_flow_call_edges_add): Update.
6163 (insert_cond_bb): Update.
6164 (execute_fixup_cfg): Update.
6165 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6166 * tree-complex.c (expand_complex_div_wide): Update.
6167 * tree-eh.c (lower_resx): Update.
6168 (unsplit_eh): Update.
6169 (cleanup_empty_eh_move_lp): Update.
6170 * tree-inline.c (copy_edges_for_bb): Update.
6171 (freqs_to_counts): Update.
6172 (copy_cfg_body): Update.
6173 * tree-ssa-dce.c (remove_dead_stmt): Update.
6174 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6175 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6176 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6177 (unloop_loops): Update.
6178 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6179 * tree-ssa-loop-split.c (connect_loops): Update.
6180 (split_loop): Update.
6181 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6182 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6183 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6184 * tree-ssa-reassoc.c (branch_fixup): Update.
6185 * tree-ssa-tail-merge.c (replace_block_by): Update.
6186 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6187 (compute_path_counts): Update.
6188 (update_profile): Update.
6189 (recompute_probabilities): Update.
6190 (update_joiner_offpath_counts): Update.
6191 (estimated_freqs_path): Update.
6192 (freqs_to_counts_path): Update.
6193 (clear_counts_path): Update.
6194 (ssa_fix_duplicate_block_edges): Update.
6195 (duplicate_thread_path): Update.
6196 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6197 (case_bit_test_cmp): Update.
6198 (collect_switch_conv_info): Update.
6199 (gen_inbound_check): Update.
6200 (do_jump_if_equal): Update.
6201 (emit_cmp_and_jump_insns): Update.
6202 * tree-tailcall.c (decrease_profile): Update.
6203 (eliminate_tail_call): Update.
6204 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6205 (vect_do_peeling): Update.
6206 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6207 * ubsan.c (ubsan_expand_null_ifn): Update.
6208 (ubsan_expand_ptr_ifn): Update.
6209 * value-prof.c (gimple_divmod_fixed_value): Update.
6210 (gimple_mod_pow2): Update.
6211 (gimple_mod_subtract): Update.
6212 (gimple_ic): Update.
6213 (gimple_stringop_fixed_value): Update.
6215 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6218 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6220 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6222 PR rtl-optimization/82395
6223 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6224 based on non_spilled_static_chain_regno_p.
6226 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6228 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6229 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6230 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6232 2017-10-19 Martin Sebor <msebor@redhat.com>
6234 PR tree-optimization/82596
6235 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6237 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6239 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6240 (get_mem_refs_of_builtin_call): Likewise.
6241 * builtins.c (expand_builtin_apply): Adjust call to
6242 allocate_dynamic_stack_space.
6243 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6244 the third argument to allocate_dynamic_stack_space, otherwise -1.
6245 (expand_builtin): Deal with all alloca variants.
6246 (is_inexpensive_builtin): Likewise.
6247 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6248 * calls.c (special_function_p): Deal with all alloca variants.
6249 (initialize_argument_information): Adjust call to
6250 allocate_dynamic_stack_space.
6251 (expand_call): Likewise.
6252 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6253 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6254 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6255 use it for the stack usage computation.
6256 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6257 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6258 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6259 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6260 (in_loop_p): Remove first argument and useless check.
6261 (pass_walloca::execute): Remove useless test and adjust call to above.
6262 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6263 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6264 (gimplify_call_expr): Deal with all alloca variants.
6265 * hsa-gen.c (gen_hsa_alloca): Likewise.
6266 (gen_hsa_insns_for_call): Likewise.
6267 * ipa-pure-const.c (special_builtin_state): Likewise.
6268 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6269 * tree-object-size.c (alloc_object_size): Likewise.
6270 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6271 (call_may_clobber_ref_p_1): Likewise.
6272 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6273 (ccp_fold_stmt): Likewise.
6274 (optimize_stack_restore): Likewise.
6275 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6276 (mark_all_reaching_defs_necessary_1): Likewise.
6277 (propagate_necessity): Likewise.
6278 (eliminate_unnecessary_stmts): Likewise.
6279 * tree.c (build_common_builtin_nodes): Build
6280 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6281 (build_alloca_call_expr): New function.
6282 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6283 (CASE_BUILT_IN_ALLOCA): Likewise.
6284 (build_alloca_call_expr): Declare.
6285 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6287 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6290 * dwarf2out.c (new_die_raw): New static inline function.
6291 (new_die): Use it to create the DIE.
6292 (add_AT_external_die_ref): Likewise.
6293 (clone_die): Likewise.
6294 (clone_as_declaration): Likewise.
6295 (dwarf2out_vms_debug_main_pointer): Likewise.
6296 (base_type_die): Likewise. Remove early return for corner cases.
6297 Do not call add_pubtype on the DIE here.
6298 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6299 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6300 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6301 native order exists for base types, attach the DIE manually and call
6302 add_pubtype on it. Do not equate a reverse order DIE to the type.
6304 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6306 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6307 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6309 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6311 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6313 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6314 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6315 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6316 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6317 * opts.c (sanitizer_opts): Add builtin.
6318 * ubsan.c (instrument_builtin): New function.
6319 (pass_ubsan::execute): Call it.
6320 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6321 * doc/invoke.texi: Document -fsanitize=builtin.
6323 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6324 builtins, store max (log2 (align), 0) into uchar field instead of
6325 align into uptr field.
6326 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6327 store uchar 0 field instead of uptr 0 field.
6328 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6329 instead of passing one address of struct with 2 locations pass
6330 two addresses of structs with 1 location each.
6331 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6332 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6333 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6334 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6335 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6336 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6337 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6338 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6340 2017-10-19 Martin Liska <mliska@suse.cz>
6343 * file-find.c (remove_prefix): Remove.
6344 * file-find.h (remove_prefix): Likewise.
6345 * gcc-ar.c: Remove smartness of lookup.
6347 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6349 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6350 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6351 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6353 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6356 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6357 (setcc + and to xor + setcc): New peephole2.
6359 2017-10-19 Tom de Vries <tom@codesourcery.com>
6361 * doc/sourcebuild.texi (Test Directives, Variants of
6362 dg-require-support): Add dg-require-stack-size.
6364 2017-10-19 Martin Liska <mliska@suse.cz>
6367 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6368 that have a large alignment.
6369 (gimplify_target_expr): Likewise.
6371 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6373 PR rtl-optimization/82602
6374 * ira.c (rtx_moveable_p): Return false for volatile asm.
6376 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6379 * config/i386/i386-modes.def (CCGZ): New CC mode.
6380 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6381 * config/i386/predicates.md (ix86_comparison_operator):
6383 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6384 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6385 with double-word subtraction.
6386 (put_condition_code): Handle CCGZmode.
6388 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6390 * wide-int.cc (debug (const wide_int &)): New.
6391 (debug (const wide_int *)): New.
6392 (debug (const widest_int &)): New.
6393 (debug (const widest_int *)): New.
6395 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6398 * lra-constraints.c (curr_insn_transform): Use non-input operand
6399 instead of output one for matched reload.
6401 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6403 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6404 (tree-ssa-loop-ivopts.h): New header file.
6405 (struct builtin_info): New fields.
6406 (classify_builtin_1): Compute and record base and offset parts for
6407 memset builtin partition by calling strip_offset.
6408 (offset_cmp, fuse_memset_builtins): New functions.
6409 (finalize_partitions): Fuse adjacent memset partitions by calling
6411 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6412 Expose the interface.
6413 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6415 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6417 PR tree-optimization/82574
6418 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6419 that data reference must be executed exactly once per iteration
6420 against the outermost loop in nest.
6421 (classify_partition): Update call to above function.
6423 2017-10-18 Richard Biener <rguenther@suse.de>
6425 PR tree-optimization/82591
6426 * graphite.c (graphite_transform_loops): Move code gen message
6428 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6429 Here. Handle scop_to_isl_ast failing.
6430 (scop_to_isl_ast): Limit the number of ISL operations.
6432 2017-10-18 Richard Biener <rguenther@suse.de>
6434 * graphite-isl-ast-to-gimple.c
6435 (translate_isl_ast_to_gimple::set_rename): Simplify.
6436 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6437 (graphite_copy_stmts_from_block): ... here.
6438 (copy_bb_and_scalar_dependences): Simplify.
6439 (add_parameters_to_ivs_params): Canonicalize.
6440 (generate_entry_out_of_ssa_copies): Simplify.
6441 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6442 by passing in ISL dimension.
6443 (parameter_index_in_region_1): Rename to ...
6444 (parameter_index_in_region): ... this.
6445 (extract_affine): Adjust assert, pass down parameter index.
6446 (add_param_constraints): Use range-info when available.
6447 (build_scop_context): Adjust.
6448 * sese.c (new_sese_info): Adjust.
6449 (free_sese_info): Likewise.
6450 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6451 Remove unused typedefs.
6452 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6454 2017-10-18 Martin Liska <mliska@suse.cz>
6456 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6458 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6460 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6461 (s390_sched_init): Do not reset s390_sched_state if we entered the
6462 current basic block via a fallthru edge and all others are unlikely.
6464 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6466 * config/s390/s390.c (NUM_SIDES): New variable.
6467 (LONGRUNNING_THRESHOLD): New variable.
6468 (LATENCY_FACTOR): New variable.
6469 (s390_sched_score): Decrease score for long-running instructions on
6471 (s390_sched_variable_issue): Perform bookkeeping for long-running
6474 2017-10-18 Richard Biener <rguenther@suse.de>
6476 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6477 Simplify with removal of the parameter rename map.
6478 (set_rename): Likewise.
6479 (should_copy_to_new_region): Likewise.
6480 (graphite_copy_stmts_from_block): Likewise.
6481 (copy_bb_and_scalar_dependences): Remove initialization of
6482 unused copied_bb_map.
6484 (copy_internal_parameters): Likewise.
6485 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6486 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6487 Use INTEGRAL_TYPE_P.
6488 (parameter_index_in_region_1): Rename to ...
6489 (assign_parameter_index_in_region): ... this. Assert we have
6490 a parameter we handle.
6491 (scan_tree_for_params): Adjust.
6492 * sese.h (parameter_rename_map_t): Remove.
6493 (struct sese_info_t): Remove unused parameter_rename_map and
6494 copied_bb_map members.
6495 * sese.c (new_sese_info): Adjust.
6496 (free_sese_info): Likewise.
6498 2017-10-18 Martin Liska <mliska@suse.cz>
6501 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6502 on an abnormal edge.
6504 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6506 * doc/invoke.texi (ffunction-sections and fdata-sections):
6509 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6511 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6512 the use statement can throw internally.
6514 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6516 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6517 any RTX present on the RHS of a SET.
6518 * compare-elim.c (try_eliminate_compare): Restore comment.
6520 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6522 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6523 may be also called on tcc_type nodes.
6524 * langhooks.c (lhd_tree_size): Likewise.
6526 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6528 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6529 format_warning_at_substring.
6530 (maybe_warn): Convert source_range * param to a location_t. Pass
6531 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6532 (format_directive): Remove code to extract source_ranges and
6533 source_range * in favor of just a location_t.
6534 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6536 * substring-locations.c (format_warning_va): Convert
6537 source_range * param to a location_t.
6538 (format_warning_at_substring): Likewise.
6539 * substring-locations.h (format_warning_va): Likewise.
6540 (format_warning_at_substring): Likewise.
6542 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6544 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6546 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6547 and scatter/gather ops.
6549 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6550 vec_gather_load and vec_scatter_store.
6551 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6552 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6553 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6554 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6555 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6556 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6558 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6560 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6561 Detect FTST instruction and handle its register pops. Only pop
6562 second operand if can_pop_second_op is true.
6563 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6564 set can_pop_second_op to false in the compare_for_stack_reg call.
6566 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6567 output_fp_compare for stack register operands.
6568 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6569 instructions here. Do not emit stack register pops here. Assert
6570 that FCOMPP pops next to top stack register. Rewrite function.
6572 2017-10-17 Nathan Sidwell <nathan@acm.org>
6575 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6576 use DECL_ASSEMBLER_NAME_RAW.
6579 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6582 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6583 Wilco Dijkstra <wilco.dijkstra@arm.com>
6585 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6586 converstion to Pmode from the buf_addr.
6588 2017-10-17 Richard Biener <rguenther@suse.de>
6590 * graphite-dependences.c (scop_get_reads_and_writes): Change
6591 output parameters to references.
6593 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6595 PR 71026/tree-optimization
6596 * fold-const.c (distribute_real_division): Removed.
6597 (fold_binary_loc): Remove calls to distribute_real_divison.
6599 2017-10-17 Richard Biener <rguenther@suse.de>
6601 * graphite-scop-detection.c
6602 (scop_detection::stmt_has_simple_data_refs_p): Always use
6603 the full nest as region.
6604 (try_generate_gimple_bb): Likewise.
6605 * sese.c (scalar_evolution_in_region): Simplify now that
6606 SCEV can handle instantiation in regions.
6607 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6608 in the non-loop part of a function if requested.
6610 2017-10-17 Richard Biener <rguenther@suse.de>
6612 PR tree-optimization/82563
6613 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6615 (graphite_regenerate_ast_isl): Call it.
6616 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6618 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6620 PR tree-optimization/82549
6621 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6622 Formatting fixes. Instead of calling make_bit_field_ref with negative
6625 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6627 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6628 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6629 _mm_maskz_reduce_ss): New.
6630 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6631 __builtin_ia32_reducess_mask): Ditto..
6632 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6633 * config/i386/sse.md (reduces<mode>): Renamed to ...
6634 (reduces<mode><mask_scalar_name>): ... this.
6635 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6637 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6638 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6640 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6642 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6643 * selftest-run-tests.c (selftest::run_tests): Call
6644 selftest::unique_ptr_tests_cc_tests.
6645 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6646 * unique-ptr-tests.cc: New file.
6648 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6651 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6653 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6654 (make_hard_regno_born, make_hard_regno_dead): Update
6655 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6657 2017-10-16 Jeff Law <law@redhat.com>
6659 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6661 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6663 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6665 2017-10-16 Olivier Hainque <hainque@adacore.com>
6667 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6668 with_cpu if we were configured for an e500v2 target cpu name.
6670 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6672 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6673 * doc/invoke.texi: Document +nodsp as a valid extension for
6676 2017-10-16 Martin Liska <mliska@suse.cz>
6678 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6679 (test_set_range): Likewise.
6680 (test_range_functions): Rename to ...
6681 (test_bit_in_range): ... this.
6682 (sbitmap_c_tests): Add new test.
6684 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6686 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6688 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6689 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6691 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6693 * config/aarch64/aarch64-builtins.c
6694 (aarch64_types_quadopu_lane_qualifiers): New.
6695 (TYPES_QUADOPU_LANE): New.
6696 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6697 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6698 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6699 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6700 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6701 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6702 (Vdottype, DOTPROD): New.
6703 (sur): Add SDOT and UDOT.
6705 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6707 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6708 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6709 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6711 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6712 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6713 Enable TARGET_DOTPROD.
6714 (cortex-a75.cortex-a55): Likewise.
6715 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6717 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6719 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6720 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6722 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6724 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6725 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6726 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6727 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6728 * config/arm/types.md (neon_dot, neon_dot_q): New.
6729 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6731 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6733 * config/arm/arm.h (TARGET_DOTPROD): New.
6734 * config/arm/arm.c (arm_arch_dotprod): New.
6735 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6736 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6737 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6738 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6739 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6740 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6741 * doc/invoke.texi (armv8.2-a): Document dotprod
6743 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6745 * i386.c (ix86_vec_cost): New function.
6746 (ix86_rtx_costs): Handle vector operations better.
6747 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6748 * x86-tune-costs.h: Add new costs to all tables.
6750 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6752 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6754 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6755 divsd, sqrtss and sqrtsd
6756 * x86-tune-costs.h: Add new entries to all costs.
6757 (znver1_cost): Fix to match real instruction latencies.
6759 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6760 Michael Collison <michael.collison@arm.com>
6762 * compare-elim.c: Include emit-rtl.h.
6763 (can_merge_compare_into_arith): New function.
6764 (try_validate_parallel): Likewise.
6765 (try_merge_compare): Likewise.
6766 (try_eliminate_compare): Call the above when no previous clobber
6768 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6771 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6775 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6776 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6780 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6781 to be any operand_equal_p operands. For & (B - 1) require
6782 B to be power of 2. Recognize
6783 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6785 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6788 * optabs.c (expand_memory_blockage): Fix call of
6789 targetm.have_memory_blockage.
6791 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6794 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6795 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6796 objects to extra_objs instead of overwriting it.
6798 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6800 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6801 Use any_fp_register_operand as operand[3] predicate. Simplify
6802 equality test for operands[2] and operands[4] memory location.
6803 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6804 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6805 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6806 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6807 any_fp_register_operand as operand[1] predicate. Simplify
6808 equality test for operands[0] and operands[3] memory location.
6809 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6810 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6811 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6813 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6815 * target-insns.def: Add memory_blockage.
6816 * optabs.c (expand_memory_blockage): New function.
6817 (expand_asm_memory_barrier): Rename ...
6818 (expand_asm_memory_blockage): ... to this.
6819 (expand_mem_thread_fence): Call expand_memory_blockage
6820 instead of expand_asm_memory_barrier.
6821 (expand_mem_singnal_fence): Ditto.
6822 (expand_atomic_load): Ditto.
6823 (expand_atomic_store): Ditto.
6824 * doc/md.texi (Standard Pattern Names For Generation):
6825 Document memory_blockage instruction pattern.
6827 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6829 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6830 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6831 * config/rl78/rl78.md: New define_expand "adddi3".
6833 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6835 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6838 2017-10-13 Jeff Law <law@redhat.com>
6840 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6842 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6845 * internal-fn.c (expand_mul_overflow): If both operands have
6846 the same highpart of -1 or 0 and the topmost bit of lowpart
6847 is different, overflow is if res <= 0 rather than res < 0.
6849 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6851 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6852 TARGET_P9_VECTOR code for unaligned_load case.
6854 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6856 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6858 2017-10-13 Nathan Sidwell <nathan@acm.org>
6860 * tree-core.h (tree_contains_struct): Make bool.
6861 * tree.c (tree_contains_struct): Likewise.
6862 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6863 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6864 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6865 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6866 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6868 2017-10-13 Richard Biener <rguenther@suse.de>
6870 * graphite-isl-ast-to-gimple.c
6871 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6872 parameters and dominance check.
6873 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6874 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6875 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6876 Do not update SSA form here or do intermediate IL verification.
6877 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6878 (graphite_initialize): Remove check on the number of loops in
6879 the function and inline into graphite_transform_loops.
6880 (graphite_finalize): Inline into graphite_transform_loops.
6881 (graphite_transform_loops): Perform SSA update and IL verification
6883 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6885 2017-10-13 Richard Biener <rguenther@suse.de>
6887 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6888 graphite_expression_type_precision): Avoid global constructor
6890 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6891 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6892 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6893 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6894 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6895 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6897 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6900 * config/i386/i386.h (ix86_red_zone_size): New.
6901 * config/i386/i386.md (push peephole2s): Replace
6902 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6904 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6905 Alan Hayward <alan.hayward@arm.com>
6906 David Sherwood <david.sherwood@arm.com>
6908 * combine.c (can_change_dest_mode): Reject changes in
6909 REGMODE_NATURAL_SIZE.
6911 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6912 Alan Hayward <alan.hayward@arm.com>
6913 David Sherwood <david.sherwood@arm.com>
6915 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6916 (expand_debug_source_expr): Likewise.
6917 * combine.c (combine_simplify_rtx): Likewise.
6918 * cse.c (fold_rtx): Likewise.
6919 * fwprop.c (canonicalize_address): Likewise.
6920 * targhooks.c (default_shift_truncation_mask): Likewise.
6922 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6923 Alan Hayward <alan.hayward@arm.com>
6924 David Sherwood <david.sherwood@arm.com>
6926 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6927 (widened_mode): Likewise.
6928 (expand_unop): Likewise.
6929 * ree.c (transform_ifelse): Likewise.
6930 (merge_def_and_ext): Likewise.
6931 (combine_reaching_defs): Likewise.
6932 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6934 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6935 Alan Hayward <alan.hayward@arm.com>
6936 David Sherwood <david.sherwood@arm.com>
6938 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6939 * combine.c (gen_lowpart_for_combine): Likewise.
6940 * dwarf2out.c (rtl_for_decl_location): Likewise.
6941 * final.c (alter_subreg): Likewise.
6942 * rtlhooks.c (gen_lowpart_general): Likewise.
6943 (gen_lowpart_if_possible): Likewise.
6945 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6946 Alan Hayward <alan.hayward@arm.com>
6947 David Sherwood <david.sherwood@arm.com>
6949 * calls.c (expand_call): Use subreg_lowpart_offset.
6950 * cse.c (cse_insn): Likewise.
6951 * regcprop.c (copy_value): Likewise.
6952 (copyprop_hardreg_forward_1): Likewise.
6954 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6957 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6958 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6959 =Q constraints to +Q and into insn condition add check
6960 that operands[0] and operands[1] are equal.
6961 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6962 =Q constraints to +Q and into insn condition add check
6963 that operands[0] is equal to either operands[1] or operands[2].
6966 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6967 instead of handling MINUS_EXPR twice (once for each argument),
6968 canonicalize operand order and handle just once, use rtype where
6969 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6972 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6973 any values of __C while still being pattern recognizable as a simple
6976 2017-10-13 Richard Biener <rguenther@suse.de>
6978 PR tree-optimization/82451
6980 2017-10-02 Richard Biener <rguenther@suse.de>
6982 PR tree-optimization/82355
6983 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6984 a mapping for the enclosing loop but avoid generating one for
6986 (copy_bb_and_scalar_dependences): Remove premature codegen
6987 error on PHIs in blocks duplicated into multiple places.
6988 * graphite-scop-detection.c
6989 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6990 in the region use it as loop and nest to analyze the DR in.
6991 (try_generate_gimple_bb): Likewise.
6992 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6993 (add_loop_constraints): For blocks in a loop not in the region
6994 create a dimension with a single iteration.
6995 * sese.h (gbb_loop_at_index): Remove assert.
6997 * cfgloop.c (loop_preheader_edge): For the loop tree root
6998 return the single successor of the entry block.
6999 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7000 Reset the SCEV hashtable and niters.
7001 * graphite-scop-detection.c
7002 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7003 assert that we only have POLYNOMIAL_CHREC that vary in loops
7004 contained in the region.
7005 (scop_detection::graphite_can_represent_expr): Adjust.
7006 (scop_detection::stmt_has_simple_data_refs_p): For loops
7007 not in the region set loop to NULL. The nest is now the
7008 entry edge to the region.
7009 (try_generate_gimple_bb): Likewise.
7010 * sese.c (scalar_evolution_in_region): Adjust for
7011 instantiate_scev change.
7012 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7013 Make nest parameter the edge into the region.
7014 (create_data_ref): Likewise.
7015 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7016 entry edge into a region and adjust instantiate_scev calls.
7017 (create_data_ref): Likewise.
7018 (graphite_find_data_references_in_stmt): Likewise.
7019 (find_data_references_in_stmt): Pass the loop preheader edge
7020 from the nest argument.
7021 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7022 parameter the edge into the region.
7023 (instantiate_parameters): Use the loop preheader edge as entry.
7024 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7026 (get_instantiated_value_entry): Make instantiate_below parameter
7027 the edge into the region.
7028 (instantiate_scev_name): Likewise. Adjust dominance checks,
7029 when we cannot use loop-based instantiation instantiate by
7030 walking use-def chains.
7031 (instantiate_scev_poly): Adjust.
7032 (instantiate_scev_binary): Likewise.
7033 (instantiate_scev_convert): Likewise.
7034 (instantiate_scev_not): Likewise.
7035 (instantiate_array_ref): Remove.
7036 (instantiate_scev_3): Likewise.
7037 (instantiate_scev_2): Likewise.
7038 (instantiate_scev_1): Likewise.
7039 (instantiate_scev_r): Do not blindly handle N-operand trees.
7040 Do not instantiate array-refs. Handle all constants and invariants.
7041 (instantiate_scev): Make instantiate_below parameter
7042 the edge into the region.
7043 (resolve_mixers): Use the loop preheader edge for the region
7044 parameter to instantiate_scev_r.
7045 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7047 2017-10-13 Richard Biener <rguenther@suse.de>
7049 PR tree-optimization/82525
7050 * graphite-isl-ast-to-gimple.c
7051 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7053 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7054 Fail code generation when we cannot represent the isl integer.
7055 (binary_op_to_tree): Elide modulo operations that are no-ops
7056 in the type we code generate. Remove now superfluous code
7059 2017-10-13 Richard Biener <rguenther@suse.de>
7061 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7062 (scop_detection::harmful_loop_in_region): Remove premature
7063 IV type restriction.
7064 (scop_detection::graphite_can_represent_scev): We can handle
7065 pointer IVs just fine.
7067 2017-10-13 Alan Modra <amodra@gmail.com>
7069 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7070 "Clobbers and Scratch Registers". Add paragraph on
7071 alternative to clobbers for scratch registers and OpenBLAS
7074 2017-10-13 Alan Modra <amodra@gmail.com>
7076 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7077 example of a memory input for a string of known length. Move
7078 commentary out of table. Add a number of new examples
7079 covering array memory inputs.
7081 2017-10-12 Martin Liska <mliska@suse.cz>
7083 PR tree-optimization/82493
7084 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7085 (test_range_functions): New function.
7086 (sbitmap_c_tests): Likewise.
7087 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7088 * selftest.h (sbitmap_c_tests): New function.
7090 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7092 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7094 * config/rs6000/amo.h: Fix spacing issue.
7096 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7099 * config/i386/i386.md (*ashl<mode>3_mask_1,
7100 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7101 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7104 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7106 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7107 (profile_probability): Set max_probability
7108 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7110 * profile-count.c (profile_probability::differs_from_p): Do not
7111 rely on max_probaiblity == 10000
7113 2017-10-12 Jeff Law <law@redhat.com>
7115 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7118 2017-10-12 Martin Sebor <msebor@redhat.com>
7122 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7123 (handle_alias_pairs): Call it.
7124 * common.opt (-Wattribute-alias): New option.
7125 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7126 * doc/invoke.texi (-Wattribute-alias): Document.
7128 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7131 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7133 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7135 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7136 (make_hard_regno_born, make_hard_regno_dead): Update
7137 bb_killed_pseudos and bb_gen_pseudos.
7139 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7141 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7143 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7145 * config/alpha/alpha.c (alpha_split_conditional_move):
7146 Use std::swap instead of manually swapping.
7147 (alpha_stdarg_optimize_hook): Ditto.
7148 (alpha_canonicalize_comparison): Ditto.
7150 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7152 * tree-loop-distribution.c (struct builtin_info): New struct.
7153 (struct partition): Refactor fields into struct builtin_info.
7154 (partition_free): Free struct builtin_info.
7155 (build_size_arg_loc, build_addr_arg_loc): Delete.
7156 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7157 information from struct builtin_info.
7158 (find_single_drs): New function refactored from classify_partition.
7159 Also moved builtin validity checks to this function.
7160 (compute_access_range, alloc_builtin): New functions.
7161 (classify_builtin_st, classify_builtin_ldst): New functions.
7162 (classify_partition): Refactor code into functions find_single_drs,
7163 classify_builtin_st and classify_builtin_ldst.
7164 (distribute_loop): Don't do runtime alias check when distributing
7166 (find_seed_stmts_for_distribution): New function.
7167 (pass_loop_distribution::execute): Refactor code finding seed
7168 stmts into above function. Support distribution for the innermost
7169 two-level loop nest. Adjust dump information.
7171 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7173 * tree-loop-distribution.c: Adjust the general comment.
7174 (NUM_PARTITION_THRESHOLD): New macro.
7175 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7176 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7177 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7178 in call to build_partition_graph.
7179 (finalize_partitions): New parameter. Make loop distribution more
7180 conservative by fusing more partitions.
7181 (distribute_loop): Don't do runtime alias check in case of loop nest
7183 (find_seed_stmts_for_distribution): New function.
7184 (prepare_perfect_loop_nest): New function.
7185 (pass_loop_distribution::execute): Refactor code finding seed stmts
7186 and loop nest into above functions. Support loop nest distribution.
7187 Adjust dump information accordingly.
7189 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7191 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7192 and set PTYPE_SEQUENTIAL for merged partition.
7194 2017-10-12 Richard Biener <rguenther@suse.de>
7196 PR tree-optimization/69728
7198 2017-09-19 Richard Biener <rguenther@suse.de>
7200 PR tree-optimization/69728
7201 * graphite-sese-to-poly.c (schedule_error): New global.
7202 (add_loop_schedule): Handle empty domain by failing the
7204 (build_original_schedule): Handle schedule_error.
7206 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7207 domain by returning an unchanged schedule.
7209 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7211 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7212 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7214 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7216 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7219 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7222 * expr.c (store_field): Don't optimize away bitsize == 0 store
7223 from CALL_EXPR with addressable return type.
7225 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7227 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7228 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7229 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7230 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7231 TARGET_ISEL instead of TARGET_ISEL<sel>.
7233 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7235 * config/rs6000/rs6000.c
7236 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7238 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7240 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7242 (reg_or_cint_operand): Fix comment.
7243 (reg_or_zero_operand): New predicate.
7244 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7245 * config/rs6000/rs6000.c (output_isel): Delete.
7246 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7247 instead of reg_or_cint_operand. Output instruction directly (not via
7249 (isel_unsigned_<mode>): Ditto.
7250 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7251 gpc_reg_operand. Add an instruction alternative for this. Output
7252 instruction directly.
7253 (*isel_reversed_unsigned_<mode>): Ditto.
7255 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7257 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7258 (TARGET_CANONICALIZE_COMPARISON): Define.
7260 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7263 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7264 Check whether the dest is REG before adding REG_EQUIV note.
7266 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7269 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7271 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7272 (make_hard_regno_born, make_hard_regno_dead): Update
7273 bb_killed_pseudos and bb_gen_pseudos.
7275 2017-10-11 Nathan Sidwell <nathan@acm.org>
7277 * incpath.h (enum incpath_kind): Name enum, prefix values.
7278 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7279 * incpath.c (heads, tails): Use INC_MAX.
7280 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7281 (merge_include_chains, split_quote_chain,
7282 register_include_chains): Update incpath_kind names.
7283 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7284 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7286 (add_framework_path, darwin_register_objc_includes): Likewise.
7287 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7289 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7291 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7292 Do not use float_operator operator predicate.
7293 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7294 * config/i386/predicates.md (float_operator): Remove predicate.
7296 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7298 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7299 (*jccxf_i387): Ditto.
7300 (*jcc<mode>_i387): Ditto.
7301 (*jccu<mode>_i387): Ditto.
7302 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7303 (*jcc_*_i387 splitters): Remove.
7304 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7305 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7306 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7309 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7311 * profile-count.h (slow_safe_scale_64bit): New function.
7312 (safe_scale_64bit): New inline.
7313 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7314 * profile-count.c: Include wide-int.h
7315 (slow_safe_scale_64bit): New.
7317 2017-10-11 Nathan Sidwell <nathan@acm.org>
7319 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7320 HAS_DECL_ASSEMBLER_NAME_P.
7321 * gimple-expr.c (gimple_decl_printable_name: Check
7322 HAS_DECL_ASSEMBLER_NAME_P too.
7323 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7324 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7325 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7326 * passes.c (rest_of_decl_compilation): Check
7327 HAS_DECL_ASSEMBLER_NAME_P too.
7328 * recog.c (verify_changes): Likewise.
7329 * tree-pretty-print.c (dump_decl_name): Likewise.
7330 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7332 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7333 (SET_DECL_ASSEMBLER_NAME): Use it.
7334 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7335 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7336 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7338 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7340 * config.gcc (i386, x86_64): Add extra objects.
7341 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7342 (ix86_min_insn_size): Declare.
7343 (ix86_issue_rate): Declare.
7344 (ix86_adjust_cost): Declare.
7345 (ia32_multipass_dfa_lookahead): Declare.
7346 (ix86_macro_fusion_p): Declare.
7347 (ix86_macro_fusion_pair_p): Declare.
7348 (ix86_bd_has_dispatch): Declare.
7349 (ix86_bd_do_dispatch): Declare.
7350 (ix86_core2i7_init_hooks): Declare.
7351 (ix86_atom_sched_reorder): Declare.
7352 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7353 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7354 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7355 (rip_relative_addr_p): Rename to ...
7356 (ix86_rip_relative_addr_p): ... this one; export.
7357 (memory_address_length): Update.
7358 (ix86_issue_rate): Move to x86-tune-sched.c.
7359 (ix86_flags_dependent): Move to x86-tune-sched.c.
7360 (ix86_agi_dependent): Move to x86-tune-sched.c.
7361 (exact_dependency_1): Move to x86-tune-sched.c.
7362 (exact_store_load_dependency): Move to x86-tune-sched.c.
7363 (ix86_adjust_cost): Move to x86-tune-sched.c.
7364 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7365 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7366 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7367 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7368 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7369 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7370 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7371 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7372 (min_insn_size): Rename to ...
7373 (ix86_min_insn_size): ... this one; export.
7374 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7375 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7376 (core2i7_first_cycle_multipass_backtrack): Move to
7377 x86-tune-sched-core.c.
7378 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7379 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7380 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7381 (ix86_avoid_jump_mispredicts): Update.
7382 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7383 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7384 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7385 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7386 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7387 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7388 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7389 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7390 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7391 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7392 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7393 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7394 (BIG): Move to ix86-tune-sched-bd.c.
7395 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7396 (enum insn_path): Move to ix86-tune-sched-bd.c.
7397 (get_mem_group): Move to ix86-tune-sched-bd.c.
7398 (is_cmp): Move to ix86-tune-sched-bd.c.
7399 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7400 (is_branch): Move to ix86-tune-sched-bd.c.
7401 (is_prefetch): Move to ix86-tune-sched-bd.c.
7402 (init_window): Move to ix86-tune-sched-bd.c.
7403 (allocate_window): Move to ix86-tune-sched-bd.c.
7404 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7405 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7406 (process_end_window): Move to ix86-tune-sched-bd.c.
7407 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7408 (find_constant): Move to ix86-tune-sched-bd.c.
7409 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7410 (has_immediate): Move to ix86-tune-sched-bd.c.
7411 (get_insn_path): Move to ix86-tune-sched-bd.c.
7412 (get_insn_group): Move to ix86-tune-sched-bd.c.
7413 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7414 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7415 (add_insn_window): Move to ix86-tune-sched-bd.c.
7416 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7417 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7418 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7419 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7420 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7421 (do_dispatch): Move to ix86-tune-sched-bd.c.
7422 (has_dispatch): Move to ix86-tune-sched-bd.c.
7423 * config/i386/t-i386: Add new object files.
7424 * config/i386/x86-tune-costs.h: New file.
7425 * config/i386/x86-tune-sched-atom.c: New file.
7426 * config/i386/x86-tune-sched-bd.c: New file.
7427 * config/i386/x86-tune-sched-core.c: New file.
7428 * config/i386/x86-tune-sched.c: New file.
7430 2017-10-11 Liu Hao <lh_mouse@126.com>
7432 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7433 the generic version below instead.
7434 (should_colorize): Recognize Windows consoles as terminals
7436 * pretty-print.c [__MINGW32__] (write_all): New function.
7437 [__MINGW32__] (find_esc_head): Likewise.
7438 [__MINGW32__] (find_esc_terminator): Likewise.
7439 [__MINGW32__] (eat_esc_sequence): Likewise.
7440 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7442 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7445 2017-10-11 Richard Biener <rguenther@suse.de>
7447 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7448 Properly call analyze_scalar_evolution with the loop of the stmt.
7450 2017-10-11 Richard Biener <rguenther@suse.de>
7452 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7453 * tree-core.h (tree_base): Add chrec_var union member.
7454 * tree.h (CHREC_VAR): Remove.
7455 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7456 * tree-chrec.h (build_polynomial_chrec): Adjust.
7457 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7458 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7460 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7462 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7463 * match.pd: ... here.
7464 ((T) X == (T) Y): Relax condition.
7466 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7468 PR tree-optimization/82472
7469 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7471 (break_alias_scc_partitions): Update postorder number.
7473 2017-10-11 Martin Liska <mliska@suse.cz>
7476 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7478 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7480 2017-10-11 Martin Liska <mliska@suse.cz>
7482 * print-rtl.c (print_insn): Move declaration of idbuf
7483 to same scope as name.
7485 2017-10-11 Martin Liska <mliska@suse.cz>
7490 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7493 2017-10-11 Martin Liska <mliska@suse.cz>
7496 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7499 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7501 * config.gcc (default_gnu_indirect_function): Default to yes for
7502 arm*-*-linux* with glibc.
7504 2017-10-11 Richard Biener <rguenther@suse.de>
7506 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7507 default-defs and types we do not want to analyze.
7508 (interpret_loop_phi): Replace unreachable code with an assert.
7509 (compute_scalar_evolution_in_loop): Remove and inline ...
7510 (analyze_scalar_evolution_1): ... here, replacing condition with
7511 what makes the intent clearer. Remove handling of cases
7512 get_scalar_evolution now handles.
7514 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7516 PR rtl-optimization/81434
7517 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7518 comment for main loop. In sched_group_found if, also add checks for
7519 pass and min_cost_group.
7521 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7523 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7524 (rs6000_insn_cost): New function.
7525 * config/rs6000/rs6000.md (cost): New attribute.
7527 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7528 H.J. Lu <hongjiu.lu@intel.com>
7532 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7533 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7534 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7535 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7536 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7537 OPTION_MASK_ISA_MMX and some other bit set require both
7538 mmx and the ISAs without the mmx bit.
7539 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7540 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7541 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7542 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7543 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7544 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7545 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7546 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7547 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7548 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7549 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7550 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7551 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7552 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7553 Add OPTION_MASK_ISA_MMX.
7555 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7557 * config.gcc (armv7*-*-freebsd*): New target.
7558 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7560 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7562 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7563 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7566 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7568 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7569 a template parameter.
7570 (WIDE_INT_REF_FOR): Update accordingly.
7571 * tree.h (wi::int_traits <const_tree>): Delete.
7572 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7573 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7574 (wi::tree_to_wide_ref): New typedef.
7575 (wi::to_wide): New function.
7576 * calls.c (get_size_range): Use wi::to_wide when operating on
7578 * cgraph.c (cgraph_node::create_thunk): Likewise.
7579 * config/i386/i386.c (ix86_data_alignment): Likewise.
7580 (ix86_local_alignment): Likewise.
7581 * dbxout.c (stabstr_O): Likewise.
7582 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7583 * expr.c (const_vector_from_tree): Likewise.
7584 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7585 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7586 (fold_negate_expr_1, int_const_binop_1, const_binop)
7587 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7588 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7589 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7590 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7591 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7592 (fold_not_const, round_up_loc): Likewise.
7593 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7594 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7595 (alloca_call_type): Likewise.
7596 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7597 * godump.c (go_output_typedef): Likewise.
7598 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7599 * internal-fn.c (get_min_precision): Likewise.
7600 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7601 * ipa-polymorphic-call.c
7602 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7603 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7604 (ipa_modify_call_arguments): Likewise.
7605 * match.pd: Likewise.
7606 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7607 * print-tree.c (print_node_brief, print_node): Likewise.
7608 * stmt.c (expand_case): Likewise.
7609 * stor-layout.c (layout_type): Likewise.
7610 * tree-affine.c (tree_to_aff_combination): Likewise.
7611 * tree-cfg.c (group_case_labels_stmt): Likewise.
7612 * tree-data-ref.c (dr_analyze_indices): Likewise.
7613 (prune_runtime_alias_test_list): Likewise.
7614 * tree-dump.c (dequeue_and_dump): Likewise.
7615 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7616 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7617 * tree-pretty-print.c (dump_generic_node): Likewise.
7618 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7619 (simple_iv_with_niters): Likewise.
7620 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7621 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7622 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7623 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7624 (refine_value_range_using_guard, number_of_iterations_ne_max)
7625 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7626 (get_cst_init_from_scev, record_nonwrapping_iv)
7627 (scev_var_range_cant_overflow): Likewise.
7628 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7629 * tree-ssa-pre.c (compute_avail): Likewise.
7630 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7631 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7632 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7633 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7634 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7635 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7636 (dump_case_nodes, try_switch_expansion): Likewise.
7637 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7638 (vect_do_peeling): Likewise.
7639 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7640 * tree-vect-stmts.c (vectorizable_load): Likewise.
7641 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7642 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7643 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7644 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7645 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7646 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7647 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7648 (evrp_dom_walker::before_dom_children): Likewise.
7649 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7650 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7651 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7652 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7653 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7654 (get_range_pos_neg): Likewise.
7655 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7656 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7657 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7658 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7659 * config/avr/avr.c (avr_fold_builtin): Likewise.
7660 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7661 * config/msp430/msp430.c (msp430_attr): Likewise.
7662 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7663 * config/powerpcspe/powerpcspe-c.c
7664 (altivec_resolve_overloaded_builtin): Likewise.
7665 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7666 (rs6000_expand_ternop_builtin): Likewise.
7667 * config/rs6000/rs6000-c.c
7668 (altivec_resolve_overloaded_builtin): Likewise.
7669 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7670 (rs6000_expand_ternop_builtin): Likewise.
7671 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7673 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7675 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7676 when copying loop nest with only one inner loop.
7678 2017-10-10 Richard Biener <rguenther@suse.de>
7680 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7681 blocks if SCEV is active.
7682 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7684 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7685 (scev_initialize): Assert we are not yet initialized.
7687 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7689 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7690 inner loop's exit stmt by making it always exit the loop, otherwise
7691 we would generate an infinite empty loop.
7693 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7695 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7696 renaming variables in new preheader if it's deleted.
7698 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7700 * tree-loop-distribution.c (struct partition): Remove unused field
7701 loops of the structure.
7702 (partition_alloc, partition_free): Ditto.
7703 (build_rdg_partition_for_vertex): Ditto.
7705 2017-10-09 Jeff Law <law@redhat.com>
7707 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7708 return type to match prototype and documentation.
7710 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7712 * config/rs6000/rs6000.c (processor_costs): Move to ...
7713 * config/rs6000/rs6000.h: ... here.
7714 (rs6000_cost): Declare.
7716 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7718 * except.c (setjmp_fn): New global variable.
7719 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7720 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7721 if DONT_USE_BUILTIN_SETJMP is defined.
7723 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7725 * target.def (insn_cost): New hook.
7726 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7727 * doc/tm.texi: Regenerate.
7728 * rtlanal.c (insn_cost): Use the new hook.
7730 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7732 * combine.c (combine_validate_cost): Compute the new insn_cost,
7733 not just pattern_cost.
7734 (try_combine): Adjust comment.
7736 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7738 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7740 * combine.c (uid_insn_cost): Adjust comment.
7741 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7743 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7744 * dse.c (find_shift_sequence): Ditto.
7745 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7746 (bb_valid_for_noce_process_p): Use pattern_cost.
7747 * rtl.h (insn_rtx_cost): Delete.
7748 (pattern_cost): New prototype.
7749 (insn_cost): New prototype.
7750 * rtlanal.c (insn_rtx_cost): Rename to...
7751 (pattern_cost): ... this.
7754 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7756 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7757 (*jcc<mode>_0_r_i387): Ditto.
7758 (*jccxf_r_i387): Ditto.
7759 (*jcc<mode>_r_i387): Ditto.
7760 (*jccu<mode>_r_i387): Ditto.
7761 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7762 (*jcc): Rename from *jcc_1.
7764 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7766 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7767 deferred rescans after the lvx/stvx recombination pre-pass.
7769 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7771 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7772 memory operation instruction support.
7773 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7774 (rs6000-ibm-aix[789]*): Likewise.
7775 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7776 Document new functions.
7778 2017-10-09 Richard Biener <rguenther@suse.de>
7780 PR tree-optimization/82397
7781 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7782 equality only for semantically equal trees.
7784 2017-10-09 Richard Biener <rguenther@suse.de>
7786 PR tree-optimization/82449
7787 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7788 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7789 allow constant addresses.
7790 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7793 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7795 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7798 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7801 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7804 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7807 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7809 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7812 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7813 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7815 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7817 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7818 (WI_BINARY_PREDICATE_RESULT): Likewise.
7819 (wi::binary_traits::operator_result): New type.
7820 (wi::binary_traits::predicate_result): Likewise.
7821 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7822 (generic_wide_int::operator==, generic_wide_int::operator!=)
7823 (generic_wide_int::operator&, generic_wide_int::and_not)
7824 (generic_wide_int::operator|, generic_wide_int::or_not)
7825 (generic_wide_int::operator^, generic_wide_int::operator+
7826 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7828 (operator~, unary operator-, operator==, operator!=, operator&)
7829 (operator|, operator^, operator+, binary operator-, operator*): New
7831 * expr.c (get_inner_reference): Use wi::bit_and_not.
7832 * fold-const.c (fold_binary_loc): Likewise.
7833 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7834 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7835 (bit_value_binop): Likewise.
7836 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7837 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7838 (extract_range_from_binary_expr_1): Likewise.
7839 (masked_increment): Likewise.
7840 (simplify_bit_ops_using_ranges): Likewise.
7842 2017-10-09 Martin Jambor <mjambor@suse.cz>
7845 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7846 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7847 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7849 (hsa_fixup_mov_insn_type): New function.
7850 (hsa_op_with_type::get_in_type): Use it.
7851 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7852 immediates in an assert.
7853 (hsa_op_with_type::extend_int_to_32bit): New method.
7854 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7855 types. Convert to dest type if necessary.
7856 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7857 (reg_for_gimple_ssa): Pass false as min32int to
7858 hsa_type_for_scalar_tree_type.
7859 (gen_hsa_addr): Fixup type when creating addresable temporary.
7860 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7861 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7862 necessary. Call hsa_fixup_mov_insn_type.
7863 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7864 extend operands and convert to dest type if necessary.
7865 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7866 to dest type if necessary.
7867 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7868 if conversion nt necessary and size matches.
7869 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7870 to dest type if necessary.
7871 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7872 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7874 (gen_hsa_clrsb): Likewise.
7875 (gen_hsa_ffs): Likewise.
7876 (gen_hsa_divmod): Extend operands and convert to dest type if
7878 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7880 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7882 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7883 Remove empty default arguments. Use a brace block as output
7885 (conditional return): Ditto.
7887 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7888 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7890 (group_ending_nop): Ditto.
7891 (doloop_end): Ditto.
7892 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7893 (splitters for those): Ditto.
7895 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7897 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7898 a conditional jump (and the compare for it) so that pc_rtx is the
7900 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7901 for the deleted and renamed ctr<mode>_internal[234] patterns.
7902 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7903 Delete second conditional return pattern.
7904 (ctr<mode>_internal2): Delete this second bdnz pattern.
7905 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7906 (ctr<mode>_internal4): Delete this second bdz pattern.
7908 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7910 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7911 (always_initialized_rtx_for_ssa_name_p): New predicate.
7912 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7913 (finish_out_of_ssa): Free new field of SA.
7914 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7915 * tree-ssa-coalesce.c: Include tree-ssa.h.
7916 (get_parm_default_def_partitions): Remove extern keyword.
7917 (get_undefined_value_partitions): New function.
7918 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7919 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7921 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7923 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7925 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7927 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7929 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7930 for targets that preffer 128bit.
7932 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7934 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7936 2017-10-08 Olivier Hainque <hainque@adacore.com>
7938 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7939 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7941 (thumb_set_return_address): Likewise.
7943 2017-10-08 Olivier Hainque <hainque@adacore.com>
7945 * common/config/arm/arm-common.c (arm_except_unwind_info):
7946 Handle DWARF2_UNWIND_INFO.
7948 2017-10-07 Michael Collison <michael.collison@arm.com>
7950 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7953 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7955 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7956 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7957 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7958 defined, force the creation of a new block for a dispatch label.
7960 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7962 * invoke.texi (Wsuggest-attribute=cold): Document.
7963 * common.opt (Wsuggest-attribute=cold): New
7964 * ipa-pure-const.c (warn_function_cold): New function.
7965 * predict.c (compute_function_frequency): Use it.
7966 * predict.h (warn_function_cold): Declare.
7968 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7970 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7973 2017-10-06 Martin Liska <mliska@suse.cz>
7975 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7976 keyword for member functions.
7977 (struct sanopt_tree_couple): New struct.
7978 (struct sanopt_tree_couple_hash): New function.
7979 (struct sanopt_ctx): Add new hash_map.
7980 (has_dominating_ubsan_ptr_check): New function.
7981 (record_ubsan_ptr_check_stmt): Likewise.
7982 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7983 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7984 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7986 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7989 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7990 aarch64_simd_valid_immediate on CONST_VECTORs.
7991 (aarch64_reg_or_bic_imm): Likewise.
7993 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7995 PR rtl-optimization/82396
7996 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7998 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8000 * graphite-dependences.c (scop_get_reads): Move code to...
8001 (scop_get_must_writes): Move code to...
8002 (scop_get_may_writes): Move code to...
8003 (scop_get_reads_and_writes): ... here.
8004 (scop_get_dependences): Call scop_get_reads_and_writes.
8006 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8008 PR tree-optimization/82434
8009 * fold-const.h (can_native_encode_type_p,
8010 can_native_encode_string_p): Remove.
8011 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8012 don't encode anything, just return what would be otherwise returned.
8013 (native_encode_fixed, native_encode_complex, native_encode_vector):
8015 (native_encode_string): Likewise. Inline by hand
8016 can_native_encode_string_p.
8017 (can_native_encode_type_p): Remove.
8018 (can_native_encode_string_p): Remove.
8019 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8020 STRING_CSTs using can_native_encode_string_p, test all
8021 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8022 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8023 argument from native_encode_expr.
8024 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8025 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8028 2017-10-06 Richard Biener <rguenther@suse.de>
8030 PR tree-optimization/82397
8031 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8032 operand_equal_p but rely on data_ref_compare_tree for detecting
8034 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8035 to match up with dr_group_sort_cmp.
8037 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8040 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8042 * config/s390/s390-builtin-types.def: Regenerate.
8044 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8047 * config/s390/s390-builtin-types.def: Regenerate.
8048 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8049 Change flag from B_VXE to B_VX.
8050 (s390_vec_min_dbl): Remove B_VXE flag.
8052 2017-10-06 Richard Biener <rguenther@suse.de>
8054 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8055 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8056 translate_isl_ast_to_gimple::is_valid_rename,
8057 translate_isl_ast_to_gimple::get_rename,
8058 translate_isl_ast_to_gimple::get_def_bb_for_const,
8059 translate_isl_ast_to_gimple::get_new_name,
8060 translate_isl_ast_to_gimple::collect_all_ssa_names,
8061 translate_isl_ast_to_gimple::copy_loop_phi_args,
8062 translate_isl_ast_to_gimple::collect_all_ssa_names,
8063 translate_isl_ast_to_gimple::copy_loop_phi_args,
8064 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8065 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8066 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8067 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8068 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8069 translate_isl_ast_to_gimple::copy_cond_phi_args,
8070 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8071 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8072 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8073 translate_isl_ast_to_gimple::rename_uses,
8074 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8075 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8076 (set_rename_for_each_def): Likewise.
8077 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8078 here. Handle rewriting SCEV analyzable uses here.
8079 (copy_bb_and_scalar_dependences): Generate code for PHI
8081 (graphite_regenerate_ast_isl): Adjust.
8082 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8083 (add_write, add_read): New functions.
8084 (build_cross_bb_scalars_def): Use it and simplify.
8085 (build_cross_bb_scalars_use): Likewise.
8086 (graphite_find_cross_bb_scalar_vars): Inline into...
8087 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8088 simulating out-of-SSA. Compute liveout and add dependencies.
8089 (build_scops): Force an empty entry block.
8090 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8092 (sese_build_liveouts): Declare.
8093 (sese_trivially_empty_bb_p): Likewise.
8094 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8095 compute liveout and debug_liveout.
8096 (sese_bad_liveouts_use): Remove.
8097 (sese_reset_debug_liveouts_bb): Likewise.
8098 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8099 (sese_build_liveouts): Build liveout and debug_liveout and store
8101 (new_sese_info): Adjust.
8102 (free_sese_info): Likewise.
8103 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8104 do not build liveout here.
8105 (move_sese_in_condition): Adjust region entry.
8106 (scev_analyzable_p): Match up with chrec_apply requirements.
8107 (sese_trivially_empty_bb_p): New.
8108 * tree-into-ssa.c (get_reaching_def): Properly support generating
8109 default-defs for incremental rewrite of anonymous names.
8111 2017-10-06 Richard Biener <rguenther@suse.de>
8113 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8114 precision do not perform modulo reduction.
8116 2017-10-06 Richard Biener <rguenther@suse.de>
8118 PR tree-optimization/82436
8119 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8120 conservatively choose the vectorization factor when checking
8121 whether we can perform the required load permutation.
8122 (vect_transform_slp_perm_load): Assert when we may not fail.
8124 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8126 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8127 message for incompatible -msdata=* and -mcall-* options.
8129 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8131 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8132 rate for post-reload scheduling.
8134 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8136 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8138 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8140 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8141 to improve monte carlo in scimark.
8143 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8145 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8146 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8147 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8148 pentium4_cost, nocona_cost): Set reassociation width to 1.
8149 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8150 width to 2 for fp operations and 1 otherwise.
8151 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8153 (atom_cost): Set reassociation width to 2.
8154 (slm_cost, generic_cost): Set fp reassociation width
8155 to 2 and 1 otherwise.
8156 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8157 (core_cost): Set fp reassociation width to 4 and vector to 2.
8158 (ix86_reassociation_width): Rewrite using cost table; special case
8159 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8160 and TARGET_AVX128_OPTIMAL.
8161 * config/i386/i386.h (processor_costs): Add
8162 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8163 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8164 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8165 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8166 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8167 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8169 2017-10-05 Nathan Sidwell <nathan@acm.org>
8171 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8173 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8175 * config/arm/arm.c (arm_test_fpu_data): New.
8176 (arm_run_selftests): Call arm_test_fpu_data.
8178 2017-10-04 Nathan Sidwell <nathan@acm.org>
8180 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8182 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8184 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8186 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8187 check type for aarch64_simd_valid_immediate.
8188 (aarch64_output_simd_mov_immediate): Update prototype.
8189 (aarch64_simd_valid_immediate): Update prototype.
8190 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8191 support for ORR-immediate.
8192 (and<mode>3): modified pattern to add support for BIC-immediate.
8193 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8194 now checks for valid immediate for BIC and ORR based on new enum
8196 (aarch64_output_simd_mov_immediate): Function now used to output
8197 BIC/ORR imm as well based on new enum argument.
8198 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8200 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8201 (aarch64_reg_or_bic_imm): Likewise.
8203 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8205 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8206 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8208 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8212 PR rtl-optimization/82396
8213 * haifa-sched.c (autopref_multipass_init): Simplify
8215 (autopref_rank_data): Simplify sort order.
8216 * sched-int.h (autopref_multipass_data_): Remove
8217 multi_mem_insn_p, min_offset and max_offset.
8219 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8221 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8223 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8225 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8228 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8230 * doc/sourcebuild.texi: Document vect_long_mult.
8232 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8234 PR tree-optimization/82413
8235 * fold-const.c (build_range_check): Use widest_int when comparing
8236 the maximum ETYPE value with HIGH.
8238 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 Jakub Jelinek <jakub@redhat.com>
8249 PR tree-optimization/82381
8250 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8251 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8254 PR tree-optimization/82374
8255 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8256 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8257 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8258 current_function_decl to the new decl.
8260 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8262 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8263 helper macro for IEEE float128 hardware built-in functions.
8264 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8266 (TRUNCF128_ODD): Likewise.
8267 (ADDF128_ODD): Likewise.
8268 (SUBF128_ODD): Likewise.
8269 (MULF128_ODD): Likewise.
8270 (DIVF128_ODD): Likewise.
8271 (FMAF128_ODD): Likewise.
8272 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8273 UNSPEC_TRUNC_ROUND_TO_ODD.
8274 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8275 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8276 floating point round to odd instructions.
8277 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8278 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8279 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8280 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8281 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8282 (trunc<mode>sf2_hw): Change the truncate with round to odd
8283 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8284 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8285 to odd hardware instructions.
8286 (sub<mode>3_odd): Likewise.
8287 (mul<mode>3_odd): Likewise.
8288 (div<mode>3_odd): Likewise.
8289 (sqrt<mode>2_odd): Likewise.
8290 (fma<mode>4_odd): Likewise.
8291 (fms<mode>4_odd): Likewise.
8292 (nfma<mode>4_odd): Likewise.
8293 (nfms<mode>4_odd): Likewise.
8294 (trunc<mode>df2_odd): Change the truncate with round to odd
8295 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8297 * doc/extend.texi (PowerPC built-in functions): Update documentation
8298 for existing IEEE float128-bit built-in functions. Add built-in
8299 functions that generate the IEEE 128-bit floating point round to
8302 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8304 PR rtl-optimization/77729
8305 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8306 to (X&(C1&~C2))|C2 transformations.
8308 2017-10-03 Martin Jambor <mjambor@suse.cz>
8310 PR tree-optimization/82363
8311 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8312 mismatch, mark lacc written regardless of racc.
8314 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8316 PR tree-optimization/82381
8317 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8318 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8319 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8320 fallthrough into reassoc_stmt_dominates_stmt_p.
8323 * combine.c (combine_instructions): Don't combine in unreachable
8326 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8330 function to not use the have_cpu variable. Do not set cpu_index,
8331 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8333 (rs6000_valid_attribute_p): Remove duplicate initializations of
8334 old_optimize and func_optimize.
8335 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8336 (rs6000_activate_target_options): Make global.
8337 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8340 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8342 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8343 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8344 Set *poffset to 0 and *psize and *pmax_size to -1 if
8345 *poffset + *psize overflows in HOST_WIDE_INT.
8347 PR tree-optimization/82387
8348 PR tree-optimization/82388
8349 PR tree-optimization/82389
8350 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8351 instead of live_bytes non-NULL.
8353 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8356 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8359 2017-10-02 Richard Biener <rguenther@suse.de>
8361 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8362 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8363 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8365 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8367 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8368 requested precision matches the type's.
8369 * calls.c (alloc_max_size): Calculate the new candidate size as
8370 a widest_int and use wi::to_widest when comparing it with the
8371 current candidate size.
8372 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8373 zero rather than integer_zero_node.
8374 * match.pd: Check for a no-op conversion before using wi::add
8375 rather than after. Use tree_to_uhwi when summing small shift
8376 counts into an unsigned int.
8378 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8379 Alan Hayward <alan.hayward@arm.com>
8380 David Sherwood <david.sherwood@arm.com>
8383 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8384 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8385 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8386 POINTER_AND_FP_REGS.
8388 2017-10-02 Richard Biener <rguenther@suse.de>
8390 PR tree-optimization/82355
8391 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8392 a mapping for the enclosing loop but avoid generating one for
8394 (copy_bb_and_scalar_dependences): Remove premature codegen
8395 error on PHIs in blocks duplicated into multiple places.
8396 * graphite-scop-detection.c
8397 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8398 in the region use it as loop and nest to analyze the DR in.
8399 (try_generate_gimple_bb): Likewise.
8400 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8401 (add_loop_constraints): For blocks in a loop not in the region
8402 create a dimension with a single iteration.
8403 * sese.h (gbb_loop_at_index): Remove assert.
8405 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8407 * omp-expand.c (adjust_context_scope): New function.
8408 (expand_parallel_call): Call adjust_context_scope.
8410 2017-10-01 Jeff Law <law@redhat.com>
8412 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8413 dom_opt_dom_walker class with direct access to private members.
8414 Add comments. Call test_for_singularity.
8415 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8416 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8417 m_dummy_cond anymore.
8418 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8420 (pass_dominator:execute): Build the dummy_cond here and pass it
8421 to the dom_opt_dom_walker ctor.
8422 (test_for_singularity): New function.
8424 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8425 Maya Rashish <coypu@sdf.org>
8427 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8428 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8429 (alpha*-*-netbsd*) Use nbsd_tm_file.
8430 (arm*-*-netbsdelf*) Likewise.
8431 (i[34567]86-*-netbsdelf*) Likewise.
8432 (x86_64-*-netbsd*) Likewise.
8433 (mips*-*-netbsd*) Likewise.
8434 (powerpc-*-netbsd*) Likewise.
8435 (sh*-*-netbsd*) Likewise.
8436 (sparc-*-netbsdelf*) Likewise.
8437 (sparc64-*-netbsd*) Likewise.
8438 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8440 (vax-*-netbsdelf*) Likewise.
8441 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8442 (UINT_FAST8_TYPE) Likewise.
8443 (INT_FAST16_TYPE) Check CHAR_FAST16.
8444 (UINT_FAST16_TYPE) Likewise.
8446 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8449 * config/i386/i386.md
8450 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8451 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8452 *divmodsi4_zext_2): New define_insn_and_split.
8453 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8454 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8455 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8456 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8457 New define_insn_and_split.
8458 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8459 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8460 operands[1] having DImode when mode is SImode.
8462 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8463 always SImode for DIV and MOD in REG_EQUAL notes.
8465 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8468 * match.pd: Fix handling of NaNs in pattern.
8470 2017-09-29 Jeff Law <law@redhat.com>
8472 * sbitmap.c (bitmap_bit_in_range_p): New function.
8473 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8474 * tree-ssa-dse.c (live_bytes_read): New function.
8475 (dse_classify_store): Ignore reads of dead bytes.
8477 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8478 typos and whitespace errors.
8479 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8480 * config/s390/s390.c (s390_emit_prologue): Likewise.
8482 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8485 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8486 with a symbol for LRA.
8488 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8490 PR rtl-optimization/82338
8491 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8493 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8495 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8496 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8497 (qsort_chk): Declare.
8498 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8499 (qsort_chk): New function.
8501 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8503 PR tree-optimization/82337
8504 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8505 phi definition if the PHI result appears in an abnormal PHI.
8506 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8507 basis appears in an abnormal PHI.
8509 2017-09-29 Richard Biener <rguenther@suse.de>
8511 * graphite-isl-ast-to-gimple.c
8512 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8513 (binary_op_to_tree): Use it.
8514 (get_rename_from_scev): Likewise.
8515 (copy_loop_phi_nodes): Likewise.
8516 (copy_bb_and_scalar_dependences): Likewise.
8517 (translate_pending_phi_nodes): Likewise.
8519 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8522 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8523 for movabsq $(i32 << shift), r64.
8525 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8527 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8528 index when encoding %esp as %rsp to avoid 0x67 prefix.
8530 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8532 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8533 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8535 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8537 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8538 Extensions with more than 16 double VFP registers.
8539 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8540 to_clear_mask and all code related to it. Replace the remaining
8541 entry by a sbitmap and adapt code accordingly.
8543 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8545 * brig-builtins.def: Change pure attributes to const.
8547 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8549 * config.gcc (default_gnu_indirect_function): Default to yes for
8550 sparc*-*-linux* with glibc.
8552 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8554 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8555 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8556 when creating .init_array and .fini_array sections with priority
8559 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8562 * config/aarch64/aarch64.c
8563 (aarch64_builtin_support_vector_misalignment): Always return false
8564 when misalignment is unknown.
8566 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8568 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8569 this function to return false if the definition used by the swap
8570 instruction is artificial, or if the memory address from which the
8571 constant value is loaded is not represented by a base address held
8572 in a register or if the base address register is a frame or stack
8573 pointer. Additionally, return false if the base address of the
8574 loaded constant is a SYMBOL_REF but is not considered to be a
8576 (replace_swapped_load_constant): New function.
8577 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8578 loaded constant vector with a load of a swapped constant vector.
8580 2017-09-27 Carl Love <cel@us.ibm.com>
8582 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8583 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8584 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8587 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8589 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8590 first, always call autopref_rank_data otherwise.
8592 2017-09-27 Richard Biener <rguenther@suse.de>
8594 * graphite-scop-detection.c (find_scop_parameters): Move
8595 loop bound handling ...
8596 (gather_bbs::before_dom_children): ... here, avoiding the need
8597 to build scop_info->loop_nest.
8598 (record_loop_in_sese): Remove.
8599 * sese.h (sese_info_t::loop_nest): Remove.
8600 * sese.c (new_sese_info): Do not allocate loop_nest.
8601 (free_sese_info): Do not free loop_nest.
8603 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8606 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8607 lhs from calls if the lhs has addressable type.
8609 2017-09-27 Richard Biener <rguenther@suse.de>
8611 * graphite.h (scop::max_alias_set): New member.
8612 * graphite-scop-detection.c: Remove references to non-existing
8613 --param in comments.
8614 (build_alias_sets): Record the maximum alias set used for drs.
8615 (build_scops): Support zero as unlimited for
8616 --param graphite-max-arrays-per-scop.
8617 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8619 (build_poly_sr_1): ... here. Compute alias set based on the
8620 maximum alias set used for drs rather than
8621 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8623 2017-09-27 Richard Biener <rguenther@suse.de>
8625 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8626 --param loop-block-tile-size=0 to disable tiling.
8628 2017-09-27 Richard Biener <rguenther@suse.de>
8630 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8631 (graphite-max-nb-scop-params): Document special value zero.
8632 * domwalk.h (dom_walker::STOP): New symbolical constant.
8633 (dom_walker::dom_walker): Add optional parameter for bb to
8635 (dom_walker::~dom_walker): Declare.
8636 (dom_walker::before_dom_children): Document STOP return value.
8637 (dom_walker::m_user_bb_to_rpo): New member.
8638 (dom_walker::m_bb_to_rpo): Likewise.
8639 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8640 mapping here if not provided by the user.
8641 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8642 provided by the user.
8643 (dom_walker::STOP): Define.
8644 (dom_walker::walk): Do not compute bb to RPO mapping here.
8645 Support STOP return value from before_dom_children to stop
8647 * graphite-optimize-isl.c (optimize_isl): If the schedule
8648 is the same still generate code if -fgraphite-identity
8649 or -floop-parallelize-all are given.
8650 * graphite-scop-detection.c: Include cfganal.h.
8651 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8653 (gather_bbs::before_dom_children): Return STOP for BBs
8655 (build_scops): Compute bb to RPO mapping and pass it to
8656 the domwalk. Treat --param graphite-max-nb-scop-params=0
8657 as not limiting the number of params.
8658 * graphite.c (graphite_initialize): Remove limit on the
8659 number of basic-blocks in a function.
8660 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8661 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8662 default value of 10.
8664 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8666 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8667 Adjust code to eliminate needing to do the shift right 32-bits
8668 operation after XSCVDPSPN.
8670 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8672 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8673 ((X / Y) != 0 -> X >= Y): Likewise.
8675 2017-09-26 Carl Love <cel@us.ibm.com>
8677 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8678 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8679 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8680 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8681 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8682 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8683 definitions and overloading.
8684 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8685 statement for P9V_BUILTIN_XST_LEN_R.
8686 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8687 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8688 define_expand and define_insn for the instructions and builtins.
8689 * doc/extend.texi: Update the built-in documentation file for the new
8691 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8692 define_insn for the instructions
8694 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8697 * gcc/config/netbsd-protos.h: New file.
8698 * gcc/config/netbsd.c: New file.
8699 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8700 * gcc/config/t-netbsd: New file.
8701 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8702 (tmake_file) Add t-netbsd.
8703 (extra_objs) Add netbsd.o.
8705 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8709 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8711 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8713 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8714 sign extension from a vector register to a GPR by doing a 32-bit
8715 direct move and then an EXTSW.
8716 (extendsi<mode>2 splitter): Likewise.
8717 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8718 right or vector extract after doing XSCVDPSPN. Use
8719 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8721 (movdi_from_sf_zero_ext): Likewise.
8722 (reload_gpr_from_vsxsf): Likewise.
8723 (p8_mfvsrd_4_disf): Delete, no longer used.
8724 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8725 then needing to move the SFmode to a GPR to use the XSCVDPSP
8726 instruction instead of FRSP and XSCVDPSPN.
8727 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8728 it is adjacent to the other XSCVSPDP insns.
8729 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8730 SFmode to be in traditional Altivec registers.
8731 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8732 (vsx_xscvspdpn): Likewise.
8733 (vsx_xscvspdpn_scalar): Likewise.
8735 2017-09-26 Martin Jambor <mjambor@suse.cz>
8737 * tree-sra.c (compare_access_positions): Put integral types first,
8738 stabilize sorting of integral types, remove conditions putting
8739 non-full-precision integers last.
8740 (sort_and_splice_var_accesses): Disable scalarization if a
8741 non-integert would be represented by a non-full-precision integer.
8743 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8745 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8746 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8747 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8748 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8749 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8750 conditionals inside the function instead of around it. Call
8751 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8752 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8754 2017-09-26 Richard Biener <rguenther@suse.de>
8756 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8758 (scop_detection::build_scop_breadth): ... this. Removed.
8759 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8760 (scop_detection::harmful_stmt_in_bb): Likewise.
8761 (scop_detection::graphite_can_represent_stmt): Likewise.
8762 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8763 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8764 (scop_detection::can_represent_loop_1): ... this. Removed.
8765 (scop_detection::harmful_loop_in_region): Simplify after inlining
8766 the above and remove more quadraticness.
8767 (build_scops): Adjust.
8768 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8771 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8774 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8775 REGNO (base) == SP_REG if base is a REG.
8778 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8779 if it is different SSA_NAME.
8780 (optimize_range_tests_cmp_bitwise): New function.
8781 (optimize_range_tests): Call it.
8783 2017-09-26 Richard Biener <rguenther@suse.de>
8785 PR tree-optimization/82321
8786 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8787 for the def being inside the loop.
8789 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8791 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8793 * config/s390/s390-builtins.def: Fix constraint on op4.
8795 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8797 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8798 independent expanders.
8799 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8800 ("vec_ordered", "vec_unordered"): New expanders.
8802 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8804 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8807 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8809 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8810 vec_unpacks_lo_v16qi.
8811 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8813 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8815 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8816 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8817 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8819 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8821 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8823 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8825 ("*vec_slb<mode>"): New insn pattern.
8826 ("vec_shr_<mode>"): New expander.
8827 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8828 and force the shift count operand to V16QImode.
8829 ("vec_srb<mode>"): Set shift count mode to V16QI.
8831 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8833 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8834 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8835 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8837 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8840 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8842 2017-09-26 Richard Biener <rguenther@suse.de>
8844 PR tree-optimization/82320
8845 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8848 2017-09-25 Jeff Law <law@redhat.com>
8850 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8851 prototype for new argument.
8852 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8853 mostly extracted from rs6000_emit_allocate_stack.
8854 (rs6000_emit_probe_stack_range_stack_clash): New function.
8855 (rs6000_emit_allocate_stack): Call
8856 rs6000_emit_probe_stack_range_stack_clash as needed.
8857 (rs6000_emit_probe_stack_range): Add additional argument
8858 to call to gen_probe_stack_range{si,di}.
8859 (output_probe_stack_range): New.
8860 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8861 (output_probe_stack_range_stack_clash): New.
8862 (rs6000_emit_prologue): Emit notes into dump file as requested.
8863 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8864 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8865 Add additional operand and pass it to output_probe_stack_range.
8867 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8869 PR tree-optimization/82163
8870 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8871 (checking_verify_loop_closed_ssa): New parameter.
8872 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8873 (check_loop_closed_ssa_stmt): Delete.
8874 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8875 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8876 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8879 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8881 * brig-builtins.def: Treat HSAIL barrier builtins as
8882 setjmp/longjump style functions.
8884 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8886 * target.def (constant_alignment): New hook.
8887 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8888 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8889 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8890 * doc/tm.texi: Regenerate.
8891 * targhooks.h (default_constant_alignment): Declare.
8892 (constant_alignment_word_strings): Likewise.
8893 * targhooks.c (default_constant_alignment): New function.
8894 (constant_alignment_word_strings): Likewise.
8895 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8896 instead of CONSTANT_ALIGNMENT.
8897 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8898 (force_const_mem): Likewise.
8899 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8900 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8901 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8902 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8903 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8905 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8906 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8907 constant_alignment_word_strings.
8908 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8909 (CONSTANT_ALIGNMENT): Likewise.
8910 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8911 (arm_constant_alignment): New function.
8912 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8913 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8914 constant_alignment_word_strings.
8915 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8916 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8917 constant_alignment_word_strings.
8918 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8919 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8920 (cris_constant_alignment): New function.
8921 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8922 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8923 (epiphany_constant_alignment): New function.
8924 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8925 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8926 constant_alignment_word_strings.
8927 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8928 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8929 constant_alignment_word_strings.
8930 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8931 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8932 constant_alignment_word_strings.
8933 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8934 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8935 * config/i386/i386.c (ix86_constant_alignment): Make static.
8936 Use the same interface as the target hook.
8937 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8938 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8939 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8940 constant_alignment_word_strings.
8941 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8942 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8943 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8944 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8945 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8946 constant_alignment_word_strings.
8947 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8948 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8949 constant_alignment_word_strings.
8950 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8951 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8952 constant_alignment_word_strings.
8953 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8954 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8956 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8957 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8958 * config/mips/mips.c (mips_constant_alignment): New function.
8959 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8960 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8961 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8962 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8963 (mmix_constant_alignment): Make static. Use the same interface
8965 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8966 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8967 constant_alignment_word_strings.
8968 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8969 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8970 constant_alignment_word_strings.
8971 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8972 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8973 constant_alignment_word_strings.
8974 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8975 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8976 (rs6000_constant_alignment): New function.
8977 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8978 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8979 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8980 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8981 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8982 (rs6000_constant_alignment): New function.
8983 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8984 * config/s390/s390.c (s390_constant_alignment): New function.
8985 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8986 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8987 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8988 constant_alignment_word_strings.
8989 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8990 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8991 (sparc_constant_alignment): New function.
8992 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8993 * config/spu/spu.c (spu_constant_alignment): New function.
8994 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8995 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8996 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8997 constant_alignment_word_strings.
8998 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8999 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9000 constant_alignment_word_strings.
9001 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9002 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9003 constant_alignment_word_strings.
9004 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9005 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9006 (visium_constant_alignment): New function.
9007 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9008 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9009 (xtensa_constant_alignment): New function.
9010 * system.h (CONSTANT_ALIGNMENT): Poison.
9012 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9014 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9015 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9016 (rs6000_builtin_valid_without_lhs): New helper function.
9017 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9018 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9020 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9022 * target.h (vec_perm_indices): Use unsigned short rather than
9024 (auto_vec_perm_indices): Likewise.
9025 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9026 Use unsigned int rather than unsigned char.
9027 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9029 2017-09-25 Richard Biener <rguenther@suse.de>
9031 * cfgloop.h (sort_sibling_loops): Declare.
9032 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9033 (sort_sibling_loops): New function sorting the sibling loop list
9035 * graphite.c (graphite_transform_loops): Sort sibling loops.
9037 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9039 * target.def (vec_perm_const_ok): Change sel parameter to
9041 * optabs-query.c (can_vec_perm_p): Update accordingly.
9042 * doc/tm.texi: Regenerate.
9043 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9044 auto_vec_perm_indices and remove separate nelt field.
9045 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9046 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9047 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9048 (aarch64_expand_vec_perm_const): Update accordingly.
9049 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9050 to vec_perm_indices.
9051 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9052 auto_vec_perm_indices and remove separate nelt field.
9053 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9054 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9055 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9057 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9058 to vec_perm_indices.
9059 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9060 sel to vec_perm_indices.
9061 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9062 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9063 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9065 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9068 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9071 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9072 on the FUNCTION_DECL function context if it has a DIE that is a
9075 2017-09-25 Richard Biener <rguenther@suse.de>
9077 PR tree-optimization/82285
9078 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9081 2017-09-25 Tom de Vries <tom@codesourcery.com>
9085 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9088 2017-09-25 Richard Biener <rguenther@suse.de>
9090 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9091 ISL errors other than isl_error_quota happen. Dump if the
9092 schedule is the same.
9093 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9094 errors instead of aborting inside ISL.
9096 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9099 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9100 of libgcc_eh for m64.
9101 * config/i386/darwin64.h: Likewise.
9103 2017-09-25 Richard Biener <rguenther@suse.de>
9106 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9107 attribute for incomplete types nor twice for complete ones.
9109 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9112 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9113 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9116 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9119 * config/i386/i386.opt (mprefer-avx256): Use
9120 ix86_target_flags variable.
9121 * config/i386/i386.c (ix86_target_string): Move
9122 -mprefer-avx256 to flag2_opts.
9124 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9127 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9128 and x != -1 | y != -1 into (x & y) != -1.
9130 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9132 * config.gcc: Add new case statement to set
9133 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9134 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9135 s390x-*-linux* case statements. Added aarch64 to the list of
9136 supported architectures.
9138 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9140 PR tree-optimization/82289
9141 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9142 STMT_VINFO_RELEVANT_P.
9144 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9145 Alan Hayward <alan.hayward@arm.com>
9146 David Sherwood <david.sherwood@arm.com>
9148 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9149 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9150 (extract_range_from_binary_expr_1): Don't call
9151 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9153 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9154 Alan Hayward <alan.hayward@arm.com>
9155 David Sherwood <david.sherwood@arm.com>
9157 * target.def (preferred_vector_alignment): New hook.
9158 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9160 * doc/tm.texi: Regenerate.
9161 * targhooks.h (default_preferred_vector_alignment): Declare.
9162 * targhooks.c (default_preferred_vector_alignment): New function.
9163 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9165 (DR_TARGET_ALIGNMENT): New macro.
9166 (aligned_access_p): Update commentary.
9167 (vect_known_alignment_in_bytes): New function.
9168 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9170 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9171 Calculate the misalignment based on the target alignment rather than
9173 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9174 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9175 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9176 the target alignment, rather than masking the element misalignment
9177 with the number of elements in a vector. Also use the target
9178 alignment when calculating the maximum number of peels.
9179 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9180 instead of TYPE_ALIGN_UNIT.
9181 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9182 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9183 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9184 (vect_create_data_ref_ptr): Likewise.
9185 (vect_setup_realignment): Realign by ANDing with
9186 -DR_TARGET_MISALIGNMENT.
9187 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9188 the number of peels based on DR_TARGET_ALIGNMENT.
9189 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9190 with the guaranteed alignment boundary when deciding whether
9192 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9193 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9194 (ensure_base_align): Remove stmt_info parameter. Get the
9195 target base alignment from DR_TARGET_ALIGNMENT.
9196 (vectorizable_store): Update call accordingly. Interpret
9197 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9199 (vectorizable_load): Likewise.
9201 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9202 Alan Hayward <alan.hayward@arm.com>
9203 David Sherwood <david.sherwood@arm.com>
9205 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9206 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9207 (vect_enhance_data_refs_alignment): Likewise.
9209 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9211 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9212 error. Only quit immediately if parsing is complete.
9213 (BEGIN): Initialize fatal_err and parse_done.
9214 (begin fpu, end fpu): Check number of arguments.
9215 (begin arch, end arch): Likewise.
9216 (begin cpu, end cpu): Likewise.
9217 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9218 (optalias): Likewise.
9220 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9222 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9223 * config/arm/arm-isa.h: Delete. Move definitions to ...
9224 * arm-cpus.in: ... here. Use new feature and fgroup values.
9225 * config/arm/arm.c (arm_option_override): Use lower case for feature
9227 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9228 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9229 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9231 (print_isa_bits_for): New function.
9232 (gen_isa): New function.
9233 (gen_comm_data): Use print_isa_bits_for.
9234 (define feature): New keyword.
9235 (define fgroup): New keyword.
9236 * config/arm/t-arm (TM_H): Remove.
9237 (GTM_H): Add arm-isa.h.
9238 (arm-isa.h): Add rule to generate file.
9239 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9240 case for feature bit names.
9242 2017-09-22 Richard Biener <rguenther@suse.de>
9244 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9246 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9247 print of no dependency loops ...
9248 * graphite.c (graphite_transform_loops): ... here.
9249 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9251 (same_close_phi_node, remove_duplicate_close_phi,
9252 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9253 (canonicalize_loop_closed_ssa): ... here and simplify.
9254 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9255 (optimize_isl): Use dump_printf_loc to tell when we stopped
9256 optimizing because of an ISL timeout.
9258 2017-09-22 Richard Biener <rguenther@suse.de>
9260 PR tree-optimization/82291
9261 * tree-if-conv.c (predicate_mem_writes): Make sure to
9262 remove writes in blocks predicated with false.
9264 2017-09-22 Richard Biener <rguenther@suse.de>
9266 * sese.c: Include cfganal.h.
9267 (if_region_set_false_region): Remove.
9268 (create_if_region_on_edge): Likewise.
9269 (move_sese_in_condition): Re-implement without destroying
9272 2017-09-22 Richard Biener <rguenther@suse.de>
9274 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9275 Verify both BBs contain loop PHI nodes before dispatching to
9277 (graphite_regenerate_ast_isl): Do not recompute dominators,
9278 do not verify three times. Restructure for clarity.
9279 * graphite-scop-detection.c (same_close_phi_node,
9280 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9281 defined_in_loop_p, canonicalize_loop_closed_ssa,
9282 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9283 checking and SSA rewrite, move to ...
9284 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9285 (graphite_initialize): Do not pass in ctx, do not reset the
9286 SCEV cache, compute only dominators.
9287 (graphite_transform_loops): Allocate ISL ctx after
9288 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9289 Maintain post-dominators only around build_scops.
9290 * sese.c (if_region_set_false_region): Make static. Free
9291 and recompute dominators.
9292 (move_sese_in_condition): Assert we don't get called with
9293 post-dominators computed.
9294 * sese.h (if_region_set_false_region): Remove.
9296 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9298 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9299 mode attribute for TARGET_AVX512VL.
9301 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9303 * config/i386/i386.opt (mprefer-avx256): New option.
9304 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9306 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9307 for TARGET_PREFER_AVX256.
9308 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9310 2017-09-21 Jeff Law <law@redhat.com>
9312 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9313 Fix dump output if the only stack space is for pushed registers.
9315 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9317 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9320 2017-09-21 Martin Sebor <msebor@redhat.com>
9323 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9324 code (in C++) or code that triggers warnings.
9326 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9328 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9330 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9332 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9333 * sched-rgn.c: Ditto.
9334 * sel-sched-ir.c: Ditto.
9336 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9338 * toplev.h (set_random_seed): Adjust return type.
9339 * toplev.c (init_local_tick): Move eager initialization of random_seed
9340 to get_random_seed. Adjust comment.
9341 (init_random_seed): Inline to get_random_seed, delete.
9342 (get_random_seed): Initialize random_seed lazily.
9343 (set_random_seed): Do not return previous value.
9344 (print_switch_value): Do not call get_random_seed.
9346 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9348 * cgraph.c (delete_function_version): New, broken out from...
9349 (cgraph_node::delete_function_version): ...here. Rename to
9350 cgraph_node::delete_function_version_by_decl. Update all uses.
9351 (cgraph_node::remove): Call delete_function_version.
9353 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9356 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9357 VAR_DECLs to which addressable non-volatile parameters are mapped
9358 and for id->retvar after the return value assignment. Clear
9359 id->retval and id->retbnd after inlining.
9361 2017-09-21 Richard Biener <rguenther@suse.de>
9363 PR tree-optimization/82276
9364 PR tree-optimization/82244
9365 * tree-vrp.c (build_assert_expr_for): Set
9366 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9368 (remove_range_assertions): Revert earlier change.
9370 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9373 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9375 2017-09-21 Richard Biener <rguenther@suse.de>
9377 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9378 Restore valid IL after code generation errors.
9379 * graphite.c (graphite_transform_loops): Diagnose code
9380 generation issues as MSG_MISSED_OPTIMIZATION and continue
9381 with processing SCOPs.
9383 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9384 Alan Hayward <alan.hayward@arm.com>
9385 David Sherwood <david.sherwood@arm.com>
9387 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9388 rather than choosing between plus_constant and gen_rtx_<CODE>.
9389 * expr.c (emit_push_insn): Likewise.
9390 (expand_expr_real_2): Likewise.
9392 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9393 Alan Hayward <alan.hayward@arm.com>
9394 David Sherwood <david.sherwood@arm.com>
9396 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9398 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9399 Alan Hayward <alan.hayward@arm.com>
9400 David Sherwood <david.sherwood@arm.com>
9402 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9403 calling tree_to_uhwi.
9405 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9406 Alan Hayward <alan.hayward@arm.com>
9407 David Sherwood <david.sherwood@arm.com>
9409 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9410 INTEGER_CST rather than a negative test for ADDR_EXPR.
9412 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9413 Alan Hayward <alan.hayward@arm.com>
9414 David Sherwood <david.sherwood@arm.com>
9416 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9417 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9419 2017-09-21 Richard Biener <rguenther@suse.de>
9421 PR tree-optimization/71351
9422 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9423 graphite_create_new_loop_guard): Remove, fold remaining parts
9425 (translate_isl_ast_node_for): ... here and simplify.
9427 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9430 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9431 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9432 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9433 alternative always use QI mode, for -Os imov (=R,R) alternative
9434 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9437 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9438 Jeff Law <law@redhat.com>
9440 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9441 (allocate_stack_space): New function, partially extracted from
9443 (s390_emit_prologue): Track offset to most recent stack probe.
9444 Code to allocate space moved into allocate_stack_space.
9445 Dump actions when no stack is allocated.
9446 (s390_prologue_plus_offset): New function.
9447 (s390_emit_stack_probe): Likewise.
9449 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9451 * common.opt (Wa, Wl, Wp, g, gz=): Add
9453 (gno-column-info): Remove.
9454 (gcolumn-info): Drop RejectNegative.
9456 (gno-record-gcc-switches): Remove.
9457 (grecord-gcc-switches): Drop RejectNegative.
9458 (gno-split-dwarf): Remove.
9459 (gsplit-dwarf): Drop RejectNegative.
9460 (gno-strict-dwarf): Remove.
9461 (gstrict-dwarf): Drop RejectNegative.
9462 * config/darwin.opt (gfull, gused): Add RejectNegative.
9463 * dwarf2out.c (gen_producer_string): Drop
9464 gno-record-gcc-switches handler.
9465 * optc-gen.awk: Add g to prefixes with negative forms.
9466 * opts-common.c (remapping_prefix_p): New.
9467 (find_opt): Check it.
9468 (generate_canonical_option): Test g prefix.
9469 (option_map): Add -gno- mapping.
9470 (add_misspelling_candidates): Check remapping_prefix_p.
9472 2017-09-20 Jeff Law <law@redhat.com>
9474 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9475 thinko in stack clash protection support.
9477 * explow.c (compute_stack_clash_protection_loop_data): Use
9478 CONST_INT_P instead of explicit test. Verify object is a
9479 CONST_INT_P before looking at INTVAL.
9480 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9481 instead of explicit test.
9483 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9486 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9487 address instead of to r1 and r11.
9489 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9491 * config.gcc: Support "knm".
9492 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9493 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9495 * config/i386/i386.c (m_KNM): Define.
9496 (processor_target_table): Add "knm".
9498 (ix86_option_override_internal): Add "knm".
9499 (ix86_issue_rate): Add PROCESSOR_KNM.
9500 (ix86_adjust_cost): Ditto.
9501 (ia32_multipass_dfa_lookahead): Ditto.
9502 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9503 (fold_builtin_cpu): Add M_INTEL_KNM.
9504 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9505 (processor_type): Add PROCESSOR_KNM.
9506 * config/i386/x86-tune.def: Add m_KNM.
9507 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9509 2017-09-20 Richard Biener <rguenther@suse.de>
9511 PR tree-optimization/80213
9512 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9513 are allowed in empty BBs as well.
9514 (canonicalize_loop_closed_ssa): Also look for other complex
9516 (scop_detection::get_sese): Include the loop-closed PHI block
9518 (scop_detection::merge_sese): Remove code adding extra blocks.
9519 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9520 (build_scops): Assert the final returned scop is invalid.
9522 2017-09-20 Richard Biener <rguenther@suse.de>
9524 PR tree-optimization/82264
9525 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9527 (vn_phi_lookup): Likewise.
9528 (vn_phi_insert): Likewise.
9530 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9532 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9533 that fits into uhwi or shwi, add DW_AT_const_value regardless
9534 of early_dwarf without going through RTL, using add_AT_unsigned
9537 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9538 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9539 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9540 (DEBUG_MACRO_SECTION): Likewise.
9541 (DEBUG_LTO_MACRO_SECTION): Likewise.
9542 (DEBUG_STR_DWO_SECTION): Likewise.
9543 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9544 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9545 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9546 (DEBUG_LTO_LINE_STR_SECTION): Define.
9547 (init_sections_and_labels): Initialize debug_line_str_section
9548 variable. Initialize debug_loc_section for -gdwarf-5 to
9549 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9551 2017-09-20 Richard Biener <rguenther@suse.de>
9553 * graphite-sese-to-poly.c (extract_affine): Properly handle
9554 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9556 2017-09-20 Richard Biener <rguenther@suse.de>
9558 PR tree-optimization/81373
9559 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9560 Force SESE live-out defs to be handled even if they are
9563 2017-09-19 Jeff Law <law@redhat.com>
9565 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9566 nothing for stack adjustments with REG_STACK_CHECK.
9567 * sched-deps.c (parse_add_or_inc): Reject insns with
9568 REG_STACK_CHECK from dependency breaking.
9569 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9570 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9571 * reg-notes.def (STACK_CHECK): New note.
9573 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9574 (ix86_expand_prologue): Dump stack clash info as needed.
9575 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9577 * function.c (dump_stack_clash_frame_info): New function.
9578 * function.h (dump_stack_clash_frame_info): Prototype.
9579 (enum stack_clash_probes): New enum.
9581 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9582 flag_stack_clash_protection.
9583 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9584 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9585 (arm_frame_pointer_required): Likewise.
9586 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9587 (ia64_expand_prologue): Likewise.
9588 * config/mips/mips.c (mips_expand_prologue): Likewise.
9589 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9590 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9591 (sparc_flat_expand_prologue): Likewise.
9592 * config/spu/spu.c (spu_expand_prologue): Likewise.
9594 * explow.c: Include "params.h".
9595 (anti_adjust_stack_and_probe_stack_clash): New function.
9596 (get_stack_check_protect): Likewise.
9597 (compute_stack_clash_protection_loop_data): Likewise.
9598 (emit_stack_clash_protection_loop_start): Likewise.
9599 (emit_stack_clash_protection_loop_end): Likewise.
9600 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9601 Use anti_adjust_stack_and_probe_stack_clash.
9602 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9603 (emit_stack_clash_protection_loop_start): Likewise.
9604 (emit_stack_clash_protection_loop_end): Likewise.
9605 * rtl.h (get_stack_check_protect): Prototype.
9606 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9607 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9608 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9610 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9612 * doc/tm.texi: Rebuilt.
9613 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9614 get_stack_check_protect.
9615 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9616 * config/arm/arm.c (arm_expand_prologue): Likewise.
9617 (arm_frame_pointer_required): Likewise.
9618 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9619 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9620 * config/mips/mips.c (mips_expand_prologue): Likewise.
9621 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9622 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9623 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9624 (sparc_flat_expand_prologue): Likewise.
9626 * common.opt (-fstack-clash-protection): New option.
9627 * flag-types.h (enum stack_check_type): Note difference between
9628 -fstack-check= and -fstack-clash-protection.
9629 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9630 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9631 * toplev.c (process_options): Issue warnings/errors for cases
9632 not handled with -fstack-clash-protection.
9633 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9634 (-fstack-check): Note additional problem with -fstack-check=generic.
9635 Note that -fstack-check is primarily for Ada and refer users
9636 to -fstack-clash-protection for stack-clash-protection.
9637 Document new params for stack clash protection.
9639 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9641 * config/i386/i386.c (ix86_split_long_move): Do not handle
9642 address used for LEA in a special way.
9644 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9646 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9648 2017-09-19 Martin Sebor <msebor@redhat.com>
9651 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9652 of incompatible types.
9654 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9656 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9657 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9658 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9659 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9661 2017-09-19 Richard Biener <rguenther@suse.de>
9663 PR tree-optimization/82244
9664 * tree-vrp.c (remove_range_assertions): Do not propagate
9665 a constant to abnormals but replace the assert with a copy.
9667 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9669 PR rtl-optimization/57878
9670 PR rtl-optimization/68988
9671 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9672 avoidance test involving non_reload_pseudos. Move frequency test
9673 below the general fragmentation avoidance test.
9675 2017-09-19 Richard Biener <rguenther@suse.de>
9677 PR tree-optimization/69728
9678 * graphite-sese-to-poly.c (schedule_error): New global.
9679 (add_loop_schedule): Handle empty domain by failing the
9681 (build_original_schedule): Handle schedule_error.
9683 2017-09-19 Richard Biener <rguenther@suse.de>
9685 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9686 Do not iterate to sibling loops but only to siblings of inner
9689 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9692 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9693 that the registers are different.
9695 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9697 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9698 to processor_model and "amdfam17h" to arch_names_table.
9699 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9701 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9704 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9706 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9707 Alan Hayward <alan.hayward@arm.com>
9708 David Sherwood <david.sherwood@arm.com>
9710 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9712 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9714 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9715 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9716 vect_schedule_slp_instance.
9717 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9718 Update call to vect_slp_analyze_node_operations. Simplify return
9720 (vect_slp_analyze_bb_1): Update call accordingly.
9721 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9722 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9723 (vect_schedule_slp): Update call accordingly.
9725 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9726 Alan Hayward <alan.hayward@arm.com>
9727 David Sherwood <david.sherwood@arm.com>
9729 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9730 with types that aren't in fact scalar.
9732 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9734 * tree-vect-slp.c (vect_record_max_nunits): New function,
9736 (vect_build_slp_tree_1): ...here.
9737 (vect_build_slp_tree_2): Call it for phis too.
9739 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9741 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9742 to vect_get_vec_def_for_operand when getting the mask operand.
9744 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9745 Alan Hayward <alan.hayward@arm.com>
9746 David Sherwood <david.sherwood@arm.com>
9748 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9751 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9752 Alan Hayward <alan.hayward@arm.com>
9753 David Sherwood <david.sherwood@arm.com>
9755 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9756 calculation for vector booleans.
9758 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9759 Alan Hayward <alan.hayward@arm.com>
9760 David Sherwood <david.sherwood@arm.com>
9762 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9764 (vect_transform_stmt): ...here.
9765 (vect_analyze_stmt): Use it instead of calling
9766 vectorizable_live_operation directly.
9768 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9770 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9771 non-SIMT targets in acc vector loops.
9773 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9775 * configure.ac: Add arc and check if assembler supports gdwarf2.
9776 * configure: Regenerate.
9778 2017-09-18 Richard Biener <rguenther@suse.de>
9780 PR tree-optimization/82220
9781 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9782 epilogue niters from the min_profitable_iters compute.
9784 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9787 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9788 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9789 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9791 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9794 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9795 switching to a new text section.
9797 2017-09-18 Richard Biener <rguenther@suse.de>
9799 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9801 (build_alias_set): Reject aliases with no access function.
9803 2017-09-18 Richard Biener <rguenther@suse.de>
9805 PR tree-optimization/79622
9806 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9808 (build_cross_bb_scalars_use): Likewise.
9810 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9812 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9814 2017-09-18 Alan Modra <amodra@gmail.com>
9817 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9818 stack_pointer_rtx for count 0. Update comments. Break up
9819 large rtl expression.
9821 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9823 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9824 Increase to 20 bytes.
9825 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9826 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9827 or avx version of the stub.
9829 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9832 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9833 compute the minimum stack alignment. Also update preferred stack
9834 boundary for leaf functions.
9836 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9838 PR tree-optimization/82228
9839 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9842 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9844 * common/config/nds32/nds32-common.c
9845 (nds32_option_optimization_table): Refine formatting.
9846 (nds32_option_optimization_table): Use -fsched-pressure and
9847 -fomit-frame-pointer for specific optimization level.
9849 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9851 * config/nds32/nds32.c: Refine formatting and comments.
9852 * config/nds32/nds32.h: Likewise.
9853 * config/nds32/nds32.md: Likewise.
9854 * config/nds32/nds32-cost.c: Likewise.
9855 * config/nds32/nds32-isr.c: Likewise.
9856 * config/nds32/nds32-md-auxiliary.c: Likewise.
9857 * config/nds32/nds32-multiple.md: Likewise.
9858 * config/nds32/nds32-predicates.c: Likewise.
9860 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9861 Jakub Jelinek <jakub@redhat.com>
9863 Add support for -std=c++2a.
9864 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9866 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9868 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9871 * doc/extend.texi (Common Function Attributes): Add
9872 references to ARM, AArch64, and S/390 specific attributes.
9873 (Function Specific Option Pragmas): Add AArch64 and S/390
9874 to list of back ends that support the target pragma.
9876 2017-09-15 Nathan Sidwell <nathan@acm.org>
9878 * doc/standards.texi: Fix C++17 description. Update URLs for
9881 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9883 * common.opt (Wcast-align=strict): New warning option.
9884 * doc/invoke.texi: Document -Wcast-align=strict.
9886 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9888 * cgraph.h (cgraph_thunk_info): Add comments.
9889 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9890 assert for VIRTUAL_* arguments stricter.
9892 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9894 PR tree-optimization/71026
9895 * match.pd: Move RDIV patterns from fold-const.c
9896 * fold-const.c (distribute_real_division): Removed.
9897 (fold_binary_loc): Remove calls to distribute_real_divison.
9899 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9901 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9902 c++1z and gnu++1z as deprecated. Change other references to
9903 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9904 Change -Wc++1z-compat to -Wc++17-compat.
9905 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9906 * dwarf2out.c (highest_c_language): Handle C++17.
9907 (gen_compile_unit_die): Likewise.
9909 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9911 PR rtl-optimization/82192
9912 * combine.c (make_extraction): Don't look through non-paradoxical
9913 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9916 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9917 Alan Hayward <alan.hayward@arm.com>
9918 David Sherwood <david.sherwood@arm.com>
9920 * target.def (function_arg_offset): New hook.
9921 * targhooks.h (default_function_arg_offset): Declare.
9922 * targhooks.c (default_function_arg_offset): New function.
9923 * function.c (locate_and_pad_parm): Use
9924 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9925 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9926 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9927 * doc/tm.texi: Regenerate.
9928 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9929 * config/spu/spu.c (spu_function_arg_offset): New function.
9930 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9931 * system.h (FUNCTION_ARG_OFFSET): Poison.
9933 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9934 Alan Hayard <alan.hayward@arm.com>
9935 David Sherwood <david.sherwood@arm.com>
9937 * target.def (truly_noop_truncation): New hook.
9938 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9939 than TRULY_NOOP_TRUNCATION.
9940 * hooks.h (hook_bool_uint_uint_true): Declare.
9941 * hooks.c (hook_bool_uint_uint_true): New function.
9942 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9943 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9944 * doc/tm.texi: Regenerate.
9945 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9946 rather than TRULY_NOOP_TRUNCATION in comments.
9947 (simplify_comparison): Likewise.
9948 (record_truncated_value): Likewise.
9949 * expmed.c (extract_bit_field_1): Likewise.
9950 (extract_split_bit_field): Likewise.
9951 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9952 instead of TRULY_NOOP_TRUNCATION.
9953 * function.c (assign_parm_setup_block): Likewise.
9954 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9955 * rtlhooks.c: Include target.h.
9956 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9957 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9958 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9959 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9960 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9961 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9962 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9963 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9964 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9965 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9966 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9967 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9968 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9969 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9970 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9971 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9972 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9973 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9974 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9975 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9976 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9977 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9978 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9979 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9980 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9981 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9982 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9983 rather than TRULY_NOOP_TRUNCATION in comments.
9984 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9985 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9986 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9987 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9988 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9989 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9990 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9991 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9992 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9993 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9994 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9995 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9996 rather than TRULY_NOOP_TRUNCATION in comments.
9997 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9998 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9999 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10000 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10001 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10002 TRULY_NOOP_TRUNCATION condition.
10003 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10004 (TRULY_NOOP_TRUNCATION): Delete.
10005 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10006 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10007 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10008 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10009 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10010 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10011 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10012 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10013 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10014 rather than TRULY_NOOP_TRUNCATION in comments.
10015 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10016 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10017 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10018 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10019 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10020 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10022 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10025 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10026 (*cmp_ior): Likewise.
10027 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10028 (*ior_scc_scc_cmp): Likewise.
10029 (*and_scc_scc): Likewise.
10030 (*and_scc_scc_cmp): Likewise.
10032 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10033 Alan Hayard <alan.hayward@arm.com>
10034 David Sherwood <david.sherwood@arm.com>
10036 * target.def (can_change_mode_class): New hook.
10037 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10038 (hard_regno_nregs): Likewise.
10039 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10040 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10041 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10042 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10043 (LOAD_EXTEND_OP): Update accordingly.
10044 * doc/tm.texi: Regenerate.
10045 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10046 CANNOT_CHANGE_MODE_CLASS.
10047 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10048 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10049 * combine.c (simplify_set): Update accordingly.
10050 * emit-rtl.c (validate_subreg): Likewise.
10051 * recog.c (general_operand): Likewise.
10052 * regcprop.c (mode_change_ok): Likewise.
10053 * reload1.c (choose_reload_regs): Likewise.
10054 (inherit_piecemeal_p): Likewise.
10055 * rtlanal.c (simplify_subreg_regno): Likewise.
10056 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10057 instead of CANNOT_CHANGE_MODE_CLASS.
10058 (reload_cse_simplify_operands): Likewise.
10059 * reload.c (push_reload): Use targetm.can_change_mode_class
10060 instead of CANNOT_CHANGE_MODE_CLASS.
10061 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10062 REG_CANNOT_CHANGE_MODE_P.
10063 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10064 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10065 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10066 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10067 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10068 (arm_can_change_mode_class): New function.
10069 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10070 than CANNOT_CHANGE_MODE_CLASS in comments.
10071 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10072 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10073 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10074 (ix86_can_change_mode_class): ...this new function, inverting the
10075 sense of the return value.
10076 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10077 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10078 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10079 (ia64_can_change_mode_class): New function.
10080 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10081 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10082 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10083 (m32c_can_change_mode_class): ...this new function, inverting the
10084 sense of the return value.
10085 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10086 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10087 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10088 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10089 (mips_can_change_mode_class): ...this new function, inverting the
10090 sense of the return value.
10091 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10092 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10093 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10094 (msp430_can_change_mode_class): New function.
10095 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10096 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10097 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10098 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10099 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10100 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10101 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10102 (pa_cannot_change_mode_class): Replace with...
10103 (pa_can_change_mode_class): ...this new function, inverting the
10104 sense of the return value.
10105 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10106 than CANNOT_CHANGE_MODE_CLASS in comments.
10107 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10108 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10109 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10110 (pdp11_cannot_change_mode_class): Replace with...
10111 (pdp11_can_change_mode_class): ...this new function, inverting the
10112 sense of the return value.
10113 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10114 * config/powerpcspe/powerpcspe-protos.h
10115 (rs6000_cannot_change_mode_class_ptr): Delete.
10116 * config/powerpcspe/powerpcspe.c
10117 (rs6000_cannot_change_mode_class_ptr): Delete.
10118 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10119 (rs6000_option_override_internal): Assign to
10120 targetm.can_change_mode_class instead of
10121 rs6000_cannot_change_mode_class_ptr.
10122 (rs6000_cannot_change_mode_class): Replace with...
10123 (rs6000_can_change_mode_class): ...this new function, inverting the
10124 sense of the return value.
10125 (rs6000_debug_cannot_change_mode_class): Replace with...
10126 (rs6000_debug_can_change_mode_class): ...this new function.
10127 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10128 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10129 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10130 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10131 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10133 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10134 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10135 (rs6000_option_override_internal): Assign to
10136 targetm.can_change_mode_class instead of
10137 rs6000_cannot_change_mode_class_ptr.
10138 (rs6000_cannot_change_mode_class): Replace with...
10139 (rs6000_can_change_mode_class): ...this new function, inverting the
10140 sense of the return value.
10141 (rs6000_debug_cannot_change_mode_class): Replace with...
10142 (rs6000_debug_can_change_mode_class): ...this new function.
10143 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10144 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10145 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10146 (s390_can_change_mode_class): ...this new function, inverting the
10147 sense of the return value.
10148 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10149 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10150 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10151 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10152 (sh_cannot_change_mode_class): Replace with...
10153 (sh_can_change_mode_class): ...this new function, inverting the
10154 sense of the return value.
10155 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10156 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10157 (sparc_can_change_mode_class): New function.
10158 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10159 * config/spu/spu.c (spu_can_change_mode_class): New function.
10160 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10161 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10162 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10163 (visium_can_change_mode_class): New function.
10164 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10166 2017-09-15 Richard Biener <rguenther@suse.de>
10168 PR tree-optimization/82217
10169 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10170 but not undefined case.
10172 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10175 * postreload.c (reload_cse_simplify_operands): Skip
10176 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10178 2017-09-15 Richard Biener <rguenther@suse.de>
10180 PR tree-optimization/68823
10181 * graphite-scop-detection.c (build_alias_set): If we have a
10182 possible dependence check whether we can handle them by just
10183 looking at the DRs DR_ACCESS_FNs.
10184 (build_scops): If build_alias_set fails, fail the SCOP.
10186 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10188 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10189 to support float128 built-in functions that require the ISA 3.0
10191 (BU_FLOAT128_3_HW): Likewise.
10192 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10193 built-in functions.
10194 (FMAF128): Likewise.
10196 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10197 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10198 floating point instructions.
10199 (rs6000_invalid_builtin): Likewise.
10200 (rs6000_builtin_mask_names): Likewise.
10201 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10202 (RS6000_BTM_FLOAT128_HW): Likewise.
10203 (RS6000_BTM_COMMON): Likewise.
10204 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10206 * doc/extend.texi (RS/6000 built-in functions): Document the
10207 IEEE 128-bit floating point square root and fused multiply-add
10208 built-in functions.
10210 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10212 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10213 reg (r2) isn't in the set of registers defined in the prologue.
10215 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10216 Alan Hayward <alan.hayward@arm.com>
10217 David Sherwood <david.sherwood@arm.com>
10219 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10220 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10221 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10222 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10223 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10225 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10226 max_vectorization_factor.
10227 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10229 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10230 Alan Hayward <alan.hayward@arm.com>
10231 David Sherwood <david.sherwood@arm.com>
10233 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10234 (vect_worthwhile_without_simd_p): Declare.
10235 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10236 (vectorizable_reduction): Use it.
10237 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10238 (vectorizable_operation): Likewise.
10240 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10241 Alan Hayward <alan.hayward@arm.com>
10242 David Sherwood <david.sherwood@arm.com>
10244 * tree-vectorizer.h (vect_get_num_copies): New function.
10245 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10246 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10247 (vectorizable_induction): Likewise.
10248 (vectorizable_live_operation): Likewise.
10249 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10250 (vectorizable_bswap): Likewise.
10251 (vectorizable_call): Likewise.
10252 (vectorizable_conversion): Likewise.
10253 (vectorizable_assignment): Likewise.
10254 (vectorizable_shift): Likewise.
10255 (vectorizable_operation): Likewise.
10256 (vectorizable_store): Likewise.
10257 (vectorizable_load): Likewise.
10258 (vectorizable_condition): Likewise.
10259 (vectorizable_comparison): Likewise.
10260 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10262 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10263 Alan Hayward <alan.hayward@arm.com>
10264 David Sherwood <david.sherwood@arm.com>
10266 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10267 of vect_init_vector.
10269 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10270 Alan Hayward <alan.hayward@arm.com>
10271 David Sherwood <david.sherwood@arm.com>
10273 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10274 an inline wrapper that provides a location.
10275 (gimple_build_vector): Likewise.
10276 * gimple-fold.c (gimple_build_vector_from_val): New function.
10277 (gimple_build_vector): Likewise.
10278 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10279 functions to build the initial value. Always return a gimple value.
10280 (get_initial_defs_for_reduction): Likewise. Only compute
10282 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10283 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10284 (vectorizable_induction): Use gimple_build_vector rather than
10287 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10288 Alan Hayward <alan.hayward@arm.com>
10289 David Sherwood <david.sherwood@arm.com>
10291 * target.h (vec_perm_indices): New typedef.
10292 (auto_vec_perm_indices): Likewise.
10293 * optabs-query.h: Include target.h
10294 (can_vec_perm_p): Take a vec_perm_indices *.
10295 * optabs-query.c (can_vec_perm_p): Likewise.
10296 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10297 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10298 * tree-vect-generic.c (lower_vec_perm): Likewise.
10299 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10300 (vect_grouped_load_supported): Likewise.
10301 (vect_shift_permute_load_chain): Likewise.
10302 (vect_permute_store_chain): Use auto_vec_perm_indices.
10303 (vect_permute_load_chain): Likewise.
10304 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10305 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10306 Update uses of can_vec_perm_p.
10307 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10308 mode with a number of elements. Take a vec_perm_indices *.
10309 (vect_create_epilog_for_reduction): Update accordingly.
10310 Use auto_vec_perm_indices.
10311 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10312 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10313 (vect_transform_slp_perm_load): Likewise.
10314 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10315 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10316 (vect_gen_perm_mask_checked): Likewise.
10317 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10318 (vect_gen_perm_mask_checked): Likewise.
10319 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10320 (vectorizable_store): Likewise.
10321 (vectorizable_load): Likewise.
10322 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10323 (vectorizable_bswap): Likewise.
10325 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10326 Alan Hayward <alan.hayward@arm.com>
10327 David Sherwood <david.sherwood@arm.com>
10329 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10330 * tree.c (build_vector): Likewise.
10331 (build_vector_from_ctor): Update accordingly.
10332 (build_vector_from_val): Likewise.
10333 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10334 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10335 * tree-vect-generic.c (add_rshift): Likewise.
10336 (expand_vector_divmod): Likewise.
10337 (optimize_vector_constructor): Likewise.
10338 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10339 (vect_transform_slp_perm_load): Likewise.
10340 (vect_schedule_slp_instance): Likewise.
10341 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10342 (vectorizable_call): Likewise.
10343 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10344 * expmed.c (make_tree): Likewise.
10345 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10346 a vector passed to build_vector.
10347 (fold_convert_const): Likewise.
10348 (exact_inverse): Likewise.
10349 (fold_ternary_loc): Likewise.
10350 (fold_relational_const): Likewise.
10351 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10352 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10353 (const_unop): Likewise. Store the reduction accumulator in a
10354 variable rather than an array.
10355 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10356 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10357 the new vector, rather than constructing it after the input arrays.
10358 (native_interpret_vector): Use auto_vec<tree> when building
10359 a vector passed to build_vector. Add elements in order.
10360 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10361 auto_vec<tree> when building a vector passed to build_vector.
10362 (vect_create_epilog_for_reduction): Likewise.
10363 (vectorizable_induction): Likewise.
10364 (get_initial_def_for_reduction): Likewise. Fix indentation of
10366 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10368 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10369 passed to build_vector.
10371 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10372 Alan Hayward <alan.hayward@arm.com>
10373 David Sherwood <david.sherwood@arm.com>
10375 * tree-core.h (tree_base::u): Add an "nelts" field.
10376 (tree_vector): Use VECTOR_CST_NELTS as the length.
10377 * tree.c (tree_size): Likewise.
10378 (make_vector): Initialize VECTOR_CST_NELTS.
10379 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10380 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10381 TYPE_VECTOR_SUBPARTS.
10382 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10383 as the number of units, setting it from VECTOR_CST_NELTS.
10384 (const_vector_from_tree): Likewise.
10385 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10386 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10387 (fold_negate_expr_1): Likewise.
10388 (fold_convert_const): Likewise.
10389 (const_binop): Likewise. Differentiate the number of output and
10391 (const_unop): Likewise.
10392 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10393 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10394 in cases that did the opposite.
10396 2017-09-14 Richard Biener <rguenther@suse.de>
10398 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10401 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10403 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10405 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10408 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10409 if and where to split a bb, except for splitting before debug insn
10410 sequences followed by non-label real insn. Delete debug insns
10411 in between basic blocks.
10413 * combine.c (make_compound_operation_int): Formatting fixes.
10415 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10416 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10417 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10418 * config/sol2.h (LINK_EH_SPEC): Likewise.
10419 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10420 * config/s390/linux.h (LINK_SPEC): Likewise.
10421 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10422 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10423 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10424 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10425 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10426 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10427 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10428 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10430 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10432 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10434 (ENDFILE_LINUX_SPEC): Likewise.
10435 (LINK_EH_SPEC): Likewise.
10436 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10437 (LINK_OS_LINUX_SPEC32): Likewise.
10438 (LINK_OS_LINUX_SPEC64): Likewise.
10439 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10440 (LINK_OS_LINUX_SPEC): Likewise.
10442 2017-09-13 Martin Liska <mliska@suse.cz>
10444 PR middle-end/82154
10445 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10446 CASE_HIGH is NULL_TREE.
10448 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10449 Alan Hayward <alan.hayward@arm.com>
10450 David Sherwood <david.sherwood@arm.com>
10452 * target.def (secondary_memory_needed): New hook.
10453 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10454 instead of SECONDARY_MEMORY_NEEDED.
10455 (secondary_memory_needed_mode): Likewise.
10456 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10457 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10458 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10459 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10460 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10461 * doc/tm.texi: Regenerate.
10462 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10463 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10464 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10465 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10466 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10467 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10468 mode argument first and change the reg_class arguments to reg_class_t.
10469 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10470 Make static. Update the call to inline_secondary_memory_needed.
10471 (ix86_register_move_cost): Update the call to
10472 inline_secondary_memory_needed.
10473 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10474 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10476 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10477 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10479 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10480 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10481 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10482 and match hook interface. Add comment from mips.h.
10483 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10484 * config/mmix/mmix.md (truncdfsf2): Refer to
10485 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10487 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10488 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10489 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10490 (pa_secondary_memory_needed): New function.
10491 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10492 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10493 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10494 (pdp11_secondary_memory_needed): Make static and match hook interface.
10495 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10496 * config/powerpcspe/powerpcspe-protos.h
10497 (rs6000_secondary_memory_needed_ptr): Delete.
10498 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10500 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10501 (rs6000_option_override_internal): Assign to
10502 targetm.secondary_memory_needed rather than
10503 rs6000_secondary_memory_needed_ptr.
10504 (rs6000_secondary_memory_needed): Match hook interface.
10505 (rs6000_debug_secondary_memory_needed): Likewise.
10506 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10507 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10508 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10509 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10510 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10511 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10513 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10514 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10515 (rs6000_option_override_internal): Assign to
10516 targetm.secondary_memory_needed rather than
10517 rs6000_secondary_memory_needed_ptr.
10518 (rs6000_secondary_memory_needed): Match hook interface.
10519 (rs6000_debug_secondary_memory_needed): Likewise.
10520 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10521 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10522 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10523 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10524 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10525 (sparc_secondary_memory_needed): New function.
10526 * lra-constraints.c (check_and_process_move): Refer to
10527 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10529 (curr_insn_transform): Likewise.
10530 (process_alt_operands): Use targetm.secondary_memory_needed
10531 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10532 (check_secondary_memory_needed_p): Likewise.
10533 (choose_split_class): Likewise.
10534 * reload.c: Unconditionally include code that was previously
10535 conditional on SECONDARY_MEMORY_NEEDED.
10536 (push_secondary_reload): Use targetm.secondary_memory_needed
10537 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10538 (push_reload): Likewise.
10539 * reload1.c: Unconditionally include code that was previously
10540 conditional on SECONDARY_MEMORY_NEEDED.
10541 (choose_reload_regs): Use targetm.secondary_memory_needed
10542 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10543 (gen_reload): Likewise.
10544 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10546 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10547 Alan Hayward <alan.hayward@arm.com>
10548 David Sherwood <david.sherwood@arm.com>
10550 * target.def (secondary_memory_needed_mode): New hook:
10551 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10552 * targhooks.h (default_secondary_memory_needed_mode): New function.
10553 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10554 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10555 * doc/tm.texi: Regenerate.
10556 * lra-constraints.c (check_and_process_move): Use
10557 targetm.secondary_memory_needed_mode instead of
10558 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10559 (curr_insn_transform): Likewise.
10560 * reload.c (get_secondary_mem): Likewise.
10561 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10562 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10564 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10565 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10566 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10567 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10568 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10570 * config/powerpcspe/powerpcspe-protos.h
10571 (rs6000_secondary_memory_needed_mode): Delete.
10572 * config/powerpcspe/powerpcspe.c
10573 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10574 (rs6000_secondary_memory_needed_mode): Make static.
10575 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10576 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10578 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10580 (rs6000_secondary_memory_needed_mode): Make static.
10581 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10582 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10583 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10584 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10585 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10587 (sparc_secondary_memory_needed_mode): New function.
10588 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10590 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10592 * config/aarch64/constraints.md (Umq): New constraint.
10593 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10595 (mov<mode>): Update condition.
10597 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10599 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10600 when bitposition is the same.
10602 2017-09-13 Richard Biener <rguenther@suse.de>
10604 * dwarf2out.c (output_die_symbol): Remove.
10605 (output_die): Do not output a DIEs symbol.
10607 2017-09-13 Richard Biener <rguenther@suse.de>
10609 PR middle-end/82128
10610 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10611 default-def to avoid breaking iterator update with the weird
10612 interaction with cgraph_update_edges_for_call_stmt_node.
10614 2017-09-13 Richard Biener <rguenther@suse.de>
10616 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10617 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10618 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10619 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10621 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10623 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10624 Disable pc relative literal load irrespective of
10625 TARGET_FIX_ERR_A53_84341 for default.
10627 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10629 * config/sparc/sparc.c (output_return): Output the source location of
10630 the insn in the delay slot, if any.
10631 (output_sibcall): Likewise.
10633 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10636 * common.opt (-static-pie): New alias.
10637 (shared): Negate static-pie.
10638 (-no-pie): Update help text.
10640 (static-pie): New option.
10641 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10642 -static-pie support.
10643 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10644 (LINK_EH_SPEC): Likewise.
10645 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10646 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10647 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10648 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10649 (init_gcc_specs): Likewise.
10650 (init_spec): Likewise.
10651 (display_help): Update help message for -pie.
10652 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10655 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10657 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10658 (movdi_aarch64): Likewise.
10659 (movti_aarch64): Likewise.
10661 2017-09-12 Simon Wright <simon@pushface.org>
10664 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10665 calculation of the minor version, always output as 0.
10667 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10670 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10671 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10672 on it early, rather than manual conversion late. For
10673 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10674 instead of performing manual conversion.
10676 2017-09-12 Carl Love <cel@us.ibm.com>
10678 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10679 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10681 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10682 VMULOSW): Add definitions.
10683 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10684 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10685 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10687 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10689 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10691 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10693 (movti_aarch64): Likewise.
10694 (movdf_aarch64): Likewise.
10695 (movtf_aarch64): Likewise.
10696 (load_pairdi): Likewise.
10697 (store_pairdi): Likewise.
10698 (load_pairdf): Likewise.
10699 (store_pairdf): Likewise.
10700 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10701 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10702 (ldr_got_small_<mode>): Likewise.
10703 (ldr_got_small_28k_<mode>): Likewise.
10704 (ldr_got_tiny): Likewise.
10705 * config/aarch64/iterators.md (ldst_sz): New.
10706 (ldpstp_sz): Likewise.
10707 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10709 (thunderx_load): Split load_8 to load_16.
10710 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10712 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10713 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10714 (xgene1_store_pair): Split store_8 to store_16.
10715 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10716 (falkor_st_0_st_sd): Split store_8 to store_16.
10718 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10720 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10721 and store1/2/3/4 to store_4/8/12/16.
10722 * config/aarch64/aarch64.md: Update for rename.
10723 * config/arm/arm.md: Likewise.: Likewise.
10724 * config/arm/arm.c: Likewise.
10725 * config/arm/thumb1.md: Likewise.
10726 * config/arm/thumb2.md: Likewise.
10727 * config/arm/vfp.md: Likewise.
10728 * config/arm/arm-generic.md: Likewise.
10729 * config/arm/arm1020e.md: Likewise.
10730 * config/arm/arm1026ejs.md: Likewise.
10731 * config/arm/arm1136jfs.md: Likewise.
10732 * config/arm/arm926ejs.md: Likewise.
10733 * config/arm/cortex-a15.md: Likewise.
10734 * config/arm/cortex-a17.md: Likewise.
10735 * config/arm/cortex-a5.md: Likewise.
10736 * config/arm/cortex-a53.md: Likewise.
10737 * config/arm/cortex-a57.md: Likewise.
10738 * config/arm/cortex-a7.md: Likewise.
10739 * config/arm/cortex-a8.md: Likewise.
10740 * config/arm/cortex-a9.md: Likewise.
10741 * config/arm/cortex-m4.md: Likewise.
10742 * config/arm/cortex-m7.md: Likewise.
10743 * config/arm/cortex-r4.md: Likewise.
10744 * config/arm/exynos-m1.md: Likewise.
10745 * config/arm/fa526.md: Likewise.
10746 * config/arm/fa606te.md: Likewise.
10747 * config/arm/fa626te.md: Likewise.
10748 * config/arm/fa726te.md: Likewise.
10749 * config/arm/fmp626.md: Likewise.
10750 * config/arm/iwmmxt.md: Likewise.
10751 * config/arm/ldmstm.md: Likewise.
10752 * config/arm/marvell-pj4.md: Likewise.
10753 * config/arm/xgene1.md: Likewise.
10754 * config/aarch64/thunderx.md: Likewise.
10755 * config/aarch64/thunderx2t99.md: Likewise.
10756 * config/aarch64/falkor.md: Likewise.
10758 2017-09-12 Martin Liska <mliska@suse.cz>
10760 * attribs.c (private_lookup_attribute): New function.
10761 * attribs.h (private_lookup_attribute): Declared here.
10762 (lookup_attribute): Called from this place.
10764 2017-09-12 Richard Biener <rguenther@suse.de>
10766 PR tree-optimization/82157
10767 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10768 stmts with side-effects.
10770 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10771 Alan Hayward <alan.hayward@arm.com>
10772 David Sherwood <david.sherwood@arm.com>
10774 * target.def (hard_regno_nregs): New hook.
10775 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10776 * targhooks.h (default_hard_regno_nregs): Declare.
10777 * targhooks.c (default_hard_regno_nregs): New function.
10778 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10779 (TARGET_HARD_REGNO_NREGS): ...this hook.
10780 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10781 (CLASS_MAX_NREGS): Likewise.
10782 * doc/tm.texi: Regenerate.
10783 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10784 instead of HARD_REGNO_NREGS.
10785 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10786 HARD_REGNO_NREGS in the comment.
10787 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10788 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10789 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10790 Return an unsigned int.
10791 (TARGET_HARD_REGNO_NREGS): Redefine.
10792 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10793 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10794 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10795 (arc_hard_regno_nregs): New function.
10796 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10797 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10798 (arm_hard_regno_nregs): New function.
10799 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10800 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10801 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10802 (TARGET_HARD_REGNO_NREGS): Redefine.
10803 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10804 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10805 (HARD_REGNO_NREGS): Delete.
10806 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10807 (cr16_hard_regno_nregs): New function.
10808 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10809 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10810 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10811 (cris_hard_regno_nregs): New function.
10812 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10813 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10814 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10815 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10816 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10817 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10818 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10819 (frv_hard_regno_nregs): Make static. Take and return an
10821 (frv_class_max_nregs): Remove outdated copy of documentation.
10822 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10823 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10824 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10825 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10826 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10827 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10828 (TARGET_HARD_REGNO_NREGS): Redefine.
10829 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10830 (CLASS_MAX_NREGS): Update comment.
10831 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10832 (ia64_hard_regno_nregs): New function.
10833 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10834 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10835 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10836 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10837 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10839 (m32c_hard_regno_nregs): Likewise. Make static.
10840 (TARGET_HARD_REGNO_NREGS): Redefine.
10841 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10842 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10843 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10844 (m68k_hard_regno_nregs): New function.
10845 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10846 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10847 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10848 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10849 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10850 Take and return an unsigned int.
10851 (TARGET_HARD_REGNO_NREGS): Redefine.
10852 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10853 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10854 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10855 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10856 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10857 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10858 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10859 (msp430_hard_regno_nregs): Make static. Take and return an
10861 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10862 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10863 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10864 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10865 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10866 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10867 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10868 (TARGET_HARD_REGNO_NREGS): Redefine.
10869 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10870 (PA_HARD_REGNO_NREGS): ...this.
10871 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10872 (PA_HARD_REGNO_NREGS): ...this.
10873 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10874 (pa_hard_regno_nregs): New function.
10875 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10876 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10877 (pdp11_hard_regno_nregs): New function.
10878 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10879 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10880 (rs6000_hard_regno_nregs_hook): New function.
10881 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10882 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10883 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10884 Take and return an unsigned int. Move earlier in file.
10885 (TARGET_HARD_REGNO_NREGS): Redefine.
10886 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10887 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10888 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10889 (rl78_hard_regno_nregs): Make static. Take and return an
10891 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10892 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10893 (rs6000_hard_regno_nregs_hook): New function.
10894 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10895 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10896 (TARGET_HARD_REGNO_NREGS): Redefine.
10897 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10898 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10899 instead of HARD_REGNO_NREGS.
10900 (s390_hard_regno_nregs): New function.
10901 (s390_hard_regno_mode_ok): Add comment from s390.h.
10902 (TARGET_HARD_REGNO_NREGS): Redefine.
10903 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10904 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10905 (sh_hard_regno_nregs): New function.
10906 (sh_pass_in_reg_p): Use it.
10907 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10908 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10909 (sparc_hard_regno_nregs): New function.
10910 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10911 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10912 (spu_function_arg_advance): Use it, supplying a valid register number.
10913 (TARGET_HARD_REGNO_NREGS): Redefine.
10914 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10915 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10916 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10917 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10918 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10919 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10920 (CLASS_MAX_NREGS): Remove copy of old documentation.
10921 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10922 (visium_hard_regno_nregs): New function.
10923 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10924 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10925 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10926 xtensa_hard_regno_nregs): New function.
10927 * system.h (HARD_REGNO_NREGS): Poison.
10929 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10931 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10932 hard_regno_nregs instead of HARD_REGNO_NREGS.
10933 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10934 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10935 (c6x_expand_epilogue): Likewise.
10936 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10937 (frv_read_iacc_argument): Likewise.
10938 * config/sh/sh.c: Include regs.h.
10939 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10940 (regs_used): Likewise.
10941 (output_stack_adjust): Likewise.
10942 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10943 * expmed.c: Include regs.h.
10944 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10945 * ree.c: Include regs.h.
10946 (combine_reaching_defs): Use hard_regno_nregs instead of
10948 (add_removable_extension): Likewise.
10950 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10952 * regs.h (hard_regno_nregs): Turn into a function.
10953 (end_hard_regno): Update accordingly.
10954 * caller-save.c (setup_save_areas): Likewise.
10955 (save_call_clobbered_regs): Likewise.
10956 (replace_reg_with_saved_mem): Likewise.
10957 (insert_restore): Likewise.
10958 (insert_save): Likewise.
10959 * combine.c (can_change_dest_mode): Likewise.
10960 (move_deaths): Likewise.
10961 (distribute_notes): Likewise.
10962 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10963 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10964 (rs6000_split_multireg_move): Likewise.
10965 (rs6000_register_move_cost): Likewise.
10966 (rs6000_memory_move_cost): Likewise.
10967 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10968 (rs6000_split_multireg_move): Likewise.
10969 (rs6000_register_move_cost): Likewise.
10970 (rs6000_memory_move_cost): Likewise.
10971 * cselib.c (cselib_reset_table): Likewise.
10972 (cselib_lookup_1): Likewise.
10973 * emit-rtl.c (set_mode_and_regno): Likewise.
10974 * function.c (aggregate_value_p): Likewise.
10975 * ira-color.c (setup_profitable_hard_regs): Likewise.
10976 (check_hard_reg_p): Likewise.
10977 (calculate_saved_nregs): Likewise.
10978 (assign_hard_reg): Likewise.
10979 (improve_allocation): Likewise.
10980 (calculate_spill_cost): Likewise.
10981 * ira-emit.c (modify_move_list): Likewise.
10982 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10983 (ira_hard_reg_in_set_p): Likewise.
10984 * ira.c (setup_reg_mode_hard_regset): Likewise.
10985 (clarify_prohibited_class_mode_regs): Likewise.
10986 (check_allocation): Likewise.
10987 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10988 (lra_setup_reg_renumber): Likewise.
10989 (setup_try_hard_regno_pseudos): Likewise.
10990 (spill_for): Likewise.
10991 (assign_hard_regno): Likewise.
10992 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10993 * lra-constraints.c (in_class_p): Likewise.
10994 (lra_constraint_offset): Likewise.
10995 (simplify_operand_subreg): Likewise.
10996 (lra_constraints): Likewise.
10997 (split_reg): Likewise.
10998 (split_if_necessary): Likewise.
10999 (invariant_p): Likewise.
11000 (inherit_in_ebb): Likewise.
11001 * lra-lives.c (process_bb_lives): Likewise.
11002 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11003 (get_hard_regs): Likewise.
11004 (do_remat): Likewise.
11005 * lra-spills.c (assign_spill_hard_regs): Likewise.
11006 * mode-switching.c (create_pre_exit): Likewise.
11007 * postreload.c (reload_combine_recognize_pattern): Likewise.
11008 * recog.c (peep2_find_free_register): Likewise.
11009 * regcprop.c (kill_value_regno): Likewise.
11010 (set_value_regno): Likewise.
11011 (copy_value): Likewise.
11012 (maybe_mode_change): Likewise.
11013 (find_oldest_value_reg): Likewise.
11014 (copyprop_hardreg_forward_1): Likewise.
11015 * regrename.c (check_new_reg_p): Likewise.
11016 (regrename_do_replace): Likewise.
11017 * reload.c (push_reload): Likewise.
11018 (combine_reloads): Likewise.
11019 (find_dummy_reload): Likewise.
11020 (operands_match_p): Likewise.
11021 (find_reloads): Likewise.
11022 (find_equiv_reg): Likewise.
11023 (reload_adjust_reg_for_mode): Likewise.
11024 * reload1.c (count_pseudo): Likewise.
11025 (count_spilled_pseudo): Likewise.
11026 (find_reg): Likewise.
11027 (clear_reload_reg_in_use): Likewise.
11028 (free_for_value_p): Likewise.
11029 (allocate_reload_reg): Likewise.
11030 (choose_reload_regs): Likewise.
11031 (reload_adjust_reg_for_temp): Likewise.
11032 (emit_reload_insns): Likewise.
11033 (delete_output_reload): Likewise.
11034 * rtlanal.c (subreg_get_info): Likewise.
11035 * sched-deps.c (sched_analyze_reg): Likewise.
11036 * sel-sched.c (init_regs_for_mode): Likewise.
11037 (mark_unavailable_hard_regs): Likewise.
11038 (choose_best_reg_1): Likewise.
11039 (verify_target_availability): Likewise.
11040 * valtrack.c (dead_debug_insert_temp): Likewise.
11041 * var-tracking.c (track_loc_p): Likewise.
11042 (emit_note_insn_var_location): Likewise.
11043 * varasm.c (make_decl_rtl): Likewise.
11044 * reginfo.c (choose_hard_reg_mode): Likewise.
11045 (init_reg_modes_target): Refer directly to
11046 this_target_regs->x_hard_regno_nregs.
11048 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11050 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11051 instead of hard_regno_nregs.
11053 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11055 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11056 end_hard_regno instead of hard_regno_nregs.
11057 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11058 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11059 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11060 * ira-color.c (improve_allocation): Likewise.
11061 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11062 * lra-lives.c (mark_regno_live): Likewise.
11063 (mark_regno_dead): Likewise.
11064 * lra-remat.c (operand_to_remat): Likewise.
11065 * lra.c (collect_non_operand_hard_regs): Likewise.
11066 * postreload.c (reload_combine_note_store): Likewise.
11067 (move2add_valid_value_p): Likewise.
11068 * reload.c (regno_clobbered_p): Likewise.
11070 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11072 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11074 * config/v850/v850.c (v850_reorg): Likewise.
11075 * reload.c (refers_to_regno_for_reload_p): Likewise.
11076 (find_equiv_reg): Likewise.
11077 * reload1.c (reload_reg_reaches_end_p): Likewise.
11079 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11081 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11083 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11084 * config/arm/arm.c (output_move_neon): Likewise.
11085 (arm_attr_length_move_neon): Likewise.
11086 (neon_split_vcombine): Likewise.
11087 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11088 (c6x_mark_reg_written): Likewise.
11089 (c6x_dwarf_register_span): Likewise.
11090 * config/i386/i386.c (ix86_save_reg): Likewise.
11091 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11092 (rws_access_reg): Likewise.
11093 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11094 * mode-switching.c (create_pre_exit): Likewise.
11095 * ree.c (combine_reaching_defs): Likewise.
11096 (add_removable_extension): Likewise.
11097 * regcprop.c (find_oldest_value_reg): Likewise.
11098 (copyprop_hardreg_forward_1): Likewise.
11099 * reload.c (reload_inner_reg_of_subreg): Likewise.
11100 (push_reload): Likewise.
11101 (combine_reloads): Likewise.
11102 (find_dummy_reload): Likewise.
11103 (reload_adjust_reg_for_mode): Likewise.
11104 * reload1.c (find_reload_regs): Likewise.
11105 (forget_old_reloads_1): Likewise.
11106 (reload_reg_free_for_value_p): Likewise.
11107 (reload_adjust_reg_for_temp): Likewise.
11108 (emit_reload_insns): Likewise.
11109 (delete_output_reload): Likewise.
11110 * sel-sched.c (choose_best_reg_1): Likewise.
11111 (choose_best_pseudo_reg): Likewise.
11113 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11114 Alan Hayward <alan.hayward@arm.com>
11115 David Sherwood <david.sherwood@arm.com>
11117 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11118 * target.def (slow_unaligned_access): New hook.
11119 * targhooks.h (default_slow_unaligned_access): Declare.
11120 * targhooks.c (default_slow_unaligned_access): New function.
11121 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11122 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11123 * doc/tm.texi: Regenerate.
11124 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11125 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11126 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11128 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11129 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11131 (rs6000_slow_unaligned_access): New function.
11132 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11133 (expand_block_compare): Likewise.
11134 (expand_strn_compare): Likewise.
11135 (rs6000_rtx_costs): Likewise.
11136 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11137 (riscv_slow_unaligned_access): Likewise.
11138 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11139 (riscv_slow_unaligned_access_p): ...this and make static.
11140 (riscv_option_override): Update accordingly.
11141 (riscv_slow_unaligned_access): New function.
11142 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11143 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11144 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11145 (rs6000_slow_unaligned_access): New function.
11146 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11147 (rs6000_rtx_costs): Likewise.
11148 * config/rs6000/rs6000-string.c (expand_block_compare)
11149 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11150 of SLOW_UNALIGNED_ACCESS.
11151 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11152 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11153 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11154 of SLOW_UNALIGNED_ACCESS.
11155 * expmed.c (simple_mem_bitfield_p): Likewise.
11156 * expr.c (alignment_for_piecewise_move): Likewise.
11157 (emit_group_load_1): Likewise.
11158 (emit_group_store): Likewise.
11159 (copy_blkmode_from_reg): Likewise.
11160 (emit_push_insn): Likewise.
11161 (expand_assignment): Likewise.
11162 (store_field): Likewise.
11163 (expand_expr_real_1): Likewise.
11164 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11165 * lra-constraints.c (simplify_operand_subreg): Likewise.
11166 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11167 * gimple-ssa-store-merging.c: Likewise in block comment at start
11169 * tree-ssa-strlen.c: Include target.h.
11170 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11171 of SLOW_UNALIGNED_ACCESS.
11172 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11174 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11176 PR rtl-optimization/82185
11177 * expmed.c (emit_store_flag_int): Only test tem if it has been
11180 2017-09-12 Richard Biener <rguenther@suse.de>
11182 PR middle-end/82149
11183 * match.pd ((FTYPE) N CMP CST): Fix typo.
11185 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11187 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11189 (mips_near_type_p): Add 'short_call' attribute as a synonym
11191 * doc/extend.texi (short_call): Document new function attribute.
11193 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11196 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11197 assertion check that in the condition.
11198 (get_atomic_generic_size): Likewise. Before testing if parameter
11199 has pointer type, if it has array type, call for C++
11200 default_conversion to perform array-to-pointer conversion.
11202 2017-09-12 Richard Biener <rguenther@suse.de>
11204 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11205 for operations we cannot scalarize.
11207 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11209 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11210 vectors heap vectors. Clean up comments.
11211 Make visited_bbs a reference.
11212 (profitable_jump_thread_path): Make GC
11213 vectors heap vectors. Clean up comments.
11215 (convert_and_register_jump_thread_path): Make GC vectors heap
11217 (check_subpath_and_update_thread_path): Same. Clean up comments.
11218 Make visited_bbs a reference.
11219 (handle_phi): Abstract common code to to
11220 register_jump_thread_path_if_profitable.
11221 Rename VAR_BB to DEF_BB.
11223 Make GC vectors heap vectors.
11224 Make visited_bbs a reference.
11225 (handle_assignment): Same.
11226 (register_jump_thread_path_if_profitable): New.
11227 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11229 Make GC vectors heap vectors. Clean up comments.
11230 Make visited_bbs a reference.
11231 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11232 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11235 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11238 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11239 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11241 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11243 Revert r251800 and r251799.
11245 2017-09-11 Martin Jambor <mjambor@suse.cz>
11248 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11249 arguments in advance.
11250 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11251 use it to find predecessor edges.
11252 (naive_outof_ssa): Collect vector of predecessors.
11254 2017-09-08 Jason Merrill <jason@redhat.com>
11256 PR c++/70029 - ICE with ref-qualifier and -flto
11257 * langhooks.h (struct lang_hooks_for_types): Add
11258 copy_lang_qualifiers.
11259 * attribs.c (build_type_attribute_qual_variant): Use it.
11260 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11262 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11263 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11265 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11268 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11269 (*mulsi3_sp64): New instruction.
11270 (mulsi3): New expander.
11272 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11274 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11276 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11278 * sancov.c: Include memmodel.h.
11280 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11283 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11286 2017-09-07 Carl Love <cel@us.ibm.com>
11288 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11289 the sldi instruction.
11291 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11293 * sancov.c: Include tm_p.h.
11295 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11298 * output.h (switch_to_other_text_partition): New declaration.
11299 * varasm.c (switch_to_other_text_partition): New function.
11300 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11301 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11302 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11303 to the other text partition before emitting LCL label and switch back
11304 after emitting the word after it.
11306 2017-09-07 Richard Biener <rguenther@suse.de>
11308 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11309 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11310 critical edges here, after loop init.
11311 (pass_data_pre): Remove PROP_no_crit_edges flags.
11312 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11313 for valueization of call args to avoid leaking VN_TOP.
11314 (visit_use): Assert we do not visit default defs.
11315 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11316 Use error_mark_node to more easily detect leaking VN_TOP.
11317 All default-defs are varying, not VN_TOP. Mark them visited.
11318 (run_scc_vn): Make code match comment.
11320 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11322 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11323 OPTION_MASK_FLOAT128_KEYWORD.
11324 (POWERPC_MASKS): Likewise.
11325 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11326 support for the -mfloat128-type option, and make -mfloat128
11327 default on PowerPC Linux systems. Define or undefine
11328 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11329 Define __float128 to be __ieee128 if IEEE 128-bit support is
11330 enabled, or undefine it.
11331 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11332 Delete defining __FLOAT128_TYPE__.
11333 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11334 -mfloat128-type option and make -mfloat128 default on PowerPC
11336 (TARGET_FLOAT128_TYPE): Likewise.
11337 (-mfloat128-type): Likewise.
11338 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11339 Delete the -mfloat128-type option and make -mfloat128 default on
11340 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11341 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11342 128-bit floating point is enabled. Change tests from using
11343 -mfloat128-type to -mfloat128.
11344 (rs6000_mangle_type): Use the correct mangling for the __float128
11345 type even if normal long double is restricted to 64-bits.
11346 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11348 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11349 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11350 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11352 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11353 documentation for -mfloat128.
11355 2017-09-06 Olivier Hainque <hainque@adacore.com>
11357 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11359 2017-09-06 Wish Wu <wishwu007@gmail.com>
11360 Jakub Jelinek <jakub@redhat.com>
11362 * asan.c (initialize_sanitizer_builtins): Add
11363 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11364 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11365 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11366 BT_FN_VOID_UINT64_PTR variables.
11367 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11368 (BT_FN_VOID_UINT16_UINT16): Likewise.
11369 (BT_FN_VOID_UINT32_UINT32): Likewise.
11370 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11371 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11372 (BT_FN_VOID_UINT64_PTR): Likewise.
11373 * common.opt (flag_sanitize_coverage): New variable.
11374 (fsanitize-coverage=trace-pc): Remove.
11375 (fsanitize-coverage=): Add.
11376 * flag-types.h (enum sanitize_coverage_code): New enum.
11377 * fold-const.c (fold_range_test): Disable non-short-circuit
11378 optimization if flag_sanitize_coverage.
11379 (fold_truth_andor): Likewise.
11380 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11381 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11382 (coverage_sanitizer_opts): New array.
11383 (get_closest_sanitizer_option): Add OPTS argument, handle also
11384 OPT_fsanitize_coverage_.
11385 (parse_sanitizer_options): Adjusted to also handle
11386 OPT_fsanitize_coverage_.
11387 (common_handle_option): Add OPT_fsanitize_coverage_.
11388 * sancov.c (instrument_comparison, instrument_switch): New function.
11389 (sancov_pass): Add trace-cmp support.
11390 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11391 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11392 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11393 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11394 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11395 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11396 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11397 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11398 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11399 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11401 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11403 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11404 error. Only quit immediately if parsing is complete.
11405 (BEGIN): Initialize fatal_err and parse_done.
11406 (begin fpu, end fpu): Check number of arguments.
11407 (begin arch, end arch): Likewise.
11408 (begin cpu, end cpu): Likewise.
11409 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11410 (optalias): Likewise.
11412 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11414 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11415 * config/arm/arm-isa.h: Delete. Move definitions to ...
11416 * arm-cpus.in: ... here. Use new feature and fgroup values.
11417 * config/arm/arm.c (arm_option_override): Use lower case for feature
11419 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11420 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11421 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11423 (print_isa_bits_for): New function.
11424 (gen_isa): New function.
11425 (gen_comm_data): Use print_isa_bits_for.
11426 (define feature): New keyword.
11427 (define fgroup): New keyword.
11428 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11429 (arm-isa.h): Add rule to generate file.
11430 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11431 case for feature bit names.
11433 2017-09-06 Richard Biener <rguenther@suse.de>
11435 * tree-ssa-pre.c (NECESSARY): Remove.
11436 (create_expression_by_pieces): Do not touch pass-local flags.
11437 (insert_into_preds_of_block): Likewise.
11438 (do_pre_regular_insertion): Likewise.
11439 (eliminate_insert): Likewise.
11440 (eliminate_dom_walker::before_dom_children): Likewise.
11441 (fini_eliminate): Do not look at inserted_exprs.
11442 (mark_operand_necessary): Remove.
11443 (remove_dead_inserted_code): Replace with simple work-list
11444 algorithm based on inserted_exprs and SSA uses.
11445 (pass_pre::execute): Re-order fini_eliminate and
11446 remove_dead_inserted_code.
11448 2017-09-06 Olivier Hainque <hainque@adacore.com>
11450 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11451 for VxWorks 7. Adjust surrounding comments.
11453 2017-09-06 Richard Biener <rguenther@suse.de>
11455 * gimple-ssa-strength-reduction.c
11456 (find_candidates_dom_walker::before_dom_children): Also allow
11459 2017-09-06 Richard Biener <rguenther@suse.de>
11461 PR tree-optimization/82108
11462 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11463 for gap in the non-permutation SLP case.
11465 2017-09-06 Martin Jambor <mjambor@suse.cz>
11467 PR tree-optimization/82078
11468 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11469 add_access_to_work_queue...
11470 (build_accesses_from_assign): ...here.
11471 (propagate_all_subaccesses): Make sure racc is the group
11472 representative, if there is one.
11474 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11476 PR middle-end/82095
11477 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11480 2017-09-06 Richard Biener <rguenther@suse.de>
11482 * gimple-ssa-strength-reduction.c
11483 (find_candidates_dom_walker::before_doom_children): Use a
11484 type and not a mode check.
11486 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11489 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11491 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11493 2017-09-05 Jeff Law <law@redhat.com>
11495 PR tree-optimization/64910
11496 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11497 cases where we have 3 or more operands.
11499 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11501 PR middle-end/81768
11502 * omp-low.c (lower_omp_for): Recompute tree invariant if
11503 gimple_omp_for_initial/final is ADDR_EXPR.
11505 PR middle-end/81768
11506 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11507 into gimple val before gimplification fo the COND_EXPR.
11509 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11511 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11512 REGION_COPY argument.
11513 (thread_through_all_blocks): Remove unused argument to
11514 duplicate_thread_path.
11516 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11517 Alan Hayward <alan.hayward@arm.com>
11518 David Sherwood <david.sherwood@arm.com>
11520 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11521 Take a scalar_mode rather than a machine_mode.
11522 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11523 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11524 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11525 (aarch64_gen_adjusted_ldpstp): Likewise.
11526 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11528 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11529 Alan Hayward <alan.hayward@arm.com>
11530 David Sherwood <david.sherwood@arm.com>
11532 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11533 Take a scalar_int_mode instead of a machine_mode.
11534 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11535 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11536 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11537 (aarch64_simd_attr_length_rglist): Delete.
11538 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11539 a scalar_int_mode instead of a machine_mode.
11540 (aarch64_add_offset): Likewise.
11541 (aarch64_internal_mov_immediate): Likewise
11542 (aarch64_add_constant_internal): Likewise.
11543 (aarch64_add_constant): Likewise.
11544 (aarch64_movw_imm): Likewise.
11545 (aarch64_rtx_arith_op_extract_p): Likewise.
11546 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11547 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11548 Remove assert that the mode isn't a vector.
11549 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11550 (aarch64_expand_mov_immediate): Update calls after above changes.
11551 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11552 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11553 (aarch64_move_imm): Likewise.
11554 (aarch64_can_const_movi_rtx_p): Likewise.
11555 (aarch64_strip_extend): Likewise.
11556 (aarch64_extr_rtx_p): Likewise.
11557 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11558 a CONST_INT when the mode parameter is VOIDmode.
11559 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11561 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11563 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11564 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11565 (bitwise_type_for_mode): Update accordingly.
11567 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11569 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11570 * stor-layout.c (mode_for_size_tree): Likewise.
11571 (mode_for_array): Update accordingly.
11572 (layout_decl): Likewise.
11573 (compute_record_mode): Likewise. Only set the mode once.
11575 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11577 * target.def (get_mask_mode): Change return type to opt_mode.
11579 * doc/tm.texi: Regenerate.
11580 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11581 * targhooks.c (default_get_mask_mode): Likewise.
11582 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11583 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11584 targetm.get_mask_mode.
11585 * tree.c (build_truth_vector_type): Likewise.
11587 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11589 * machmode.h (mode_for_vector): Return an opt_mode.
11590 * stor-layout.c (mode_for_vector): Likewise.
11591 (mode_for_int_vector): Update accordingly.
11592 (layout_type): Likewise.
11593 * config/i386/i386.c (emit_memmov): Likewise.
11594 (ix86_expand_set_or_movmem): Likewise.
11595 (ix86_expand_vector_init): Likewise.
11596 (ix86_get_mask_mode): Likewise.
11597 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11599 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11600 * expmed.c (extract_bit_field_1): Likewise.
11601 * expr.c (expand_expr_real_2): Likewise.
11602 * optabs-query.c (can_vec_perm_p): Likewise.
11603 (can_vec_mask_load_store_p): Likewise.
11604 * optabs.c (expand_vec_perm): Likewise.
11605 * targhooks.c (default_get_mask_mode): Likewise.
11606 * tree-vect-stmts.c (vectorizable_store): Likewise.
11607 (vectorizable_load): Likewise.
11608 (get_vectype_for_scalar_type_and_size): Likewise.
11610 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11612 * machmode.h (mode_for_int_vector): New function.
11613 * stor-layout.c (mode_for_int_vector): Likewise.
11614 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11615 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11616 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11617 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11618 (s390_expand_vcond): Likewise.
11620 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11622 * machmode.h (opt_machine_mode): New type.
11623 (opt_mode<T>): Allow construction from anything that can be
11625 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11626 (mode_for_size): Return an opt_machine_mode.
11627 * stor-layout.c (mode_for_size): Likewise.
11628 (mode_for_size_tree): Update call accordingly.
11629 (bitwise_mode_for_mode): Likewise.
11630 (make_fract_type): Likewise.
11631 (make_accum_type): Likewise.
11632 * caller-save.c (replace_reg_with_saved_mem): Update call
11634 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11635 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11636 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11637 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11638 * expmed.c (extract_bit_field_1): Likewise.
11639 * reload.c (get_secondary_mem): Likewise.
11640 * varasm.c (assemble_integer): Likewise.
11641 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11644 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11646 * machmode.h (decimal_float_mode_for_size): New function.
11647 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11648 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11649 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11650 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11651 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11652 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11654 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11656 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11657 (get_builtin_sync_mode): Likewise.
11658 (expand_ifn_atomic_compare_exchange): Likewise.
11659 (expand_builtin_atomic_clear): Likewise.
11660 (expand_builtin_atomic_test_and_set): Likewise.
11661 (fold_builtin_atomic_always_lock_free): Likewise.
11662 * calls.c (compute_argument_addresses): Likewise.
11663 (emit_library_call_value_1): Likewise.
11664 (store_one_arg): Likewise.
11665 * combine.c (combine_instructions): Likewise.
11666 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11667 * config/arm/arm.c (arm_function_value): Likewise.
11668 (aapcs_allocate_return_reg): Likewise.
11669 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11670 * config/i386/i386.c (construct_container): Likewise.
11671 (ix86_gimplify_va_arg): Likewise.
11672 (ix86_expand_sse_cmp): Likewise.
11673 (emit_memmov): Likewise.
11674 (emit_memset): Likewise.
11675 (expand_small_movmem_or_setmem): Likewise.
11676 (ix86_expand_pextr): Likewise.
11677 (ix86_expand_pinsr): Likewise.
11678 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11679 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11681 * config/mips/mips.c (mips_function_value_1) Likewise.
11682 (mips_block_move_straight): Likewise.
11683 (mips_expand_ins_as_unaligned_store): Likewise.
11684 * config/powerpcspe/powerpcspe.c
11685 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11686 (rs6000_darwin64_record_arg_flush): Likewise.
11687 * config/rs6000/rs6000.c
11688 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11689 (rs6000_darwin64_record_arg_flush): Likewise.
11690 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11691 (sparc_function_value_1): Likewise.
11692 * config/spu/spu.c (adjust_operand): Likewise.
11693 (spu_emit_branch_or_set): Likewise.
11694 (arith_immediate_p): Likewise.
11695 * emit-rtl.c (gen_lowpart_common): Likewise.
11696 * expr.c (expand_expr_real_1): Likewise.
11697 * function.c (assign_parm_setup_block): Likewise.
11698 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11699 * reload1.c (alter_reg): Likewise.
11700 * stor-layout.c (mode_for_vector): Likewise.
11701 (layout_type): Likewise.
11703 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11705 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11706 (spu_convert_move): Likewise.
11707 * lower-subreg.c (resolve_simple_move): Likewise.
11709 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11712 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11713 define_insn to a define_expand.
11714 (altivec_vsum2sws_direct): New define_insn.
11715 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11717 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11719 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11722 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11727 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11728 realign stack if stack alignment needed is less than incoming
11731 2017-09-05 Marek Polacek <polacek@redhat.com>
11734 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11737 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11739 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11741 2017-09-05 Richard Biener <rguenther@suse.de>
11743 PR tree-optimization/82084
11744 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11746 2017-09-05 Richard Biener <rguenther@suse.de>
11748 PR tree-optimization/82102
11749 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11751 2017-09-05 Martin Liska <mliska@suse.cz>
11753 PR tree-optimization/82032
11754 * tree-cfg.c (generate_range_test): New function.
11755 * tree-cfg.h (generate_range_test): Declared here.
11756 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11757 (cleanup_control_expr_graph): Use it.
11758 * tree-switch-conversion.c (try_switch_expansion): Remove
11760 (emit_case_nodes): Use generate_range_test.
11762 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11765 * config/i386/i386.md (*<btsc><mode>_mask): Add
11766 TARGET_USE_BT to insn constraint.
11767 (*btr<mode>_mask): Ditto.
11769 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11771 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11772 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11774 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11777 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11778 TARGET_NEON and TARGET_IWMMXT.
11779 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11780 TARGET_NEON and TARGET_IWMMXT.
11781 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11783 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11785 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11786 (ix86_rewrite_tls_address): Ditto.
11787 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11788 (ix86_rewrite_tls_address_1): Ditto.
11789 (ix86_rewrite_tls_address): Ditto.
11790 * config/i386/predicates.md (tls_address_pattern): New predicate.
11791 * config/i386/i386.md (TLS address splitter): New splitter.
11793 2017-09-04 Richard Biener <rguenther@suse.de>
11795 PR tree-optimization/82084
11796 * fold-const.h (can_native_encode_string_p): Declare.
11797 * fold-const.c (can_native_encode_string_p): Factor out from ...
11798 (native_encode_string): ... here.
11799 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11800 vectorizing stores from constants we later cannot handle.
11802 2017-09-04 Marek Polacek <polacek@redhat.com>
11805 * doc/invoke.texi: Update -Wtautological-compare documentation.
11807 2017-09-04 Jeff Law <law@redhat.com>
11809 PR tree-optimization/64910
11810 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11811 swap the first and last operand if the last is a constant.
11813 2017-09-04 Marek Polacek <polacek@redhat.com>
11816 * convert.c (do_narrow): When sanitizing signed integer overflows,
11817 bail out for signed types.
11818 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11820 2017-09-04 Richard Biener <rguenther@suse.de>
11822 PR tree-optimization/82060
11823 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11824 Move devirtualization after stmt folding and before EH/AB/noreturn
11825 cleanup to get the stmt refs canonicalized. Use a bool instead
11826 of gimple_modified_p since that doesn't work for NOPs. Schedule
11827 NOPs generated by folding for removal.
11829 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11830 Alan Hayward <alan.hayward@arm.com>
11831 David Sherwood <david.sherwood@arm.com>
11833 * coretypes.h (pad_direction): New enum.
11834 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11835 (FUNCTION_ARG_PADDING): Likewise.
11836 * target.def (function_arg_padding): New hook.
11837 * targhooks.h (default_function_arg_padding): Declare.
11838 * targhooks.c (default_function_arg_padding): New function.
11839 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11840 (TARGET_FUNCTION_ARG_PADDING): ...this.
11841 * doc/tm.texi: Regenerate.
11842 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11843 instead of direction.
11844 (compute_argument_addresses): Likewise.
11845 (load_register_parameters): Likewise.
11846 (emit_library_call_value_1): Likewise.
11847 (store_one_arg): Use targetm.calls.function_arg_padding instead
11848 of FUNCTION_ARG_PADDING.
11849 (must_pass_in_stack_var_size_or_pad): Likewise.
11850 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11851 (emit_group_store): Likewise.
11852 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11853 instead of FUNCTION_ARG_PADDING.
11854 (emit_push_insn): Likewise, and propagate enum change throughout
11856 * function.h (direction): Delete.
11857 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11859 * function.c (assign_parm_find_stack_rtl): Likewise.
11860 (assign_parm_setup_block_p): Likewise.
11861 (assign_parm_setup_block): Likewise.
11862 (gimplify_parameters): Likewise.
11863 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11864 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11866 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11867 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11868 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11869 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11870 (aarch64_function_arg_padding): ...this new function.
11871 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11872 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11873 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11874 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11875 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11876 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11877 (arm_pad_arg_upward): Replace with...
11878 (arm_function_arg_padding): ...this new function.
11879 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11881 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11882 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11883 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11884 (ia64_hpux_function_arg_padding): Replace with...
11885 (ia64_function_arg_padding): ...this new function. Use pad_direction
11886 instead of direction. Check for TARGET_HPUX.
11887 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11888 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11889 (iq2000_function_arg_padding): New function.
11890 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11891 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11892 (mips_function_arg_padding): ...this new function.
11893 (mips_pad_reg_upward): Update accordingly.
11894 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11895 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11896 targetm.calls.function_arg_padding.
11897 (FUNCTION_ARG_PADDING): Delete.
11898 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11899 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11900 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11901 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11902 (nios2_block_reg_padding): Return pad_direction instead of direction.
11903 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11904 instead of direction.
11905 (nios2_function_arg_padding): Likewise. Make static.
11906 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11907 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11908 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11909 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11910 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11911 (pa_function_arg_padding): Make static. Return pad_direction instead
11913 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11914 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11915 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11916 instead of direction. Use targetm.calls.function_arg_padding.
11917 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11918 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11919 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11920 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11921 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11923 (function_arg_padding): Rename to...
11924 (rs6000_function_arg_padding): ...this. Make static. Return
11925 pad_direction instead of direction.
11926 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11927 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11928 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11929 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11930 instead of direction. Use targetm.calls.function_arg_padding.
11931 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11932 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11933 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11934 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11935 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11936 (function_arg_padding): Rename to...
11937 (rs6000_function_arg_padding): ...this. Make static. Return
11938 pad_direction instead of direction.
11939 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11940 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11941 * config/s390/s390.c (s390_function_arg_padding): New function.
11942 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11943 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11944 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11945 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11946 (function_arg_padding): Rename to...
11947 (sparc_function_arg_padding): ...this. Make static. Return
11948 pad_direction instead of direction.
11949 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11950 * config/spu/spu.c (spu_function_arg_padding): New function.
11951 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11952 * system.h (FUNCTION_ARG_PADDING): Poison.
11954 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11955 Alan Hayward <alan.hayward@arm.com>
11956 David Sherwood <david.sherwood@arm.com>
11958 * target.def (modes_tieable_p): New hook.
11959 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11960 (TARGET_MODES_TIEABLE_P): ...this.
11961 * doc/tm.texi.in: Regenerate.
11962 * hooks.h (hook_bool_mode_mode_true): Declare.
11963 * hooks.c (hook_bool_mode_mode_true): New function.
11964 * combine.c (subst): Use targetm.modes_tieable_p instead of
11966 * dse.c (find_shift_sequence): Likewise.
11967 * expmed.c (extract_low_bits): Likewise.
11968 * lower-subreg.c: Include target.h.
11969 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11971 * rtlanal.c (rtx_cost): Likewise.
11972 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11973 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11974 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11975 (TARGET_MODES_TIEABLE_P): Redefine.
11976 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11977 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11978 (TARGET_MODES_TIEABLE_P): Redefine.
11979 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11980 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11981 (arc_modes_tieable_p): New function.
11982 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11983 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11984 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11985 (arm_modes_tieable_p): Make static.
11986 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11987 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11988 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11989 (TARGET_MODES_TIEABLE_P): Redefine.
11990 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11991 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11992 (TARGET_MODES_TIEABLE_P): Redefine.
11993 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11994 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11995 (cr16_modes_tieable_p): New function.
11996 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11997 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11998 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11999 (TRULY_NOOP_TRUNCATION): Update comment.
12000 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12001 (TRULY_NOOP_TRUNCATION): Update comment.
12002 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12003 (frv_modes_tieable_p): New function.
12004 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12005 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12006 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12007 (TARGET_MODES_TIEABLE_P): Redefine.
12008 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12009 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12010 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12011 (TARGET_MODES_TIEABLE_P): Redefine.
12012 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12013 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12014 (ia64_modes_tieable_p): New function.
12015 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12016 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12017 (iq2000_modes_tieable_p): New function.
12018 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12019 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12020 (lm32_modes_tieable_p): New function.
12021 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12022 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12023 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12024 (TARGET_MODES_TIEABLE_P): Redefine.
12025 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12026 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12027 (m32r_modes_tieable_p): New function.
12028 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12029 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12030 (m68k_modes_tieable_p): New function.
12031 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12032 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12033 (mcore_modes_tieable_p): New function.
12034 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12035 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12037 (TARGET_MODES_TIEABLE_P): Redefine.
12038 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12039 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12040 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12041 (TARGET_MODES_TIEABLE_P): Redefine.
12042 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12043 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12044 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12045 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12046 (mn10300_modes_tieable_p): ...this and make static.
12047 (TARGET_MODES_TIEABLE_P): Redefine.
12048 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12049 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12050 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12051 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12052 (msp430_modes_tieable_p): Make static.
12053 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12054 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12055 (TARGET_MODES_TIEABLE_P): Redefine.
12056 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12057 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12058 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12059 (TARGET_MODES_TIEABLE_P): Redefine.
12060 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12061 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12062 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12063 (TARGET_MODES_TIEABLE_P): Redefine.
12064 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12065 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12066 (pdp11_modes_tieable_p): New function.
12067 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12068 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12069 (rs6000_modes_tieable_p): New function.
12070 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12071 * config/powerpcspe/powerpcspe.md: Update comment.
12072 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12073 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12074 (TARGET_MODES_TIEABLE_P): Redefine.
12075 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12076 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12077 (rl78_modes_tieable_p): New function.
12078 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12079 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12080 (rs6000_modes_tieable_p): New function.
12081 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12082 * config/rs6000/rs6000.md: Update comment.
12083 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12084 * config/rx/rx.c (rx_modes_tieable_p): New function.
12085 (TARGET_MODES_TIEABLE_P): Redefine.
12086 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12087 * config/s390/s390.c (s390_modes_tieable_p): New function.
12088 (TARGET_MODES_TIEABLE_P): Redefine.
12089 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12090 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12091 (sh_modes_tieable_p): New function.
12092 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12093 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12094 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12095 (sparc_modes_tieable_p): Make static.
12096 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12097 * config/spu/spu.c (spu_modes_tieable_p): New function.
12098 (TARGET_MODES_TIEABLE_P): Redefine.
12099 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12100 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12101 (TARGET_MODES_TIEABLE_P): Redefine.
12102 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12103 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12104 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12105 * config/v850/v850.c (v850_modes_tieable_p): New function.
12106 (TARGET_MODES_TIEABLE_P): Redefine.
12107 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12108 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12109 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12110 (visium_modes_tieable_p): New function.
12111 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12112 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12113 (xtensa_modes_tieable_p): New function.
12114 * system.h (MODES_TIEABLE_P): Poison.
12116 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12117 Alan Hayward <alan.hayward@arm.com>
12118 David Sherwood <david.sherwood@arm.com>
12120 * target.def (hard_regno_mode_ok): New hook.
12121 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12122 (TARGET_HARD_REGNO_MODE_OK): ...this.
12123 * doc/tm.texi.in: Regenerate.
12124 * hooks.h (hook_bool_uint_mode_true): Declare.
12125 * hooks.c (hook_bool_uint_mode_true): New function.
12126 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12127 HARD_REGNO_MODE_OK.
12128 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12129 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12130 instead of HARD_REGNO_MODE_OK.
12131 * caller-save.c: Include target.h.
12132 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12133 HARD_REGNO_MODE_OK.
12134 * combine.c (can_combine_p): Likewise.
12135 (combinable_i3pat): Likewise.
12136 (can_change_dest_mode): Likewise.
12137 * expr.c (init_expr_target): Likewise.
12138 (convert_move): Likewise.
12139 (convert_modes): Likewise.
12140 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12141 (setup_prohibited_mode_move_regs): Likewise.
12142 * ira.h (target_ira): Likewise.
12143 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12144 * lra-constraints.c (process_alt_operands): Likewise.
12145 (split_reg): Likewise.
12146 * recog.c (peep2_find_free_register): Likewise.
12147 * ree.c (combine_reaching_defs): Likewise.
12148 * regcprop.c (maybe_mode_change): Likewise.
12149 * reginfo.c (init_reg_sets_1): Likewise.
12150 (choose_hard_reg_mode): Likewise.
12151 (simplifiable_subregs): Likewise.
12152 * regrename.c (check_new_reg_p): Likewise.
12153 * reload.c (find_valid_class): Likewise.
12154 (find_valid_class_1): Likewise.
12155 (reload_inner_reg_of_subreg): Likewise.
12156 (push_reload): Likewise.
12157 (combine_reloads): Likewise.
12158 (find_dummy_reload): Likewise.
12159 (find_reloads): Likewise.
12160 * reload1.c (find_reg): Likewise.
12161 (set_reload_reg): Likewise.
12162 (allocate_reload_reg): Likewise.
12163 (choose_reload_regs): Likewise.
12164 (reload_adjust_reg_for_temp): Likewise.
12165 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12166 (simplify_subreg_regno): Likewise.
12167 * sel-sched.c (init_regs_for_mode): Likewise.
12168 * varasm.c (make_decl_rtl): Likewise.
12169 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12170 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12171 HARD_REGNO_MODE_OK.
12172 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12173 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12174 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12175 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12176 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12177 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12178 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12179 (arc_mode_class): Delete.
12180 (HARD_REGNO_MODE_OK): Delete.
12181 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12182 (arc_hard_regno_mode_ok): Rename old array to...
12183 (arc_hard_regno_mode_ok_modes): ...this.
12184 (arc_conditional_register_usage): Update accordingly.
12185 (arc_mode_class): Make static.
12186 (arc_hard_regno_mode_ok): New function.
12187 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12188 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12189 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12190 (arm_hard_regno_mode_ok): Make static.
12191 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12192 HARD_REGNO_MODE_OK.
12193 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12194 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12195 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12197 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12198 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12199 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12200 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12201 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12203 * config/bfin/predicates.md (valid_reg_operand): Use
12204 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12205 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12206 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12208 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12209 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12210 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12211 (cr16_hard_regno_mode_ok): Make static and return a bool.
12212 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12213 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12214 (cris_hard_regno_mode_ok): New function.
12215 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12216 (epiphany_mode_class): Delete.
12217 (HARD_REGNO_MODE_OK): Delete.
12218 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12219 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12220 (hard_regno_mode_ok): Rename to...
12221 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12222 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12223 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12224 HARD_REGNO_MODE_OK.
12225 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12226 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12227 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12228 (frv_hard_regno_mode_ok): Make static and return a bool.
12229 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12230 HARD_REGNO_MODE_OK.
12231 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12232 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12233 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12234 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12236 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12237 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12238 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12239 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12242 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12243 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12244 (ia64_hard_regno_mode_ok): New function.
12245 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12246 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12247 (iq2000_hard_regno_mode_ok): New function.
12248 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12249 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12250 (lm32_hard_regno_mode_ok): New function.
12251 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12252 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12253 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12254 instead of HARD_REGNO_MODE_OK.
12255 (m32c_hard_regno_ok): Rename to...
12256 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12257 (m32c_cannot_change_mode_class): Update accordingly.
12258 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12259 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12260 (m32r_mode_class): Delete.
12261 (HARD_REGNO_MODE_OK): Delete.
12262 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12263 (m32r_hard_regno_mode_ok): Rename to...
12264 (m32r_hard_regno_modes): ...this.
12265 (m32r_mode_class): Make static.
12266 (m32r_hard_regno_mode_ok): New function.
12267 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12268 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12269 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12270 (m68k_hard_regno_mode_ok): Make static.
12271 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12272 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12273 (mcore_hard_regno_mode_ok): New function.
12274 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12275 (HARD_REGNO_MODE_OK): Delete.
12276 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12278 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12279 (microblaze_hard_regno_mode_ok): New function.
12280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12281 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12282 (mips_hard_regno_mode_ok): Delete.
12283 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12284 (mips_hard_regno_mode_ok_p): ...this and make static.
12285 (mips_hard_regno_mode_ok_p): Rename to...
12286 (mips_hard_regno_mode_ok_uncached): ...this.
12287 (mips_hard_regno_mode_ok): New function.
12288 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12289 of HARD_REGNO_MODE_OK.
12290 (mips_option_override): Update after above name changes.
12291 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12292 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12293 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12294 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12295 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12296 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12297 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12298 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12299 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12300 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12301 (msp430_hard_regno_mode_ok): Make static and return a bool.
12302 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12303 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12304 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12307 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12308 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12309 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12310 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12311 (PA_HARD_REGNO_MODE_OK): ...this
12312 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12313 (PA_HARD_REGNO_MODE_OK): ...this.
12314 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12315 (pa_hard_regno_mode_ok): New function.
12316 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12317 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12318 (pdp11_hard_regno_mode_ok): New function.
12319 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12320 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12322 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12325 (rs6000_hard_regno_mode_ok): Rename to...
12326 (rs6000_hard_regno_mode_ok_uncached): ...this.
12327 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12328 (rs6000_hard_regno_mode_ok): New function.
12329 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12330 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12331 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12332 (riscv_hard_regno_mode_ok): ...this and make static.
12333 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12334 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12335 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12336 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12337 (rl78_hard_regno_mode_ok): Make static and return bool.
12338 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12339 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12341 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12343 (rs6000_hard_regno_mode_ok): Rename to...
12344 (rs6000_hard_regno_mode_ok_uncached): ...this.
12345 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12346 (rs6000_hard_regno_mode_ok): New function.
12347 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12348 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12349 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12350 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12351 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12352 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12353 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12354 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12355 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12356 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12357 (sh_hard_regno_mode_ok): Make static.
12358 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12359 instead of HARD_REGNO_MODE_OK.
12360 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12361 (sparc_mode_class): Delete.
12362 (HARD_REGNO_MODE_OK): Delete.
12363 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12364 (hard_regno_mode_classes): Make static.
12365 (sparc_mode_class): Likewise.
12366 (sparc_hard_regno_mode_ok): New function.
12367 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12368 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12369 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12371 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12372 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12373 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12374 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12375 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12376 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12377 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12378 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12379 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12380 (visium_hard_regno_mode_ok): New function.
12381 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12382 instead of HARD_REGNO_MODE_OK.
12383 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12384 (HARD_REGNO_MODE_OK): Delete.
12385 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12386 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12387 (xtensa_option_override): Update accordingly.
12388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12389 (xtensa_hard_regno_mode_ok): New function.
12390 * system.h (HARD_REGNO_MODE_OK): Poison.
12392 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12393 Alan Hayward <alan.hayward@arm.com>
12394 David Sherwood <david.sherwood@arm.com>
12396 * target.def (hard_regno_call_part_clobbered): New hook.
12397 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12398 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12399 * doc/tm.texi: Regenerate.
12400 * hooks.h (hook_bool_uint_mode_false): Declare.
12401 * hooks.c (hook_bool_uint_mode_false): New function.
12402 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12403 * cselib.c (cselib_process_insn): Use
12404 targetm.hard_regno_call_part_clobbered instead of
12405 HARD_REGNO_CALL_PART_CLOBBERED.
12406 * ira-conflicts.c (ira_build_conflicts): Likewise.
12407 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12408 * lra-constraints.c (need_for_call_save_p): Likewise.
12409 * lra-lives.c: Include target.h.
12410 (check_pseudos_live_through_calls): Use
12411 targetm.hard_regno_call_part_clobbered instead of
12412 HARD_REGNO_CALL_PART_CLOBBERED.
12413 * regcprop.c: Include target.h.
12414 (copyprop_hardreg_forward_1): Use
12415 targetm.hard_regno_call_part_clobbered instead of
12416 HARD_REGNO_CALL_PART_CLOBBERED.
12417 * reginfo.c (choose_hard_reg_mode): Likewise.
12418 * regrename.c (check_new_reg_p): Likewise.
12419 * reload.c (find_equiv_reg): Likewise.
12420 * reload1.c (emit_reload_insns): Likewise.
12421 * sched-deps.c (deps_analyze_insn): Likewise.
12422 * sel-sched.c (init_regs_for_mode): Likewise.
12423 (mark_unavailable_hard_regs): Likewise.
12424 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12425 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12426 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12428 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12429 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12430 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12432 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12434 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12435 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12436 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12438 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12439 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12440 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12442 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12443 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12445 * config/powerpcspe/powerpcspe.c
12446 (rs6000_hard_regno_call_part_clobbered): New function.
12447 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12448 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12449 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12451 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12452 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12453 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12455 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12456 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12457 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12459 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12460 Alan Hayward <alan.hayward@arm.com>
12461 David Sherwood <david.sherwood@arm.com>
12463 * rtl.h (subreg_memory_offset): Declare.
12464 * emit-rtl.c (subreg_memory_offset): New function.
12465 * expmed.c (store_bit_field_1): Use it.
12466 * expr.c (undefined_operand_subword_p): Likewise.
12467 * simplify-rtx.c (simplify_subreg): Likewise.
12469 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12471 PR rtl-optimization/57448
12474 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12475 using atomic_load pattern.
12476 (expand_atomic_store): Likewise.
12478 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12481 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12482 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12483 handling. Use replace_call_with_value with NULL instead of
12484 gsi_replace, unlink_stmt_vdef and release_defs.
12486 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12489 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12491 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12494 * rtl.h (emit_library_call_value_1): Declare.
12495 (emit_library_call): Replace declaration with a series of overloads.
12496 Remove the parameter count argument.
12497 (emit_library_call_value): Likewise.
12498 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12499 with an "rtx_mode_t *".
12500 (emit_library_call_value): Delete.
12501 (emit_library_call): Likewise.
12502 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12503 (asan_emit_allocas_unpoison): Likewise.
12504 * builtins.c (expand_builtin_powi): Likewise.
12505 (expand_asan_emit_allocas_unpoison): Likewise.
12506 * cfgexpand.c (expand_main_function): Likewise.
12507 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12508 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12509 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12510 * config/arm/arm.c (arm_trampoline_init): Likewise.
12511 (arm_call_tls_get_addr): Likewise.
12512 (arm_expand_divmod_libfunc): Likewise.
12513 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12514 (smulsi3_highpart): Likewise.
12515 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12516 (c6x_expand_compare): Likewise.
12517 (c6x_expand_movmem): Likewise.
12518 * config/frv/frv.c (frv_trampoline_init): Likewise.
12519 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12520 (ix86_expand_divmod_libfunc): Likewise.
12521 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12522 (ia64_expand_compare): Likewise.
12523 (ia64_profile_hook): Likewise.
12524 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12525 (nonlocal_goto): Likewise.
12526 (restore_stack_nonlocal): Likewise.
12527 * config/m32r/m32r.c (block_move_call): Likewise.
12528 (m32r_trampoline_init): Likewise.
12529 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12530 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12531 (m68k_call_m68k_read_tp): Likewise.
12532 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12533 (microblaze_expand_divide): Likewise.
12534 * config/mips/mips.h (mips_args): Likewise.
12535 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12536 (MIPS_ICACHE_SYNC): Likewise.
12537 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12538 (nios2_trampoline_init): Likewise.
12539 * config/pa/pa.c (hppa_tls_call): Likewise.
12540 (pa_trampoline_init): Likewise.
12541 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12542 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12543 (expand_strn_compare): Likewise.
12544 (rs6000_generate_compare): Likewise.
12545 (rs6000_expand_float128_convert): Likewise.
12546 (output_profile_hook): Likewise.
12547 (rs6000_trampoline_init): Likewise.
12548 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12549 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12550 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12551 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12552 (rs6000_generate_compare): Likewise.
12553 (rs6000_expand_float128_convert): Likewise.
12554 (output_profile_hook): Likewise.
12555 (rs6000_trampoline_init): Likewise.
12556 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12557 * config/sh/sh.c (sh_trampoline_init): Likewise.
12558 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12559 (sparc_emit_float_lib_cmp): Likewise.
12560 (sparc32_initialize_trampoline): Likewise.
12561 (sparc64_initialize_trampoline): Likewise.
12562 (sparc_profile_hook): Likewise.
12563 * config/spu/spu.c (ea_load_store): Likewise.
12564 * config/spu/spu.md (floatunssidf2): Likewise.
12565 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12566 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12567 * config/visium/visium.c (expand_block_move_4): Likewise.
12568 (expand_block_move_2): Likewise.
12569 (expand_block_move_1): Likewise.
12570 (expand_block_set_4): Likewise.
12571 (expand_block_set_2): Likewise.
12572 (expand_block_set_1): Likewise.
12573 (visium_trampoline_init): Likewise.
12574 (visium_profile_hook): Likewise.
12575 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12576 (xtensa_setup_frame_addresses): Likewise.
12577 (xtensa_trampoline_init): Likewise.
12578 * except.c (sjlj_emit_function_enter): Likewise.
12579 (sjlj_emit_function_exit): Likewise.
12580 * explow.c (allocate_dynamic_stack_space): Likewise.
12581 (probe_stack_range): Likewise.
12582 * expr.c (convert_mode_scalar): Likewise.
12583 * optabs.c (expand_binop): Likewise.
12584 (expand_twoval_binop_libfunc): Likewise.
12585 (expand_unop): Likewise.
12586 (prepare_cmp_insn): Likewise.
12587 (prepare_float_lib_cmp): Likewise.
12588 (expand_float): Likewise.
12589 (expand_fix): Likewise.
12590 (expand_fixed_convert): Likewise.
12591 (maybe_emit_sync_lock_test_and_set): Likewise.
12592 (expand_atomic_compare_and_swap): Likewise.
12593 (expand_mem_thread_fence): Likewise.
12594 (expand_atomic_fetch_op): Likewise.
12596 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12598 * doc/generic.texi (OpenACC): Adjust URL.
12599 * doc/invoke.texi (C Dialect Options): Ditto.
12601 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12603 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12604 predicate for operand 1. Add (m,<S>) constraint.
12605 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12606 Prevent memory operand 1 with register operand 2.
12608 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12610 PR rtl-optimization/82024
12611 * combine.c (try_combine): If the combination result is a PARALLEL,
12612 and we only need to retain the SET in there that would be placed
12613 at I2, check that we can place that at I3 instead, before doing so.
12615 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12618 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12620 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12621 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12624 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12625 Jeff Law <law@redhat.com>
12627 * varasm.c (bss_initializer_p): Do not put constants into .bss
12628 (categorize_decl_for_section): Handle bss_initializer_p returning
12629 false when DECL_INITIAL is NULL.
12631 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12634 * config/s390/s390.c (s390_can_inline_p): New function.
12636 2017-09-01 Jeff Law <law@redhat.com>
12638 PR tree-optimization/82052
12639 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12640 Always initialize the returned slot after a hash table miss
12641 when INSERT is true.
12643 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12645 * config/s390/s390.md (mem_signal_fence): Remove.
12646 * doc/md.texi (mem_signal_fence): Remove.
12647 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12649 * target-insns.def (mem_signal_fence): Remove.
12651 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12654 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12657 * asan.c (create_odr_indicator): Strip name encoding from assembler
12658 name before appending it after __odr_asan_.
12660 2017-09-01 Martin Liska <mliska@suse.cz>
12662 PR tree-optimization/82059
12663 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12664 frequency only when an edge is redirected.
12666 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12668 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12669 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12670 (arc_conditional_register_usage): Remove ARC600 lp_count
12672 (arc_file_start): Emit Tag_ARC_CPU_variation.
12673 (arc_can_use_doloop_p): New conditions to use ZOLs.
12674 (hwloop_fail): New function.
12675 (hwloop_optimize): Likewise.
12676 (hwloop_pattern_reg): Likewise.
12677 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12678 (arc_reorg_loops): New function, calls reorg_loops.
12679 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12680 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12682 (arc_loop_hazard): Remove function, functionality moved into
12684 (arc_hazard): Remove arc_loop_hazard call.
12685 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12686 into hwloop_optimize.
12687 (arc_label_align): Remove ZOL handling.
12688 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12689 * config/arc/arc.md (doloop_begin): Remove pattern.
12690 (doloop_begin_i): Likewise.
12691 (doloop_end_i): Likewise.
12692 (doloop_fallback): Likewise.
12693 (doloop_fallback_m): Likewise.
12694 (doloop_end): Reimplement expand.
12695 (arc_lp): New pattern for LP instruction.
12696 (loop_end): New pattern.
12697 (loop_fail): Likewise.
12698 (decrement_and_branch_until_zero): Likewise.
12699 * config/arc/arc.opt (mlpc-width): New option.
12700 * doc/invoke.texi (mlpc-width): Document option.
12702 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12704 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12705 (arc_ccfsm_advance): Fix checking for delay slots.
12706 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12708 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12710 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12712 (movhi_insn): Update store instruction constraint which are saving
12713 6-bit short immediates.
12714 (movsi_insn): Consider also short scaled load operations.
12715 (zero_extendhisi2_i): Use Usd constraint instead of T.
12716 (extendhisi2_i): Add q constraint.
12717 (arc_clzsi2): Add type and length attributes.
12718 (arc_ctzsi2): Likewise.
12719 * config/arc/constraints.md (Usc): Update constraint, the
12720 assembler can parse two relocations for a single instruction.
12722 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12724 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12725 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12727 2017-08-31 Olivier Hainque <hainque@adacore.com>
12729 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12730 match as powerpc-wrs-vxworks*.
12732 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12734 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12735 register constraint for by-element operand.
12736 (aarch64_mls_elt_merge<mode>): Likewise.
12738 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12740 * config/arc/arc.c (arc_can_follow_jump): Check for short
12743 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12745 * config.gcc: Use g.opt for arc.
12746 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12747 functionality moved to ...
12748 (legitimate_scaled_address_p): New function, ...here.
12749 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12750 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12751 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12753 (arc_override_options): Handle G option.
12754 (arc_output_pic_addr_const): Correct function definition.
12755 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12756 (arc_decl_anon_ns_mem_p): Delete.
12757 (arc_in_small_data_p): Overhaul this function to take into
12758 consideration the value given via G option.
12759 (arc_rewrite_small_data_1): Renamed and corrected old
12760 arc_rewrite_small_data function.
12761 (arc_rewrite_small_data): New function.
12762 (small_data_pattern): Don't use pic_offset_table_rtx.
12763 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12764 * config/arc/simdext.md (movmisalignv2hi): Use
12765 prepare_move_operands function.
12767 (movmisalign*): Likewise.
12768 * doc/invoke.texi (ARC options): Document -G option.
12770 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12772 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12774 * config/arc/arc.c (arc_print_operand): Output scalled address for
12775 sdata whenever is possible.
12776 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12777 load/stores are available.
12778 (compact_sda_memory_operand): Check for the alignment required by
12779 code density instructions.
12780 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12782 * config/arc/constraints.md (Usd): Update constraint.
12783 (Us0): New constraint.
12784 (Usc): Update constraint.
12786 2017-08-31 Richard Biener <rguenther@suse.de>
12788 PR middle-end/82054
12789 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12790 function only once.
12792 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12794 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12795 Resize type_signature.
12797 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12798 Alan Hayward <alan.hayward@arm.com>
12799 David Sherwood <david.sherwood@arm.com>
12801 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12802 subregs whose inner modes can be stored in GPRs.
12803 (aarch64_classify_index): Likewise.
12805 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12806 Alan Hayward <alan.hayward@arm.com>
12807 David Sherwood <david.sherwood@arm.com>
12809 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12810 (V_INT_EQUIV): ...this.
12811 (v_cmp_result): Rename to...
12812 (v_int_equiv): ...this.
12813 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12814 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12815 (copysign<mode>3): Likewise.
12816 (aarch64_simd_bsl<mode>_internal): Likewise.
12817 (aarch64_simd_bsl<mode>): Likewise.
12818 (vec_cmp<mode><mode>): Likewise.
12819 (vcond<mode><mode>): Likewise.
12820 (vcond<v_cmp_mixed><mode>): Likewise.
12821 (vcondu<mode><v_cmp_mixed>): Likewise.
12822 (aarch64_cm<optab><mode>): Likewise.
12823 (aarch64_cmtst<mode>): Likewise.
12824 (aarch64_fac<optab><mode>): Likewise.
12825 (vec_perm_const<mode>): Likewise.
12826 (vcond_mask_<mode><v_cmp_result>): Rename to...
12827 (vcond_mask_<mode><v_int_equiv>): ...this.
12828 (vec_cmp<mode><v_cmp_result>): Rename to...
12829 (vec_cmp<mode><v_int_equiv>): ...this.
12831 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12832 Alan Hayward <alan.hayward@arm.com>
12833 David Sherwood <david.sherwood@arm.com>
12835 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12837 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12838 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12839 (UNSPEC_LD4_DREG): New unspecs.
12840 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12841 (aarch64_ld2<mode>_dreg_be): Replace with...
12842 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12844 (aarch64_ld3<mode>_dreg_le)
12845 (aarch64_ld3<mode>_dreg_be): Replace with...
12846 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12848 (aarch64_ld4<mode>_dreg_le)
12849 (aarch64_ld4<mode>_dreg_be): Replace with...
12850 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12853 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12855 PR tree-optimization/81987
12856 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12857 insert an initializer in a location not dominated by the stride
12860 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12862 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12863 on the entire header of the finally block in the fallthru case.
12865 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12867 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12869 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12871 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12872 rs6000_emit_move_from_cr and call renamed function.
12873 (rs6000_emit_prologue): Call renamed functions.
12874 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12875 movesi_from_cr, remove volatile CRs.
12877 2017-08-30 Jon Beniston <jon@beniston.com>
12878 Richard Biener <rguenther@suse.de>
12880 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12881 of VECTOR_MODE_P check.
12882 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12883 element vector types.
12885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12887 * df.h (df_read_modify_subreg_p): Remove in favor of...
12888 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12889 const_rtx instead of an rtx.
12890 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12891 * df-problems.c (df_word_lr_mark_ref): Likewise.
12892 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12893 (mark_pseudo_reg_dead): Likewise.
12894 (mark_ref_dead): Likewise.
12895 * reginfo.c (init_subregs_of_mode): Likewise.
12896 * sched-deps.c (sched_analyze_1): Likewise.
12897 * df-scan.c (df_def_record_1): Likewise.
12898 (df_uses_record): Likewise.
12899 (df_read_modify_subreg_p): Remove in favor of...
12900 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12901 const_rtx instead of an rtx.
12903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12904 Alan Hayward <alan.hayward@arm.com>
12905 David Sherwood <david.sherwood@arm.com>
12907 * rtl.h (partial_subreg_p): New function.
12908 * caller-save.c (save_call_clobbered_regs): Use it.
12909 * calls.c (expand_call): Likewise.
12910 * combine.c (combinable_i3pat): Likewise.
12911 (simplify_set): Likewise.
12912 (make_extraction): Likewise.
12913 (make_compound_operation_int): Likewise.
12914 (gen_lowpart_or_truncate): Likewise.
12915 (force_to_mode): Likewise.
12916 (make_field_assignment): Likewise.
12917 (reg_truncated_to_mode): Likewise.
12918 (record_truncated_value): Likewise.
12919 (move_deaths): Likewise.
12920 * cse.c (record_jump_cond): Likewise.
12921 (cse_insn): Likewise.
12922 * cselib.c (cselib_lookup_1): Likewise.
12923 * expmed.c (extract_bit_field_using_extv): Likewise.
12924 * function.c (assign_parm_setup_reg): Likewise.
12925 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12926 * ira-build.c (create_insn_allocnos): Likewise.
12927 * lra-coalesce.c (merge_pseudos): Likewise.
12928 * lra-constraints.c (match_reload): Likewise.
12929 (simplify_operand_subreg): Likewise.
12930 (curr_insn_transform): Likewise.
12931 * lra-lives.c (process_bb_lives): Likewise.
12932 * lra.c (new_insn_reg): Likewise.
12933 (lra_substitute_pseudo): Likewise.
12934 * regcprop.c (mode_change_ok): Likewise.
12935 (maybe_mode_change): Likewise.
12936 (copyprop_hardreg_forward_1): Likewise.
12937 * reload.c (push_reload): Likewise.
12938 (find_reloads): Likewise.
12939 (find_reloads_subreg_address): Likewise.
12940 * reload1.c (alter_reg): Likewise.
12941 (eliminate_regs_1): Likewise.
12942 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12944 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12946 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12947 back to if statements, including unpack.
12949 2017-08-30 Martin Liska <mliska@suse.cz>
12951 PR inline-asm/82001
12952 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12954 (func_checker::compare_asm_inputs_outputs): ... this function.
12955 (func_checker::compare_gimple_asm): Use the function to compare
12956 also ASM constrains.
12957 * ipa-icf-gimple.h: Rename the function.
12959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12960 Alan Hayward <alan.hayward@arm.com>
12961 David Sherwood <david.sherwood@arm.com>
12963 * coretypes.h (complex_mode): New type.
12964 * gdbhooks.py (build_pretty_printer): Handle it.
12965 * machmode.h (complex_mode): New class.
12966 (complex_mode::includes_p): New function.
12967 (is_complex_int_mode): Likewise.
12968 (is_complex_float_mode): Likewise.
12969 * genmodes.c (get_mode_class): Handle complex mode classes.
12970 * function.c (expand_function_end): Use is_complex_int_mode.
12972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12973 Alan Hayward <alan.hayward@arm.com>
12974 David Sherwood <david.sherwood@arm.com>
12976 * coretypes.h (scalar_mode_pod): New typedef.
12977 * gdbhooks.py (build_pretty_printer): Handle it.
12978 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12979 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12980 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12981 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12982 as_a <scalar_mode>.
12984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12985 Alan Hayward <alan.hayward@arm.com>
12986 David Sherwood <david.sherwood@arm.com>
12988 * machmode.h (mode_for_vector): Take a scalar_mode instead
12990 * stor-layout.c (mode_for_vector): Likewise.
12991 * explow.c (promote_mode): Use as_a <scalar_mode>.
12992 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12995 Alan Hayward <alan.hayward@arm.com>
12996 David Sherwood <david.sherwood@arm.com>
12998 * target.def (preferred_simd_mode): Take a scalar_mode
12999 instead of a machine_mode.
13000 * targhooks.h (default_preferred_simd_mode): Likewise.
13001 * targhooks.c (default_preferred_simd_mode): Likewise.
13002 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13003 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13004 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13005 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13006 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13007 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13008 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13009 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13011 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13012 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13013 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13014 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13015 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13016 * doc/tm.texi: Regenerate.
13017 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13021 Alan Hayward <alan.hayward@arm.com>
13022 David Sherwood <david.sherwood@arm.com>
13024 * target.def (scalar_mode_supported_p): Take a scalar_mode
13025 instead of a machine_mode.
13026 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13027 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13028 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13029 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13030 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13031 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13032 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13033 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13034 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13035 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13036 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13037 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13038 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13039 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13041 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13042 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13043 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13044 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13045 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13047 * doc/tm.texi: Regenerate.
13049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13050 Alan Hayward <alan.hayward@arm.com>
13051 David Sherwood <david.sherwood@arm.com>
13053 * coretypes.h (opt_scalar_mode): New typedef.
13054 * gdbhooks.py (build_pretty_printers): Handle it.
13055 * machmode.h (mode_iterator::get_2xwider): Add overload for
13057 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13059 * expr.c (convert_mode_scalar): Likewise.
13060 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13061 * optabs.c (expand_float): Likewise.
13062 (expand_fix): Likewise.
13063 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13066 Alan Hayward <alan.hayward@arm.com>
13067 David Sherwood <david.sherwood@arm.com>
13069 * optabs.c (expand_float): Explicitly check for scalars before
13070 using a branching expansion.
13071 (expand_fix): Likewise.
13073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13074 Alan Hayward <alan.hayward@arm.com>
13075 David Sherwood <david.sherwood@arm.com>
13077 * expr.c (convert_mode): Split scalar handling out into...
13078 (convert_mode_scalar): ...this new function. Treat the modes
13081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13082 Alan Hayward <alan.hayward@arm.com>
13083 David Sherwood <david.sherwood@arm.com>
13085 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13087 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13090 Alan Hayward <alan.hayward@arm.com>
13091 David Sherwood <david.sherwood@arm.com>
13093 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13094 rather than a machine_mode.
13095 (fixed_from_string): Likewise.
13096 (fixed_convert): Likewise.
13097 (fixed_convert_from_int): Likewise.
13098 (fixed_convert_from_real): Likewise.
13099 (real_convert_from_fixed): Likewise.
13100 * fixed-value.c (fixed_from_double_int): Likewise.
13101 (fixed_from_string): Likewise.
13102 (fixed_convert): Likewise.
13103 (fixed_convert_from_int): Likewise.
13104 (fixed_convert_from_real): Likewise.
13105 (real_convert_from_fixed): Likewise.
13106 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13109 Alan Hayward <alan.hayward@arm.com>
13110 David Sherwood <david.sherwood@arm.com>
13112 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13113 of separate mode class checks. Do not allow vector modes here.
13114 (immed_wide_int_const): Use as_a <scalar_mode>.
13115 * explow.c (trunc_int_for_mode): Likewise.
13116 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13117 (wi::shwi): Likewise.
13118 (wi::min_value): Likewise.
13119 (wi::max_value): Likewise.
13120 * dwarf2out.c (loc_descriptor): Likewise.
13121 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13122 for CONST_WIDE_INT.
13124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13125 Alan Hayward <alan.hayward@arm.com>
13126 David Sherwood <david.sherwood@arm.com>
13128 * tree.h (SCALAR_TYPE_MODE): New macro.
13129 * expr.c (expand_expr_addr_expr_1): Use it.
13130 (expand_expr_real_2): Likewise.
13131 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13132 (fold_convert_const_fixed_from_int): Likewise.
13133 (fold_convert_const_fixed_from_real): Likewise.
13134 (native_encode_fixed): Likewise
13135 (native_encode_complex): Likewise
13136 (native_encode_vector): Likewise.
13137 (native_interpret_fixed): Likewise.
13138 (native_interpret_real): Likewise.
13139 (native_interpret_complex): Likewise.
13140 (native_interpret_vector): Likewise.
13141 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13142 (simd_clone_adjust_argument_types): Likewise.
13143 (simd_clone_init_simd_arrays): Likewise.
13144 (simd_clone_adjust): Likewise.
13145 * stor-layout.c (layout_type): Likewise.
13146 * tree.c (build_minus_one_cst): Likewise.
13147 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13148 * tree-inline.c (estimate_move_cost): Likewise.
13149 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13150 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13151 (vectorizable_reduction): Likewise.
13152 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13153 (vect_recog_mixed_size_cond_pattern): Likewise.
13154 (check_bool_pattern): Likewise.
13155 (adjust_bool_pattern): Likewise.
13156 (search_type_for_mask_1): Likewise.
13157 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13158 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13159 (vectorizable_load): Likewise.
13160 (vectorizable_store): Likewise.
13161 * ubsan.c (ubsan_encode_value): Likewise.
13162 * varasm.c (output_constant): Likewise.
13164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13165 Alan Hayward <alan.hayward@arm.com>
13166 David Sherwood <david.sherwood@arm.com>
13168 * coretypes.h (scalar_mode): New class.
13169 * machmode.h (scalar_mode): Likewise.
13170 (scalar_mode::includes_p): New function.
13171 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13172 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13173 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13174 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13175 * expmed.c (store_bit_field_1): Likewise.
13176 (extract_bit_field_1): Likewise.
13177 * expr.c (write_complex_part): Likewise.
13178 (read_complex_part): Likewise.
13179 (emit_move_complex_push): Likewise.
13180 (expand_expr_real_2): Likewise.
13181 * function.c (assign_parm_setup_reg): Likewise.
13182 (assign_parms_unsplit_complex): Likewise.
13183 * optabs.c (expand_binop): Likewise.
13184 * rtlanal.c (subreg_get_info): Likewise.
13185 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13186 * varasm.c (output_constant_pool_2): Likewise.
13188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13189 Alan Hayward <alan.hayward@arm.com>
13190 David Sherwood <david.sherwood@arm.com>
13192 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13194 (expmed_mult_highpart_optab): Likewise.
13195 (expmed_mult_highpart): Likewise.
13197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13198 Alan Hayward <alan.hayward@arm.com>
13199 David Sherwood <david.sherwood@arm.com>
13201 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13202 instead of a machine_mode.
13203 (builtin_memset_read_str): Likewise.
13204 * builtins.c (c_readstr): Likewise.
13205 (builtin_memcpy_read_str): Likewise.
13206 (builtin_strncpy_read_str): Likewise.
13207 (builtin_memset_read_str): Likewise.
13208 (builtin_memset_gen_str): Likewise.
13209 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13210 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13211 instead of a machine_mode.
13212 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13214 (make_extraction): Likewise.
13215 (try_widen_shift_mode): Take and return scalar_int_modes instead
13217 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13218 a scalar_int_mode instead of a machine_mode.
13219 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13220 (avr_addr_space_pointer_mode): Likewise.
13221 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13222 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13223 (msp430_unwind_word_mode): Likewise.
13224 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13225 (spu_addr_space_pointer_mode): Likewise.
13226 (spu_addr_space_address_mode): Likewise.
13227 (spu_libgcc_cmp_return_mode): Likewise.
13228 (spu_libgcc_shift_count_mode): Likewise.
13229 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13230 (rl78_addr_space_pointer_mode): Likewise.
13231 (fl78_unwind_word_mode): Likewise.
13232 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13234 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13235 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13236 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13237 (mips_valid_pointer_mode): Likewise.
13238 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13239 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13240 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13242 (ft32_addr_space_address_mode): Likewise.
13243 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13244 scalar_int_mode instead of a machine_mode.
13245 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13247 (m32c_addr_space_address_mode): Likewise.
13248 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13249 (rs6000_eh_return_filter_mode): Likewise.
13250 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13251 (rs6000_eh_return_filter_mode): Likewise.
13252 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13253 (s390_libgcc_shift_count_mode): Likewise.
13254 (s390_unwind_word_mode): Likewise.
13255 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13257 * target.def (mode_rep_extended): Likewise.
13258 (valid_pointer_mode): Likewise.
13259 (addr_space.valid_pointer_mode): Likewise.
13260 (eh_return_filter_mode): Return a scalar_int_mode rather than
13262 (libgcc_cmp_return_mode): Likewise.
13263 (libgcc_shift_count_mode): Likewise.
13264 (unwind_word_mode): Likewise.
13265 (addr_space.pointer_mode): Likewise.
13266 (addr_space.address_mode): Likewise.
13267 * doc/tm.texi: Regenerate.
13268 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13270 (do_jump): Use scalar_int_mode for local variables.
13271 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13272 rather than a machine_mode.
13273 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13274 (scompare_loc_descriptor_wide): Likewise.
13275 (scompare_loc_descriptor_narrow): Likewise.
13276 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13278 * except.c (sjlj_emit_dispatch_table): Likewise.
13279 (expand_builtin_eh_copy_values): Likewise.
13280 * explow.c (convert_memory_address_addr_space_1): Likewise.
13281 Take a scalar_int_mode rather than a machine_mode.
13282 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13283 than a machine_mode.
13284 (memory_address_addr_space): Use scalar_int_mode for local variables.
13285 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13286 rather than a machine_mode.
13287 * expmed.c (mask_rtx): Likewise.
13288 (init_expmed_one_conv): Likewise.
13289 (expand_mult_highpart_adjust): Likewise.
13290 (extract_high_half): Likewise.
13291 (expmed_mult_highpart_optab): Likewise.
13292 (expmed_mult_highpart): Likewise.
13293 (expand_smod_pow2): Likewise.
13294 (expand_sdiv_pow2): Likewise.
13295 (emit_store_flag_int): Likewise.
13296 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13298 (extract_low_bits): Likewise.
13299 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13301 * expr.c (pieces_addr::adjust): Likewise.
13302 (can_store_by_pieces): Likewise.
13303 (store_by_pieces): Likewise.
13304 (clear_by_pieces_1): Likewise.
13305 (expand_expr_addr_expr_1): Likewise.
13306 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13307 (expand_expr_real_1): Likewise.
13308 (try_casesi): Likewise.
13309 * final.c (shorten_branches): Likewise.
13310 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13311 type of "mode" to machine_mode.
13312 * internal-fn.c (expand_arith_overflow_result_store): Take a
13313 scalar_int_mode rather than a machine_mode.
13314 (expand_mul_overflow): Use scalar_int_mode for local variables.
13315 * loop-doloop.c (doloop_modify): Likewise.
13316 (doloop_optimize): Likewise.
13317 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13318 than a machine_mode.
13319 (expand_doubleword_shift_condmove): Likewise.
13320 (expand_doubleword_shift): Likewise.
13321 (expand_doubleword_clz): Likewise.
13322 (expand_doubleword_popcount): Likewise.
13323 (expand_doubleword_parity): Likewise.
13324 (expand_absneg_bit): Use scalar_int_mode for local variables.
13325 (prepare_float_lib_cmp): Likewise.
13326 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13327 rather than a machine_mode.
13328 (convert_memory_address_addr_space): Likewise.
13329 (get_mode_bounds): Likewise.
13330 (get_address_mode): Return a scalar_int_mode rather than a
13332 * rtlanal.c (get_address_mode): Likewise.
13333 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13334 than a machine_mode.
13335 * targhooks.c (default_mode_rep_extended): Likewise.
13336 (default_valid_pointer_mode): Likewise.
13337 (default_addr_space_valid_pointer_mode): Likewise.
13338 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13339 than a machine_mode.
13340 (default_libgcc_cmp_return_mode): Likewise.
13341 (default_libgcc_shift_count_mode): Likewise.
13342 (default_unwind_word_mode): Likewise.
13343 (default_addr_space_pointer_mode): Likewise.
13344 (default_addr_space_address_mode): Likewise.
13345 * targhooks.h (default_eh_return_filter_mode): Likewise.
13346 (default_libgcc_cmp_return_mode): Likewise.
13347 (default_libgcc_shift_count_mode): Likewise.
13348 (default_unwind_word_mode): Likewise.
13349 (default_addr_space_pointer_mode): Likewise.
13350 (default_addr_space_address_mode): Likewise.
13351 (default_mode_rep_extended): Take a scalar_int_mode rather than
13353 (default_valid_pointer_mode): Likewise.
13354 (default_addr_space_valid_pointer_mode): Likewise.
13355 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13357 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13358 rather than a machine_mode.
13359 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13360 for local variables.
13361 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13362 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13363 than a machine_mode.
13365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13366 Alan Hayward <alan.hayward@arm.com>
13367 David Sherwood <david.sherwood@arm.com>
13369 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13370 the mode argument to scalar_int_mode.
13371 (do_jump_by_parts_zero_rtx): Likewise.
13372 (do_jump_by_parts_equality_rtx): Likewise.
13373 (do_jump_by_parts_greater): Take a mode argument.
13374 (do_jump_by_parts_equality): Likewise.
13375 (do_jump_1): Update calls accordingly.
13377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13378 Alan Hayward <alan.hayward@arm.com>
13379 David Sherwood <david.sherwood@arm.com>
13381 * is-a.h (safe_dyn_cast): New function.
13382 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13383 (jump_table_for_label): Likewise.
13384 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13385 instead of an rtx_insn *.
13386 (shorten_branches): Use dyn_cast instead of LABEL_P and
13387 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13388 rtx_jump_table_data::get_data_mode.
13389 (final_scan_insn): Likewise.
13391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13392 Alan Hayward <alan.hayward@arm.com>
13393 David Sherwood <david.sherwood@arm.com>
13395 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13396 trying to combine a full-register integer set with a subreg
13399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13400 Alan Hayward <alan.hayward@arm.com>
13401 David Sherwood <david.sherwood@arm.com>
13403 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13404 that is always either address_mode or pointer_mode.
13406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13407 Alan Hayward <alan.hayward@arm.com>
13408 David Sherwood <david.sherwood@arm.com>
13410 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13411 when the two are known to be equal.
13413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13414 Alan Hayward <alan.hayward@arm.com>
13415 David Sherwood <david.sherwood@arm.com>
13417 * simplify-rtx.c (simplify_const_unary_operation): Use
13418 is_a <scalar_int_mode> instead of checking for a nonzero
13419 precision. Forcibly convert op_mode to a scalar_int_mode
13420 in that case. More clearly differentiate the operand and
13421 result modes and use the former when deciding what the value
13422 of a count-bits operation should be. Use is_int_mode instead
13423 of checking for a MODE_INT. Remove redundant check for whether
13424 this mode has a zero precision.
13426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13427 Alan Hayward <alan.hayward@arm.com>
13428 David Sherwood <david.sherwood@arm.com>
13430 * optabs.c (widen_leading): Change the type of the mode argument
13431 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13432 (widen_bswap): Likewise.
13433 (expand_parity): Likewise.
13434 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13435 (expand_ffs): Likewise.
13436 (epand_unop): Check for scalar integer modes before calling the
13439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13440 Alan Hayward <alan.hayward@arm.com>
13441 David Sherwood <david.sherwood@arm.com>
13443 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13445 (expand_expr_real_1): Update call accordingly.
13447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13448 Alan Hayward <alan.hayward@arm.com>
13449 David Sherwood <david.sherwood@arm.com>
13451 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13452 value_mode arguments. Use scalar_int_mode internally.
13453 (store_bit_field_1): Rename the new integer mode from imode
13454 to op0_mode and use it instead of GET_MODE (op0). Update calls
13455 to store_split_bit_field, store_bit_field_using_insv and
13456 store_fixed_bit_field.
13457 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13458 Use scalar_int_mode internally. Use a bit count rather than a mode
13459 when calculating the largest bit size for get_best_mode.
13460 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13461 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13462 Remove assertion that OP0 has a scalar integer mode.
13463 (store_split_bit_field): Add op0_mode and value_mode arguments.
13464 Update calls to extract_fixed_bit_field.
13465 (extract_bit_field_using_extv): Add an op0_mode argument.
13466 Use scalar_int_mode internally.
13467 (extract_bit_field_1): Rename the new integer mode from imode to
13468 op0_mode and use it instead of GET_MODE (op0). Update calls to
13469 extract_split_bit_field, extract_bit_field_using_extv and
13470 extract_fixed_bit_field.
13471 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13472 to extract_split_bit_field and extract_fixed_bit_field_1.
13473 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13474 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13475 on the target mode.
13476 (extract_split_bit_field): Add an op0_mode argument. Update call
13477 to extract_fixed_bit_field.
13479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13480 Alan Hayward <alan.hayward@arm.com>
13481 David Sherwood <david.sherwood@arm.com>
13483 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13484 * explow.c (hard_function_value): Likewise.
13485 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13486 convert_to_mode call outside the loop.
13487 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13488 for the mode iterator. Require the mode specified by max_pieces
13490 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13492 (copy_blkmode_to_reg): Likewise.
13493 (set_storage_via_setmem): Likewise.
13494 * optabs.c (prepare_cmp_insn): Likewise.
13495 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13496 * stor-layout.c (finish_bitfield_representative): Likewise.
13498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13499 Alan Hayward <alan.hayward@arm.com>
13500 David Sherwood <david.sherwood@arm.com>
13502 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13503 * expr.c (convert_move): Use them.
13504 (convert_modes): Likewise.
13505 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13512 parameter for the mode of "x". Remove the "known_x", "known_mode"
13513 and "known_ret" arguments. Change the type of the mode argument
13514 to scalar_int_mode.
13515 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13516 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13517 (reg_num_sign_bit_copies_for_combine): Likewise.
13518 * rtlanal.c (nonzero_bits1): Likewise.
13519 (num_sign_bit_copies1): Likewise.
13520 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13521 (reg_num_sign_bit_copies_general): Likewise.
13522 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13523 (reg_nonzero_bits_general): Likewise.
13525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13526 Alan Hayward <alan.hayward@arm.com>
13527 David Sherwood <david.sherwood@arm.com>
13529 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13530 than in subroutines. Return 1 for non-integer modes.
13531 (cached_num_sign_bit_copies): Change the type of the mode parameter
13532 to scalar_int_mode.
13533 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13534 classes. Handle CONST_INT_P first and then check whether X also
13535 has a scalar integer mode. Check the same thing for inner registers
13536 of a SUBREG and for values that are being extended or truncated.
13538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13539 Alan Hayward <alan.hayward@arm.com>
13540 David Sherwood <david.sherwood@arm.com>
13542 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13543 in subroutines. Return the mode mask for non-integer modes.
13544 (cached_nonzero_bits): Change the type of the mode parameter
13545 to scalar_int_mode.
13546 (nonzero_bits1): Likewise. Remove early exit for other mode
13547 classes. Handle CONST_INT_P first and then check whether X
13548 also has a scalar integer mode.
13550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13551 Alan Hayward <alan.hayward@arm.com>
13552 David Sherwood <david.sherwood@arm.com>
13554 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13555 Return a scalar_int_mode and assert that the size is greater than
13557 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13558 (op_by_pieces_d::op_by_pieces_d): Likewise.
13559 (op_by_pieces_d::run): Likewise.
13560 (can_store_by_pieces): Likewise.
13562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13563 Alan Hayward <alan.hayward@arm.com>
13564 David Sherwood <david.sherwood@arm.com>
13566 * combine.c (extract_left_shift): Add a mode argument and update
13568 (make_compound_operation_int): Change the type of the mode parameter
13569 to scalar_int_mode and update the call to extract_left_shift.
13571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13572 Alan Hayward <alan.hayward@arm.com>
13573 David Sherwood <david.sherwood@arm.com>
13575 * combine.c (simplify_and_const_int): Change the type of the mode
13576 parameter to scalar_int_mode.
13577 (simplify_and_const_int_1): Likewise. Update recursive call.
13579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13580 Alan Hayward <alan.hayward@arm.com>
13581 David Sherwood <david.sherwood@arm.com>
13583 * combine.c (simplify_compare_const): Check that the mode is a
13584 scalar_int_mode (rather than VOIDmode) before testing its
13586 (simplify_comparison): Move COMPARISON_P handling out of the
13587 loop and restrict the latter part of the loop to scalar_int_modes.
13588 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13589 and when considering SUBREG_REGs. Use is_int_mode instead of
13590 checking GET_MODE_CLASS against MODE_INT.
13592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13593 Alan Hayward <alan.hayward@arm.com>
13594 David Sherwood <david.sherwood@arm.com>
13596 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13597 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13598 shift_unit_mode and for modes involved in scalar shifts.
13600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13601 Alan Hayward <alan.hayward@arm.com>
13602 David Sherwood <david.sherwood@arm.com>
13604 * combine.c (force_int_to_mode): New function, split out from...
13605 (force_to_mode): ...here. Keep xmode up-to-date and use it
13606 instead of GET_MODE (x).
13608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13609 Alan Hayward <alan.hayward@arm.com>
13610 David Sherwood <david.sherwood@arm.com>
13612 * optabs-query.h (extraction_insn::struct_mode): Change type to
13613 opt_scalar_int_mode and update comment.
13614 (extraction_insn::field_mode): Change type to scalar_int_mode.
13615 (extraction_insn::pos_mode): Likewise.
13616 * combine.c (make_extraction): Update accordingly.
13617 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13618 (get_optab_extraction_insn): Likewise.
13619 * recog.c (simplify_while_replacing): Likewise.
13620 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13621 parameter to opt_scalar_int_mode.
13623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13624 Alan Hayward <alan.hayward@arm.com>
13625 David Sherwood <david.sherwood@arm.com>
13627 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13628 to a scalar_int_mode instead of a machine_mode.
13629 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13630 (get_best_mode): Return a boolean and use a pointer argument to store
13631 the selected mode. Replace the limit mode parameter with a bit limit.
13632 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13633 for the values returned by bit_field_mode_iterator::next_mode.
13634 (store_bit_field): Update call to get_best_mode.
13635 (store_fixed_bit_field): Likewise.
13636 (extract_fixed_bit_field): Likewise.
13637 * expr.c (optimize_bitfield_assignment_op): Likewise.
13638 * fold-const.c (optimize_bit_field_compare): Likewise.
13639 (fold_truth_andor_1): Likewise.
13640 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13641 Update for new type of m_mode.
13642 (get_best_mode): As above.
13644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13645 Alan Hayward <alan.hayward@arm.com>
13646 David Sherwood <david.sherwood@arm.com>
13648 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13649 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13650 (store_bit_field): Check is_a <scalar_int_mode> before calling
13651 strict_volatile_bitfield_p.
13652 (extract_bit_field): Likewise.
13654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13655 Alan Hayward <alan.hayward@arm.com>
13656 David Sherwood <david.sherwood@arm.com>
13658 * target.def (cstore_mode): Return a scalar_int_mode.
13659 * doc/tm.texi: Regenerate.
13660 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13661 * targhooks.h (default_cstore_mode): Likewise.
13662 * targhooks.c (default_cstore_mode): Likewise, using a forced
13664 * expmed.c (emit_cstore): Expect the target of the cstore to be
13667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13668 Alan Hayward <alan.hayward@arm.com>
13669 David Sherwood <david.sherwood@arm.com>
13671 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13673 (niter_desc): Likewise mode.
13674 (iv_analyze): Add a mode parameter.
13676 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13677 and change its type to scalar_int_mode.
13678 * loop-iv.c: Update commentary at head of file.
13679 (iv_constant): Pass the mode paraeter before the rtx it describes
13680 and change its type to scalar_int_mode. Remove VOIDmode handling.
13681 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13682 (iv_extend): Likewise.
13683 (shorten_into_mode): Likewise.
13684 (iv_add): Use scalar_int_mode.
13685 (iv_mult): Likewise.
13686 (iv_shift): Likewise.
13687 (canonicalize_iv_subregs): Likewise.
13688 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13689 it describes and change its mode to scalar_int_mode. Also change
13690 the type of the returned inner_mode to scalar_int_mode.
13691 (get_biv_step): Likewise, turning outer_mode from a pointer
13692 into a direct parameter. Update call to get_biv_step_1.
13693 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13694 iv_constant and get_biv_step.
13695 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13696 and change its type to scalar_int_mode. Don't initialise iv->mode
13697 to VOIDmode and remove later checks for its still being VOIDmode.
13698 Update calls to iv_analyze_op and iv_analyze_expr. Check
13699 is_a <scalar_int_mode> when changing the mode under consideration.
13700 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13701 Update call to iv_analyze_expr.
13702 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13703 inner register is not also a scalar_int_mode. Update call to
13705 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13706 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13707 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13708 separate mode class checks. Update calls to iv_analyze. Remove
13709 fix-up of VOIDmodes after iv_analyze_biv.
13710 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13711 don't have a scalar_int_mode. Update call to biv_p.
13713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13714 Alan Hayward <alan.hayward@arm.com>
13715 David Sherwood <david.sherwood@arm.com>
13717 * cfgexpand.c (convert_debug_memory_address): Use
13718 as_a <scalar_int_mode>.
13719 * combine.c (expand_compound_operation): Likewise.
13720 (make_extraction): Likewise.
13721 (change_zero_ext): Likewise.
13722 (simplify_comparison): Likewise.
13723 * cse.c (cse_insn): Likewise.
13724 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13725 (mem_loc_descriptor): Likewise.
13726 (loc_descriptor): Likewise.
13727 * expmed.c (init_expmed_one_mode): Likewise.
13728 (synth_mult): Likewise.
13729 (emit_store_flag_1): Likewise.
13730 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13731 of a comparison with size.
13732 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13733 (reduce_to_bit_field_precision): Likewise.
13734 * function.c (expand_function_end): Likewise.
13735 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13736 * loop-doloop.c (doloop_modify): Likewise.
13737 * optabs.c (expand_binop): Likewise.
13738 (expand_unop): Likewise.
13739 (expand_copysign_absneg): Likewise.
13740 (prepare_cmp_insn): Likewise.
13741 (maybe_legitimize_operand): Likewise.
13742 * recog.c (const_scalar_int_operand): Likewise.
13743 * rtlanal.c (get_address_mode): Likewise.
13744 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13745 (simplify_cond_clz_ctz): Likewise.
13746 * tree-nested.c (get_nl_goto_field): Likewise.
13747 * tree.c (build_vector_type_for_mode): Likewise.
13748 * var-tracking.c (use_narrower_mode): Likewise.
13750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13751 Alan Hayward <alan.hayward@arm.com>
13752 David Sherwood <david.sherwood@arm.com>
13754 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13755 * builtins.c (expand_builtin_signbit): Use it.
13756 * cfgexpand.c (expand_debug_expr): Likewise.
13757 * dojump.c (do_jump): Likewise.
13758 (do_compare_and_jump): Likewise.
13759 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13760 * expmed.c (make_tree): Likewise.
13761 * expr.c (expand_expr_real_2): Likewise.
13762 (expand_expr_real_1): Likewise.
13763 (try_casesi): Likewise.
13764 * fold-const-call.c (fold_const_call_ss): Likewise.
13765 * fold-const.c (unextend): Likewise.
13766 (extract_muldiv_1): Likewise.
13767 (fold_single_bit_test): Likewise.
13768 (native_encode_int): Likewise.
13769 (native_encode_string): Likewise.
13770 (native_interpret_int): Likewise.
13771 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13772 * internal-fn.c (expand_addsub_overflow): Likewise.
13773 (expand_neg_overflow): Likewise.
13774 (expand_mul_overflow): Likewise.
13775 (expand_arith_overflow): Likewise.
13776 * match.pd: Likewise.
13777 * stor-layout.c (layout_type): Likewise.
13778 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13779 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13780 * tree-ssanames.c (get_range_info): Likewise.
13781 * tree-switch-conversion.c (array_value_type) Likewise.
13782 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13783 (vect_recog_divmod_pattern): Likewise.
13784 (vect_recog_mixed_size_cond_pattern): Likewise.
13785 * tree-vrp.c (extract_range_basic): Likewise.
13786 (simplify_float_conversion_using_ranges): Likewise.
13787 * tree.c (int_fits_type_p): Likewise.
13788 * ubsan.c (instrument_bool_enum_load): Likewise.
13789 * varasm.c (mergeable_string_section): Likewise.
13790 (narrowing_initializer_constant_valid_p): Likewise.
13791 (output_constant): Likewise.
13793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13794 Alan Hayward <alan.hayward@arm.com>
13795 David Sherwood <david.sherwood@arm.com>
13797 * machmode.h (NARROWEST_INT_MODE): New macro.
13798 * expr.c (alignment_for_piecewise_move): Use it instead of
13799 GET_CLASS_NARROWEST_MODE (MODE_INT).
13800 (push_block): Likewise.
13801 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13803 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13806 Alan Hayward <alan.hayward@arm.com>
13807 David Sherwood <david.sherwood@arm.com>
13809 * postreload.c (move2add_valid_value_p): Change the type of the
13810 mode parameter to scalar_int_mode.
13811 (move2add_use_add2_insn): Add a mode parameter and use it instead
13813 (move2add_use_add3_insn): Likewise.
13814 (reload_cse_move2add): Update accordingly.
13816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13817 Alan Hayward <alan.hayward@arm.com>
13818 David Sherwood <david.sherwood@arm.com>
13820 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13822 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13823 * optabs.c (expand_unop): Likewise.
13825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13826 Alan Hayward <alan.hayward@arm.com>
13827 David Sherwood <david.sherwood@arm.com>
13829 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13830 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13831 (popcount_loc_descriptor): Likewise.
13832 (bswap_loc_descriptor): Likewise.
13833 (rotate_loc_descriptor): Likewise.
13834 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13835 calling the functions above.
13837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13838 Alan Hayward <alan.hayward@arm.com>
13839 David Sherwood <david.sherwood@arm.com>
13841 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13843 (try_combine): Likewise.
13844 (simplify_if_then_else): Likewise.
13845 * cse.c (cse_insn): Likewise.
13846 * dwarf2out.c (mem_loc_descriptor): Likewise.
13847 * emit-rtl.c (gen_lowpart_common): Likewise.
13848 * simplify-rtx.c (simplify_truncation): Likewise.
13849 (simplify_binary_operation_1): Likewise.
13850 (simplify_const_relational_operation): Likewise.
13851 (simplify_ternary_operation): Likewise.
13852 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13855 Alan Hayward <alan.hayward@arm.com>
13856 David Sherwood <david.sherwood@arm.com>
13858 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13859 * reload.c (push_reload): Likewise.
13860 (find_reloads): Likewise.
13862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13863 Alan Hayward <alan.hayward@arm.com>
13864 David Sherwood <david.sherwood@arm.com>
13866 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13867 (make_compound_operation_int): Likewise.
13868 (change_zero_ext): Likewise.
13869 * expr.c (convert_move): Likewise.
13870 (convert_modes): Likewise.
13871 * fwprop.c (forward_propagate_subreg): Likewise.
13872 * loop-iv.c (get_biv_step_1): Likewise.
13873 * optabs.c (widen_operand): Likewise.
13874 * postreload.c (move2add_valid_value_p): Likewise.
13875 * recog.c (simplify_while_replacing): Likewise.
13876 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13877 (simplify_binary_operation_1): Likewise. Remove redundant
13878 mode equality check.
13880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13881 Alan Hayward <alan.hayward@arm.com>
13882 David Sherwood <david.sherwood@arm.com>
13884 * combine.c (combine_simplify_rtx): Add checks for
13885 is_a <scalar_int_mode>.
13886 (simplify_if_then_else): Likewise.
13887 (make_field_assignment): Likewise.
13888 (simplify_comparison): Likewise.
13889 * ifcvt.c (noce_try_bitop): Likewise.
13890 * loop-invariant.c (canonicalize_address_mult): Likewise.
13891 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13894 Alan Hayward <alan.hayward@arm.com>
13895 David Sherwood <david.sherwood@arm.com>
13897 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13898 is_a <scalar_int_mode> instead of != BLKmode.
13900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13901 Alan Hayward <alan.hayward@arm.com>
13902 David Sherwood <david.sherwood@arm.com>
13904 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13905 instead of != VOIDmode.
13906 * combine.c (if_then_else_cond): Likewise.
13907 (change_zero_ext): Likewise.
13908 * dwarf2out.c (mem_loc_descriptor): Likewise.
13909 (loc_descriptor): Likewise.
13910 * rtlanal.c (canonicalize_condition): Likewise.
13911 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13914 Alan Hayward <alan.hayward@arm.com>
13915 David Sherwood <david.sherwood@arm.com>
13917 * simplify-rtx.c (simplify_binary_operation_1): Use
13918 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13921 Alan Hayward <alan.hayward@arm.com>
13922 David Sherwood <david.sherwood@arm.com>
13924 * wide-int.h (int_traits<unsigned char>) New class.
13925 (int_traits<unsigned short>) Likewise.
13926 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13927 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13929 * combine.c (set_nonzero_bits_and_sign_copies): Use
13930 is_a <scalar_int_mode>.
13931 (find_split_point): Likewise.
13932 (combine_simplify_rtx): Likewise.
13933 (simplify_logical): Likewise.
13934 (expand_compound_operation): Likewise.
13935 (expand_field_assignment): Likewise.
13936 (make_compound_operation): Likewise.
13937 (extended_count): Likewise.
13938 (change_zero_ext): Likewise.
13939 (simplify_comparison): Likewise.
13940 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13941 (ucompare_loc_descriptor): Likewise.
13942 (minmax_loc_descriptor): Likewise.
13943 (mem_loc_descriptor): Likewise.
13944 (loc_descriptor): Likewise.
13945 * expmed.c (init_expmed_one_mode): Likewise.
13946 * lra-constraints.c (lra_constraint_offset): Likewise.
13947 * optabs.c (prepare_libcall_arg): Likewise.
13948 * postreload.c (move2add_note_store): Likewise.
13949 * reload.c (operands_match_p): Likewise.
13950 * rtl.h (load_extend_op): Likewise.
13951 * rtlhooks.c (gen_lowpart_general): Likewise.
13952 * simplify-rtx.c (simplify_truncation): Likewise.
13953 (simplify_unary_operation_1): Likewise.
13954 (simplify_binary_operation_1): Likewise.
13955 (simplify_const_binary_operation): Likewise.
13956 (simplify_const_relational_operation): Likewise.
13957 (simplify_subreg): Likewise.
13958 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13959 * var-tracking.c (adjust_mems): Likewise.
13960 (prepare_call_arguments): Likewise.
13962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13963 Alan Hayward <alan.hayward@arm.com>
13964 David Sherwood <david.sherwood@arm.com>
13966 * machmode.h (is_int_mode): New fuction.
13967 * combine.c (find_split_point): Use it.
13968 (combine_simplify_rtx): Likewise.
13969 (simplify_if_then_else): Likewise.
13970 (simplify_set): Likewise.
13971 (simplify_shift_const_1): Likewise.
13972 (simplify_comparison): Likewise.
13973 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13974 * cse.c (notreg_cost): Likewise.
13975 (cse_insn): Likewise.
13976 * cselib.c (cselib_lookup_1): Likewise.
13977 * dojump.c (do_jump_1): Likewise.
13978 (do_compare_rtx_and_jump): Likewise.
13979 * dse.c (get_call_args): Likewise.
13980 * dwarf2out.c (rtl_for_decl_init): Likewise.
13981 (native_encode_initializer): Likewise.
13982 * expmed.c (emit_store_flag_1): Likewise.
13983 (emit_store_flag): Likewise.
13984 * expr.c (convert_modes): Likewise.
13985 (store_field): Likewise.
13986 (expand_expr_real_1): Likewise.
13987 * fold-const.c (fold_read_from_constant_string): Likewise.
13988 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13989 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13990 * optabs.c (expand_binop): Likewise.
13991 (expand_unop): Likewise.
13992 (expand_abs_nojump): Likewise.
13993 (expand_one_cmpl_abs_nojump): Likewise.
13994 * simplify-rtx.c (mode_signbit_p): Likewise.
13995 (val_signbit_p): Likewise.
13996 (val_signbit_known_set_p): Likewise.
13997 (val_signbit_known_clear_p): Likewise.
13998 (simplify_relational_operation_1): Likewise.
13999 * tree.c (vector_type_mode): Likewise.
14001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14002 Alan Hayward <alan.hayward@arm.com>
14003 David Sherwood <david.sherwood@arm.com>
14005 * machmode.h (smallest_mode_for_size): Fix formatting.
14006 (smallest_int_mode_for_size): New function.
14007 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14008 instead of smallest_mode_for_size.
14009 * combine.c (make_extraction): Likewise.
14010 * config/arc/arc.c (arc_expand_movmem): Likewise.
14011 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14012 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14013 * config/s390/s390.c (s390_expand_insv): Likewise.
14014 * config/sparc/sparc.c (assign_int_registers): Likewise.
14015 * config/spu/spu.c (spu_function_value): Likewise.
14016 (spu_function_arg): Likewise.
14017 * coverage.c (get_gcov_type): Likewise.
14018 (get_gcov_unsigned_t): Likewise.
14019 * dse.c (find_shift_sequence): Likewise.
14020 * expmed.c (store_bit_field_1): Likewise.
14021 * expr.c (convert_move): Likewise.
14022 (store_field): Likewise.
14023 * internal-fn.c (expand_arith_overflow): Likewise.
14024 * optabs-query.c (get_best_extraction_insn): Likewise.
14025 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14026 * stor-layout.c (layout_type): Likewise.
14027 (initialize_sizetypes): Likewise.
14028 * targhooks.c (default_get_mask_mode): Likewise.
14029 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14032 Alan Hayward <alan.hayward@arm.com>
14033 David Sherwood <david.sherwood@arm.com>
14035 * machmode.h (opt_mode::else_blk): New function.
14036 (int_mode_for_mode): Declare.
14037 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14038 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14040 * cfgexpand.c (expand_debug_expr): Likewise.
14041 * combine.c (gen_lowpart_or_truncate): Likewise.
14042 (gen_lowpart_for_combine): Likewise.
14043 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14044 * config/avr/avr.c (avr_to_int_mode): Likewise.
14045 (avr_out_plus_1): Likewise.
14046 (avr_out_plus): Likewise.
14047 (avr_out_round): Likewise.
14048 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14049 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14050 (s390_expand_vcond): Likewise.
14051 * config/spu/spu.c (spu_split_immediate): Likewise.
14052 (spu_expand_mov): Likewise.
14053 * dse.c (get_stored_val): Likewise.
14054 * expmed.c (store_bit_field_1): Likewise.
14055 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14057 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14058 (extract_low_bits): Likewise.
14059 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14060 handling rather than repeating the check.
14061 (emit_group_store): Likewise.
14062 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14063 * optabs.c (expand_absneg_bit): Likewise.
14064 (expand_copysign_absneg): Likewise.
14065 (expand_copysign_bit): Likewise.
14066 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14067 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14068 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14069 * var-tracking.c (prepare_call_arguments): Likewise.
14070 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14071 int_mode_for_mode instead of mode_for_size.
14072 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14075 Alan Hayward <alan.hayward@arm.com>
14076 David Sherwood <david.sherwood@arm.com>
14078 * machmode.h (int_mode_for_size): New function.
14079 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14080 instead of mode_for_size.
14081 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14083 * combine.c (expand_field_assignment): Use int_mode_for_size
14084 instead of mode_for_size.
14085 (make_extraction): Likewise.
14086 (simplify_shift_const_1): Likewise.
14087 (simplify_comparison): Likewise.
14088 * dojump.c (do_jump): Likewise.
14089 * dwarf2out.c (mem_loc_descriptor): Likewise.
14090 * emit-rtl.c (init_derived_machine_modes): Likewise.
14091 * expmed.c (flip_storage_order): Likewise.
14092 (convert_extracted_bit_field): Likewise.
14093 * expr.c (copy_blkmode_from_reg): Likewise.
14094 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14095 * internal-fn.c (expand_mul_overflow): Likewise.
14096 * lower-subreg.c (simple_move): Likewise.
14097 * optabs-libfuncs.c (init_optabs): Likewise.
14098 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14099 * tree.c (vector_type_mode): Likewise.
14100 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14101 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14102 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14103 * tree-vect-stmts.c (vectorizable_load): Likewise.
14104 (vectorizable_store): Likewise.
14106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14107 Alan Hayward <alan.hayward@arm.com>
14108 David Sherwood <david.sherwood@arm.com>
14110 * coretypes.h (pod_mode): New type.
14111 (scalar_int_mode_pod): New typedef.
14112 * machmode.h (pod_mode): New class.
14113 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14114 * genmodes.c (emit_mode_int_n): Update accordingly.
14115 * lower-subreg.h (target_lower_subreg): Change type to
14116 scalar_int_mode_pod.
14117 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14118 scalar_int_mode_pod.
14120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14121 Alan Hayward <alan.hayward@arm.com>
14122 David Sherwood <david.sherwood@arm.com>
14124 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14125 machine_mode to scalar_int_mode.
14126 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14127 (rs6000_option_override_internal): Remove cast to int.
14128 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14129 machine_mode to scalar_int_mode.
14130 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14131 (rs6000_option_override_internal): Remove cast to int.
14132 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14133 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14135 * config/s390/s390.c (s390_expand_builtin): Likewise.
14136 * coretypes.h (scalar_int_mode): New type.
14137 (opt_scalar_int_mode): New typedef.
14138 * machmode.h (scalar_int_mode): New class.
14139 (scalar_int_mode::includes_p): New function.
14140 (byte_mode): Change type to scalar_int_mode.
14141 (word_mode): Likewise.
14142 (ptr_mode): Likewise.
14143 * emit-rtl.c (byte_mode): Likewise.
14144 (word_mode): Likewise.
14145 (ptr_mode): Likewise.
14146 (init_derived_machine_modes): Update accordingly.
14147 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14148 and MODE_PARTIAL_INT.
14149 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14150 opt_scalar_int_mode.
14152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14153 Alan Hayward <alan.hayward@arm.com>
14154 David Sherwood <david.sherwood@arm.com>
14156 * target.def (libgcc_floating_mode_supported_p): Take a
14158 * doc/tm.texi: Regenerate.
14159 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14161 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14162 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14166 Alan Hayward <alan.hayward@arm.com>
14167 David Sherwood <david.sherwood@arm.com>
14169 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14170 * doc/tm.texi: Regenerate.
14171 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14172 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14173 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14174 * targhooks.h (default_floatn_mode): Likewise.
14175 * targhooks.c (default_floatn_mode): Likewise.
14176 * tree.c (build_common_tree_nodes): Update accordingly.
14178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14179 Alan Hayward <alan.hayward@arm.com>
14180 David Sherwood <david.sherwood@arm.com>
14182 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14183 (mode_iterator::iterate_p): Likewise.
14184 (mode_iterator::get_wider): Likewise.
14185 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14188 Alan Hayward <alan.hayward@arm.com>
14189 David Sherwood <david.sherwood@arm.com>
14191 * coretypes.h (opt_scalar_float_mode): New typedef.
14192 * machmode.h (float_mode_for_size): New function.
14193 * emit-rtl.c (double_mode): Delete.
14194 (init_emit_once): Use float_mode_for_size.
14195 * stor-layout.c (layout_type): Likewise.
14196 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14199 Alan Hayward <alan.hayward@arm.com>
14200 David Sherwood <david.sherwood@arm.com>
14202 * output.h (assemble_real): Take a scalar_float_mode.
14203 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14204 * config/arm/arm.md (consttable_4): Likewise.
14205 (consttable_8): Likewise.
14206 (consttable_16): Likewise.
14207 * config/mips/mips.md (consttable_float): Likewise.
14208 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14209 * varasm.c (assemble_real): Take a scalar_float_mode.
14210 (output_constant_pool_2): Update accordingly.
14211 (output_constant): Likewise.
14213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14214 Alan Hayward <alan.hayward@arm.com>
14215 David Sherwood <david.sherwood@arm.com>
14217 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14218 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14219 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14220 (native_encode_real): Likewise.
14221 (native_interpret_real): Likewise.
14222 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14223 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14226 Alan Hayward <alan.hayward@arm.com>
14227 David Sherwood <david.sherwood@arm.com>
14229 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14230 <scalar_float_mode>. Simplify.
14231 (gen_extend_conv_libfunc): Likewise.
14233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14234 Alan Hayward <alan.hayward@arm.com>
14235 David Sherwood <david.sherwood@arm.com>
14237 * coretypes.h (scalar_float_mode): New type.
14238 * machmode.h (mode_traits::from_int): Use machine_mode if
14239 USE_ENUM_MODES is defined.
14240 (is_a): New function.
14242 (dyn_cast): Likewise.
14243 (scalar_float_mode): New class.
14244 (scalar_float_mode::includes_p): New function.
14245 (is_float_mode): Likewise.
14246 * gdbhooks.py (MachineModePrinter): New class.
14247 (build_pretty_printer): Use it for scalar_float_mode.
14248 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14249 (format_helper::format_helper): Turn into a template.
14250 * genmodes.c (get_mode_class): New function.
14251 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14252 or machine_mode if none.
14253 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14254 as_a <scalar_float_mode>.
14255 * dwarf2out.c (mem_loc_descriptor): Likewise.
14256 (insert_float): Likewise.
14257 (add_const_value_attribute): Likewise.
14258 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14259 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14260 (expand_unop): Update accordingly.
14261 (expand_abs_nojump): Likewise.
14262 (expand_copysign_absneg): Take a scalar_float_mode.
14263 (expand_copysign_bit): Likewise.
14264 (expand_copysign): Update accordingly.
14266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14267 Alan Hayward <alan.hayward@arm.com>
14268 David Sherwood <david.sherwood@arm.com>
14270 * coretypes.h (opt_mode): New class.
14271 * machmode.h (opt_mode): Likewise.
14272 (opt_mode::else_void): New function.
14273 (opt_mode::require): Likewise.
14274 (opt_mode::exists): Likewise.
14275 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14276 (GET_MODE_2XWIDER_MODE): Likewise.
14277 (mode_iterator::get_wider): Update accordingly.
14278 (mode_iterator::get_2xwider): Likewise.
14279 (mode_iterator::get_known_wider): Likewise, turning into a template.
14280 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14281 forcing a wider mode to exist.
14282 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14283 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14284 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14285 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14286 * lower-subreg.c (init_lower_subreg): Likewise.
14287 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14288 on the final iteration.
14289 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14290 a wider mode exists before asking for a move pattern.
14291 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14292 forcing a wider mode to exist.
14293 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14294 returning false if no such mode exists.
14295 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14296 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14297 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14298 Avoid checking for a MODE_INT if we already know the mode is not a
14300 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14301 forcing a wider mode to exist.
14302 (expmed_mult_highpart_optab): Likewise.
14303 (expmed_mult_highpart): Likewise.
14304 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14306 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14307 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14308 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14309 * internal-fn.c (expand_mul_overflow): Update use of
14310 GET_MODE_2XWIDER_MODE.
14311 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14312 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14313 GET_MODE_WIDER_MODE.
14314 (convert_plusminus_to_widen): Likewise.
14315 * tree-switch-conversion.c (array_value_type): Likewise.
14316 * var-tracking.c (emit_note_insn_var_location): Likewise.
14317 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14318 Return false inside rather than outside the loop if no wider mode
14320 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14321 and GET_MODE_2XWIDER_MODE
14322 (can_compare_p): Use else_void.
14323 * gdbhooks.py (OptMachineModePrinter): New class.
14324 (build_pretty_printer): Use it for opt_mode.
14326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14327 Alan Hayward <alan.hayward@arm.com>
14328 David Sherwood <david.sherwood@arm.com>
14330 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14331 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14334 Alan Hayward <alan.hayward@arm.com>
14335 David Sherwood <david.sherwood@arm.com>
14337 * machmode.h (mode_traits): New structure.
14338 (get_narrowest_mode): New function.
14339 (mode_iterator::start): Likewise.
14340 (mode_iterator::iterate_p): Likewise.
14341 (mode_iterator::get_wider): Likewise.
14342 (mode_iterator::get_known_wider): Likewise.
14343 (mode_iterator::get_2xwider): Likewise.
14344 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14345 (FOR_EACH_MODE): Likewise.
14346 (FOR_EACH_MODE_FROM): Likewise.
14347 (FOR_EACH_MODE_UNTIL): Likewise.
14348 (FOR_EACH_WIDER_MODE): Likewise.
14349 (FOR_EACH_2XWIDER_MODE): Likewise.
14350 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14351 * combine.c (simplify_comparison): Likewise
14352 * config/i386/i386.c (type_natural_mode): Likewise.
14353 * cse.c (cse_insn): Likewise.
14354 * dse.c (find_shift_sequence): Likewise.
14355 * emit-rtl.c (init_derived_machine_modes): Likewise.
14356 (init_emit_once): Likewise.
14357 * explow.c (hard_function_value): Likewise.
14358 * expmed.c (extract_fixed_bit_field_1): Likewise.
14359 (extract_bit_field_1): Likewise.
14360 (expand_divmod): Likewise.
14361 (emit_store_flag_1): Likewise.
14362 * expr.c (init_expr_target): Likewise.
14363 (convert_move): Likewise.
14364 (alignment_for_piecewise_move): Likewise.
14365 (widest_int_mode_for_size): Likewise.
14366 (emit_block_move_via_movmem): Likewise.
14367 (copy_blkmode_to_reg): Likewise.
14368 (set_storage_via_setmem): Likewise.
14369 (compress_float_constant): Likewise.
14370 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14371 * optabs-query.c (get_best_extraction_insn): Likewise.
14372 * optabs.c (expand_binop): Likewise.
14373 (expand_twoval_unop): Likewise.
14374 (expand_twoval_binop): Likewise.
14375 (widen_leading): Likewise.
14376 (widen_bswap): Likewise.
14377 (expand_parity): Likewise.
14378 (expand_unop): Likewise.
14379 (prepare_cmp_insn): Likewise.
14380 (prepare_float_lib_cmp): Likewise.
14381 (expand_float): Likewise.
14382 (expand_fix): Likewise.
14383 (expand_sfix_optab): Likewise.
14384 * postreload.c (move2add_use_add2_insn): Likewise.
14385 * reg-stack.c (reg_to_stack): Likewise.
14386 * reginfo.c (choose_hard_reg_mode): Likewise.
14387 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14388 * stor-layout.c (mode_for_size): Likewise.
14389 (smallest_mode_for_size): Likewise.
14390 (mode_for_vector): Likewise.
14391 (finish_bitfield_representative): Likewise.
14392 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14393 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14394 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14395 * var-tracking.c (prepare_call_arguments): Likewise.
14397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14398 Alan Hayward <alan.hayward@arm.com>
14399 David Sherwood <david.sherwood@arm.com>
14401 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14402 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14403 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14404 * machmode.h (mode_size): Move earlier in file.
14405 (mode_precision): Likewise.
14406 (mode_inner): Likewise.
14407 (mode_nunits): Likewise.
14408 (mode_unit_size): Likewise.
14409 (unit_unit_precision): Likewise.
14410 (mode_wider): Likewise.
14411 (mode_2xwider): Likewise.
14412 (machine_mode): New class.
14413 (mode_to_bytes): New function.
14414 (mode_to_bits): Likewise.
14415 (mode_to_precision): Likewise.
14416 (mode_to_inner): Likewise.
14417 (mode_to_unit_size): Likewise.
14418 (mode_to_unit_precision): Likewise.
14419 (mode_to_nunits): Likewise.
14420 (GET_MODE_SIZE): Use mode_to_bytes.
14421 (GET_MODE_BITSIZE): Use mode_to_bits.
14422 (GET_MODE_PRECISION): Use mode_to_precision.
14423 (GET_MODE_INNER): Use mode_to_inner.
14424 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14425 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14426 (GET_MODE_NUNITS): Use mode_to_nunits.
14427 * system.h (ALWAYS_INLINE): New macro.
14428 * config/powerpcspe/powerpcspe-c.c
14429 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14430 int for arg1_mode and arg2_mode.
14432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14433 Alan Hayward <alan.hayward@arm.com>
14434 David Sherwood <david.sherwood@arm.com>
14436 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14437 Prefix mode names with E_ in case statements.
14438 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14439 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14440 (aarch64_split_simd_move): Likewise.
14441 (aarch64_gen_storewb_pair): Likewise.
14442 (aarch64_gen_loadwb_pair): Likewise.
14443 (aarch64_gen_store_pair): Likewise.
14444 (aarch64_gen_load_pair): Likewise.
14445 (aarch64_get_condition_code_1): Likewise.
14446 (aarch64_constant_pool_reload_icode): Likewise.
14447 (get_rsqrte_type): Likewise.
14448 (get_rsqrts_type): Likewise.
14449 (get_recpe_type): Likewise.
14450 (get_recps_type): Likewise.
14451 (aarch64_gimplify_va_arg_expr): Likewise.
14452 (aarch64_simd_container_mode): Likewise.
14453 (aarch64_emit_load_exclusive): Likewise.
14454 (aarch64_emit_store_exclusive): Likewise.
14455 (aarch64_expand_compare_and_swap): Likewise.
14456 (aarch64_gen_atomic_cas): Likewise.
14457 (aarch64_emit_bic): Likewise.
14458 (aarch64_emit_atomic_swap): Likewise.
14459 (aarch64_emit_atomic_load_op): Likewise.
14460 (aarch64_evpc_trn): Likewise.
14461 (aarch64_evpc_uzp): Likewise.
14462 (aarch64_evpc_zip): Likewise.
14463 (aarch64_evpc_ext): Likewise.
14464 (aarch64_evpc_rev): Likewise.
14465 (aarch64_evpc_dup): Likewise.
14466 (aarch64_gen_ccmp_first): Likewise.
14467 (aarch64_gen_ccmp_next): Likewise.
14468 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14469 (alpha_emit_xfloating_libcall): Likewise.
14470 (emit_insxl): Likewise.
14471 (alpha_arg_type): Likewise.
14472 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14473 (arc_preferred_simd_mode): Likewise.
14474 (arc_secondary_reload): Likewise.
14475 (get_arc_condition_code): Likewise.
14476 (arc_print_operand): Likewise.
14477 (arc_legitimate_constant_p): Likewise.
14478 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14479 * config/arc/arc.md (casesi_load): Likewise.
14480 (casesi_compact_jump): Likewise.
14481 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14482 (cc_use_register): Likewise.
14483 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14484 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14485 (arm_init_iwmmxt_builtins): Likewise.
14486 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14487 (neon_expand_vector_init): Likewise.
14488 (arm_attr_length_move_neon): Likewise.
14489 (maybe_get_arm_condition_code): Likewise.
14490 (arm_emit_vector_const): Likewise.
14491 (arm_preferred_simd_mode): Likewise.
14492 (arm_output_iwmmxt_tinsr): Likewise.
14493 (thumb1_output_casesi): Likewise.
14494 (thumb2_output_casesi): Likewise.
14495 (arm_emit_load_exclusive): Likewise.
14496 (arm_emit_store_exclusive): Likewise.
14497 (arm_expand_compare_and_swap): Likewise.
14498 (arm_evpc_neon_vuzp): Likewise.
14499 (arm_evpc_neon_vzip): Likewise.
14500 (arm_evpc_neon_vrev): Likewise.
14501 (arm_evpc_neon_vtrn): Likewise.
14502 (arm_evpc_neon_vext): Likewise.
14503 (arm_validize_comparison): Likewise.
14504 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14505 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14506 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14507 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14508 (c6x_preferred_simd_mode): Likewise.
14509 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14510 (epiphany_rtx_costs): Likewise.
14511 * config/epiphany/predicates.md (proper_comparison_operator):
14513 * config/frv/frv.c (condexec_memory_operand): Likewise.
14514 (frv_emit_move): Likewise.
14515 (output_move_single): Likewise.
14516 (output_condmove_single): Likewise.
14517 (frv_hard_regno_mode_ok): Likewise.
14518 (frv_matching_accg_mode): Likewise.
14519 * config/h8300/h8300.c (split_adds_subs): Likewise.
14520 (h8300_rtx_costs): Likewise.
14521 (h8300_print_operand): Likewise.
14522 (compute_mov_length): Likewise.
14523 (output_logical_op): Likewise.
14524 (compute_logical_op_length): Likewise.
14525 (compute_logical_op_cc): Likewise.
14526 (h8300_shift_needs_scratch_p): Likewise.
14527 (output_a_shift): Likewise.
14528 (compute_a_shift_length): Likewise.
14529 (compute_a_shift_cc): Likewise.
14530 (expand_a_rotate): Likewise.
14531 (output_a_rotate): Likewise.
14532 * config/i386/i386.c (classify_argument): Likewise.
14533 (function_arg_advance_32): Likewise.
14534 (function_arg_32): Likewise.
14535 (function_arg_64): Likewise.
14536 (function_value_64): Likewise.
14537 (ix86_gimplify_va_arg): Likewise.
14538 (ix86_legitimate_constant_p): Likewise.
14539 (put_condition_code): Likewise.
14540 (split_double_mode): Likewise.
14541 (ix86_avx256_split_vector_move_misalign): Likewise.
14542 (ix86_expand_vector_logical_operator): Likewise.
14543 (ix86_split_idivmod): Likewise.
14544 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14545 (ix86_build_const_vector): Likewise.
14546 (ix86_build_signbit_mask): Likewise.
14547 (ix86_match_ccmode): Likewise.
14548 (ix86_cc_modes_compatible): Likewise.
14549 (ix86_expand_branch): Likewise.
14550 (ix86_expand_sse_cmp): Likewise.
14551 (ix86_expand_sse_movcc): Likewise.
14552 (ix86_expand_int_sse_cmp): Likewise.
14553 (ix86_expand_vec_perm_vpermi2): Likewise.
14554 (ix86_expand_vec_perm): Likewise.
14555 (ix86_expand_sse_unpack): Likewise.
14556 (ix86_expand_int_addcc): Likewise.
14557 (ix86_split_to_parts): Likewise.
14558 (ix86_vectorize_builtin_gather): Likewise.
14559 (ix86_vectorize_builtin_scatter): Likewise.
14560 (avx_vpermilp_parallel): Likewise.
14561 (inline_memory_move_cost): Likewise.
14562 (ix86_tieable_integer_mode_p): Likewise.
14563 (x86_maybe_negate_const_int): Likewise.
14564 (ix86_expand_vector_init_duplicate): Likewise.
14565 (ix86_expand_vector_init_one_nonzero): Likewise.
14566 (ix86_expand_vector_init_one_var): Likewise.
14567 (ix86_expand_vector_init_concat): Likewise.
14568 (ix86_expand_vector_init_interleave): Likewise.
14569 (ix86_expand_vector_init_general): Likewise.
14570 (ix86_expand_vector_set): Likewise.
14571 (ix86_expand_vector_extract): Likewise.
14572 (emit_reduc_half): Likewise.
14573 (ix86_emit_i387_round): Likewise.
14574 (ix86_mangle_type): Likewise.
14575 (ix86_expand_round_sse4): Likewise.
14576 (expand_vec_perm_blend): Likewise.
14577 (canonicalize_vector_int_perm): Likewise.
14578 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14579 (expand_vec_perm_1): Likewise.
14580 (expand_vec_perm_interleave3): Likewise.
14581 (expand_vec_perm_even_odd_pack): Likewise.
14582 (expand_vec_perm_even_odd_1): Likewise.
14583 (expand_vec_perm_broadcast_1): Likewise.
14584 (ix86_vectorize_vec_perm_const_ok): Likewise.
14585 (ix86_expand_vecop_qihi): Likewise.
14586 (ix86_expand_mul_widen_hilo): Likewise.
14587 (ix86_expand_sse2_abs): Likewise.
14588 (ix86_expand_pextr): Likewise.
14589 (ix86_expand_pinsr): Likewise.
14590 (ix86_preferred_simd_mode): Likewise.
14591 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14592 * config/i386/sse.md (*andnot<mode>3): Likewise.
14593 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14594 (*<code><mode>3): Likewise.
14595 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14596 (ia64_expand_atomic_op): Likewise.
14597 (ia64_arg_type): Likewise.
14598 (ia64_mode_to_int): Likewise.
14599 (ia64_scalar_mode_supported_p): Likewise.
14600 (ia64_vector_mode_supported_p): Likewise.
14601 (expand_vec_perm_broadcast): Likewise.
14602 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14603 (iq2000_function_arg_advance): Likewise.
14604 (iq2000_function_arg): Likewise.
14605 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14606 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14607 (m68k_libcall_value): Likewise.
14608 (m68k_function_value): Likewise.
14609 (sched_attr_op_type): Likewise.
14610 * config/mcore/mcore.c (mcore_output_move): Likewise.
14611 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14613 (microblaze_function_arg): Likewise.
14614 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14615 (mips_print_operand): Likewise.
14616 (mips_mode_ok_for_mov_fmt_p): Likewise.
14617 (mips_vector_mode_supported_p): Likewise.
14618 (mips_preferred_simd_mode): Likewise.
14619 (mips_expand_vpc_loongson_even_odd): Likewise.
14620 (mips_expand_vec_unpack): Likewise.
14621 (mips_expand_vi_broadcast): Likewise.
14622 (mips_expand_vector_init): Likewise.
14623 (mips_expand_vec_reduc): Likewise.
14624 (mips_expand_msa_cmp): Likewise.
14625 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14626 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14627 (cc_flags_for_mode): Likewise.
14628 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14629 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14630 (nds32_output_casesi_pc_relative): Likewise.
14631 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14632 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14633 (nvptx_gen_unpack): Likewise.
14634 (nvptx_gen_pack): Likewise.
14635 (nvptx_gen_shuffle): Likewise.
14636 (nvptx_gen_wcast): Likewise.
14637 (nvptx_preferred_simd_mode): Likewise.
14638 * config/pa/pa.c (pa_secondary_reload): Likewise.
14639 * config/pa/predicates.md (base14_operand): Likewise.
14640 * config/powerpcspe/powerpcspe-c.c
14641 (altivec_resolve_overloaded_builtin): Likewise.
14642 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14644 (rs6000_preferred_simd_mode): Likewise.
14645 (output_vec_const_move): Likewise.
14646 (rs6000_expand_vector_extract): Likewise.
14647 (rs6000_split_vec_extract_var): Likewise.
14648 (reg_offset_addressing_ok_p): Likewise.
14649 (rs6000_legitimate_offset_address_p): Likewise.
14650 (rs6000_legitimize_address): Likewise.
14651 (rs6000_emit_set_const): Likewise.
14652 (rs6000_const_vec): Likewise.
14653 (rs6000_emit_move): Likewise.
14654 (spe_build_register_parallel): Likewise.
14655 (rs6000_darwin64_record_arg_recurse): Likewise.
14656 (swap_selector_for_mode): Likewise.
14657 (spe_init_builtins): Likewise.
14658 (paired_init_builtins): Likewise.
14659 (altivec_init_builtins): Likewise.
14660 (do_load_for_compare): Likewise.
14661 (rs6000_generate_compare): Likewise.
14662 (rs6000_expand_float128_convert): Likewise.
14663 (emit_load_locked): Likewise.
14664 (emit_store_conditional): Likewise.
14665 (rs6000_output_function_epilogue): Likewise.
14666 (rs6000_handle_altivec_attribute): Likewise.
14667 (rs6000_function_value): Likewise.
14668 (emit_fusion_gpr_load): Likewise.
14669 (emit_fusion_p9_load): Likewise.
14670 (emit_fusion_p9_store): Likewise.
14671 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14672 (fusion_gpr_mem_load): Likewise.
14673 (fusion_addis_mem_combo_load): Likewise.
14674 (fusion_addis_mem_combo_store): Likewise.
14675 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14676 (fusion_gpr_mem_load): Likewise.
14677 (fusion_addis_mem_combo_load): Likewise.
14678 (fusion_addis_mem_combo_store): Likewise.
14679 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14681 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14682 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14683 (rs6000_preferred_simd_mode): Likewise.
14684 (output_vec_const_move): Likewise.
14685 (rs6000_expand_vector_extract): Likewise.
14686 (rs6000_split_vec_extract_var): Likewise.
14687 (reg_offset_addressing_ok_p): Likewise.
14688 (rs6000_legitimate_offset_address_p): Likewise.
14689 (rs6000_legitimize_address): Likewise.
14690 (rs6000_emit_set_const): Likewise.
14691 (rs6000_const_vec): Likewise.
14692 (rs6000_emit_move): Likewise.
14693 (rs6000_darwin64_record_arg_recurse): Likewise.
14694 (swap_selector_for_mode): Likewise.
14695 (paired_init_builtins): Likewise.
14696 (altivec_init_builtins): Likewise.
14697 (rs6000_expand_float128_convert): Likewise.
14698 (emit_load_locked): Likewise.
14699 (emit_store_conditional): Likewise.
14700 (rs6000_output_function_epilogue): Likewise.
14701 (rs6000_handle_altivec_attribute): Likewise.
14702 (rs6000_function_value): Likewise.
14703 (emit_fusion_gpr_load): Likewise.
14704 (emit_fusion_p9_load): Likewise.
14705 (emit_fusion_p9_store): Likewise.
14706 * config/rx/rx.c (rx_gen_move_template): Likewise.
14707 (flags_from_mode): Likewise.
14708 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14709 (s390_slb_comparison): Likewise.
14710 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14711 (s390_vector_mode_supported_p): Likewise.
14712 (s390_cc_modes_compatible): Likewise.
14713 (s390_match_ccmode_set): Likewise.
14714 (s390_canonicalize_comparison): Likewise.
14715 (s390_emit_compare_and_swap): Likewise.
14716 (s390_branch_condition_mask): Likewise.
14717 (s390_rtx_costs): Likewise.
14718 (s390_secondary_reload): Likewise.
14719 (__SECONDARY_RELOAD_CASE): Likewise.
14720 (s390_expand_cs): Likewise.
14721 (s390_preferred_simd_mode): Likewise.
14722 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14723 * config/sh/sh.c (sh_print_operand): Likewise.
14724 (dump_table): Likewise.
14725 (sh_secondary_reload): Likewise.
14726 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14727 * config/sh/sh.md (casesi_worker_1): Likewise.
14728 (casesi_worker_2): Likewise.
14729 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14730 (fcc_comparison_operator): Likewise.
14731 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14732 (emit_soft_tfmode_cvt): Likewise.
14733 (sparc_preferred_simd_mode): Likewise.
14734 (output_cbranch): Likewise.
14735 (sparc_print_operand): Likewise.
14736 (sparc_expand_vec_perm_bmask): Likewise.
14737 (vector_init_bshuffle): Likewise.
14738 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14739 (spu_vector_mode_supported_p): Likewise.
14740 (spu_expand_insv): Likewise.
14741 (spu_emit_branch_or_set): Likewise.
14742 (spu_handle_vector_attribute): Likewise.
14743 (spu_builtin_splats): Likewise.
14744 (spu_builtin_extract): Likewise.
14745 (spu_builtin_promote): Likewise.
14746 (spu_expand_sign_extend): Likewise.
14747 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14748 (tilegx_simd_int): Likewise.
14749 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14750 (tilepro_simd_int): Likewise.
14751 * config/v850/v850.c (const_double_split): Likewise.
14752 (v850_print_operand): Likewise.
14753 (ep_memory_offset): Likewise.
14754 * config/vax/vax.c (vax_rtx_costs): Likewise.
14755 (vax_output_int_move): Likewise.
14756 (vax_output_int_add): Likewise.
14757 (vax_output_int_subtract): Likewise.
14758 * config/visium/predicates.md (visium_branch_operator): Likewise.
14759 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14760 (visium_print_operand_address): Likewise.
14761 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14762 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14763 (xtensa_expand_conditional_branch): Likewise.
14764 (xtensa_copy_incoming_a7): Likewise.
14765 (xtensa_output_literal): Likewise.
14766 * dfp.c (decimal_real_maxval): Likewise.
14767 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14770 Alan Hayward <alan.hayward@arm.com>
14771 David Sherwood <david.sherwood@arm.com>
14773 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14774 (mode_nunits_inline): Likewise.
14775 (mode_inner_inline): Likewise.
14776 (mode_unit_size_inline): Likewise.
14777 (mode_unit_precision_inline): Likewise.
14778 (emit_insn_modes_h): Likewise. Also emit a #define of the
14780 (emit_mode_wider): Add an E_ prefix to mode names.
14781 (emit_mode_complex): Likewise.
14782 (emit_mode_inner): Likewise.
14783 (emit_mode_adjustments): Likewise.
14784 (emit_mode_int_n): Likewise.
14785 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14786 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14787 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14788 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14789 (CRC32_BUILTIN, ENTRY): Likewise.
14790 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14791 (aarch64_pop_regs): Likewise.
14792 (aarch64_process_components): Likewise.
14793 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14794 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14795 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14796 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14797 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14798 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14800 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14801 (pcmpistr): Likewise.
14802 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14803 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14804 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14806 * config/rl78/rl78.c (mduc_regs): Likewise.
14807 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14808 (htm_expand_builtin): Likewise.
14809 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14810 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14811 * config/xtensa/xtensa.c (print_operand): Likewise.
14812 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14813 (NUM_MODE_VECTOR_INT): Likewise.
14814 * genoutput.c (null_operand): Likewise.
14815 (output_operand_data): Likewise.
14816 * genrecog.c (print_parameter_value): Likewise.
14817 * lra.c (debug_operand_data): Likewise.
14819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14821 * dwarf2out.c (scompare_loc_descriptor_wide)
14822 (scompare_loc_descriptor_narrow): New functions, split out from...
14823 (scompare_loc_descriptor): ...here.
14824 * expmed.c (emit_store_flag_int): New function, split out from...
14825 (emit_store_flag): ...here.
14827 2017-08-30 Richard Biener <rguenther@suse.de>
14829 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14830 (dwarf2out_early_finish): Move setting of AT_pubnames from
14831 early debug output to early finish.
14833 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14835 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14836 and -mdata-region to the assembler.
14838 2017-08-30 Richard Biener <rguenther@suse.de>
14840 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14842 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14843 (dwarf2out_early_global_decl): Only generate a DIE for the
14844 abstract origin if it doesn't already exist or is a declaration DIE.
14845 (resolve_addr): Do not add the linkage name twice when
14846 generating a stub DIE for the DW_TAG_GNU_call_site target.
14848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14850 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14851 Use machine_mode rather than int for arg1_mode.
14853 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14856 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14857 that the second argument of the built-in functions to unpack
14858 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14859 switch statement instead a lot of if statements.
14860 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14861 Allow 64-bit values to be in Altivec registers as well as
14862 traditional floating point registers.
14863 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14865 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14867 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14868 MAX_REGS_PER_ADDRESS == 1.
14870 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14872 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14873 * config/i386/i386.c (ix86_option_override_internal): Simplify
14874 setting of opts->x_flag_entry.
14876 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14877 Jakub Jelinek <jakub@redhat.com>
14878 Richard Biener <rguenther@suse.de>
14880 PR tree-optimization/81503
14881 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14882 folded constant fits in the target type; reorder tests for clarity.
14884 2017-08-29 Martin Liska <mliska@suse.cz>
14886 * passes.def: Include pass_lower_switch.
14887 * stmt.c (dump_case_nodes): Remove and move to
14888 tree-switch-conversion.
14889 (case_values_threshold): Likewise.
14890 (expand_switch_as_decision_tree_p): Likewise.
14891 (emit_case_decision_tree): Likewise.
14892 (expand_case): Likewise.
14893 (balance_case_nodes): Likewise.
14894 (node_has_low_bound): Likewise.
14895 (node_has_high_bound): Likewise.
14896 (node_is_bounded): Likewise.
14897 (emit_case_nodes): Likewise.
14898 (struct simple_case_node): New struct.
14899 (add_case_node): Remove.
14900 (emit_case_dispatch_table): Use vector instead of case_list.
14901 (reset_out_edges_aux): Remove.
14902 (compute_cases_per_edge): Likewise.
14903 (expand_case): Build list of simple_case_node.
14904 (expand_sjlj_dispatch_table): Use it.
14905 * tree-switch-conversion.c (struct case_node): Moved from
14906 stmt.c and adjusted.
14907 (emit_case_nodes): Likewise.
14908 (node_has_low_bound): Likewise.
14909 (node_has_high_bound): Likewise.
14910 (node_is_bounded): Likewise.
14911 (case_values_threshold): Likewise.
14912 (reset_out_edges_aux): Likewise.
14913 (compute_cases_per_edge): Likewise.
14914 (add_case_node): Likewise.
14915 (dump_case_nodes): Likewise.
14916 (balance_case_nodes): Likewise.
14917 (expand_switch_as_decision_tree_p): Likewise.
14918 (emit_jump): Likewise.
14919 (emit_case_decision_tree): Likewise.
14920 (try_switch_expansion): Likewise.
14921 (do_jump_if_equal): Likewise.
14922 (emit_cmp_and_jump_insns): Likewise.
14923 (fix_phi_operands_for_edge): New function.
14924 (record_phi_operand_mapping): Likewise.
14925 (class pass_lower_switch): New pass.
14926 (pass_lower_switch::execute): New function.
14927 (make_pass_lower_switch): Likewise.
14928 (conditional_probability):
14929 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14930 * tree-pass.h: Add make_pass_lower_switch.
14932 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14935 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14938 2017-08-29 Richard Biener <rguenther@suse.de>
14940 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14941 we do not add a DW_AT_inline attribute twice.
14942 (gen_subprogram_die): Remove code setting DW_AT_inline on
14943 DECL_ABSTRACT_P nodes.
14945 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14947 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14948 calls to internal functions.
14949 (gimplify_modify_expr): Likewise.
14950 * tree-call-cdce.c (use_internal_fn): Likewise.
14951 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14952 (convert_to_divmod): Set the nothrow flag.
14953 * tree-if-conv.c (predicate_mem_writes): Likewise.
14954 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14955 (vectorizable_call): Likewise.
14956 (vectorizable_store): Likewise.
14957 (vectorizable_load): Likewise.
14958 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14959 (vect_recog_mask_conversion_pattern): Likewise.
14961 2017-08-29 Martin Liska <mliska@suse.cz>
14964 * gcc.c (driver_handle_option): Add new argument.
14965 * opts-common.c (handle_option): Pass
14966 target_option_override_hook.
14967 * opts-global.c (lang_handle_option): Add new option.
14968 (set_default_handlers): Add new argument.
14969 (decode_options): Likewise.
14970 * opts.c (target_handle_option): Likewise.
14971 (common_handle_option): Call target_option_override_hook.
14972 * opts.h (struct cl_option_handler_func): Add hook for
14973 target option override.
14974 (struct cl_option_handlers): Likewise.
14975 (set_default_handlers): Add new argument.
14976 (decode_options): Likewise.
14977 (common_handle_option): Likewise.
14978 (target_handle_option): Likewise.
14979 * toplev.c (toplev::main): Pass targetm.target_option.override
14982 2017-08-29 Richard Biener <rguenther@suse.de>
14983 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14985 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14986 life to the active subtree.
14988 2017-08-28 Jeff Law <law@redhat.com>
14990 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14991 derive_equivalences.
14992 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14993 Code moved into....
14994 (edge_info::derive_equivalences): New private member function
14996 * tree-ssa-dom.c (class edge_info): Changed from a struct
14997 to a class. Add ctor/dtor, methods and data members.
14998 (edge_info::edge_info): Renamed from allocate_edge_info.
14999 Initialize additional members.
15000 (edge_info::~edge_info): New.
15001 (free_dom_edge_info): Delete the edge info.
15002 (record_edge_info): Use new class & associated member functions.
15003 Tighten forms for testing for edge equivalences.
15004 (record_temporary_equivalences): Iterate over the simple
15005 equivalences rather than assuming there's only one per edge.
15006 (cprop_into_successor_phis): Iterate over the simple
15007 equivalences rather than assuming there's only one per edge.
15008 (optimize_stmt): Use operand_equal_p rather than pointer
15009 equality for mini-DSE code.
15011 2017-08-28 Nathan Sidwell <nathan@acm.org>
15013 * gcc.c (execute): Fold SIGPIPE handling into switch
15014 statement. Adjust internal error message.
15016 2017-08-28 Richard Biener <rguenther@suse.de>
15019 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15020 Do nothing for removed DIEs.
15022 2017-08-28 Richard Biener <rguenther@suse.de>
15024 PR tree-optimization/81977
15025 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15028 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15031 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15032 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15033 using targetm.gen_mem_thread_fence.
15035 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15038 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15039 predicate to register_operand. Reorder operands.
15040 (*btr<mode>): Ditto.
15041 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15042 (*btr<mode>_mask): Ditto.
15044 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15046 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15047 * config/rs6000/xmmintrin.h: New file.
15048 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15050 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15053 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15054 parameter and_insn and return it.
15055 (recombine_lvx_pattern): Insert a copy to ensure availability of
15056 the base register of the copied masking operation at the point of
15057 the instruction replacement.
15058 (recombine_stvx_pattern): Likewise.
15060 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15062 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15063 undocumented switches.
15064 (-mpower9-dform-vector): Likewise.
15065 (-mpower9-dform): Likewise.
15066 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15067 comments to delete references to -mpower9-dform* switches.
15068 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15069 Delete reference to -mpower9-dform* switches, test for
15070 -mpower9-vector instead.
15071 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15072 (OTHER_P9_VECTOR_MASKS): Likewise.
15073 (POWERPC_MASKS): Likewise.
15074 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15075 tests against -mpower9-dform* to -mpower9-vector. Delete code
15076 that checked for -mpower9-dform* consistancy with other options.
15077 Add test for -mpower9-misc to enable other power9 switches.
15078 (rs6000_init_hard_regno_mode_ok): Likewise.
15079 (rs6000_option_override_internal): Likewise.
15080 (rs6000_emit_prologue): Likewise.
15081 (rs6000_emit_epilogue): Likewise.
15082 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15083 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15084 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15086 (emit_fusion_p9_store): Likewise.
15087 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15088 resetting these macros if the assembler does not support ISA 3.0
15090 (TARGET_P9_DFORM_VECTOR): Likewise.
15091 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15092 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15094 2017-08-25 Alan Modra <amodra@gmail.com>
15096 PR rtl-optimization/81747
15097 * cse.c (cse_extended_basic_block): Don't attempt to record
15098 equivalences for degenerate conditional branches that branch
15099 to their fall-through.
15101 2017-08-24 Martin Sebor <msebor@redhat.com>
15103 PR middle-end/81908
15104 * gimple-fold.c (size_must_be_zero_p): New function.
15105 (gimple_fold_builtin_memory_op): Call it.
15107 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15109 * config/rs6000/mm_malloc.h: New file.
15111 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15113 PR tree-optimization/81913
15114 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15115 analysis when either IVs in condition can wrap.
15117 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15119 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15120 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15122 2017-08-24 Richard Biener <rguenther@suse.de>
15125 * targhooks.c (default_target_can_inline_p): Properly
15126 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15127 is present and always compare.
15128 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15129 infer -mfpmath=sse from TARGET_SSE_P.
15130 (ix86_can_inline_p): Properly use target_option_default_node when
15131 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15133 2017-08-24 Richard Biener <rguenther@suse.de>
15136 * dwarf2out.c (output_die): Handle flag_generate_offload like
15138 (output_comp_unit): Likewise.
15139 (gen_array_type_die): Likewise.
15140 (dwarf2out_early_finish): Likewise.
15141 (note_variable_value_in_expr): Likewise.
15142 (dwarf2out_finish): Likewise. Adjust assert.
15143 * cgraphunit.c (symbol_table::compile): Move setting of
15144 flag_generate_offload earlier ...
15145 (symbol_table::finalize_compilation_unit): ... here, before
15146 early debug finalization.
15148 2017-08-24 Richard Biener <rguenther@suse.de>
15150 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15151 and ipa-fnsummary.h.
15152 (ix86_can_inline_p): When ix86_fpmath flags do not match
15153 check whether the callee uses FP math at all.
15155 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15157 PR middle-end/81931
15158 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15159 instead of TYPE_PRECISION.
15161 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15162 Alan Hayward <alan.hayward@arm.com>
15163 David Sherwood <david.sherwood@arm.com>
15165 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15167 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15169 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15171 (ix86_frame::stack_realign_allocate): New field.
15172 (struct machine_frame_state): Modify comments.
15173 (machine_frame_state::sp_realigned_fp_end): New field.
15174 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15175 layout calculation.
15176 (sp_valid_at): Add assertion to assure no attempt to access invalid
15177 offset of a realigned stack.
15178 (fp_valid_at): Likewise.
15179 (choose_baseaddr): Modify comments.
15180 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15181 ix86_expand_prologue.
15182 (ix86_expand_prologue): Modify stack realignment and allocation.
15183 (ix86_expand_epilogue): Modify comments.
15184 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15185 avx2_runtime, avx512f, and avx512f_runtime.
15187 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15189 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15190 (mstackrealign): Do not init to -1.
15191 * config/i386/i386.c (ix86_option_override_internal):
15192 Check opts_set, not opts when setting default value of
15193 opts->x_ix86_force_align_arg_pointer.
15195 2017-08-23 Richard Biener <rguenther@suse.de>
15197 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15198 lang_hooks.decl_printable_name.
15199 * print-rtl-function.c (print_rtx_function): Likewise.
15200 * tree-pretty-print.c (dump_function_header): Likewise.
15202 2017-08-23 Richard Biener <rguenther@suse.de>
15205 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15206 -g0 at compile-time.
15208 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15210 PR middle-end/19706
15211 * doc/sourcebuild.texi (Other hardware attributes):
15214 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15216 PR middle-end/19706
15217 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15218 Fix single-use check.
15220 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15222 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15224 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15226 * doc/install.texi: Modify to add more details on running selected
15229 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15231 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15232 is combined with -mabi=ms.
15233 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15236 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15238 PR tree-optimization/81488
15239 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15240 and cached_basis fields.
15241 (MAX_SPREAD): New constant.
15242 (alloc_cand_and_find_basis): Initialize new fields.
15243 (clear_visited): New function.
15244 (create_phi_basis_1): Rename from create_phi_basis, set visited
15245 and cached_basis fields.
15246 (create_phi_basis): New wrapper function.
15247 (phi_add_costs_1): Rename from phi_add_costs, add spread
15248 parameter, set visited field, short-circuit when limits reached.
15249 (phi_add_costs): New wrapper function.
15250 (record_phi_increments_1): Rename from record_phi_increments, set
15252 (record_phi_increments): New wrapper function.
15253 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15254 (phi_incr_cost): New wrapper function.
15255 (all_phi_incrs_profitable_1): Rename from
15256 all_phi_incrs_profitable, set visited field.
15257 (all_phi_incrs_profitable): New wrapper function.
15259 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayward <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15264 that takes the outer and inner modes.
15265 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15266 comparison as the canonical test for a paradoxical subreg.
15267 * combine.c (simplify_set): Use paradoxical_subreg_p.
15268 (make_extraction): Likewise.
15269 (force_to_mode): Likewise.
15270 (rtx_equal_for_field_assignment_p): Likewise.
15271 (gen_lowpart_for_combine): Likewise.
15272 (simplify_comparison): Likewise.
15273 * cse.c (equiv_constant): Likewise.
15274 * expmed.c (store_bit_field_1): Likewise.
15275 * final.c (alter_subreg): Likewise.
15276 * fwprop.c (propagate_rtx): Likewise.
15277 (forward_propagate_subreg): Likewise.
15278 * ira-conflicts.c (ira_build_conflicts): Likewise.
15279 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15280 * lra-constraints.c (curr_insn_transform): Likewise.
15281 (split_reg): Likewise.
15282 * lra-eliminations.c (move_plus_up): Likewise.
15283 (lra_eliminate_regs_1): Likewise.
15284 * recog.c (general_operand): Likewise.
15285 * ree.c (combine_reaching_defs): Likewise.
15286 * reload.c (push_reload): Likewise.
15287 (find_reloads): Likewise.
15288 * reload1.c (elimination_effects): Likewise.
15289 (compute_reload_subreg_offset): Likewise.
15290 (choose_reload_regs): Likewise.
15291 * rtlanal.c (subreg_lsb_1): Likewise.
15292 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15293 (simplify_subreg): Likewise.
15294 * var-tracking.c (track_loc_p): Likewise.
15295 * emit-rtl.c (byte_lowpart_offset): Likewise.
15296 (paradoxical_subreg_p): Delete out-of-line definition.
15298 2017-08-22 Jeff Law <law@redhat.com>
15300 PR tree-optimization/81741
15301 PR tree-optimization/71947
15302 * tree-ssa-dom.c: Include tree-inline.h.
15303 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15304 equivalences if one is more expensive to compute than the other.
15305 * tree-ssa-scopedtables.h (class const_or_copies): Make
15306 record_const_or_copy_raw method private.
15307 (class avail_exprs_stack): New method simplify_binary_operation.
15308 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15309 avail_exprs_stack::simplify_binary_operation as needed.
15310 (avail_exprs_stack::simplify_binary_operation): New function.
15312 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15314 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15315 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15316 (DOT_SYMBOLS): Likewise.
15317 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15318 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15319 (RS6000_ABI_NAME): Likewise.
15320 (TARGET_CMODEL): Likewise.
15321 (TOC_SECTION_ASM_OP): Likewise.
15322 (SET_CMODEL): New macro.
15323 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15325 2017-08-22 Richard Biener <rguenther@suse.de>
15327 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15328 to allow for free-lang-data replacements similar to verify_type_variant.
15330 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15335 VOIDmode for the LABEL_REF.
15337 2017-08-22 Richard Biener <rguenther@suse.de>
15339 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15341 2017-08-22 Richard Biener <rguenther@suse.de>
15343 * common.opt (feliminate-dwarf2-dups): Ignore.
15344 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15345 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15346 same_die_p_wrap, compute_section_prefix,
15347 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15348 (comdat_symbol_id, comdat_symbol_number): Likewise.
15349 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15351 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15352 (output_die): Mark unreachable path unreachable.
15353 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15354 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15355 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15356 (dwarf2out_early_finish): Likewise.
15358 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15360 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15362 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15365 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15366 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15367 (avr_attribute_table) <io, io_low, address>: Initialize
15368 .decl_required with true.
15370 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15372 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15373 undocumented debugging options.
15374 (-mvsx-scalar-double): Likewise.
15375 (-mallow-df-permute): Likewise.
15376 (-mvectorize-builtins): Likewise.
15377 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15378 (rs6000_builtin_vectorized_function): Likewise.
15379 (rs6000_builtin_md_vectorized_function): Likewise.
15380 (rs6000_opt_vars): Likewise.
15382 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15385 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15386 (*btrq_imm): Rename from *btrq.
15387 (*btcq_imm): Rename from *btcq.
15388 (btsc): New code attribute.
15389 (*<btsc><mode>): New insn pattern.
15390 (*btr<mode>): Ditto.
15391 (*<btsc><mode>_mask): New insn_and_split pattern.
15392 (*btr<mode>_mask): Ditto.
15394 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15395 Alan Hayward <alan.hayward@arm.com>
15396 David Sherwood <david.sherwood@arm.com>
15398 * function.c (pad_below): Simplify padding calculation.
15400 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15401 Alan Hayward <alan.hayward@arm.com>
15402 David Sherwood <david.sherwood@arm.com>
15404 * target.def (function_prologue): Remove frame size argument.
15405 (function_epilogue): Likewise.
15406 * doc/tm.texi: Regenerate.
15407 * final.c (final_start_function): Update call to function_prologue.
15408 (final_end_function): Update call to function_epilogue.
15409 (default_function_pro_epilogue): Remove frame size argument.
15410 * output.h (default_function_pro_epilogue): Likewise.
15411 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15412 (arm_output_function_prologue): Likewise.
15413 * config/frv/frv.c (frv_function_prologue): Likewise.
15414 (frv_function_epilogue): Likewise.
15415 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15416 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15417 (ia64_output_function_epilogue): Likewise.
15418 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15419 (m32r_output_function_epilogue): Likewise.
15420 * config/microblaze/microblaze.c (microblaze_function_prologue)
15421 (microblaze_function_epilogue): Likewise.
15422 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15423 (mips_output_function_epilogue): Likewise.
15424 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15425 (mmix_target_asm_function_epilogue): Likewise.
15426 * config/msp430/msp430.c (msp430_start_function): Likewise.
15427 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15428 (nds32_asm_function_epilogue): Likewise.
15429 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15430 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15431 (pa_output_function_epilogue): Likewise.
15432 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15433 (rs6000_output_function_epilogue): Likewise.
15434 * config/rl78/rl78.c (rl78_start_function): Likewise.
15435 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15436 (rs6000_output_function_epilogue): Likewise.
15437 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15438 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15439 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15440 (sparc_asm_function_epilogue): Likewise.
15442 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15444 * tree.h (type_has_mode_precision_p): New function.
15445 * convert.c (convert_to_integer_1): Use it.
15446 * expr.c (expand_expr_real_2): Likewise.
15447 (expand_expr_real_1): Likewise.
15448 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15449 * match.pd: Likewise.
15450 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15451 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15452 * tree-tailcall.c (process_assignment): Likewise.
15453 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15454 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15455 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15456 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15457 (vectorizable_assignment): Likewise.
15458 (vectorizable_shift): Likewise.
15459 (vectorizable_operation): Likewise.
15460 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15462 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15464 * match.pd: Add pow (C, x) simplification.
15466 2017-08-21 Richard Biener <rguenther@suse.de>
15468 PR tree-optimization/81900
15469 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15470 for blocks with abnormal predecessors.
15471 (compute_antic): Do not set visited flag prematurely.
15473 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15476 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15478 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15480 * stor-layout.h (vector_type_mode): Move to...
15481 * tree.h (vector_type_mode): ...here.
15482 * stor-layout.c (vector_type_mode): Move to...
15483 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15485 2017-08-21 Richard Biener <rguenther@suse.de>
15487 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15488 register_external_die hooks.
15489 (debug_false_tree_charstarstar_uhwistar): Declare.
15490 (debug_nothing_tree_charstar_uhwi): Likewise.
15491 * debug.c (do_nothing_debug_hooks): Adjust.
15492 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15493 (debug_nothing_tree_charstar_uhwi): Likewise.
15494 * dbxout.c (dbx_debug_hooks): Adjust.
15495 (xcoff_debug_hooks): Likewise.
15496 * sdbout.c (sdb_debug_hooks): Likewise.
15497 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15498 * dwarf2out.c (macinfo_label_base): New global.
15499 (dwarf2out_register_external_die): New function for the
15500 register_external_die hook.
15501 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15502 (dwarf2_debug_hooks): Use them.
15503 (dwarf2_lineno_debug_hooks): Adjust.
15504 (struct die_struct): Add with_offset flag.
15505 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15506 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15507 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15508 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15509 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15510 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15511 defining section names for the early LTO debug variants.
15512 (reset_indirect_string): New helper.
15513 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15514 (print_dw_val): Add support for offsetted symbol references.
15515 (get_ultimate_context): Split out from is_cxx.
15516 (is_cxx): Use get_ultimate_context.
15517 (is_fortran): Add decl overload.
15518 (compute_comp_unit_symbol): Split out worker from
15519 compute_section_prefix.
15520 (compute_section_prefix): Call compute_comp_unit_symbol and
15521 set comdat_type_p here.
15522 (output_die): Skip DIE symbol output for the LTO added one.
15523 Handle DIE symbol references with offset.
15524 (output_comp_unit): Guard section name mangling properly.
15525 For LTO debug sections emit a symbol at the section beginning
15526 which we use to refer to its DIEs.
15527 (add_abstract_origin_attribute): For DIEs registered via
15528 dwarf2out_register_external_die directly refer to the early
15529 DIE rather than indirectly through the shadow one we created.
15530 Remove obsolete call to dwarf2out_abstract_function for
15531 non-function/block origins.
15532 (gen_array_type_die): When generating early LTO debug do
15533 not emit DW_AT_string_length.
15534 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15535 late when in LTO. As suggested place a gcc_unreachable for
15536 the DECL_ABSTRACT_P case.
15537 (gen_subprogram_die): Avoid another specification DIE
15538 for early built declarations/definitions for the late LTO case.
15539 (gen_variable_die): Add type references for late duplicated VLA dies
15541 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15542 we have the abstract instance already.
15543 (process_scope_var): Adjust decl DIE contexts in LTO which
15544 first puts them in limbo.
15545 (gen_decl_die): Do not generate type DIEs late apart from
15546 types for VLAs or for decls we do not yet have a DIE. Do not
15547 call dwarf2out_abstract_function late.
15548 (dwarf2out_early_global_decl): Make sure to create DIEs
15549 for abstract instances of a decl first.
15550 (dwarf2out_late_global_decl): Adjust comment.
15551 (output_macinfo_op): With multiple macro sections use
15552 macinfo_label_base to distinguish labels.
15553 (output_macinfo): Likewise. Update macinfo_label_base.
15554 Pass in the line info label.
15555 (note_variable_value_in_expr): When generating LTO resolve
15556 all variable values here by generating DIEs as needed.
15557 (init_sections_and_labels): Add early LTO debug flag parameter
15558 and generate different sections and names if set. Add generation
15559 counter for the labels so we can have multiple of them.
15560 (reset_dies): Helper to allow DIEs to be output multiple times.
15561 (dwarf2out_finish): When outputting DIEs to the fat part of an
15562 LTO object first reset DIEs.
15563 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15564 (modified_type_die): Check for decl_ultimate_origin being self
15566 (gen_type_die_with_usage): Likewise.
15567 (gen_typedef_die): Allow decl_ultimate_origin being self.
15568 (set_decl_abstract_flags): Remove.
15569 (set_block_abstract_flags): Likewise.
15570 (dwarf2out_abstract_function): Treat the early generated DIEs
15571 as the abstract copy and only add DW_AT_inline and
15572 DW_AT_artificial here and call set_decl_origin_self.
15573 If the DIE has an abstract origin don't do anything.
15574 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15575 if we have none yet (Go fails to build one, PR78628).
15576 (variably_modified_type_p): Prevent endless recursion for Ada
15577 cyclic pointer types.
15578 * lto-streamer-in.c: Include debug.h.
15579 (dref_queue): New global.
15580 (lto_read_tree_1): Stream in DIE references.
15581 (lto_input_tree): Register DIE references.
15582 (input_function): Stream DECL_DEBUG_ARGS.
15583 * lto-streamer-out.c: Include debug.h.
15584 (lto_write_tree_1): Output DIE references.
15585 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15586 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15587 (output_function): Stream DECL_DEBUG_ARGS.
15588 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15589 Stream DECL_ABSTRACT_ORIGIN.
15590 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15591 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15592 DECL_CONTEXT for file-scope decls.
15593 * lto-streamer.h (struct dref_entry): Declare.
15594 (dref_queue): Likewise.
15595 * cfgexpand.c (pass_expand::execute): Do not call the
15596 outlining_inline_function hook here.
15597 * lto-wrapper.c (debug_obj): New global.
15598 (tool_cleanup): Unlink it if required.
15599 (debug_objcopy): New function.
15600 (run_gcc): Handle early debug sections in the IL files by
15601 extracting them to separate files, partially linkin them and
15602 feeding the result back as result to the linker.
15603 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15604 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15605 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15606 sections into a separate segment.
15607 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15609 (darwin_asm_dwarf_section): Likewise.
15610 (darwin_asm_output_dwarf_offset): Likewise.
15611 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15613 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15614 Alan Hayward <alan.hayward@arm.com>
15615 David Sherwood <david.sherwood@arm.com>
15617 * read-md.h (md_reader::record_potential_iterator_use): Replace
15618 pointer argument with an rtx and an index.
15619 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15620 (apply_mode_iterator): Likewise.
15621 (apply_code_iterator): Likewise.
15622 (apply_int_iterator): Likewise.
15623 (apply_subst_iterator): Likewise.
15624 (record_iterator_use): Likewise.
15625 (record_attribute_use): Likewise.
15626 (md_reader::record_potential_iterator_use): Likewise. Update calls
15627 to record_iterator_use and apply_iterator.
15628 (iterator_use): Replace ptr with x and index.
15629 (attribute_use): Likewise.
15630 (apply_attribute_uses): Update calls to apply_iterator.
15631 (apply_iterators): Likewise. Update initialization of iterator_use.
15632 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15633 and record_potential_iterator_use.
15634 (rtx_reader::read_rtx_operand): Likewise.
15636 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15637 Alan Hayward <alan.hayward@arm.com>
15638 David Sherwood <david.sherwood@arm.com>
15640 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15643 2017-08-21 Richard Biener <rguenther@suse.de>
15645 PR middle-end/81884
15646 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15647 at struct end conservatively when comparing common bases.
15649 2017-08-21 Richard Biener <rguenther@suse.de>
15651 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15652 (mem_ref_in_stmt): Remove.
15653 (determine_max_movement): Use ref index to get at the reference.
15654 (invariantness_dom_walker::before_dom_children): Deal with
15655 lim data already initialized.
15656 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15658 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15660 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15661 (TARGET_ISA_ROUND): Ditto.
15662 (TARGET_ROUND): Ditto.
15663 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15664 * config/i386/i386.md: Ditto.
15665 * config/i386/sse.md: Ditto.
15666 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15667 with OPTION_MASK_ISA_SSE4_1.
15669 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15672 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15673 __builtin_ia32_lzcnt_u16.
15675 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15678 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15679 (rs6000_set_current_function): Rewrite function to use it.
15681 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15684 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15685 and TYPE_WARN_IF_NOT_ALIGN.
15686 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15687 (handle_warn_if_not_align): New.
15688 (place_union_field): Call handle_warn_if_not_align.
15689 (place_field): Call handle_warn_if_not_align.
15690 Copy TYPE_WARN_IF_NOT_ALIGN.
15691 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15692 (layout_type): Likewise.
15693 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15695 (tree_decl_common): Add warn_if_not_align.
15696 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15697 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15698 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15699 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15700 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15701 * doc/extend.texi: Document warn_if_not_aligned attribute.
15702 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15704 2017-08-17 Martin Liska <mliska@suse.cz>
15707 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15708 (get_data_dependence): Use it as pointer type.
15709 (distribute_loop): Likewise.
15711 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15713 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15714 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15715 (p8_vmrgow_<mode>_direct): New define_insn.
15716 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15717 handle endianness for vmrgew and vmrgow permute patterns.
15719 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15721 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15722 * config/rs6000/rs6000-cpus.def: Remove comment.
15723 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15724 (POWERPC_MASKS): Likewise.
15725 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15726 use of TARGET_VSX_TIMODE.
15727 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15728 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15729 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15730 (rs6000_option_override_internal): Remove dead code.
15731 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15732 (rs6000_legitimize_reload_address): Likewise.
15733 (rs6000_legitimate_address_p): Likewise.
15734 (rs6000_opt_masks): Delete "vsx-timode".
15735 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15736 from function comment.
15737 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15738 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15739 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15741 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15742 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15743 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15744 (VSX_TI): Likewise.
15746 (define_peephole2): Likewise.
15748 2017-08-17 Martin Sebor <msebor@redhat.com>
15751 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15752 past the end of an array.
15753 (test_pp_format): Add test cases.
15755 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15757 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15758 missing ECF_NOTHROW flags.
15760 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15763 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15764 operands residing in integer registers.
15765 (*vsx_le_perm_load_<mode>): Likewise.
15766 (*vsx_le_perm_store_<mode>): Likewise.
15767 (define_peephole2): Add peepholes to optimize the above.
15769 2017-08-17 Marek Polacek <polacek@redhat.com>
15771 PR middle-end/81814
15772 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15773 to mimic what shorten_compare did. Change the return type to bool.
15774 (fold_cond_expr_with_comparison): Update call to
15775 operand_equal_for_comparison_p.
15776 (fold_ternary_loc): Likewise.
15778 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15780 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15782 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15784 2017-08-17 Richard Biener <rguenther@suse.de>
15786 * tree-ssa-structalias.c (solve_graph): When propagating
15787 to successors update the graphs succ edges and avoid duplicate work.
15789 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15792 * config/i386/i386.c (ix86_option_override_internal): Save target
15793 specific options after ix86_stack_protector_guard_reg was changed.
15795 2017-08-17 Richard Biener <rguenther@suse.de>
15797 PR tree-optimization/81827
15798 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15799 (new_var_info): Initialize it conservatively.
15800 (get_call_vi): Mark register vars.
15801 (new_scalar_tmp_constraint_exp): Likewise.
15802 (handle_rhs_call): Likewise.
15803 (handle_const_call): Likewise.
15804 (create_function_info_for): Likewise.
15805 (solve_constraints): Sort varinfos to separate register from
15806 non-register vars to pack points-to solution bitmaps during
15809 2017-08-17 Marek Polacek <polacek@redhat.com>
15811 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15813 2017-08-17 Richard Biener <rguenther@suse.de>
15815 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15816 to true when overflow is undefined and we saturated the result.
15818 2017-08-17 Alan Modra <amodra@gmail.com>
15821 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15822 Don't use store multiple if only one reg needs saving.
15823 (interesting_frame_related_regno): New function.
15824 (rs6000_frame_related): Don't emit frame info for regs that
15826 (rs6000_emit_epilogue): Likewise.
15828 2017-08-16 Nathan Sidwell <nathan@acm.org>
15830 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15831 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15832 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15833 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15834 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15835 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15836 (verify_type): Adjust for TYPE_BINFO move.
15837 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15838 process TYPE_BINFO directly.
15839 (hash_tree): Likewise.
15840 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15842 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15845 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15847 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15849 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15852 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15853 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15854 Add ix86_binary_operator_ok to insn constraint.
15855 (*iordi_1_bts): Ditto.
15856 (*xordi_1_btc): Ditto.
15857 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15858 Update corresponding peephole2 pattern.
15862 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15864 PR tree-optimization/81832
15865 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15866 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15868 2017-08-16 Marek Polacek <polacek@redhat.com>
15871 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15872 regarding size_zero_node.
15874 2017-08-16 Martin Liska <mliska@suse.cz>
15877 * config.gcc: Respect previously set extra_objs in case
15880 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15882 PR tree-optimization/81835
15883 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15884 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15885 not depend on the phi.
15887 2017-08-16 Alan Modra <amodra@gmail.com>
15889 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15892 2017-08-16 Alan Modra <amodra@gmail.com>
15894 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15895 (save_reg_p): ..into this. Update all callers.
15896 (first_reg_to_save): Simplify.
15898 2017-08-16 Alan Modra <amodra@gmail.com>
15900 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15903 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15907 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15908 constant count if that count is less than 32.
15910 2017-08-15 Nathan Sidwell <nathan@acm.org>
15912 * gcc.c (execute): Emit friendlier message if inferior is killed
15913 by an external cause.
15915 2017-08-15 Richard Biener <rguenther@suse.de>
15917 PR tree-optimization/81790
15918 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15919 CONSTRUCTORs from simplifying and VN.
15921 2017-08-14 Martin Sebor <msebor@redhat.com>
15923 * builtin-attrs.def: Add comments.
15925 2017-08-14 Martin Sebor <msebor@redhat.com>
15928 * doc/extend.texi (attribute nonstring): Document new attribute.
15930 2017-08-14 Martin Sebor <msebor@redhat.com>
15933 * tree-diagnostic.c (default_tree_printer): Handle %G.
15934 * gimple-pretty-print.h (percent_G_format): Declare new function.
15935 * gimple-pretty-print.c (percent_G_format): Define.
15936 * tree-pretty-print.c (percent_K_format): Add argument.
15938 2017-08-14 Martin Sebor <msebor@redhat.com>
15940 PR translation/79998
15941 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15942 Remove a stray space.
15944 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15947 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15948 (*iordi_1_bts): Ditto.
15949 (*xordi_1_btc): Ditto.
15951 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15954 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15955 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15957 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15958 quoted strings, and make more translator-friendly.
15959 (darwin_rs6000_override_options): Likewise.
15960 (rs6000_option_override_internal): Likewise.
15961 (rs6000_return_in_memory): Fix overlong line.
15962 (init_cmulative_args): Use quoted strings, and make more
15963 translator-friendly.
15964 (rs6000_pass_by_reference): Fix overlong line.
15965 (def_builtin): Use quoted strings.
15966 (altivec_expand_predicate_builtin): Use quoted strings, and make
15967 more translator-friendly.
15968 (htm_expand_builtin): Use quoted strings.
15969 (cpu_expand_builtin): Use quoted strings, and make more
15970 translator-friendly.
15971 (altivec_expand_builtin): Likewise.
15972 (paired_expand_predicate_builtin): Likewise.
15973 (rs6000_invalid_builtin): Likewise.
15974 (builtin_function_type): Use quoted strings.
15975 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15976 more translator-friendly.
15977 (rs6000_trampoline_init): Likewise.
15978 (rs6000_handle_altivec_attribute): Likewise.
15979 (rs6000_inner_target_options): Use quoted strings.
15980 (rs6000_disable_incompatible_switches): Likewise.
15981 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15982 strings, and make more translator-friendly.
15983 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15985 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15987 PR tree-optimization/81799
15988 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15989 cond_expr to simple gimple operand.
15991 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15993 PR middle-end/46932
15994 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15996 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16000 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16001 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16002 TARGET_GASISR_PROLOGUES.
16003 * config/avr/avr.c (avr_option_override): Same.
16004 (avr_pass_pre_proep::execute): Same.
16006 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16009 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16010 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16012 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16014 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16015 define_insn_and_split. Split to a memory load from 0 in
16016 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16017 using PTR mode iterator.
16018 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16019 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16020 (*add_tp_<mode>): Redefine as define_insn_and_split.
16021 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16022 address space. Merge with *add_tp_x32 using PTR mode iterator.
16023 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16024 Split to an add with a memory load from 0 in
16025 DEFAULT_TLS_SEG_REG address space.
16027 2017-08-12 Andrew Pinski <apinski@cavium.com>
16029 * config/aarch64/aarch64-option-extensions.def (rdma):
16030 Fix feature string to what Linux prints out in /proc/cpuinfo.
16032 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16035 * dwarf2out.c (modified_type_die): For C typedef types that have
16036 an ultimate origin, process the ultimate origin instead of the
16038 (gen_typedef_die): Assert that input DECLs have no ultimate
16040 (gen_type_die_with_usage): For typedef variants that have an
16041 ultimate origin, just call gen_decl_die on the original DECL.
16042 (process_scope_var): Avoid creating DIEs for local typedefs and
16043 concrete static variables.
16045 2017-08-12 Alan Modra <amodra@gmail.com>
16049 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16050 match gnu-user.h startfile.
16051 (ENDFILE_LINUX_SPEC): Similarly.
16053 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16056 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16058 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16060 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16061 * config/aarch64/aarch64.md (mov<mode>): Change.
16062 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16063 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16064 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16066 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16068 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16069 for storage order barriers.
16071 2017-08-11 Martin Liska <mliska@suse.cz>
16074 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16075 variables of void type.
16077 2017-08-11 Martin Liska <mliska@suse.cz>
16079 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16080 TARGET_SUPPORTS_ALIASES.
16081 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16082 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16083 (optimize_weakref): Likewise.
16084 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16085 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16086 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16088 2017-08-11 Martin Liska <mliska@suse.cz>
16091 * config/i386/i386.c (make_resolver_func): Do complete
16092 refactoring of the function.
16094 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16097 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16098 * config/i386/i386.c (ix86_stack_protect_guard): Use
16099 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16100 * doc/invoke.texi (x86 Options): Document
16101 -mstack-protector-guard-symbol= option.
16103 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16105 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16106 * config/i386/i386.c (ix86_split_stack_guard): New function.
16107 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16108 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16109 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16110 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16111 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16112 (split_stack_space_check): Call ix86_split_stack_guard.
16114 2017-08-10 Martin Sebor <msebor@redhat.com>
16116 * print-tree.c (print_node): Print location using the established
16118 Replace spaces with colons.
16119 (debug_raw, debug): Ditto.
16121 2017-08-10 Martin Sebor <msebor@redhat.com>
16124 * pretty-print.c (pp_format): Correct the handling of %s precision.
16126 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16129 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16131 (ix86_finalize_stack_frame_flags): This. Also clear
16132 frame_pointer_needed if -fno-omit-frame-pointer is used without
16134 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16135 with ix86_finalize_stack_frame_flags.
16136 (ix86_expand_epilogue): Likewise.
16137 (ix86_expand_split_stack_prologue): Likewise.
16138 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16140 2017-08-10 Martin Liska <mliska@suse.cz>
16143 * c-attribs.c (handle_target_attribute):
16144 Report warning for an empty string argument of target attribute.
16146 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16149 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16151 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16152 or DECL_NONLOCAL labels.
16153 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16154 or DECL_NONLOCAL labels here.
16156 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16159 to indicate when early gimple folding has been disabled.
16160 (rs6000_gimple_fold_builtin): Add debug content.
16161 (rs6000_invalid_builtin): Fix whitespace.
16162 (rs6000_expand_builtin): Fix whitespace.
16163 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16165 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16168 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16169 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16171 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16173 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16174 (qdf24xx): Likewise.
16175 * config/aarch64/aarch64.md: Include falkor.md.
16176 * config/aarch64/falkor.md: New.
16178 2017-08-09 Marek Polacek <polacek@redhat.com>
16181 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16182 * diagnostic.c (emit_diagnostic): Add a comment.
16183 (emit_diagnostic_valist): New function.
16185 2017-08-09 Marek Polacek <polacek@redhat.com>
16188 * input.c (make_location): New overload.
16189 * input.h (make_location): Declare.
16191 2017-08-08 Alan Modra <amodra@gmail.com>
16192 H.J. Lu <hongjiu.lu@intel.com>
16195 * gcc.c (NO_PIE_SPEC): Delete.
16196 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16198 (LINK_PIE_SPEC): ..to here.
16199 (LINK_COMMAND_SPEC): Support -no-pie.
16200 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16201 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16203 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16204 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16205 (ENDFILE_CRTEND_SPEC): Similarly.
16207 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16210 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16211 (mstack-protector-guard-offset=): Ditto.
16212 * config/i386/i386.c (ix86_option_override): Handle
16213 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16215 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16216 ix86_stack_protect_guard_offset variables.
16217 (TARGET_STACK_PROTECT_GUARD): Always define.
16218 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16219 and -mstack-protector-guard-offset= options.
16221 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16223 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16224 boundary case for the last candidate.
16226 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16228 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16229 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16231 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16233 PR middle-end/19706
16234 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16235 * config/aarch64/aarch64-builtins.c
16236 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16237 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16238 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16240 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16241 Andrew Pinski <pinskia@gmail.com>
16243 PR middle-end/19706
16244 * internal-fn.def (XORSIGN): New.
16245 * optabs.def (xorsign_optab): New.
16246 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16247 (convert_expand_mult_copysign): New.
16248 (pass_optimize_widening_mul::execute): Call
16249 convert_expand_mult_copysign.
16251 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16253 PR tree-optimization/81354
16254 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16255 Insert on edges rather than explicitly creating landing pads.
16256 (analyze_candidates_and_replace): Commit edge inserts.
16258 2017-08-08 Richard Biener <rguenther@suse.de>
16260 PR middle-end/81719
16261 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16262 (expand_simple_operations): Also look through ADDR_EXPRs with
16263 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16265 2017-08-08 Richard Biener <rguenther@suse.de>
16267 PR tree-optimization/81723
16268 * tree-vect-slp.c (struct bst_traits): New hash traits.
16269 (bst_fail): New global.
16270 (vect_build_slp_tree_2): New worker, split out from ...
16271 (vect_build_slp_tree): ... this now wrapping it with using
16272 bst_fail set to cache SLP tree build fails. Properly handle
16274 (vect_analyze_slp_instance): Allocate and free bst_fail.
16276 2017-08-08 Martin Liska <mliska@suse.cz>
16279 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16280 LABEL_DECLs that can be from a different function.
16282 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16284 PR tree-optimization/81744
16285 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16286 loop's number of iterations.
16288 2017-08-08 Martin Liska <mliska@suse.cz>
16290 * asan.c: Include header files.
16291 * attribs.c (build_decl_attribute_variant): New function moved
16293 (build_type_attribute_qual_variant): Likewise.
16294 (cmp_attrib_identifiers): Likewise.
16295 (simple_cst_list_equal): Likewise.
16296 (omp_declare_simd_clauses_equal): Likewise.
16297 (attribute_value_equal): Likewise.
16298 (comp_type_attributes): Likewise.
16299 (build_type_attribute_variant): Likewise.
16300 (lookup_ident_attribute): Likewise.
16301 (remove_attribute): Likewise.
16302 (merge_attributes): Likewise.
16303 (merge_type_attributes): Likewise.
16304 (merge_decl_attributes): Likewise.
16305 (merge_dllimport_decl_attributes): Likewise.
16306 (handle_dll_attribute): Likewise.
16307 (attribute_list_equal): Likewise.
16308 (attribute_list_contained): Likewise.
16309 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16310 (lookup_attribute_by_prefix): Likewise.
16311 * bb-reorder.c: Include header files.
16312 * builtins.c: Likewise.
16313 * calls.c: Likewise.
16314 * cfgexpand.c: Likewise.
16315 * cgraph.c: Likewise.
16316 * cgraphunit.c: Likewise.
16317 * convert.c: Likewise.
16318 * dwarf2out.c: Likewise.
16319 * final.c: Likewise.
16320 * fold-const.c: Likewise.
16321 * function.c: Likewise.
16322 * gimple-expr.c: Likewise.
16323 * gimple-fold.c: Likewise.
16324 * gimple-pretty-print.c: Likewise.
16325 * gimple.c: Likewise.
16326 * gimplify.c: Likewise.
16327 * hsa-common.c: Likewise.
16328 * hsa-gen.c: Likewise.
16329 * internal-fn.c: Likewise.
16330 * ipa-chkp.c: Likewise.
16331 * ipa-cp.c: Likewise.
16332 * ipa-devirt.c: Likewise.
16333 * ipa-fnsummary.c: Likewise.
16334 * ipa-inline.c: Likewise.
16335 * ipa-visibility.c: Likewise.
16337 * lto-cgraph.c: Likewise.
16338 * omp-expand.c: Likewise.
16339 * omp-general.c: Likewise.
16340 * omp-low.c: Likewise.
16341 * omp-offload.c: Likewise.
16342 * omp-simd-clone.c: Likewise.
16343 * opts-global.c: Likewise.
16344 * passes.c: Likewise.
16345 * predict.c: Likewise.
16346 * sancov.c: Likewise.
16347 * sanopt.c: Likewise.
16348 * symtab.c: Likewise.
16349 * toplev.c: Likewise.
16350 * trans-mem.c: Likewise.
16351 * tree-chkp.c: Likewise.
16352 * tree-eh.c: Likewise.
16353 * tree-into-ssa.c: Likewise.
16354 * tree-object-size.c: Likewise.
16355 * tree-parloops.c: Likewise.
16356 * tree-profile.c: Likewise.
16357 * tree-ssa-ccp.c: Likewise.
16358 * tree-ssa-live.c: Likewise.
16359 * tree-ssa-loop.c: Likewise.
16360 * tree-ssa-sccvn.c: Likewise.
16361 * tree-ssa-structalias.c: Likewise.
16362 * tree-ssa.c: Likewise.
16363 * tree-streamer-in.c: Likewise.
16364 * tree-vectorizer.c: Likewise.
16365 * tree-vrp.c: Likewise.
16366 * tsan.c: Likewise.
16367 * ubsan.c: Likewise.
16368 * varasm.c: Likewise.
16369 * varpool.c: Likewise.
16370 * tree.c: Remove functions moved to attribs.[ch].
16371 * tree.h: Likewise.
16372 * config/aarch64/aarch64.c: Add attrs.h header file.
16373 * config/alpha/alpha.c: Likewise.
16374 * config/arc/arc.c: Likewise.
16375 * config/arm/arm.c: Likewise.
16376 * config/avr/avr.c: Likewise.
16377 * config/bfin/bfin.c: Likewise.
16378 * config/c6x/c6x.c: Likewise.
16379 * config/cr16/cr16.c: Likewise.
16380 * config/cris/cris.c: Likewise.
16381 * config/darwin.c: Likewise.
16382 * config/epiphany/epiphany.c: Likewise.
16383 * config/fr30/fr30.c: Likewise.
16384 * config/frv/frv.c: Likewise.
16385 * config/ft32/ft32.c: Likewise.
16386 * config/h8300/h8300.c: Likewise.
16387 * config/i386/winnt.c: Likewise.
16388 * config/ia64/ia64.c: Likewise.
16389 * config/iq2000/iq2000.c: Likewise.
16390 * config/lm32/lm32.c: Likewise.
16391 * config/m32c/m32c.c: Likewise.
16392 * config/m32r/m32r.c: Likewise.
16393 * config/m68k/m68k.c: Likewise.
16394 * config/mcore/mcore.c: Likewise.
16395 * config/microblaze/microblaze.c: Likewise.
16396 * config/mips/mips.c: Likewise.
16397 * config/mmix/mmix.c: Likewise.
16398 * config/mn10300/mn10300.c: Likewise.
16399 * config/moxie/moxie.c: Likewise.
16400 * config/msp430/msp430.c: Likewise.
16401 * config/nds32/nds32-isr.c: Likewise.
16402 * config/nds32/nds32.c: Likewise.
16403 * config/nios2/nios2.c: Likewise.
16404 * config/nvptx/nvptx.c: Likewise.
16405 * config/pa/pa.c: Likewise.
16406 * config/pdp11/pdp11.c: Likewise.
16407 * config/powerpcspe/powerpcspe.c: Likewise.
16408 * config/riscv/riscv.c: Likewise.
16409 * config/rl78/rl78.c: Likewise.
16410 * config/rx/rx.c: Likewise.
16411 * config/s390/s390.c: Likewise.
16412 * config/sh/sh.c: Likewise.
16413 * config/sol2.c: Likewise.
16414 * config/sparc/sparc.c: Likewise.
16415 * config/spu/spu.c: Likewise.
16416 * config/stormy16/stormy16.c: Likewise.
16417 * config/tilegx/tilegx.c: Likewise.
16418 * config/tilepro/tilepro.c: Likewise.
16419 * config/v850/v850.c: Likewise.
16420 * config/vax/vax.c: Likewise.
16421 * config/visium/visium.c: Likewise.
16422 * config/xtensa/xtensa.c: Likewise.
16424 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16427 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16428 constraints since the -mupper-regs-* switches have been
16430 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16431 into a vector from a double word element that was extracted from
16432 another vector, and eliminate extra XXPERMDI instructions.
16433 (vsx_concat_<mode>_2): Likewise.
16434 (vsx_concat_<mode>_3): Likewise.
16435 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16436 concat to allow optimizing inserts from previous extracts.
16438 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16440 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16441 memory reference to a SSP offset in TLS address space.
16442 (ix86_print_operand) <case '@'>: Remove.
16443 (ix86_print_operand_punct_valid_p): Remove '@' code.
16444 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16445 UNSPEC_SP_TLS_TEST.
16446 (stack_tls_protect_set_<mode>): Remove.
16447 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16448 (stack_tls_protect_test_<mode>): Remove.
16449 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16451 2017-08-07 Olivier Hainque <hainque@adacore.com>
16454 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16456 2017-08-07 Douglas Rupp <rupp@adacore.com>
16458 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16459 variable was referenced as multidir in command.
16461 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16464 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16467 2017-08-07 Martin Liska <mliska@suse.cz>
16469 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16470 * config/rl78/rl78.c: Add include of attribs.h.
16471 * config/sh/sh.c: Likewise.
16472 * config/v850/v850.c: Likewise.
16474 2017-08-07 Tom de Vries <tom@codesourcery.com>
16476 PR middle-end/78266
16477 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16479 2017-08-07 Martin Liska <mliska@suse.cz>
16481 * config/mips/mips.c: Include attribs.h.
16483 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16486 * doc/invoke.texi: Document change in behvaior for -Ofast for
16489 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16491 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16493 (aarch64_pop_regs): Likewise.
16494 (aarch64_gen_load_pair): Likewise.
16495 (aarch64_save_callee_saves): Likewise.
16496 (aarch64_restore_callee_saves): Likewise.
16498 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16500 * config/i386/i386.c: Revert the last change.
16502 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16505 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16507 (ix86_finalize_stack_frame_flags): This. Also clear
16508 frame_pointer_needed if -fno-omit-frame-pointer is used without
16510 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16511 with ix86_finalize_stack_frame_flags.
16512 (ix86_expand_epilogue): Likewise.
16513 (ix86_expand_split_stack_prologue): Likewise.
16515 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16518 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16519 to P_AES for Westmere.
16521 2017-08-07 Jonathan Yong <10walls@gmail.com>
16523 * config/i386/mingw.opt (fset-stack-executable): Removed.
16524 * config/i386/cygming.opt (fset-stack-executable): Moved
16526 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16528 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16530 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16532 2017-08-07 Marek Polacek <polacek@redhat.com>
16534 PR middle-end/81737
16535 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16537 2017-08-07 Martin Liska <mliska@suse.cz>
16539 * attribs.h (canonicalize_attr_name): New function.
16540 (cmp_attribs): Move from c-format.c and adjusted.
16541 (is_attribute_p): Moved from tree.h.
16542 * tree-inline.c: Add new includes.
16543 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16544 (private_is_attribute_p): Remove.
16545 (private_lookup_attribute): Likewise.
16546 (private_lookup_attribute_by_prefix): Simplify.
16547 (remove_attribute): Use is_attribute_p.
16548 * tree.h: Remove removed declarations.
16550 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16552 PR middle-end/81698
16553 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16554 instead of computing it in the function. Formatting fix.
16555 (expand_case): Don't rely on default_edge being the first edge,
16556 clear it if removing it, pass default_edge to
16557 emit_case_dispatch_table.
16558 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16561 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16563 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16564 insn in the function, emit NOP after the insn.
16566 2017-08-06 Tom de Vries <tom@codesourcery.com>
16568 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16571 2017-08-06 Tom de Vries <tom@codesourcery.com>
16573 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16576 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16578 PR tree-optimization/57371
16579 * match.pd: New pattern.
16581 2017-08-04 Marek Polacek <polacek@redhat.com>
16583 PR middle-end/81695
16584 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16585 perform the computation in offset_int.
16587 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16589 PR tree-optimization/81136
16590 * tree-vectorizer.h: Include tree-hash-traits.h.
16591 (vec_base_alignments): New typedef.
16592 (vec_info): Add a base_alignments field.
16593 (vect_record_base_alignments): Declare.
16594 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16596 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16597 (create_data_ref): Add an is_conditional_in_stmt argument.
16598 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16599 the is_conditional_in_stmt field.
16600 (data_ref_loc): Add an is_conditional_in_stmt field.
16601 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16602 (find_data_references_in_stmt): Update call to create_data_ref.
16603 (graphite_find_data_references_in_stmt): Likewise.
16604 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16605 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16606 (vect_record_base_alignment): New function.
16607 (vect_record_base_alignments): Likewise.
16608 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16609 for nested statements even if we fail to compute a misalignment.
16610 Use pooled base alignments for unconditional references.
16611 (vect_find_same_alignment_drs): Compare base addresses instead
16613 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16614 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16616 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16618 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16619 Add an explicit name for the enum. Use auto_vec for slp_instances
16620 and grouped_stores.
16621 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16623 (_bb_vec_info): Add a constructor and destructor.
16624 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16625 (destroy_loop_vec_info): Delete.
16626 (vect_destroy_datarefs): Likewise.
16627 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16628 (vec_info::vec_info): New function.
16629 (vec_info::~vec_info): Likewise.
16630 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16631 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16632 destroy_loop_vec_info.
16633 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16634 (_loop_vec_info::_loop_vec_info): ...this.
16635 (destroy_loop_vec_info): Replace with...
16636 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16637 the stmt_vec_infos. Leave handling of vec_info information to its
16638 destructor. Remove explicit vector releases.
16639 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16640 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16641 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16642 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16643 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16644 (destroy_bb_vec_info): Replace with...
16645 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16646 information to its destructor.
16647 (vect_slp_analyze_bb_1): Use new and delete instead of
16648 new_bb_vec_info and destroy_bb_vec_info.
16649 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16652 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16654 * tree-data-ref.h (subscript): Add access_fn field.
16655 (data_dependence_relation): Add could_be_independent_p.
16656 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16657 (same_access_functions): Move to tree-data-ref.c.
16658 * tree-data-ref.c (ref_contains_union_access_p): New function.
16659 (access_fn_component_p): Likewise.
16660 (access_fn_components_comparable_p): Likewise.
16661 (dr_analyze_indices): Add a reference to access_fn_component_p.
16662 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16664 (constant_access_functions): Likewise.
16665 (add_other_self_distances): Likewise.
16666 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16667 (initialize_data_dependence_relation): Use XCNEW and remove
16668 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16669 of access functions that have the same type. Allow the
16670 subsequence to end with different bases in some circumstances.
16671 Record the chosen access functions in SUB_ACCESS_FN.
16672 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16673 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16674 (subscript_dependence_tester_1): Likewise dra and drb.
16675 (build_classic_dist_vector): Update calls accordingly.
16676 (subscript_dependence_tester): Likewise.
16677 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16678 DDR_COULD_BE_INDEPENDENT_P.
16679 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16680 comp_alias_ddrs instead of may_alias_ddrs.
16681 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16683 (vect_analyze_data_ref_dependence): Use it if
16684 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16685 distance vectors if that fails.
16686 (dependence_distance_ge_vf): New function.
16687 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16688 LOOP_VINFO_MAY_ALIAS_DDRS.
16690 2017-08-04 Richard Biener <rguenther@suse.de>
16692 PR middle-end/81705
16693 * fold-const.c (fold_binary_loc): Properly restrict
16694 minus_var0 && minus_var1 case when associating undefined overflow
16697 2017-08-04 Tom de Vries <tom@codesourcery.com>
16699 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16701 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16703 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16704 Don't start diagnostic messages with a capital letter.
16705 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16707 (rs6000_invalid_builtin): Likewise.
16708 (rs6000_trampoline_init): Likewise.
16710 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16713 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16714 after setting changeable df flags.
16716 2017-08-03 Richard Biener <rguenther@suse.de>
16718 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16719 up if the use is in USE - X.
16721 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16723 * toplev.c (dumpfile.h): New include.
16724 (internal_error_reentered): New static function. Use it...
16725 (internal_error_function): ...here to handle reentered internal_error.
16727 2017-08-03 Richard Biener <rguenther@suse.de>
16729 PR middle-end/81148
16730 * fold-const.c (split_tree): Add minus_var and minus_con
16731 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16732 here but always use minus_*.
16733 (associate_trees): Assert we never associate with MINUS_EXPR
16734 and NULL first operand. Do not recurse for PLUS_EXPR operands
16735 when associating as MINUS_EXPR either.
16736 (fold_binary_loc): Track minus_var and minus_con.
16738 2017-08-03 Tom de Vries <tom@codesourcery.com>
16741 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16742 ACCEL_COMPILER, apply finish_options on
16743 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16745 2017-08-03 Tom de Vries <tom@codesourcery.com>
16748 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16749 function_entry_patch_area_size > 0.
16751 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16754 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16755 instead of 10??LU, perform unit multiplication in wide_int,
16756 don't change alloc_object_size_limit if the limit is larger
16759 PR tree-optimization/81655
16760 PR tree-optimization/81588
16761 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16762 the case when ranges[i].low and high are 1 for unsigned type with
16765 PR middle-end/81052
16766 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16767 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16769 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16771 * tree-vrp.h: Add include guard.
16773 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16776 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16777 (ud2): New insn pattern.
16778 * config/i386/i386.c (ix86_expand_epilogue):
16779 For naked functions, generate ud2 instead of trap insn.
16781 2017-08-02 Marek Polacek <polacek@redhat.com>
16784 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16786 2017-08-02 Tom de Vries <tom@codesourcery.com>
16787 Cesar Philippidis <cesar@codesourcery.com>
16789 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16790 Add missing edge probabilities.
16792 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16794 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16795 Correct endianness.
16797 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16799 PR middle-end/79499
16800 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16801 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16802 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16803 prologue_seq sequences - if any.
16805 2017-08-02 Richard Biener <rguenther@suse.de>
16807 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16808 via vectors if supported, integer extracts via punning if supported
16809 or otherwise vector extracts.
16811 2017-08-02 Richard Biener <rguenther@suse.de>
16813 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16815 (bitmap_insert_into_set): ... this.
16817 2017-08-02 Richard Biener <rguenther@suse.de>
16819 PR tree-optimization/81633
16821 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16823 PR tree-optimization/71752
16824 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16826 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16828 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16829 (machine_function::call_ms2sysv_pad_out): Remove field.
16830 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16831 (ix86_compute_frame_layout): Likewise.
16833 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16836 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16837 attribute with interrupt attribute.
16839 2017-08-01 Andrew Pinski <apinski@cavium.com>
16841 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16842 BIT_INSERT_EXPR's operand 1
16843 to see if the types precision matches.
16845 2017-08-01 Martin Liska <mliska@suse.cz>
16847 PR middle-end/70140
16848 * builtins.c (expand_builtin_memcpy_args): Remove.
16849 (expand_builtin_memcpy): Call newly added function
16850 expand_builtin_memory_copy_args.
16851 (expand_builtin_memcpy_with_bounds): Likewise.
16852 (expand_builtin_mempcpy): Remove last argument.
16853 (expand_builtin_mempcpy_with_bounds): Likewise.
16854 (expand_builtin_memory_copy_args): New function created from
16855 expand_builtin_mempcpy_args with small modifications.
16856 (expand_builtin_mempcpy_args): Remove.
16857 (expand_builtin_stpcpy): Remove unused argument.
16858 (expand_builtin): Likewise.
16859 (expand_builtin_with_bounds): Likewise.
16861 2017-08-01 Martin Liska <mliska@suse.cz>
16864 Make mempcpy more optimal (PR middle-end/70140).
16866 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16869 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16870 __builtin_vec_cmpne verify both arguments are compatible vectors
16871 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16872 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16873 move computation of aligned to after checking the argument types.
16877 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16878 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16881 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16882 Doug Rupp <rupp@adacore.com>
16883 Olivier Hainque <hainque@adacore.com>
16885 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16886 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16888 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16889 for TARGET_OS_CPP_BUILTIN.
16890 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16891 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16893 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16894 passing required abi options to the assembler for EABI configurations.
16895 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16896 of .text.hot and .text.unlikely sections for kernel modules when
16897 using ARM style exceptions.
16898 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16899 options. Add EXTRA_CC1_SPEC.
16900 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16902 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16904 (ARM_TARGET2_DWARF_FORMAT): Define.
16905 * config/arm/t-vxworks: Adjust multilib control to removal of the
16906 Diab command line options.
16908 2017-08-01 Martin Liska <mliska@suse.cz>
16910 PR gcov-profile/81561
16911 * gcov.c (unblock): Make unblocking safe as we need to preserve
16912 index correspondence of blocks and block_lists.
16914 2017-08-01 Richard Biener <rguenther@suse.de>
16916 PR tree-optimization/81181
16917 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16918 (compute_antic): ... end of iteration here.
16920 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16922 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16923 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16924 (ftree-slp-vectorize): Likewise.
16925 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16926 can no longer be set independent of flag_tree_loop_vectorize.
16927 * omp-general.c (emp_max_vf): Likewise.
16928 * opts.c (enable_fdo_optimizations): Remove references to
16929 flag_tree_vectorize, these are now implicit.
16930 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16931 and leave it for the options machinery.
16933 2017-08-01 Martin Liska <mliska@suse.cz>
16935 PR middle-end/70140
16936 * builtins.c (expand_builtin_memcpy_args): Remove.
16937 (expand_builtin_memcpy): Call newly added function
16938 expand_builtin_memory_copy_args.
16939 (expand_builtin_memcpy_with_bounds): Likewise.
16940 (expand_builtin_mempcpy): Remove last argument.
16941 (expand_builtin_mempcpy_with_bounds): Likewise.
16942 (expand_builtin_memory_copy_args): New function created from
16943 expand_builtin_mempcpy_args with small modifications.
16944 (expand_builtin_mempcpy_args): Remove.
16945 (expand_builtin_stpcpy): Remove unused argument.
16946 (expand_builtin): Likewise.
16947 (expand_builtin_with_bounds): Likewise.
16949 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16952 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16953 print "ds:" only for immediates in generic address space.
16955 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16958 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16959 (ix86_function_ok_for_sibcall): Return false for naked functions.
16961 2017-08-01 Richard Biener <rguenther@suse.de>
16963 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16964 (compute_antic): Seed worklist with exit block predecessors.
16965 * cfganal.c (dfs_find_deadend): For a cycle return the source
16966 of the edge closing it.
16968 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16970 * config/aarch64/aarch64.c
16971 (aarch64_can_const_movi_rtx_p): Move 0 check.
16973 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16975 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16976 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16978 * match.pd: Ditto in address comparison pattern.
16980 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16982 PR tree-optimization/81627
16983 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16984 closed ssa form for store-store chain.
16986 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16988 PR tree-optimization/81620
16989 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16990 for store-store chain.
16992 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16994 PR tree-optimization/81588
16995 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16996 ranges[i].in_p, invert comparison code ccode. For >/>=,
16997 swap rhs1 and rhs2 and comparison code unconditionally,
16998 for </<= don't do that. Don't swap rhs1/rhs2 again if
16999 ranges[i].in_p, instead invert comparison code ccode if
17000 opcode or oe->rank is BIT_IOR_EXPR.
17003 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17004 a direct optab to conversion optab.
17005 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17006 with GET_MODE_INNER as last argument instead of optab_handler.
17007 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17008 vector extraction if possible and optab is available.
17009 * expr.c (store_constructor): Use convert_optab_handler instead
17010 of optab_handler. Use vector initialization from smaller
17011 vectors if possible and optab is available.
17012 * tree-vect-stmts.c (vectorizable_load): Likewise.
17013 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17014 now have two modes.
17015 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17016 of vec_init from half-sized vectors with the same element mode.
17017 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17018 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17019 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17020 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17021 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17022 after mode in gen_vec_extract* calls.
17023 (vec_extract<mode>): Renamed to ...
17024 (vec_extract<mode><ssescalarmodelower>): ... this.
17025 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17026 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17027 element mode after mode in gen_vec_init* calls.
17028 (VEC_INIT_HALF_MODE): New mode iterator.
17029 (vec_init<mode>): Renamed to ...
17030 (vec_init<mode><ssescalarmodelower>): ... this.
17031 (vec_init<mode><ssehalfvecmodelower>): New expander.
17032 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17033 (vec_extractv2sfsf): ... this.
17034 (vec_initv2sf): Renamed to ...
17035 (vec_initv2sfsf): ... this.
17036 (vec_extractv2si): Renamed to ...
17037 (vec_extractv2sisi): ... this.
17038 (vec_initv2si): Renamed to ...
17039 (vec_initv2sisi): ... this.
17040 (vec_extractv4hi): Renamed to ...
17041 (vec_extractv4hihi): ... this.
17042 (vec_initv4hi): Renamed to ...
17043 (vec_initv4hihi): ... this.
17044 (vec_extractv8qi): Renamed to ...
17045 (vec_extractv8qiqi): ... this.
17046 (vec_initv8qi): Renamed to ...
17047 (vec_initv8qiqi): ... this.
17048 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17049 (vec_init<mode>): Renamed to ...
17050 (vec_init<mode><VEC_base_l>): ... this.
17051 (vec_extract<mode>): Renamed to ...
17052 (vec_extract<mode><VEC_base_l>): ... this.
17053 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17054 (vec_initv2sfsf): ... this.
17055 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17056 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17057 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17058 element mode after mode in gen_vec_init* calls.
17059 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17060 (vec_init<mode><Vel>): ... this.
17061 (vec_extract<mode>): Renamed to ...
17062 (vec_extract<mode><Vel>): ... this.
17063 * config/aarch64/iterators.md (Vel): New mode attribute.
17064 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17065 Add element mode after mode in gen_vec_extract* calls.
17066 * config/s390/vector.md (non_vec_l): New mode attribute.
17067 (vec_extract<mode>): Renamed to ...
17068 (vec_extract<mode><non_vec_l>): ... this.
17069 (vec_init<mode>): Renamed to ...
17070 (vec_init<mode><non_vec_l>): ... this.
17071 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17072 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17074 * config/arm/iterators.md (V_elem_l): New mode attribute.
17075 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17076 (vec_extract<mode><V_elem_l>): ... this.
17077 (vec_extractv2di): Renamed to ...
17078 (vec_extractv2didi): ... this.
17079 (vec_init<mode>): Renamed to ...
17080 (vec_init<mode><V_elem_l>): ... this.
17081 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17082 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17083 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17084 Add element mode after gen_vec_extract* calls.
17085 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17086 (vec_init<mode><unitmode>): ... this.
17087 (vec_extract<mode>): Renamed to ...
17088 (vec_extract<mode><unitmode>): ... this.
17089 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17090 (vec_init<mode><unitmode>): ... this.
17091 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17092 (vec_initv2sfsf): ... this.
17093 (vec_extractv2sf): Renamed to ...
17094 (vec_extractv2sfsf): ... this.
17095 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17096 Add element mode after gen_vec_extract* calls.
17097 * config/mips/mips.md (unitmode): New mode iterator.
17098 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17099 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17100 * config/spu/spu.md (inner_l): New mode attribute.
17101 (vec_init<mode>): Renamed to ...
17102 (vec_init<mode><inner_l>): ... this.
17103 (vec_extract<mode>): Renamed to ...
17104 (vec_extract<mode><inner_l>): ... this.
17105 * config/sparc/sparc.md (veltmode): New mode iterator.
17106 (vec_init<VMALL:mode>): Renamed to ...
17107 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17108 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17109 (vec_initv2sisi): ... this.
17110 (vec_initv2sf): Renamed to ...
17111 (vec_initv2sfsf): ... this.
17112 (vec_extractv2sf): Renamed to ...
17113 (vec_extractv2sfsf): ... this.
17114 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17115 (vec_init<mode>): Renamed to ...
17116 (vec_init<mode><VEC_base_l>): ... this.
17117 (vec_extract<mode>): Renamed to ...
17118 (vec_extract<mode><VEC_base_l>): ... this.
17119 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17120 (vec_initv2sfsf): ... this.
17121 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17122 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17123 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17124 gen_vec_init* calls.
17126 2017-08-01 Richard Biener <rguenther@suse.de>
17128 PR tree-optimization/81297
17129 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17130 TREE_OVERFLOW from INTEGER_CSTs.
17132 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17134 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17136 2017-07-31 Carl Love <cel@us.ibm.com>
17138 * config/rs6000/rs6000-c: Add support for built-in functions
17139 vector signed char vec_xl_be (signed long long, signed char *);
17140 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17141 vector signed int vec_xl_be (signed long long, signed int *);
17142 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17143 vector signed long long vec_xl_be (signed long long, signed long long *);
17144 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17145 vector signed short vec_xl_be (signed long long, signed short *);
17146 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17147 vector double vec_xl_be (signed long long, double *);
17148 vector float vec_xl_be (signed long long, float *);
17149 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17150 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17151 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17153 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17154 (altivec_expand_builtin): Add switch statement to call
17155 altivec_expand_xl_be for each builtin.
17156 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17157 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17158 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17159 __builtin_vsx_le_be_v16qi.
17160 * doc/extend.texi: Update the built-in documentation file for the
17161 new built-in functions.
17163 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17166 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17168 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17170 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17172 * config.gcc: Add z14.
17173 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17174 CPU model numbers for z13s and z14.
17175 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17177 * config/s390/s390-opts.h (enum processor_type): Rename
17178 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17179 * config/s390/s390.c (processor_table): Add field for CPU name to
17180 be passed to Binutils.
17181 (s390_asm_output_machine_for_arch): Use the new field in
17182 processor_table for Binutils.
17183 (s390_expand_builtin): Replace arch12 with z14.
17184 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17185 (s390_get_sched_attrmask): Likewise.
17186 (s390_get_unit_mask): Likewise.
17187 * config/s390/s390.opt: Add z14 to processor_type enum.
17189 2017-07-31 Martin Jambor <mjambor@suse.cz>
17192 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17193 regardless of optimization level.
17195 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17196 Martin Liska <mliska@suse.cz>
17198 * predict.def: Remove old comment and adjust probability.
17199 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17200 PREDICT statements.
17202 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17205 * config/i386/i386.c (ix86_function_naked): New function.
17206 (ix86_can_use_return_insn_p): Return false for naked functions.
17207 (ix86_expand_prologue): Skip prologue for naked functions.
17208 (ix86_expand_epilogue): Skip epilogue for naked functions
17209 and emit trap instruction.
17210 (ix86_warn_func_return): New function.
17211 (ix86_attribute_table): Add "naked" attribute specification.
17212 (TARGET_WARN_FUNC_RETURN): Define.
17213 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17215 2017-07-31 Martin Liska <mliska@suse.cz>
17217 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17218 (dump_gimple_bb_header): Always dump BB info.
17219 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17221 2017-07-31 Martin Liska <mliska@suse.cz>
17224 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17225 also with current_function_decl non-null equality.
17227 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17230 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17231 change type to the element type, instead add eltype variable and
17232 use it where we are interested in the element type.
17234 PR tree-optimization/81603
17235 * ipa-polymorphic-call.c
17236 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17237 offset arithmetic in offset_int, bail out if the resulting bit offset
17238 doesn't fit into shwi.
17240 2017-07-31 Martin Liska <mliska@suse.cz>
17242 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17243 (gimplify_save_expr): Fix comment.
17245 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17248 * config/i386/i386.c (ix86_function_arg): Update arguments for
17250 (ix86_compute_frame_layout): Set the initial stack offset to
17251 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17252 INCOMING_FRAME_SP_OFFSET.
17253 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17254 stack before exception handler returns.
17255 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17256 the 'ERROR_CODE' for exception handler.
17258 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17260 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17261 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17262 (ASM_OUTPUT_REG_POP): Ditto.
17263 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17264 instead of asm_fprintf to output pure string.
17266 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17268 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17269 to imported_module_or_decl hook.
17270 (debug_nothing_tree_tree_tree_bool): Remove.
17271 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17272 * debug.c (do_nothing_debug_hooks): Use
17273 debug_nothing_tree_tree_tree_bool_bool instead of
17274 debug_nothing_tree_tree_tree_bool.
17275 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17276 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17277 * sdbout.c (sdb_debug_hooks): Likewise.
17278 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17279 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17281 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17282 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17283 attribute, don't add anything.
17285 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17287 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17288 (fold_build2_stat_loc): Likewise.
17289 (fold_build3_stat_loc): Likewise.
17290 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17291 (fold_build1_loc): Remove macro.
17292 (fold_build2_loc): Likewise.
17293 (fold_build3_loc): Likewise.
17295 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17297 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17298 (gimple_build_debug_bind_source_stat): Likewise.
17299 * gimple.h (gimple_build_debug_bind): Remove macro.
17300 (gimple_build_debug_bind_source): Likewise.
17302 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17304 * bitmap.c (bitmap_alloc): Adjust.
17305 (bitmap_gc_alloc): Likewise.
17306 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17310 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17311 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17312 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17313 (bitmap_gc_alloc_stat): Likewise.
17314 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17316 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17318 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17319 * rtl.h (shallow_copy_rtx): Remove macro.
17321 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17323 * emit-rtl.c (gen_raw_REG): Adjust.
17324 * gengenrtl.c (gendef): Likewise.
17325 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17326 * rtl.h (rtx_alloc): Remove macro.
17328 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17330 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17331 (build_tree_list_stat): Likewise.
17332 * tree.h (build_tree_list): Remove macro.
17333 (build_tree_list_vec): Likewise.
17335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17337 * tree.c (make_vector_stat): Remove _stat from name.
17338 (build_vector_stat): Likewise.
17339 * tree.h (make_vector_stat): Remove macro.
17340 (build_vector_stat): Likewise.
17342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17344 * tree.h (build_var_debug_value): Remove prototype.
17346 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17348 * tree.c (tree_cons_stat): Remove _stat from name.
17349 * tree.h (tree_cons): Remove macro.
17351 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17353 * tree.c (build_vl_exp_stat): Remove _stat from name.
17354 * tree.h (build_vl_exp): Remove macro.
17356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17358 * tree.c (build_decl_stat): Remove _stat from name.
17359 * tree.h (build_decl): Remove macro.
17361 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17363 * gimple.c (gimple_build_with_ops_stat): Adjust.
17364 (gimple_alloc_stat): Remove _stat from name.
17365 * gimple.h (gimple_alloc): Remove macro.
17367 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17369 * tree.c (make_tree_vec_stat): Remove _stat from name.
17370 (grow_tree_vec_stat): Likewise.
17371 * tree.h (make_tree_vec_stat): Adjust prototype.
17372 (grow_tree_vec_stat): Likewise.
17373 (make_tree_vec): Remove macro.
17374 (grow_tree_vec): Likewise.
17376 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17378 * fold-const.c (fold_build1_stat_loc): Adjust.
17379 (fold_build2_stat_loc): Likewise.
17380 (fold_build3_stat_loc): Likewise.
17381 * tree.c (build0_stat): Remove _stat from name.
17382 (build1_stat): Likewise.
17383 (build2_stat): Likewise.
17384 (build3_stat): Likewise.
17385 (build4_stat): Likewise.
17386 (build5_stat): Likewise.
17387 * tree.h (build1_loc): Remove macro, and rename _stat function
17389 (build2_loc): Likewise.
17390 (build3_loc): Likewise.
17391 (build4_loc): Likewise.
17392 (build5_loc): Likewise.
17394 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17396 * tree.c (make_int_cst_stat): Remove _stat from name.
17397 * tree.h (make_int_cst_stat): Adjust prototype.
17398 (make_int_cst): Remove macro.
17400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17402 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17403 * tree.h (make_tree_binfo_stat): Adjust prototype.
17404 (make_tree_binfo): Remove.
17406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17408 * tree.c (copy_node_stat): Rename to copy_node.
17409 (build_distinct_type_copy): Adjust.
17410 * tree.h (copy_node_stat): Adjust prototype.
17411 (copy_node): Remove macro.
17413 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17415 * tree.c (make_node_stat): rename to make_node.
17416 (build_tree_list_stat): Adjust.
17417 (build0_stat): Likewise.
17418 (build2_stat): Likewise.
17419 (build3_stat): Likewise.
17420 (build4_stat): Likewise.
17421 (build5_stat): Likewise.
17422 (build_decl_stat): Likewise.
17423 * tree.h (make_node_stat): Adjust prototype.
17424 (make_node): remove macro.
17426 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17428 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17429 (PPC_FEATURE2_SCV): Likewise.
17430 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17432 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17434 * config/aarch64/aarch64.c
17435 (aarch64_internal_mov_immediate): Add new special pattern.
17436 * config/aarch64/aarch64.md (*movdi_aarch64):
17437 Add reg/32bit const mov case.
17439 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17440 Richard Sandiford <richard.sandiford@linaro.org>
17442 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17443 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17444 Add integer and movi cases.
17445 (movi-split-hf-df-sf split, fp16): New.
17446 (enabled): Added TARGET_FP_F16INST.
17447 * config/aarch64/iterators.md (GPF_HF): New.
17448 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17450 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17452 * config/aarch64/aarch64.c
17453 (aarch64_simd_container_mode): Add prototype.
17454 (aarch64_expand_mov_immediate): Add HI support.
17455 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17456 (aarch64_can_const_movi_rtx_p): New.
17457 (aarch64_preferred_reload_class):
17458 Remove restrictions of using FP registers for certain SIMD operations.
17459 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17460 (aarch64_valid_floating_const): Add integer move validation.
17461 (aarch64_simd_imm_scalar_p): Remove.
17462 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17463 (aarch64_legitimate_constant_p): Expand list of supported cases.
17464 * config/aarch64/aarch64-protos.h
17465 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17466 (aarch64_reinterpret_float_as_int): New.
17467 (aarch64_simd_imm_scalar_p): Remove.
17468 * config/aarch64/constraints.md (Uvi): New.
17469 (Dd): Split into Ds and new Dd.
17470 * config/aarch64/aarch64.md (*movsi_aarch64):
17472 (*movdi_aarch64): Add SIMD mov case.
17474 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17476 * tree-predcom.c: (struct chain): Handle store-store chain in which
17477 stores for elimination only store loop invariant values.
17478 (execute_pred_commoning_chain): Ditto.
17479 (prepare_initializers_chain_store_elim): Ditto.
17480 (prepare_finalizers): Ditto.
17481 (is_inv_store_elimination_chain): New function.
17482 (initialize_root_vars_store_elim_1): New function.
17484 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17486 * tree-predcom.c: Revise general description of the pass.
17487 (enum chain_type): New enum type for store elimination.
17488 (struct chain): New field supporting store elimination.
17489 (struct component): Ditto.
17490 (dump_chain): Dump store-stores chain.
17491 (release_chain): Release resources.
17492 (split_data_refs_to_components): Compute and create component
17493 contains only stores for elimination.
17494 (get_chain_last_ref_at): New function.
17495 (make_invariant_chain): Initialization.
17496 (make_rooted_chain): Specify chain type in parameter and record it.
17497 (add_looparound_copies): Skip for store-stores chain.
17498 (determine_roots_comp): Compute type of chain and pass it to
17500 (initialize_root_vars_store_elim_2): New function.
17501 (finalize_eliminated_stores): New function.
17502 (remove_stmt): Handle store for elimination.
17503 (execute_pred_commoning_chain): Execute predictive commoning on
17504 store-store chains.
17505 (determine_unroll_factor): Skip unroll for store-stores chain.
17506 (prepare_initializers_chain_store_elim): New function.
17507 (prepare_initializers_chain): Hanlde store-store chain.
17508 (prepare_finalizers_chain, prepare_finalizers): New function.
17509 (tree_predictive_commoning_loop): Return integer value indicating
17510 if loop is unrolled or lcssa form is corrupted.
17511 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17513 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17515 * tree-predcom.c (initialize_root): Delete.
17516 (execute_pred_commoning_chain): Initialize root vars and replace
17517 reference of non-combined chain directly, rather than call above
17520 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17522 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17523 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17525 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17527 * tree-predcom.c (struct chain): New field init_seq.
17528 (release_chain): Release init_seq.
17529 (prepare_initializers_chain): Record intialization stmts in above
17531 (insert_init_seqs): New function.
17532 (tree_predictive_commoning_loop): Call insert_init_seqs.
17534 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17536 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17538 2017-07-28 Richard Biener <rguenther@suse.de>
17540 * match.pd: Remove superfluous :c.
17541 * genmatch.c (simplify::id): Add member.
17542 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17544 (current_id): New global.
17545 (dt_node::parent): Move from ...
17546 (dt_operand::parent): ... here. Add for_id member.
17547 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17548 (decision_tree::find_node): Relax order requirement when
17549 merging DT_TRUE nodes to ones inbetween the current simplify
17550 and the one we try to merge with. Add diagnostic whenever
17551 we need to enforce pattern order by not merging.
17552 (decision_tree::insert): Set current_id.
17553 (decision_tree::print_node): Dump parent node and for_id.
17554 (parser::last_id): Add member.
17555 (parser::push_simplify): Assign unique id.
17556 (parser::parser): Initialize last_id.
17558 2017-07-28 Martin Liska <mliska@suse.cz>
17561 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17562 gimple_build_debug_bind.
17564 2017-07-28 Richard Biener <rguenther@suse.de>
17566 PR tree-optimization/81502
17567 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17569 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17571 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17572 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17573 for BIT_FIELD_REF args.
17574 * fold-const.c (make_bit_field_ref): Likewise.
17575 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17577 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17580 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17581 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17582 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17583 Or it into SANITIZER_UNDEFINED.
17584 * ubsan.c: Include gimple-fold.h and varasm.h.
17585 (ubsan_expand_ptr_ifn): New function.
17586 (instrument_pointer_overflow): New function.
17587 (maybe_instrument_pointer_overflow): New function.
17588 (instrument_object_size): Formatting fix.
17589 (pass_ubsan::execute): Call instrument_pointer_overflow
17590 and maybe_instrument_pointer_overflow.
17591 * internal-fn.c (expand_UBSAN_PTR): New function.
17592 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17593 * sanitizer.def (__ubsan_handle_pointer_overflow,
17594 __ubsan_handle_pointer_overflow_abort): New builtins.
17595 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17596 * internal-fn.def (UBSAN_PTR): New internal function.
17597 * opts.c (sanitizer_opts): Add pointer-overflow.
17598 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17599 * fold-const.c (build_range_check): Compute pointer range check in
17600 integral type if pointer arithmetics would be needed. Formatting
17603 2017-07-28 Martin Liska <mliska@suse.cz>
17606 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17607 parameters that are of a variable-length.
17609 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17611 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17613 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17614 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17615 (CRT_CALL_STATIC_FUNCTION): Likewise.
17616 (ASM_DEFAULT_SPEC): New define.
17617 (ASM_SPEC32): Likewise.
17618 (ASM_SPEC64): Likewise.
17619 (ASM_SPEC_COMMON): Likewise.
17620 (ASM_SPEC): Likewise.
17621 (INVALID_64BIT): Likewise.
17622 (LINK_OS_DEFAULT_SPEC): Likewise.
17623 (LINK_OS_SPEC32): Likewise.
17624 (LINK_OS_SPEC64): Likewise.
17625 (POWERPC_LINUX): Likewise.
17626 (PTRDIFF_TYPE): Likewise.
17627 (RESTORE_FP_PREFIX): Likewise.
17628 (RESTORE_FP_SUFFIX): Likewise.
17629 (SAVE_FP_PREFIX): Likewise.
17630 (SAVE_FP_SUFFIX): Likewise.
17631 (SIZE_TYPE): Likewise.
17632 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17633 (TARGET_64BIT): Likewise.
17634 (TARGET_64BIT): Likewise.
17635 (TARGET_AIX): Likewise.
17636 (WCHAR_TYPE_SIZE): Likewise.
17637 (WCHAR_TYPE): Undefine.
17638 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17639 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17640 (CPP_OS_RTEMS_SPEC): Delete.
17641 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17642 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17644 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17646 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17648 PR tree-optimization/81578
17649 * tree-parloops.c (build_new_reduction): Bail out if
17650 reduction_code isn't one of the standard OpenMP reductions.
17651 Move the details printing after that decision.
17653 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17655 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17656 related to reload_in_progress.
17657 (splat_input_operand): Likewise.
17658 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17660 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17662 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17663 (TARGET_INSTANTIATE_DECLS): Likewise.
17664 (legitimate_indexed_address_p): Delete reload_in_progress code.
17665 (rs6000_debug_legitimate_address_p): Likewise.
17666 (rs6000_eliminate_indexed_memrefs): Likewise.
17667 (rs6000_emit_le_vsx_store): Likewise.
17668 (rs6000_emit_move_si_sf_subreg): Likewise.
17669 (rs6000_emit_move): Likewise.
17670 (register_to_reg_type): Likewise.
17671 (rs6000_pre_atomic_barrier): Likewise.
17672 (rs6000_machopic_legitimize_pic_address): Likewise.
17673 (rs6000_allocate_stack_temp): Likewise.
17674 (rs6000_address_for_fpconvert): Likewise.
17675 (rs6000_address_for_altivec): Likewise.
17676 (rs6000_secondary_memory_needed_rtx): Delete function.
17677 (rs6000_check_sdmode): Likewise.
17678 (rs6000_alloc_sdmode_stack_slot): Likewise.
17679 (rs6000_instantiate_decls): Likewise.
17680 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17681 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17682 Delete reload_in_progress.
17683 (*vec_reload_and_plus_<mptrsize>): Likewise.
17684 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17685 (vsx_div_v2di): Likewise.
17686 (vsx_udiv_v2di): Likewise.
17688 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17690 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17691 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17692 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17693 (rs6000_debug_reg_global): Delete print of LRA status.
17694 (rs6000_option_override_internal): Delete dead LRA related code.
17695 (rs6000_lra_p): Delete function.
17696 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17698 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17700 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17701 * config/riscv/rtems.h: New file.
17703 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17704 Sudakshina Das <sudi.das@arm.com>
17706 * config/aarch64/aarch64.md
17707 (define_split for and<mode>3nr_compare): Move
17708 non aarch64_logical_operand to a register.
17709 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17710 register immediate operand to a register.
17711 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17713 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17715 PR middle-end/81564
17716 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17718 2017-07-27 Richard Biener <rguenther@suse.de>
17720 PR tree-optimization/81573
17721 PR tree-optimization/81494
17722 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17723 multi defuse cycle case.
17725 2017-07-27 Richard Biener <rguenther@suse.de>
17727 PR tree-optimization/81571
17728 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17731 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17733 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17734 earlier and only if MASK_FPU is set. Adjust formatting.
17736 2017-07-27 Martin Liska <mliska@suse.cz>
17738 * opt-functions.awk: Add validation of value of Init.
17739 * optc-gen.awk: Pass new argument.
17741 2017-07-27 Martin Liska <mliska@suse.cz>
17743 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17744 Fix wrong condition.
17746 2017-07-27 Martin Liska <mliska@suse.cz>
17748 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17749 BBs and edges seen by autoFDO.
17751 2017-07-27 Richard Biener <rguenther@suse.de>
17753 PR tree-optimization/81502
17754 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17755 with incompatible but same sized type.
17756 (execute_update_addresses_taken): Likewise.
17758 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17760 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17761 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17763 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17766 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17767 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17768 Change s_operand to memory_operand.
17770 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17772 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17773 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17774 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17775 Emit instructions rather than returning an expression. Handle TFmode
17776 and KFmode by casting to TImode.
17777 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17778 (rs6000_emit_le_vsx_store): Likewise.
17779 * config/rs6000/vsx.md (VSX_TI): New iterator.
17780 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17781 (*vsx_le_undo_permute_<mode>): Likewise.
17782 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17783 emit the split sequence.
17784 (*vsx_le_perm_store_<mode>): Likewise.
17786 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17788 PR tree-optimization/81555
17789 PR tree-optimization/81556
17790 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17791 if true, force CHANGED for the recursive invocation.
17792 (reassociate_bb): Remember original length of ops array, pass
17793 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17795 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17796 attributes for noipa attribute. For naked attribute use
17797 lookup_attribute first before lookup_attribute_spec.
17798 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17800 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17801 for functions with noipa attribute.
17802 (cgraph_externally_visible_p): Return true for functions with noipa
17804 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17805 for functions with noipa attribute.
17806 * doc/extend.texi: Document noipa function attribute.
17807 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17808 also for functions with noipa attribute.
17809 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17811 2017-07-26 Andrew Pinski <apinski@cavium.com>
17813 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17814 vec_unalign_load_cost and vec_unalign_store_cost.
17816 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17818 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17819 -mvsx-small-integer option.
17820 (ISA_3_0_MASKS_IEEE): Likewise.
17821 (OTHER_VSX_VECTOR_MASKS): Likewise.
17822 (POWERPC_MASKS): Likewise.
17823 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17824 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17825 code, only testing for DImode being allowed in non-VSX floating
17827 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17828 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17830 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17831 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17832 TARGET_P8_VECTOR test.
17833 (rs6000_secondary_reload_simple_move): Likewise.
17834 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17835 since TARGET_P9_VECTOR was already tested.
17836 (rs6000_opt_masks): Remove -mvsx-small-integer.
17837 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17838 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17840 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17841 test for TARGET_VEXTRACTUB was used, and that uses
17843 (p9 extract splitter): Likewise.
17844 (vsx_extract_<mode>_di_p9): Likewise.
17845 (vsx_extract_<mode>_store_p9): Likewise.
17846 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17847 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17848 the elimination of TARGET_VSX_SMALL_INTEGER.
17849 (vsx_extract_<mode>_p8): Likewise.
17850 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17851 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17852 (vsx_set_<mode>_p9): Likewise.
17853 (vsx_set_v4sf_p9): Likewise.
17854 (vsx_set_v4sf_p9_zero): Likewise.
17855 (vsx_insert_extract_v4sf_p9): Likewise.
17856 (vsx_insert_extract_v4sf_p9_2): Likewise.
17857 * config/rs6000/rs6000.md (sign extend splitter): Change
17858 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17859 (floatsi<mode>2_lfiwax_mem): Likewise.
17860 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17861 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17862 since a test for TARGET_P9_VECTOR was used.
17863 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17864 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17865 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17866 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17867 TARGET_P8_VECTOR test.
17868 (fix_trunc<mode>si2_stfiwx): Likewise.
17869 (fix_trunc<mode>si2_internal): Likewise.
17870 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17871 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17873 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17874 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17875 TARGET_P8_VECTOR test.
17876 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17877 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17878 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17880 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17881 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17882 since a test for TARGET_P9_VECTOR was used.
17883 (splitter for loading small constants): Likewise.
17885 2017-07-26 Andrew Pinski <apinski@cavium.com>
17887 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17890 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17893 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17894 (fp_valid_at): Likewise.
17896 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17898 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17899 (qdf24xx_addrcost_table): Likewise.
17900 (cortexa57_tunings): Update to use generic_branch_cost.
17901 (cortexa72_tunings): Likewise.
17902 (cortexa73_tunings): Likewise.
17903 (qdf24xx_tunings): Likewise.
17905 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17907 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17908 (thunderx2t99_branch_cost): Likewise.
17909 (cortexa35_tunings): Update to use generic_branch_cost.
17910 (cortexa53_tunings): Likewise.
17911 (cortexa57_tunings): Likewise.
17912 (cortexa72_tunings): Likewise.
17913 (cortexa73_tunings): Likewise.
17914 (thunderx2t99_tunings): Likewise.
17916 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17918 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17919 (sparc_option_override): Honour MASK_FSMULD.
17920 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17921 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17922 * config/sparc/sparc.opt (mfsmuld): New option.
17923 * doc/invoke.texi (mfsmuld): Document option.
17925 2017-07-26 Marek Polacek <polacek@redhat.com>
17927 PR middle-end/70992
17928 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17930 2017-07-26 Richard Biener <rguenther@suse.de>
17932 * gimple-match-head.c (do_valueize): Return OP if valueize
17934 (get_def): New helper to get at the def stmt of a SSA name
17935 if valueize allows.
17936 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17937 do_valueize to get at the def stmt.
17938 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17940 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17942 PR middle-end/46932
17943 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17945 2017-07-26 Martin Liska <mliska@suse.cz>
17948 * function.c (expand_function_start): Make expansion of
17949 nonlocal_goto_save_area after parm_birth_insn.
17951 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17953 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17954 from all CPU target flags enable members.
17956 2017-07-26 Richard Biener <rguenther@suse.de>
17958 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17959 (decision_tree::gen): Make 'type' const.
17960 (write_predicate): Likewise.
17962 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17964 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17965 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17966 (rs6000_option_override_internal): Likewise.
17967 (rs6000_expand_vector_set): Likewise.
17968 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17969 (TARGET_UPPER_REGS_SF): Likewise.
17970 (TARGET_UPPER_REGS_DI): Likewise.
17971 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17972 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17973 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17974 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17975 (Splitters for DI constants in Altivec registers): Likewise.
17976 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17977 (vsx_set_v4sf_p9): Likewise.
17978 (vsx_set_v4sf_p9_zero): Likewise.
17979 (vsx_insert_extract_v4sf_p9): Likewise.
17980 (vsx_insert_extract_v4sf_p9_2): Likewise.
17982 2017-07-25 Carl Love <cel@us.ibm.com>
17984 * doc/extend.texi: Update the built-in documentation file for the
17985 existing built-in functions
17986 vector signed char vec_cnttz (vector signed char);
17987 vector unsigned char vec_cnttz (vector unsigned char);
17988 vector signed short vec_cnttz (vector signed short);
17989 vector unsigned short vec_cnttz (vector unsigned short);
17990 vector signed int vec_cnttz (vector signed int);
17991 vector unsigned int vec_cnttz (vector unsigned int);
17992 vector signed long long vec_cnttz (vector signed long long);
17993 vector unsigned long long vec_cnttz (vector unsigned long long);
17995 2017-07-25 Andrew Pinski <apinski@cavium.com>
17997 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17998 accesses where the use is for the first operand of a BIT_INSERT.
18000 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18003 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18004 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18006 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18008 * config/i386/gstabs.h: Delete.
18009 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18011 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18013 * config/i386/i386.c (ix86_decompose_address): Do not check for
18014 register RTX when looking at index_reg or base_reg.
18015 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18017 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18019 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18020 to update EH info here.
18022 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18024 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18026 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18028 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18030 2017-07-25 Torsten Duwe <duwe@suse.de>
18032 * common.opt: Introduce -fpatchable-function-entry
18033 command line option, and its variables function_entry_patch_area_size
18034 and function_entry_patch_area_start.
18035 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18036 including a two-value parser.
18037 * target.def (print_patchable_function_entry): New target hook.
18038 * targhooks.h (default_print_patchable_function_entry): New function.
18039 * targhooks.c (default_print_patchable_function_entry): Likewise.
18040 * toplev.c (process_options): Switch off IPA-RA if
18041 patchable function entries are being generated.
18042 * varasm.c (assemble_start_function): Look at the
18043 patchable-function-entry command line switch and current
18044 function attributes and maybe generate NOP instructions by
18045 calling the print_patchable_function_entry hook.
18046 * doc/extend.texi: Document patchable_function_entry attribute.
18047 * doc/invoke.texi: Document -fpatchable_function_entry
18048 command line option.
18049 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18051 * doc/tm.texi: Re-generate.
18053 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18056 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18057 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18059 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18061 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18063 2017-07-25 Richard Biener <rguenther@suse.de>
18065 PR tree-optimization/81455
18066 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18067 not walk in cycles when looking for guards.
18069 2017-07-25 Richard Biener <rguenther@suse.de>
18071 PR tree-optimization/81529
18072 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18073 when optimizing backedge uses.
18075 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18077 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18079 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18080 to dl_section_ref. On AIX, append an expression to subtract
18081 the size of the section length to dl_section_ref.
18083 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18085 * configure.ac: If any of the config.* scripts fail, exit 1.
18086 * configure: Regenerate.
18088 2017-07-25 Richard Biener <rguenther@suse.de>
18090 PR middle-end/81546
18091 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18092 of immediate uses, be more verbose on errors.
18094 2017-07-25 Richard Biener <rguenther@suse.de>
18096 PR tree-optimization/81510
18097 * tree-vect-loop.c (vect_is_simple_reduction): When the
18098 reduction stmt is not inside the loop bail out.
18100 2017-07-25 Richard Biener <rguenther@suse.de>
18102 PR tree-optimization/81303
18103 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18104 profitability check against LOOP_VINFO_NITERSM1.
18106 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18108 * domwalk.c (cmp_bb_postorder): Simplify.
18109 (sort_bbs_postorder): New function. Use it...
18110 (dom_walker::walk): ...here to optimize common cases.
18112 2017-07-25 Martin Liska <mliska@suse.cz>
18115 * ipa-visibility.c (function_and_variable_visibility): Make the
18116 redirection just on target that supports aliasing.
18117 Fix GNU coding style.
18119 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18122 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18124 * config/arm/rtems.h: Likewise.
18125 * config/bfin/rtems.h: Likewise.
18126 * config/i386/rtemself.h: Likewise.
18127 * config/lm32/rtems.h: Likewise.
18128 * config/m32c/rtems.h: Likewise.
18129 * config/m68k/rtemself.h: Likewise.
18130 * config/microblaze/rtems.h: Likewise.
18131 * config/mips/rtems.h: Likewise.
18132 * config/moxie/rtems.h: Likewise.
18133 * config/nios2/rtems.h: Likewise.
18134 * config/powerpcspe/rtems.h: Likewise.
18135 * config/rs6000/rtems.h: Likewise.
18136 * config/rtems.h: Likewise.
18137 * config/sh/rtems.h: Likewise.
18138 * config/sh/rtemself.h: Likewise.
18139 * config/sparc/rtemself.h: Likewise.
18141 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18144 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18145 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18146 * tree-ssa-structalias.c (alias_get_name): Same.
18148 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18151 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18152 instructions if no du chain is found.
18154 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18156 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18158 2017-07-25 Richard Biener <rguenther@suse.de>
18160 PR middle-end/81505
18161 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18164 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18166 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18167 upper-regs options.
18168 (ISA_2_7_MASKS_SERVER): Likewise.
18169 (ISA_3_0_MASKS_IEEE): Likewise.
18170 (OTHER_P8_VECTOR_MASKS): Likewise.
18171 (OTHER_VSX_VECTOR_MASKS): Likewise.
18172 (POWERPC_MASKS): Likewise.
18173 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18174 duplicate list of options.
18175 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18176 explicit -mupper-regs options.
18177 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18178 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18179 alias for -mupper-regs-df.
18180 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18181 (rs6000_init_hard_regno_mode_ok): Likewise.
18182 (rs6000_option_override_internal): Likewise.
18183 (rs6000_opt_masks): Likewise.
18184 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18185 options in terms of whether -mvsx or -mpower8-vector was used.
18186 (TARGET_UPPER_REGS_DI): Likewise.
18187 (TARGET_UPPER_REGS_SF): Likewise.
18188 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18189 -mupper-regs-* options.
18191 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18193 * passes.c (emergency_dump_function): Print some empty lines and a
18194 header before the RTL dump.
18196 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18198 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18200 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18203 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18204 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18206 2017-07-24 Carl Love <cel@us.ibm.com>
18208 * config/rs6000/rs6000-c.c: Add support for built-in functions
18209 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18210 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18211 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18212 vec_extract_fp_from_shortl): Add defines for the two builtins.
18213 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18214 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18216 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18217 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18218 * doc/extend.texi: Update the built-in documentation file for the
18219 new built-in function.
18221 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18224 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18226 * doc/generic.texi: Likewise.
18227 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18228 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18230 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18232 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18233 (aarch64_mls_elt_merge<mode>): Likewise.
18235 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18237 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18238 having __cxa_atexit.
18240 2017-07-23 Michael Collison <michael.collison@arm.com>
18242 * config/arm/arm.c (arm_option_override): Deprecate
18243 use of -mstructure-size-boundary.
18244 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18245 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18247 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18250 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18251 Reduce cost estimate for direct moves.
18253 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18256 * config/i386/i386.c (ix86_option_override_internal): Disable
18257 BMI, BMI2 and TBM instructions for -m16.
18259 2017-07-21 Carl Love <cel@us.ibm.com>
18261 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18262 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18263 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18264 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18265 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18266 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18267 VMULOSW): New enum "unspec" values.
18268 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18269 altivec_vmulosw): New patterns.
18270 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18271 VMULOSW): Add definitions.
18273 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18275 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18276 (qdf24xx): Likewise.
18277 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18278 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18279 (AARCH64_FL_V8_1): Renumber.
18280 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18281 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18282 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18283 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18284 rdma to feature modifiers list.
18286 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18288 PR middle-end/56727
18289 * ipa-visibility (function_and_variable_visibility): Convert
18290 recursive PLT call to direct call if appropriate.
18292 2017-07-21 Andrew Pinski <apinski@cavium.com>
18294 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18295 operand 1 to see if the types precision matches.
18296 * fold-const.c (operand_equal_p): Likewise.
18298 2017-07-21 Richard Biener <rguenther@suse.de>
18300 PR tree-optimization/81303
18301 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18302 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18303 (vect_peeling_hash_get_lowest_cost): Adjust.
18304 (vect_enhance_data_refs_alignment): Likewise. Use
18305 vect_get_peeling_costs_all_drs to compute the penalty for no
18306 peeling to match up costs.
18308 2017-07-21 Richard Biener <rguenther@suse.de>
18310 PR tree-optimization/81500
18311 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18312 we didn't identify a reduction path.
18314 2017-07-21 Tom de Vries <tom@codesourcery.com>
18315 Cesar Philippidis <cesar@codesourcery.com>
18317 PR gcov-profile/81442
18318 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18321 2017-07-21 Tom de Vries <tom@codesourcery.com>
18324 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18326 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18327 nvptx_override_options_after_change.
18329 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18331 * dwarf2out.c (output_file_names): Avoid double testing for
18332 dwarf_version >= 5.
18334 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18336 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18338 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18340 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18342 (try_crossjump_to_edge): Do not punt on partitioned functions.
18344 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18346 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18347 Put all BBs reachable only via paths crossing cold region to cold
18349 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18351 2016-07-21 Richard Biener <rguenther@suse.de>
18353 PR tree-optimization/81303
18354 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18355 into account prologue and epilogue iterations when raising
18356 min_profitable_iters to sth at least covering one vector iteration.
18358 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18360 * config/arm/arm.c (arm_test_cpu_arch_dat):
18363 2017-07-20 Nathan Sidwell <nathan@acm.org>
18365 Remove TYPE_METHODS.
18366 * tree.h (TYPE_METHODS): Delete.
18367 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18368 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18369 (dbxout_type_methods): Scan TYPE_FIELDS.
18370 (dbxout_type): Don't check TYPE_METHODS here.
18371 * function.c (use_register_for_decl): Always ignore register for
18372 class types when not optimizing.
18373 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18374 * tree.c (free_lang_data_in_type): Stitch out member functions and
18375 templates from TYPE_FIELDS.
18376 (build_distinct_type_copy, verify_type_variant,
18377 verify_type): Member fns are on TYPE_FIELDS.
18378 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18379 * tree-pretty-print.c (dump_generic_node): Likewise.
18381 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18384 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18385 V2TImode and V4TImode.
18386 (ix86_expand_vector_extract): Likewise.
18387 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18388 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18389 (ssescalarmode): Handle V4TImode and V2TImode.
18390 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18391 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18392 (VEXTRACTI128_MODE): New mode iterator.
18393 (splitter for *vec_extractv?ti first element): New.
18394 (VEC_INIT_MODE): New mode iterator.
18395 (vec_init<mode>): Consolidate 3 expanders into one using
18396 VEC_INIT_MODE mode iterator.
18398 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18400 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18401 non_spilled_static_chain_regno_p.
18403 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18405 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18407 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18409 * bb-reorder.c (connect_traces): Allow copying of blocks within
18412 2017-07-20 Richard Biener <rguenther@suse.de>
18414 * gimple.h (gimple_phi_result): Add gphi * overload.
18415 (gimple_phi_result_ptr): Likewise.
18416 (gimple_phi_arg): Likewise. Adjust index assert to only
18417 allow actual argument accesses rather than all slots available
18419 (gimple_phi_arg_def): Add gphi * overload.
18420 * tree-phinodes.c (make_phi_node): Initialize only actual
18422 (resize_phi_node): Clear memory not covered by old node,
18423 do not initialize excess argument slots.
18424 (reserve_phi_args_for_new_edge): Initialize new argument slot
18427 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18429 PR tree-optimization/81388
18431 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18433 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18434 by removing computation of may_be_zero.
18436 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18437 Tom de Vries <tom@codesourcery.com>
18439 PR middle-end/81030
18440 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18441 when gimple level profile disagrees with what RTL expander did.
18443 2017-07-20 Richard Biener <rguenther@suse.de>
18445 PR tree-optimization/61171
18446 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18447 (vect_analyze_stmt): Add slp instance parameter.
18448 (vectorizable_reduction): Likewise.
18449 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18450 (vect_is_simple_reduction): Deal with chains not detected
18451 as SLP reduction chain, specifically not properly associated
18452 chains containing a mix of plus/minus.
18453 (get_reduction_op): Remove.
18454 (get_initial_defs_for_reduction): Simplify, pass in whether
18455 this is a reduction chain, pass in the SLP node for the PHIs.
18456 (vect_create_epilog_for_reduction): Get the SLP instance as
18458 (vectorizable_reduction): Get the SLP instance as arg.
18459 During analysis remember the SLP node with the PHIs in the
18460 instance. Simplify getting at the vectorized reduction PHIs.
18461 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18462 through SLP instance.
18463 (vect_slp_analyze_operations): Likewise.
18464 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18465 (vect_transform_stmt): Likewise.
18467 2017-07-20 Tom de Vries <tom@codesourcery.com>
18469 PR tree-optimization/81489
18470 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18471 read of phi arg location to before loop that modifies phi.
18473 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18475 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18478 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18480 PR middle-end/81331
18481 * except.c (execute): Fix ordering issue.
18483 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18485 PR rtl-optimization/81423
18486 * combine.c (make_compound_operation_int): Don't try to optimize
18487 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18489 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18491 PR rtl-optimization/81423
18492 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18493 with a constant that is -1 in the truncated to mode.
18495 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18497 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18498 (determine_unlikely_bbs): ... here.
18499 * predict.h (propagate_unlikely_bbs_forward): Declare.
18500 * cfgexpand.c (pass_expand::execute): Use it.
18501 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18503 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18504 propagate_unlikely_bbs_forward.
18506 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18508 PR middle-end/81331
18509 * except.c (maybe_add_nop_after_section_switch): New function.
18512 2017-07-19 Tom de Vries <tom@codesourcery.com>
18514 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18516 2017-07-19 Tom de Vries <tom@codesourcery.com>
18518 * gimple.h (gimple_phi_arg): Make assert more strict.
18520 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18522 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18523 * config/rs6000/mmintrin.h: New file.
18524 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18526 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18528 PR tree-optimization/81346
18529 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18531 2017-07-19 Tom de Vries <tom@codesourcery.com>
18533 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18535 2017-07-19 Tom de Vries <tom@codesourcery.com>
18537 * config/nvptx/nvptx-modes.def: Add V2DImode.
18538 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18539 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18540 (nvptx_output_mov_insn): Handle lack of mov.b128.
18541 (nvptx_print_operand): Handle 'H' and 'L' codes.
18542 (nvptx_vector_mode_supported): Allow V2DImode.
18543 (nvptx_preferred_simd_mode): New function.
18544 (nvptx_data_alignment): New function.
18545 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18546 nvptx_preferred_simd_mode.
18547 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18549 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18551 2017-07-19 Tom de Vries <tom@codesourcery.com>
18553 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18554 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18555 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18556 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18557 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18558 (mov<VECIM>_insn): New define_insn.
18559 (define_expand "mov<VECIM>): New define_expand.
18561 2017-07-19 Tom de Vries <tom@codesourcery.com>
18563 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18565 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18567 PR tree-optimization/81346
18568 * fold-const.h (fold_div_compare, range_check_type): Declare.
18569 * fold-const.c (range_check_type): New function.
18570 (build_range_check): Use range_check_type.
18571 (fold_div_compare): No longer static, rewritten into
18572 a match.pd helper function.
18573 (fold_comparison): Don't call fold_div_compare here.
18574 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18575 as helper function.
18577 2017-07-19 Nathan Sidwell <nathan@acm.org>
18579 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18580 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18581 * tree.c (find_decls_types_r, verify_type): Use
18582 TYPE_{MIN,MAX}_VALUE_RAW.
18583 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18584 (hash_tree): Likewise.
18585 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18587 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18590 2017-07-18 Tom de Vries <tom@codesourcery.com>
18592 PR middle-end/81464
18593 * omp-expand.c (expand_omp_for_static_chunk): Handle
18594 equal-argument loop exit phi.
18596 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18599 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18600 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18601 operand 2 predicate.
18602 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18603 operand 2 predicate.
18604 (ror,rol -> rorx splitters): Use const_int_operand as
18605 operand 2 predicate.
18607 2017-06-18 Richard Biener <rguenther@suse.de>
18609 PR tree-optimization/81410
18610 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18611 the gap in the ! slp_perm SLP case after each group.
18613 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18615 PR middle-end/81463
18616 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18619 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18621 PR middle-end/81462
18622 * predict.c (set_even_probabilities): Cleanup; do not affect
18623 probabilities that are already known.
18624 (combine_predictions_for_bb): Call even when count is set.
18626 2017-07-18 Nathan Sidwell <nathan@acm.org>
18628 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18631 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18634 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18635 optimization for loop niter analysis.
18637 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18640 * config/avr/avr.c (avr_optimize_casesi): Don't use
18641 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18643 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18645 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18646 body_cost_vec from _vect_peel_extended_info.
18647 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18648 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18651 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18653 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18655 2017-07-18 Richard Biener <rguenther@suse.de>
18657 PR tree-optimization/80620
18658 PR tree-optimization/81403
18659 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18660 info when re-using a VN table entry.
18662 2017-07-18 Richard Biener <rguenther@suse.de>
18664 PR tree-optimization/81418
18665 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18666 vectype_in. Verify that with lane-reducing reduction operations
18667 we have a single def-use cycle.
18669 2017-07-17 Carl Love <cel@us.ibm.com>
18671 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18673 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18674 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18675 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18677 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18678 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18679 VMULOSW): New enum "unspec" values.
18680 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18681 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18682 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18683 altivec_vmulosw): New patterns.
18684 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18685 VMULOSW): Add definitions.
18686 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18688 * config/alpha/alpha.c: Include predict.h.
18690 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18692 * tree-vrp.c (compare_assert_loc): Fix comparison function
18693 to return predictable results.
18695 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18697 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18699 (subdi3): Likewise.
18700 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18701 * doc/invoke.texi (mexpand-adddi): Update text.
18703 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18705 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18706 that also clobbers the CC register. The old expand code is moved
18708 (*arc_clzsi2): ... here.
18709 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18710 the CC register. The old expand code is moved to ...
18711 (arc_ctzsi2): ... here.
18713 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18715 * config/arc/arc.opt (mindexed-loads): Use initial value
18716 TARGET_INDEXED_LOADS_DEFAULT.
18717 (mauto-modify-reg): Use initial value
18718 TARGET_AUTO_MODIFY_REG_DEFAULT.
18719 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18720 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18721 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18722 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18724 2017-07-17 Martin Liska <mliska@suse.cz>
18727 * opts.c (finish_options): Do not allow -fgnu-tm
18728 w/ -fsanitize={kernel-,}address. Say sorry.
18730 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18733 * tree-loop-distribution.c (classify_partition): Only assert on
18734 numer of iterations.
18735 (merge_dep_scc_partitions): Delete prameter. Update function call.
18736 (distribute_loop): Remove code handling loop with unknown niters.
18737 (pass_loop_distribution::execute): Skip loop with unknown niters.
18739 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18742 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18743 function sort_partitions_by_post_order.
18745 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18747 PR tree-optimization/81374
18748 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18749 the max index of basic blocks, rather than number of basic blocks.
18751 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18753 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18755 (arc_legitimate_pic_operand_p): Likewise.
18756 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18758 (arc_needs_pcl_p): Likewise.
18759 (arc_legitimate_pc_offset_p): Likewise.
18760 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18761 function is also used in constrains.md.
18762 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18763 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18764 PLUS. Only return true/false in known cases, otherwise assert.
18765 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18766 is already called in arc_legitimate_constant_p.
18767 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18769 (LEGITIMATE_PIC_OPERAND_P): Use
18770 arc_raw_symbolic_reference_mentioned_p function.
18771 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18776 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18777 Andrew Burgess <andrew.burgess@embecosm.com>
18779 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18780 (arc_return_address_register): New function.
18781 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18782 (arc_handle_fndecl_attribute): Add naked attribute.
18783 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18784 (TARGET_WARN_FUNC_RETURN): Likewise.
18785 (arc_allocate_stack_slots_for_args): New function.
18786 (arc_warn_func_return): Likewise.
18787 (machine_function): Change type fn_type.
18788 (arc_compute_function_type): Consider new naked function type,
18789 change function return type.
18790 (arc_must_save_register): Adapt to handle new
18791 arc_compute_function_type's return type.
18792 (arc_expand_prologue): Likewise.
18793 (arc_expand_epilogue): Likewise.
18794 (arc_return_address_regs): Delete.
18795 (arc_return_address_register): New function.
18796 (arc_epilogue_uses): Use above function.
18797 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18798 (arc_function_type): Change encoding, add naked type.
18799 (ARC_INTERRUPT_P): Change to handle the new encoding.
18800 (ARC_FAST_INTERRUPT_P): Likewise.
18801 (ARC_NORMAL_P): Define.
18802 (ARC_NAKED_P): Likewise.
18803 (arc_compute_function_type): Delete prototype.
18804 * config/arc/arc.md (in_ret_delay_slot): Use
18805 arc_return_address_register function.
18806 (simple_return): Likewise.
18807 (p_return_i): Likewise.
18809 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18811 PR tree-optimization/81428
18812 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18813 can't be built for those types.
18815 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18817 Remove stuff dead since r239246.
18819 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18820 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18821 (avr_inform_devices): Remove dead stuff.
18823 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18825 * config/arm/arm_neon.h: Fix softp typo.
18827 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18829 PR tree-optimization/81365
18830 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18831 aggregate moves onto bb predecessor edges, make sure there are no
18832 loads that could alias the lhs in between the start of bb and the
18835 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18838 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18839 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18840 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18842 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18844 PR tree-optimization/81396
18845 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18846 (init_symbolic_number): Initialize it to 1.
18847 (perform_symbolic_merge): Add n_ops from both operands into the new
18849 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18850 without base_addr as useless if they need more than one operation.
18851 (bswap_replace): Handle !bswap case for NULL base_addr.
18853 2017-07-17 Tom de Vries <tom@codesourcery.com>
18856 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18859 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18861 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18862 conditional builtin define __FIX_LEON3FT_B2BST.
18864 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18866 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18867 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18870 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18872 PR rtl-optimization/81424
18873 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18874 to remove potential trapping from operands if -fnon-call-exceptions.
18876 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18878 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18879 profile_proability for scalling.
18880 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18882 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18884 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18886 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18888 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18889 fixpoint arithmetics.
18891 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18893 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18894 fixpoint arithmetics.
18896 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18898 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18899 fixpoint arithmetics.
18901 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18903 * profile-count.h (profile_probability::from_reg_br_prob_note,
18904 profile_probability::to_reg_br_prob_note): New functions.
18905 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18906 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18907 * predict.c (probability_reliable_p): Update.
18908 (edge_probability_reliable_p): Update.
18909 (br_prob_note_reliable_p): Update.
18910 (invert_br_probabilities): Update.
18911 (add_reg_br_prob_note): New function.
18912 (combine_predictions_for_insn): Update.
18913 * asan.c (asan_clear_shadow): Update.
18914 * cfgbuild.c (compute_outgoing_frequencies): Update.
18915 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18916 (update_br_prob_note): Update.
18917 (rtl_verify_edges): Update.
18918 (purge_dead_edges): Update.
18919 (fixup_reorder_chain): Update.
18920 * emit-rtl.c (try_split): Update.
18921 * ifcvt.c (cond_exec_process_insns): Update.
18922 (cond_exec_process_if_block): Update.
18923 (dead_or_predicable): Update.
18924 * internal-fn.c (expand_addsub_overflow): Update.
18925 (expand_neg_overflow): Update.
18926 (expand_mul_overflow): Update.
18927 * loop-doloop.c (doloop_modify): Update.
18928 * loop-unroll.c (compare_and_jump_seq): Update.
18929 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18930 * predict.h: Update.
18931 * reorg.c (mostly_true_jump): Update.
18933 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18934 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18935 * config/arc/arc.c: (emit_unlikely_jump): Update.
18936 * config/arm/arm.c: (emit_unlikely_jump): Update.
18937 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18938 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18939 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18940 (ix86_print_operand): Update.
18941 (ix86_split_fp_branch): Update.
18942 (predict_jump): Update.
18943 * config/ia64/ia64.c (ia64_print_operand): Update.
18944 * config/mmix/mmix.c (mmix_print_operand): Update.
18945 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18946 (rs6000_expand_split_stack_prologue): Update.
18947 * config/rs6000/rs6000.c: Update.
18948 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18949 (s390_expand_vec_movstr): Update.
18950 (s390_expand_cs_tdsi): Update.
18951 (s390_expand_split_stack_prologue): Update.
18952 * config/sh/sh.c (sh_print_operand): Update.
18953 (expand_cbranchsi4): Update.
18954 (expand_cbranchdi4): Update.
18955 * config/sparc/sparc.c (output_v9branch): Update.
18956 * config/spu/spu.c (get_branch_target): Update.
18957 (ea_load_store_inline): Update.
18958 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18959 * config/tilepro/tilepro.c: Update.
18961 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18963 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18964 (gimplify_save_expr): Likewise.
18966 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18968 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18970 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18972 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18974 * timevar.def (TV_IPA_FNSUMMARY): Define.
18976 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18978 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18979 to back store errata sensitive sequence from being generated.
18980 (sqrtdf2_fix): Likewise.
18982 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18984 * tree-ssa-threadupdate.c (compute_path_counts,
18985 update_joiner_offpath_counts): Use profile_probability.
18987 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18990 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18992 * config/arm/arm-c.c (arm_cpu_builtins): Define
18993 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18995 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18997 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18998 array entries to represent __ieee128 versions of the
18999 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19000 scalar_extract_sig, and scalar_insert_exp built-in functions.
19001 (altivec_resolve_overloaded_builtin): Add special case handling
19002 for the __builtin_scalar_insert_exp function, as represented by
19003 the P9V_BUILTIN_VEC_VSIEDP constant.
19004 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19005 exponent support for __ieee128 argument.
19006 (VSESQP): Add scalar extract signature support for __ieee128
19008 (VSTDCNQP): Add scalar test negative support for __ieee128
19010 (VSIEQP): Add scalar insert exponent support for __int128 argument
19011 with __ieee128 result.
19012 (VSIEQPF): Add scalar insert exponent support for __ieee128
19013 argument with __ieee128 result.
19014 (VSTDCQP): Add scalar test data class support for __ieee128
19016 (VSTDCNQP): Add overload support for scalar test negative with
19017 __ieee128 argument.
19018 (VSTDCQP): Add overload support for scalar test data class
19019 __ieee128 argument.
19020 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19021 UNSPEC_VSX_SXSIGDP.
19022 (UNSPEC_VSX_SIEXPQP): New constant.
19023 (xsxexpqp): New insn for VSX scalar extract exponent quad
19025 (xsxsigqp): New insn for VSX scalar extract significand quad
19027 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19028 precision with floating point argument.
19029 (xststdcqp): New expand for VSX scalar test data class quad
19031 (xststdcnegqp): New expand for VSX scalar test negative quad
19033 (xststdcqp): New insn to match expansions for VSX scalar test data
19034 class quad precision and VSX scalar test negative quad precision.
19035 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19036 special case operand checking to enforce that second operand of
19037 VSX scalar test data class with quad precision argument is a 7-bit
19039 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19040 prototypes and descriptions of __ieee128 versions of
19041 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19042 scalar_test_data_class, and scalar_test_neg built-in functions.
19044 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19046 PR tree-optimization/81162
19047 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19048 replace a negate with an add.
19050 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19052 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19054 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19056 * config/arm/arm-c.c (arm_cpu_builtins): Define
19057 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19059 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19061 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19062 (armv8-r): Set ARM Cortex-R52 as default CPU.
19063 * config/arm/arm-tables.opt: Regenerate.
19064 * config/arm/arm-tune.md: Regenerate.
19065 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19067 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19068 extension for -mcpu=cortex-r52.
19070 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19072 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19073 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19074 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19075 (fp-armv8): Define it as FP_ARMv8 only.
19076 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19077 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19079 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19080 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19081 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19082 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19083 than TARGET_FPU_ARMV8.
19084 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19085 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19086 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19088 * config/arm/neon.md (neon_vrint): Likewise.
19089 (neon_vcvt): Likewise.
19090 (neon_<fmaxmin_op><mode>): Likewise.
19091 (<fmaxmin><mode>3): Likewise.
19092 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19093 * config/arm/predicates.md (arm_cond_move_operator): Check against
19094 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19096 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19098 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19099 to top of function.
19101 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19103 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19104 loop in comment with memset.
19106 2017-07-14 Martin Liska <mliska@suse.cz>
19108 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19109 * dwarf2out.c (is_java): Remove the function.
19110 (output_pubname): Remove usage of the function.
19111 (lower_bound_default): Remove usage of DW_LANG_Java.
19112 (gen_compile_unit_die): Likewise.
19113 * gcc.c: Remove compiler defaults for .java and .zip files.
19114 * gimple-expr.c (remove_suffix): Change as there's no longer
19115 extension than 4-letter one.
19116 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19117 (gimplify_save_expr): Likewise.
19118 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19119 as it's possible even for other languages than Java.
19120 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19121 * lto-opts.c (lto_write_options): Remove reference to Java.
19122 * opts.c (strip_off_ending): Update file extension handling.
19123 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19124 * tree-eh.c (lower_resx): Likewise.
19125 * tree.c (free_lang_data_in_type): Remove dead code.
19126 (find_decls_types_r): Likewise.
19127 (build_common_builtin_nodes): Remove Java from a comment.
19128 (verify_type): Remove dead code.
19129 * varasm.c (assemble_external): Remove Java from a comment.
19131 2017-07-14 Martin Liska <mliska@suse.cz>
19133 * opts.c (finish_options): Add quotes.
19134 (common_handle_option): Likewise.
19136 2017-07-14 Martin Liska <mliska@suse.cz>
19138 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19139 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19140 Remove N_SO_PASCAL.
19141 * dwarf2out.c (lower_bound_default): Do not handle
19143 (gen_compile_unit_die): Likewise.
19144 * gcc.c: Remove default extension binding for GNU Pascal.
19145 * stmt.c: Remove Pascal language from a comment.
19146 * xcoffout.c: Likewise.
19148 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19151 * diagnostic-show-locus.c (fixit_cmp): New function.
19152 (layout::layout): Sort m_fixit_hints.
19153 (column_range::column_range): Assert that the values are valid.
19154 (struct char_span): New struct.
19155 (correction::overwrite): New method.
19156 (struct source_line): New struct.
19157 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19158 calls in terms of classes source_line and char_span, and
19159 correction::overwrite.
19160 (selftest::test_overlapped_fixit_printing_2): New function.
19161 (selftest::diagnostic_show_locus_c_tests): Call it.
19163 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19165 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19166 early if there is no lhs.
19168 2017-07-13 Martin Liska <mliska@suse.cz>
19170 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19171 (gen_reference_type_die): Likewise.
19172 * stor-layout.c: Remove Pascal-related comment.
19174 2017-07-13 Martin Liska <mliska@suse.cz>
19176 * opts.c (finish_options): Add quotes to error messages.
19177 (parse_sanitizer_options): Likewise.
19179 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19181 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19183 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19185 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19187 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19189 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19191 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19193 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19196 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19197 provides the hardware capability bits, define the macro
19198 __BUILTIN_CPU_SUPPORTS__.
19199 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19200 if GLIBC does not provide the hardware capability bits. Add a
19201 gcc_unreachable call if the built-in cpu function is neither
19202 __builtin_cpu_is nor __builtin_cpu_supports.
19203 (rs6000_get_function_versions_dispatcher): Change the warning
19204 that an old GLIBC is used which does not export the capability
19205 bits to be an error.
19206 * doc/extend.texi (target_clones attribute): Document the
19207 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19208 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19209 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19210 the macros defined by GCC if the newer GLIBC is available.
19212 2017-07-12 Jeff Law <law@redhat.com>
19214 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19215 remaining includes slightly.
19216 * config/riscv/riscv-builtins.c: Include profile-count.h.
19218 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19221 * config/avr/avr.c (avr_set_current_function): In diagnostic
19222 messages: Quote keywords and (parts of) identifiers.
19223 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19226 2017-07-12 Carl Love <cel@us.ibm.com>
19228 * config/rs6000/rs6000-c.c: Add support for built-in functions
19229 vector bool char vec_revb (vector bool char);
19230 vector bool short vec_revb (vector short char);
19231 vector bool int vec_revb (vector bool int);
19232 vector bool long long vec_revb (vector bool long long);
19233 * doc/extend.texi: Update the built-in documentation file for the
19234 new built-in functions.
19236 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19238 * config/s390/s390.md: Remove movcc splitter.
19240 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19242 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19243 load/store on condition.
19245 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19248 * config/avr/avr.c (avr_encode_section_info)
19249 [progmem && !TREE_READONLY]: Error if progmem object needs
19252 2017-07-11 Michael Collison <michael.collison@arm.com>
19254 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19257 2017-07-11 Carl Love <cel@us.ibm.com>
19259 * config/rs6000/rs6000-c.c: Add support for builtins
19260 vector unsigned int vec_parity_lsbb (vector signed int);
19261 vector unsigned int vec_parity_lsbb (vector unsigned int);
19262 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19263 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19264 vector unsigned long long vec_parity_lsbb (vector signed long long);
19265 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19266 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19267 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19268 * doc/extend.texi: Update the built-in documentation file for the
19269 new built-in functions.
19271 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19273 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19274 (layout::m_primary_loc): New field.
19275 (layout::layout): Initialize new field. Move location filtering
19276 logic from here to...
19277 (layout::maybe_add_location_range): ...this new method. Add
19278 support for filtering to just the lines already specified by other
19280 (layout::will_show_line_p): New method.
19281 (gcc_rich_location::add_location_if_nearby): New method.
19282 (selftest::test_add_location_if_nearby): New test function.
19283 (selftest::diagnostic_show_locus_c_tests): Call it.
19284 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19287 2017-07-11 Tom de Vries <tom@codesourcery.com>
19289 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19290 (bb_first_real_insn): New function.
19291 (nvptx_single): Add extra initialization of broadcasted condition
19294 2017-07-11 Nathan Sidwell <nathan@acm.org>
19296 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19298 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19300 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19301 Binutils doc as TEXI will mess them up.
19302 * doc/invoke.texi (AVR Options): Same here.
19304 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19306 * config/sparc/sparc.opt (mfix-ut700): New option.
19307 (mfix-gr712rc): Likewise.
19308 (sparc_fix_b2bst): New variable.
19309 * doc/invoke.texi (SPARC options): Document them.
19310 (ARM options): Fix warnings.
19311 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19312 instructions to prevent sequences that can trigger the store-store
19313 errata for certain LEON3FT processors.
19314 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19315 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19316 * config/sparc/sparc.md (fix_b2bst): New attribute.
19317 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19319 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19322 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19324 (*rsqrtsf2_sse): Ditto.
19326 (div<mode>3): Macroize insn from divdf3 and divsf3
19327 using MODEF mode iterator.
19329 2017-07-10 Martin Sebor <msebor@redhat.com>
19331 PR tree-optimization/80397
19332 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19333 instead of testing for equality to INTEGER_TYPE.
19335 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19337 * config.gcc: Remove uclibc from arc target spec.
19339 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19341 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19343 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19346 * lto-wrapper.c (remove_option): New function.
19347 (merge_and_complain): Merge PIC/PIE options more realistically.
19349 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19351 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19355 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19356 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19357 * config.in: Regenerate.
19358 * configure: Regenerate.
19359 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19360 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19361 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19362 (TARGET_GASISR_PROLOGUES): ...target mask.
19363 * common/config/avr/avr-common.c
19364 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19365 Set -mgas-isr-prologues.
19366 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19367 INSERT_PASS_BEFORE for it.
19368 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19369 * config/avr/avr.c (avr_option_override)
19370 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19371 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19372 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19373 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19374 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19376 (make_avr_pass_pre_proep): New function.
19377 (emit_push_sfr) <treg>: Add argument to function and use it
19378 instead of TMP_REG.
19379 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19380 and set machine->gasisr.yes.
19381 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19382 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19383 __gcc_isr.n_pushed to .L__stack_usage.
19384 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19385 (avr_asm_final_postscan_insn): ...this new static function.
19386 * config/avr/avr.h (machine_function)
19387 <is_no_gccisr, use_L__stack_usage>: New fields.
19388 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19389 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19390 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19391 (gasisr, *gasisr): New expander and insn.
19392 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19393 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19394 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19396 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19398 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19401 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19403 Move jump-tables out of .text again.
19406 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19407 (ASM_OUTPUT_ADDR_VEC): New function.
19408 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19409 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19410 INSN_ADDRESSes as asm comment.
19411 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19412 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19413 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19414 * config/avr/avr.md (*tablejump): Adjust comment.
19415 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19416 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19418 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19419 (avr_output_addr_vec): New proto.
19420 (avr_log_t) <insn_addresses>: New field.
19422 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19425 * config/i386/i386.c (ix86_function_arg_advance): Set
19426 outgoing_args_on_stack to true if there are outgoing arguments
19428 (ix86_function_arg): Likewise.
19429 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19430 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19431 * config/i386/i386.h (machine_function): Add
19432 outgoing_args_on_stack.
19434 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19436 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19437 supporting pthreds.
19438 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19440 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19442 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19443 (REAL_H): Remove $(MACHMODE_H).
19444 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19446 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19447 $(MACHMODE_H) and double-int.h.
19448 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19450 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19453 2017-07-07 Andrew Pinski <apinski@cavium.com>
19455 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19456 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19458 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19460 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19461 Add warning if GCC was not configured to link against a GLIBC that
19462 exports the hardware capability bits.
19463 (make_resolver_func): Make resolver function private and not a
19464 COMDAT function. Create the name with clone_function_name instead
19465 of make_unique_name.
19468 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19469 correct operand in doing the split.
19471 2017-07-07 Carl Love <cel@us.ibm.com>
19473 * config/rs6000/rs6000-c: Add support for built-in function
19474 vector unsigned short vec_pack_to_short_fp32 (vector float,
19476 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19477 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19478 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19479 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19480 (convert_4f32_8i16): Add define_expand.
19481 * doc/extend.texi: Update the built-in documentation file for the
19482 new built-in function.
19484 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19486 * config/sparc/m8.md: New file.
19487 * config/sparc/sparc.md: Include m8.md.
19489 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19491 * config/sparc/sparc.opt: New option -mvis4b.
19492 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19493 (sparc_option_override): Handle VIS4B.
19494 (enum sparc_builtins): Define
19495 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19496 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19497 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19498 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19499 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19500 (check_constant_argument): New function.
19501 (sparc_vis_init_builtins): Define builtins
19502 __builtin_vis_dictunpack{8,16,32},
19503 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19504 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19505 __builtin_vis_fpcmpde{8,16,32}shl and
19506 __builtin_vis_fpcmpur{8,16,32}shl.
19507 (sparc_expand_builtin): Check that the constant operands to
19508 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19509 constant and in range.
19510 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19512 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19513 (SPARC_IMM5_P): Likewise.
19514 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19515 (enabled): Handle vis4b.
19516 (UNSPEC_DICTUNPACK): New unspec.
19517 (UNSPEC_FPCMPSHL): Likewise.
19518 (UNSPEC_FPUCMPSHL): Likewise.
19519 (UNSPEC_FPCMPDESHL): Likewise.
19520 (UNSPEC_FPCMPURSHL): Likewise.
19521 (cpu_feature): New CPU feature `vis4b'.
19522 (dictunpack{8,16,32}): New insns.
19523 (FPCSMODE): New mode iterator.
19524 (fpcscond): New code iterator.
19525 (fpcsucond): Likewise.
19526 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19527 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19528 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19529 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19530 * config/sparc/constraints.md: Define constraints `q' for unsigned
19531 2-bit integer constants and `t' for unsigned 5-bit integer
19533 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19535 (imm5_operand_dictunpack16): Likewise.
19536 (imm5_operand_dictunpack32): Likewise.
19537 (imm2_operand): Likewise.
19538 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19539 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19540 ditunpack* and fpcmp*shl builtins.
19542 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19544 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19545 * config.in: Add HAVE_AS_SPARC6 define.
19546 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19548 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19550 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19551 (CPP_CPU_SPEC): Handle m8.
19552 (ASM_CPU_SPEC): Likewise.
19553 * config/sparc/sparc-opts.h (enum processor_type): Add
19555 * config/sparc/sparc.c (m8_costs): New struct.
19556 (sparc_option_override): Handle TARGET_CPU_m8.
19557 (sparc32_initialize_trampoline): Likewise.
19558 (sparc64_initialize_trampoline): Likewise.
19559 (sparc_issue_rate): Likewise.
19560 (sparc_register_move_cost): Likewise.
19561 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19562 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19563 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19564 (CPP_CPU_SPEC): Handle M8.
19565 (ASM_CPU_SPEC): Likewise.
19566 (AS_M8_FLAG): Define.
19567 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19568 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19569 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19571 * configure: Regenerate.
19572 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19575 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19577 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19579 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19580 ("*movdi_insn_sp32"): Do not set v3pipe.
19581 ("*movsi_insn"): Likewise.
19582 ("*movdi_insn_sp64"): Likewise.
19583 ("*movsf_insn"): Likewise.
19584 ("*movdf_insn_sp32"): Likewise.
19585 ("*movdf_insn_sp64"): Likewise.
19586 ("*zero_extendsidi2_insn_sp64"): Likewise.
19587 ("*sign_extendsidi2_insn"): Likewise.
19588 ("*mov<VM32:mode>_insn"): Likewise.
19589 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19590 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19591 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19592 ("<vlop:code><VL:mode>3"): Likewise.
19593 ("*not_<vlop:code><VL:mode>3"): Likewise.
19594 ("*nand<VL:mode>_vis"): Likewise.
19595 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19596 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19597 ("one_cmpl<VL:mode>2"): Likewise.
19598 ("faligndata<VM64:mode>_vis"): Likewise.
19599 ("alignaddrsi_vis"): Likewise.
19600 ("alignaddrdi_vis"): Likweise.
19601 ("alignaddrlsi_vis"): Likewise.
19602 ("alignaddrldi_vis"): Likewise.
19603 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19604 ("bmaskdi_vis"): Likewise.
19605 ("bmasksi_vis"): Likewise.
19606 ("bshuffle<VM64:mode>_vis"): Likewise.
19607 ("cmask8<P:mode>_vis"): Likewise.
19608 ("cmask16<P:mode>_vis"): Likewise.
19609 ("cmask32<P:mode>_vis"): Likewise.
19610 ("pdistn<P:mode>_vis"): Likewise.
19611 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19613 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19615 * config/sparc/sparc.md ("subtype"): New insn attribute.
19616 ("*wrgsr_sp64"): Set insn subtype.
19617 ("*rdgsr_sp64"): Likewise.
19618 ("alignaddrsi_vis"): Likewise.
19619 ("alignaddrdi_vis"): Likewise.
19620 ("alignaddrlsi_vis"): Likewise.
19621 ("alignaddrldi_vis"): Likewise.
19622 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19623 ("fexpand_vis"): Likewise.
19624 ("fpmerge_vis"): Likewise.
19625 ("faligndata<VM64:mode>_vis"): Likewise.
19626 ("bshuffle<VM64:mode>_vis"): Likewise.
19627 ("cmask8<P:mode>_vis"): Likewise.
19628 ("cmask16<P:mode>_vis"): Likewise.
19629 ("cmask32<P:mode>_vis"): Likewise.
19630 ("fchksm16_vis"): Likewise.
19631 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19632 ("fmean16_vis"): Likewise.
19633 ("fp<plusminus_insn>64_vis"): Likewise.
19634 ("<plusminus_insn>v8qi3"): Likewise.
19635 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19636 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19637 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19638 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19639 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19640 ("*movqi_insn"): Likewise.
19641 ("*movhi_insn"): Likewise.
19642 ("*movsi_insn"): Likewise.
19643 ("movsi_pic_gotdata_op"): Likewise.
19644 ("*movdi_insn_sp32"): Likewise.
19645 ("*movdi_insn_sp64"): Likewise.
19646 ("movdi_pic_gotdata_op"): Likewise.
19647 ("*movsf_insn"): Likewise.
19648 ("*movdf_insn_sp32"): Likewise.
19649 ("*movdf_insn_sp64"): Likewise.
19650 ("*zero_extendhisi2_insn"): Likewise.
19651 ("*zero_extendqihi2_insn"): Likewise.
19652 ("*zero_extendqisi2_insn"): Likewise.
19653 ("*zero_extendqidi2_insn"): Likewise.
19654 ("*zero_extendhidi2_insn"): Likewise.
19655 ("*zero_extendsidi2_insn_sp64"): Likewise.
19656 ("ldfsr"): Likewise.
19657 ("prefetch_64"): Likewise.
19658 ("prefetch_32"): Likewise.
19659 ("tie_ld32"): Likewise.
19660 ("tie_ld64"): Likewise.
19661 ("*tldo_ldub_sp32"): Likewise.
19662 ("*tldo_ldub1_sp32"): Likewise.
19663 ("*tldo_ldub2_sp32"): Likewise.
19664 ("*tldo_ldub_sp64"): Likewise.
19665 ("*tldo_ldub1_sp64"): Likewise.
19666 ("*tldo_ldub2_sp64"): Likewise.
19667 ("*tldo_ldub3_sp64"): Likewise.
19668 ("*tldo_lduh_sp32"): Likewise.
19669 ("*tldo_lduh1_sp32"): Likewise.
19670 ("*tldo_lduh_sp64"): Likewise.
19671 ("*tldo_lduh1_sp64"): Likewise.
19672 ("*tldo_lduh2_sp64"): Likewise.
19673 ("*tldo_lduw_sp32"): Likewise.
19674 ("*tldo_lduw_sp64"): Likewise.
19675 ("*tldo_lduw1_sp64"): Likewise.
19676 ("*tldo_ldx_sp64"): Likewise.
19677 ("*mov<VM32:mode>_insn"): Likewise.
19678 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19679 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19681 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19683 * config/sparc/sparc.md ("type"): New insn type viscmp.
19684 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19686 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19687 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19688 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19689 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19691 ("n7_vis_logical_11cycle"): Likewise.
19692 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19693 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19694 * config/sparc/niagara.md ("niag_vis"): Likewise.
19695 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19696 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19698 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19700 * config/sparc/sparc.md: New instruction type `bmask'.
19701 (bmaskdi_vis): Use the `bmask' type.
19702 (bmasksi_vis): Likewise.
19703 * config/sparc/ultra3.md (us3_array): Likewise.
19704 * config/sparc/niagara7.md (n7_array): Likewise.
19705 * config/sparc/niagara4.md (n4_array): Likewise.
19706 * config/sparc/niagara2.md (niag2_vis): Likewise.
19707 (niag3_vis): Likewise.
19708 * config/sparc/niagara.md (niag_vis): Likewise.
19710 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19712 * ipa-comdats.c: Remove optimize check from gate.
19713 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19714 for functions not optimized.
19715 (ipa_fn_summary_read): Skip optimize check.
19716 (ipa_fn_summary_write): Likewise.
19717 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19719 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19721 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19722 (check_callers): Give up on caller which is not optimized.
19723 (inline_small_functions): Likewise.
19724 (ipa_inline): Do not give up when not optimizing.
19725 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19726 away unoptimizes cdtors.
19727 (whole_program_function_and_variable_visibility): Do
19728 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19729 * ipa.c (process_references): Do not check optimize.
19730 (symbol_table::remove_unreachable_nodes): Update optimize check.
19731 (set_writeonly_bit): Update optimize check.
19732 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19733 (pass_ipa_single_use::gate): Remove.
19735 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19737 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19738 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19739 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19740 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19741 permute_load, permute_store, adjust_extract, adjust_splat,
19742 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19743 replace_swap_with_copy, dump_swap_insn_table,
19744 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19745 recombine_lvx_pattern, recombine_stvx_pattern,
19746 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19747 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19748 to file rs6000-p8swap.c.
19749 * config/rs6000/rs6000-p8swap.c: New file.
19750 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19751 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19752 and rs6000*-*-* targets.
19754 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19756 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19758 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19760 * lto-wrapper.c (merge_and_complain): Do not merge
19761 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19762 fsigned_zeros, ftrapping_math, fwrapv.
19763 (append_compiler_options): Do not track these options.
19764 (append_linker_options): Likewie
19766 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19768 * cgraphunit.c (cgraph_node::finalize_function): When
19769 !flag_toplevel_reorde set no_reorder flag.
19770 (varpool_node::finalize_decl): Likewise.
19771 (symbol_table::compile): Drop no toplevel reorder path.
19773 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19775 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19776 edges; zero probability is not better than uninitialized.
19778 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19780 * asan.h (asan_sanitize_allocas_p): Declare.
19781 * asan.c (asan_sanitize_allocas_p): New function.
19782 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19783 (handle_builtin_alloca): Likewise.
19784 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19785 if !asan_sanitize_allocas_p.
19786 * params.def (asan-instrument-allocas): Add new option.
19787 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19788 * opts.c (common_handle_option): Disable allocas sanitization for
19791 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19793 * asan.c: Include gimple-fold.h.
19794 (get_last_alloca_addr): New function.
19795 (handle_builtin_stackrestore): Likewise.
19796 (handle_builtin_alloca): Likewise.
19797 (asan_emit_allocas_unpoison): Likewise.
19798 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19799 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19800 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19801 (instrument_builtin_call): Pass gimple iterator to
19802 get_mem_refs_of_builtin_call.
19803 (last_alloca_addr): New global.
19804 * asan.h (asan_emit_allocas_unpoison): Declare.
19805 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19806 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19807 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19808 if function calls alloca.
19809 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19810 * gimple-fold.h (replace_call_with_value): Declare.
19811 * internal-fn.c: Include asan.h.
19812 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19813 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19815 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19817 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19818 (C_SELFTEST_FLAGS): New.
19819 (CPP_SELFTEST_FLAGS): New.
19820 (SELFTEST_DEPS): New, from deps of s-selftest.
19821 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19822 (CPP_SELFTEST_DEPS): New.
19823 (selftest): Add dependency on s-selftest-c++.
19824 (s-selftest): Rename to...
19825 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19826 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19827 than SELFTEST_FLAGS.
19828 (selftest-gdb): Rename to...
19829 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19831 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19832 (selftest-valgrind): Rename to...
19833 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19835 (selftest-valgrind): Reintroduce as an alias for
19836 selftest-c-valgrind.
19837 (s-selftest-c++): New.
19838 (selftest-c++-gdb): New.
19839 (selftest-c++-valgrind): New.
19841 2017-07-06 Olivier Hainque <hainque@adacore.com>
19843 * gcc.c (process_command): When deciding if undefined variables
19844 should be ignored when processing specs, accept "gcc -v" as well.
19846 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19848 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19849 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19851 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19853 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19854 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19855 * config/arm/arm-tables.opt: Regenerate.
19856 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19858 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19860 2017-07-06 Carl Love <cel@us.ibm.com>
19862 * ChangeLog: Clean up from mid air collision
19864 2017-07-06 Carl Love <cel@us.ibm.com>
19866 * config/rs6000/rs6000-c.c: Add support for built-in functions
19867 vector signed int vec_subc (vector signed int, vector signed int);
19868 vector signed __int128 vec_subc (vector signed __int128,
19869 vector signed __int128);
19870 vector unsigned __int128 vec_subc (vector unsigned __int128,
19871 vector unsigned __int128);
19872 vector signed int vec_sube (vector signed int, vector signed int,
19873 vector signed int);
19874 vector unsigned int vec_sube (vector unsigned int,
19875 vector unsigned int,
19876 vector unsigned int);
19877 vector signed __int128 vec_sube (vector signed __int128,
19878 vector signed __int128,
19879 vector signed__int128);
19880 vector unsigned __int128 vec_sube (vector unsigned __int128,
19881 vector unsigned __int128,
19882 vector unsigned __int128);
19883 vector signed int vec_subec (vector signed int, vector signed int,
19884 vector signed int);
19885 vector unsigned int vec_subec (vector unsigned int,
19886 vector unsigned int,
19887 vector unsigned int);
19888 vector signed __int128 vec_subec (vector signed __int128,
19889 vector signed __int128,
19890 vector signed__int128);
19891 vector unsigned __int128 vec_subec (vector unsigned __int128,
19892 vector unsigned __int128,
19893 vector unsigned __int128);
19894 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19895 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19896 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19897 BU_ALTIVEC_OVERLOAD_X definitions.
19898 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19899 * doc/extend.texi: Update the built-in documentation file for the new
19900 built-in functions.
19902 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19905 * diagnostic-show-locus.c (layout::layout): Use start and finish
19906 spelling location for the start and finish of each range.
19907 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19908 Add unused aspect param.
19909 * input.c (expand_location_1): Add "aspect" param, and use it
19910 to access the correct part of the location.
19911 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19913 (expand_location_to_spelling_point): Likewise.
19914 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19915 param, and pass it to expand_location_1.
19917 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19919 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19920 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19921 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19922 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19923 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19924 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19925 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19926 _mm_maskz_getmant_ss): New intrinsics.
19927 (__builtin_ia32_getexpss128_mask): Changed to ...
19928 __builtin_ia32_getexpss128_round ... this.
19929 (__builtin_ia32_getexpsd128_mask): Changed to ...
19930 __builtin_ia32_getexpsd128_round ... this.
19931 * config/i386/i386-builtin-types.def
19932 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19933 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19934 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19935 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19936 __builtin_ia32_getmantss_mask_round): New builtins.
19937 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19938 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19939 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19940 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19941 * config/i386/sse.md
19942 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19943 avx512f_sgetexp<mode><mask_scalar_name>
19944 <round_saeonly_scalar_name> ... this.
19945 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19946 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19947 vgetexp<ssescalarmodesuffix>
19948 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19949 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19950 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19951 avx512f_vgetmant<mode><mask_scalar_name>
19952 <round_saeonly_scalar_name> ... this.
19953 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19954 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19955 vgetmant<ssescalarmodesuffix>
19956 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19957 %0<mask_scalar_operand4>, %1, %2
19958 <round_saeonly_scalar_mask_op4>, %3} ... this.
19959 * config/i386/subst.md (mask_scalar_operand4,
19960 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19961 round_saeonly_scalar_nimm_predicate): New subst attributes.
19963 2017-07-06 Julia Koval <julia.koval@intel.com>
19965 * config/i386/i386.c (ix86_erase_embedded_rounding):
19966 Remove code for old rounding pattern.
19968 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19970 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19972 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19974 * doc/sourcebuild.texi (Test Directives, Variants of
19975 dg-require-support): Add documentation for dg-require-stack-check.
19977 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19979 * config/i386/subst.md (mask_scalar, round_scalar,
19980 round_saeonly_scalar): New meta-templates.
19981 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19982 round_scalar_mask_operand3, round_scalar_mask_op3,
19983 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19984 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19985 round_saeonly_scalar_constraint,
19986 round_saeonly_scalar_prefix): New subst attribute.
19987 * config/i386/sse.md
19988 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19989 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19990 <round_scalar_name> ... this.
19991 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19992 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19993 <round_scalar_name> ... this.
19994 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19995 <sse>_vm<code><mode>3<mask_scalar_name>
19996 <round_saeonly_scalar_name> ... this.
19997 (v<plusminus_mnemonic><ssescalarmodesuffix>
19998 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19999 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20000 v<plusminus_mnemonic><ssescalarmodesuffix>
20001 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20002 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20003 (v<multdiv_mnemonic><ssescalarmodesuffix>
20004 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20005 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20006 v<multdiv_mnemonic><ssescalarmodesuffix>
20007 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20008 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20009 (v<maxmin_float><ssescalarmodesuffix>
20010 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20011 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20012 v<maxmin_float><ssescalarmodesuffix>
20013 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20014 %0<mask_scalar_operand3>, %1, %<iptr>2
20015 <round_saeonly_scalar_mask_op3>} ... this.
20017 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20019 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20020 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20022 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20023 Alan Hayward <alan.hayward@arm.com>
20024 David Sherwood <david.sherwood@arm.com>
20026 * combine.c (simplify_if_then_else): Remove "enum" before
20028 * compare-elim.c (can_eliminate_compare): Likewise.
20029 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20031 (aarch64_lookup_simd_builtin_type): Likewise.
20032 (aarch64_simd_builtin_type): Likewise.
20033 (aarch64_init_simd_builtin_types): Likewise.
20034 (aarch64_simd_expand_args): Likewise.
20035 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20037 (aarch64_reverse_mask): Likewise.
20038 (aarch64_simd_emit_reg_reg_move): Likewise.
20039 (aarch64_gen_adjusted_ldpstp): Likewise.
20040 (aarch64_ccmp_mode_to_code): Likewise.
20041 (aarch64_operands_ok_for_ldpstp): Likewise.
20042 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20043 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20045 (aarch64_min_divisions_for_recip_mul): Likewise.
20046 (aarch64_reassociation_width): Likewise.
20047 (aarch64_get_condition_code_1): Likewise.
20048 (aarch64_simd_emit_reg_reg_move): Likewise.
20049 (aarch64_simd_attr_length_rglist): Likewise.
20050 (aarch64_reverse_mask): Likewise.
20051 (aarch64_operands_ok_for_ldpstp): Likewise.
20052 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20053 (aarch64_gen_adjusted_ldpstp): Likewise.
20054 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20056 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20057 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20058 (arm_lookup_simd_builtin_type): Likewise.
20059 (arm_simd_builtin_type): Likewise.
20060 (arm_init_simd_builtin_types): Likewise.
20061 (arm_expand_builtin_args): Likewise.
20062 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20063 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20064 (ft32_setup_incoming_varargs): Likewise.
20065 (ft32_function_arg): Likewise.
20066 (ft32_function_arg_advance): Likewise.
20067 (ft32_pass_by_reference): Likewise.
20068 (ft32_arg_partial_bytes): Likewise.
20069 (ft32_valid_pointer_mode): Likewise.
20070 (ft32_addr_space_pointer_mode): Likewise.
20071 (ft32_addr_space_legitimate_address_p): Likewise.
20072 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20074 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20075 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20076 (iamcu_alignment): Likewise.
20077 (canonicalize_vector_int_perm): Likewise.
20078 (ix86_noce_conversion_profitable_p): Likewise.
20079 (ix86_mpx_bound_mode): Likewise.
20080 (ix86_operands_ok_for_move_multiple): Likewise.
20081 * config/microblaze/microblaze-protos.h
20082 (microblaze_expand_conditional_branch_reg): Likewise.
20083 * config/microblaze/microblaze.c
20084 (microblaze_expand_conditional_branch_reg): Likewise.
20085 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20087 (rs6000_reassociation_width): Likewise.
20088 (rs6000_invalid_binary_op): Likewise.
20089 (fusion_p9_p): Likewise.
20090 (emit_fusion_p9_load): Likewise.
20091 (emit_fusion_p9_store): Likewise.
20092 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20094 (riscv_hard_regno_mode_ok_p): Likewise.
20095 (riscv_address_insns): Likewise.
20096 (riscv_split_symbol): Likewise.
20097 (riscv_legitimize_move): Likewise.
20098 (riscv_function_value): Likewise.
20099 (riscv_hard_regno_nregs): Likewise.
20100 (riscv_expand_builtin): Likewise.
20101 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20102 (riscv_build_integer): Likewise.
20103 (riscv_split_integer): Likewise.
20104 (riscv_legitimate_constant_p): Likewise.
20105 (riscv_cannot_force_const_mem): Likewise.
20106 (riscv_regno_mode_ok_for_base_p): Likewise.
20107 (riscv_valid_base_register_p): Likewise.
20108 (riscv_valid_offset_p): Likewise.
20109 (riscv_valid_lo_sum_p): Likewise.
20110 (riscv_classify_address): Likewise.
20111 (riscv_legitimate_address_p): Likewise.
20112 (riscv_address_insns): Likewise.
20113 (riscv_load_store_insns): Likewise.
20114 (riscv_force_binary): Likewise.
20115 (riscv_split_symbol): Likewise.
20116 (riscv_force_address): Likewise.
20117 (riscv_legitimize_address): Likewise.
20118 (riscv_move_integer): Likewise.
20119 (riscv_legitimize_const_move): Likewise.
20120 (riscv_legitimize_move): Likewise.
20121 (riscv_address_cost): Likewise.
20122 (riscv_subword): Likewise.
20123 (riscv_output_move): Likewise.
20124 (riscv_canonicalize_int_order_test): Likewise.
20125 (riscv_emit_int_order_test): Likewise.
20126 (riscv_function_arg_boundary): Likewise.
20127 (riscv_pass_mode_in_fpr_p): Likewise.
20128 (riscv_pass_fpr_single): Likewise.
20129 (riscv_pass_fpr_pair): Likewise.
20130 (riscv_get_arg_info): Likewise.
20131 (riscv_function_arg): Likewise.
20132 (riscv_function_arg_advance): Likewise.
20133 (riscv_arg_partial_bytes): Likewise.
20134 (riscv_function_value): Likewise.
20135 (riscv_pass_by_reference): Likewise.
20136 (riscv_setup_incoming_varargs): Likewise.
20137 (riscv_print_operand): Likewise.
20138 (riscv_elf_select_rtx_section): Likewise.
20139 (riscv_save_restore_reg): Likewise.
20140 (riscv_for_each_saved_reg): Likewise.
20141 (riscv_register_move_cost): Likewise.
20142 (riscv_hard_regno_mode_ok_p): Likewise.
20143 (riscv_hard_regno_nregs): Likewise.
20144 (riscv_class_max_nregs): Likewise.
20145 (riscv_memory_move_cost): Likewise.
20146 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20147 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20148 (rl78_addr_space_address_mode): Likewise.
20149 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20151 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20152 (rs6000_reassociation_width): Likewise.
20153 (rs6000_invalid_binary_op): Likewise.
20154 (fusion_p9_p): Likewise.
20155 (emit_fusion_p9_load): Likewise.
20156 (emit_fusion_p9_store): Likewise.
20157 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20158 (ok_for_simple_move_operands): Likewise.
20159 (ok_for_simple_move_strict_operands): Likewise.
20160 (ok_for_simple_arith_logic_operands): Likewise.
20161 (visium_legitimize_reload_address): Likewise.
20162 (visium_select_cc_mode): Likewise.
20163 (output_cbranch): Likewise.
20164 (visium_split_double_move): Likewise.
20165 (visium_expand_copysign): Likewise.
20166 (visium_expand_int_cstore): Likewise.
20167 (visium_expand_fp_cstore): Likewise.
20168 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20169 (visium_function_arg): Likewise.
20170 (visium_function_arg_advance): Likewise.
20171 (visium_libcall_value): Likewise.
20172 (visium_setup_incoming_varargs): Likewise.
20173 (visium_legitimate_constant_p): Likewise.
20174 (visium_legitimate_address_p): Likewise.
20175 (visium_legitimize_address): Likewise.
20176 (visium_secondary_reload): Likewise.
20177 (visium_register_move_cost): Likewise.
20178 (visium_memory_move_cost): Likewise.
20179 (prepare_move_operands): Likewise.
20180 (ok_for_simple_move_operands): Likewise.
20181 (ok_for_simple_move_strict_operands): Likewise.
20182 (ok_for_simple_arith_logic_operands): Likewise.
20183 (visium_function_value_1): Likewise.
20184 (rtx_ok_for_offset_p): Likewise.
20185 (visium_legitimize_reload_address): Likewise.
20186 (visium_split_double_move): Likewise.
20187 (visium_expand_copysign): Likewise.
20188 (visium_expand_int_cstore): Likewise.
20189 (visium_expand_fp_cstore): Likewise.
20190 (visium_split_cstore): Likewise.
20191 (visium_select_cc_mode): Likewise.
20192 (visium_split_cbranch): Likewise.
20193 (output_cbranch): Likewise.
20194 (visium_print_operand_address): Likewise.
20195 * expmed.c (flip_storage_order): Likewise.
20196 * expmed.h (emit_cstore): Likewise.
20197 (flip_storage_order): Likewise.
20198 * genrecog.c (validate_pattern): Likewise.
20199 * hsa-gen.c (gen_hsa_addr): Likewise.
20200 * internal-fn.c (expand_arith_overflow): Likewise.
20201 * ira-color.c (allocno_copy_cost_saving): Likewise.
20202 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20203 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20204 (process_invariant_for_inheritance): Likewise.
20205 * lra-eliminations.c (move_plus_up): Likewise.
20206 * omp-low.c (lower_oacc_reductions): Likewise.
20207 * simplify-rtx.c (simplify_subreg): Likewise.
20208 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20209 (TARGET_CHKP_BOUND_MODE): Likewise..
20210 * targhooks.c (default_chkp_bound_mode): Likewise.
20211 (default_setup_incoming_vararg_bounds): Likewise.
20212 * targhooks.h (default_chkp_bound_mode): Likewise.
20213 (default_setup_incoming_vararg_bounds): Likewise.
20214 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20215 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20216 (have_whole_vector_shift): Likewise.
20217 * tree-vect-stmts.c (vectorizable_load): Likewise.
20218 * doc/tm.texi: Regenerate.
20220 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20222 Graceful degrade if Binutils PR21472 is not available.
20225 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20226 .rodata in flash test fails.
20227 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20228 * confgure: Regenerate.
20229 * config.in: Regenerate.
20230 * config/avr/avr.c (avr_asm_named_section)
20231 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20232 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20233 (avr_asm_init_sections): Same.
20235 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20237 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20238 (fma<VH:mode>4_intrinsic): Likewise.
20239 (*fmsub<VCVTF:mode>4): Likewise.
20240 (*fmsub<VH:mode>4_intrinsic): Likewise.
20242 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20245 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20246 Don't depend on "optimize > 0".
20247 (out_movhi_r_mr, out_movqi_mr_r): Same.
20248 (out_movhi_mr_r, out_movqi_r_mr): Same.
20249 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20250 io_address_operand on "optimize > 0".
20252 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20254 * tree-loop-distribution.c: Add general explanantion on the pass.
20255 (generate_loops_for_partition): Mark distributed loop.
20256 (pg_add_dependence_edges): New parameter. Handle alias data
20257 dependence specially and record it in the parameter if asked.
20258 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20259 (init_partition_graph_vertices, add_partition_graph_edge): New.
20260 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20261 (free_partition_graph_vdata, build_partition_graph): New.
20262 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20263 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20264 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20265 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20266 (version_for_distribution_p, finalize_partitions): New.
20267 (distribute_loop): Handle alias data dependence specially. Factor
20268 out loop fusion code as functions and call these functions.
20270 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20272 * tree-loop-distribution.c (classify_partition): New parameter and
20273 better handle reduction statement.
20274 (rdg_build_partitions): Revise comment.
20275 (distribute_loop): Compute statements in all partitions and pass it
20276 to classify_partition.
20278 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20280 * tree-loop-distribution.c (enum partition_type): New.
20281 (struct partition): New field type.
20282 (partition_merge_into): Add parameter. Update partition type.
20283 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20284 (build_rdg_partition_for_vertex): Compute partition type.
20285 (rdg_build_partitions): Dump partition type.
20286 (distribute_loop): Update calls to partition_merge_into.
20288 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20290 * tree-loop-distribution.c (struct ddr_hasher): New.
20291 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20293 (classify_partition): Call get_data_dependence.
20294 (pg_add_dependence_edges): Ditto.
20295 (distribute_loop): Release data dependence hash table.
20297 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20299 * tree-loop-distribution.c (ref_base_address): Delete.
20300 (similar_memory_accesses): Rename ...
20301 (share_memory_accesses): ... to this. Check if partitions access
20302 the same memory reference.
20303 (distribute_loop): Call share_memory_accesses.
20305 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20307 * tree-loop-distribution.c (struct partition): New field recording
20308 its data reference.
20309 (partition_alloc, partition_free): Init and release data refs.
20310 (partition_merge_into): Merge data refs.
20311 (build_rdg_partition_for_vertex): Collect data refs for partition.
20312 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20314 (distribute_loop): Remve data refs from vertice data of partition
20317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20319 * tree-loop-distribution.c (params.h): Include header file.
20320 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20321 (datarefs_vec): New global var.
20322 (create_rdg_vertices): Use datarefs_vec directly.
20323 (free_rdg): Don't free data references.
20324 (build_rdg): Update use. Don't free data references.
20325 (distribute_loop): Compute global variable for data references.
20326 Bail out if there are too many data references.
20328 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20330 * tree-loop-distribution.c (loop_nest): New global var.
20331 (build_rdg): Use loop directly, rather than loop nest.
20332 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20334 (distribute_loop): Compute global variable loop nest. Update use.
20336 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20338 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20339 (partition_merge_into): New parameter. Dump reason for fusion.
20340 (distribute_loop): Update use of partition_merge_into.
20342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20344 * tree-loop-distribution.c (bb_top_order_index): New.
20345 (bb_top_order_index_size, bb_top_order_cmp): New.
20346 (stmts_from_loop): Use topological order.
20347 (pass_loop_distribution::execute): Compute and release topological
20348 order for basic blocks.
20350 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20352 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20355 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20357 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20358 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20359 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20360 * internal-fn.def (LOOP_DIST_ALIAS): New.
20361 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20362 (fold_loop_internal_call): ... this.
20363 (vect_loop_dist_alias_call): New function.
20364 (set_uid_loop_bbs): Call fold_loop_internal_call.
20365 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20368 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20371 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20372 Require dead FLAGS_REG at the beginning of a peephole.
20374 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20377 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20378 arguments in the call to __builtin_ia32_sbb_u32.
20379 (_subborrow_u64): Swap _X and _Y arguments in the call to
20380 __builtin_ia32_sbb_u64.
20382 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20385 * tree-vrp.c (compare_assert_loc): Turn into a function template
20386 with stable template parameter. Only test if a->e is NULL,
20387 !a->e == !b->e has been verified already. Use e == NULL or
20388 e != NULL instead of e or ! e tests. If stable is true, don't use
20389 iterative_hash_expr, on the other side allow a or b or both NULL
20390 and sort the NULLs last.
20391 (process_assert_insertions): Sort using compare_assert_loc<false>
20392 instead of compare_assert_loc, later sort using
20393 compare_assert_loc<true> before calling process_assert_insertions_for
20394 in a loop. Use break instead of continue once seen NULL pointer.
20396 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20398 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20399 Cortex-R7 and Cortex-R8 processors.
20401 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20403 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20404 uninitialized while src is not.
20406 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20408 * common/config/arm/arm-common.c: Adjust include path for
20410 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20411 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20412 (arm-cpu-data.h): Likewise.
20413 (arm-cpu-cdata.h): Likewise.
20414 * config/arm/arm-cpu.h: Delete.
20415 * config/arm/arm-cpu-cdata.h: Delete.
20416 * config/arm/arm-cpu-data.h: Delete.
20418 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20420 * config/arm/arm-cpus.in (cortex-a55): New.
20421 (cortex-a75): Likewise.
20422 (cortex-a75.cortex-a55): Likewise.
20423 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20425 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20426 * config/arm/arm-cpu-cdata.h: Regenerate.
20427 * config/arm/arm-cpu-data.h: Regenerate.
20428 * config/arm/arm-cpu.h: Regenerate.
20429 * config/arm/arm-tables.opt: Regenerate.
20430 * config/arm/arm-tune.md: Regenerate.
20432 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20434 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20436 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20438 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20441 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20443 PR tree-optimization/81292
20444 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20445 full_string_p, also call adjust_related_strinfos if the adjustment
20446 is simple, otherwise invalidate related strinfos.
20448 2017-07-04 Martin Liska <mliska@suse.cz>
20451 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20452 newly created variable as DECL_IGNORED_P.
20454 2017-07-04 Martin Liska <mliska@suse.cz>
20457 * ipa-inline.c (inline_small_functions):
20458 Use xstrdup_for_dump.
20460 2017-07-04 Tom de Vries <tom@codesourcery.com>
20462 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20464 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20467 * config/darwin.c (darwin_function_switched_text_sections):
20470 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20472 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20474 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20476 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20478 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20480 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20481 min_profitable_iters, and th as inclusive lower bounds.
20482 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20483 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20484 for min_profitable_iters and min_profitable_estimate.
20485 (vect_transform_loop): Treat th as an inclusive lower bound.
20486 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20488 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20491 * config/darwin.c (darwin_function_switched_text_sections):
20492 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20493 in two pieces, and suppress the use of buf.
20495 2017-07-03 Nathan Sidwell <nathan@acm.org>
20497 * hash-table.h (hash_table_mod1): Fix indentation.
20499 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20501 PR middle-end/81290
20502 * predict.c (force_edge_cold): Be more careful about propagation
20504 * profile-count.h (profile_probability::guessed,
20505 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20507 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20509 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20511 * doc/invoke.texi (rcpc architecture extension): Document it.
20513 2017-07-03 Richard Biener <rguenther@suse.de>
20515 PR tree-optimization/60510
20516 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20517 the scalar reduction PHI and use it.
20518 (vectorizable_reduction): Properly guard the single_defuse_cycle
20519 path for non-SLP reduction chains where we cannot use it.
20520 Rework reduc_def/index and vector type deduction. Rework
20521 vector operand gathering during reduction op code-gen.
20522 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20523 chains dissolve the chain and leave it to non-SLP reduction
20526 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20528 * tree-data-ref.h (dr_alignment): Declare.
20529 * tree-data-ref.c (dr_alignment): New function.
20530 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20531 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20533 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20535 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20537 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20538 and base_misalignment fields.
20539 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20540 * tree-data-ref.c: Include builtins.h.
20541 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20542 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20543 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20544 * tree-vect-data-refs.c: Include tree-cfg.h.
20545 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20546 fields instead of calculating an alignment here.
20547 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20548 innermost_loop_behavior fields.
20550 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20552 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20554 (DR_STEP_ALIGNMENT): New macro.
20555 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20556 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20557 (create_data_ref): Print it.
20558 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20559 to tell whether the step preserves vector (mis)alignment.
20560 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20561 Move the check for an integer step and generalise to all INTEGER_CST.
20562 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20563 Print the outer step alignment.
20565 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20567 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20568 with offset_alignment.
20569 (DR_ALIGNED_TO): Delete.
20570 (DR_OFFSET_ALIGNMENT): New macro.
20571 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20572 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20573 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20574 (create_data_ref): Likewise.
20575 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20576 (vect_analyze_data_refs): Likewise.
20577 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20578 creating dummy innermost behavior.
20580 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20582 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20583 with a "innermost_loop_behavior *" and refeence tree.
20584 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20585 (create_data_ref): Update call accordingly.
20586 * tree-predcom.c (find_looparound_phi): Likewise.
20588 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20590 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20591 fields with dr_wrt_vec_loop.
20592 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20593 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20594 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20595 (vect_dr_behavior): New function.
20596 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20597 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20598 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20599 track whether the step preserves the misalignment.
20600 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20601 Use vect_dr_behavior.
20602 (vect_setup_realignment): Update call accordingly.
20603 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20604 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20605 call to vect_create_addr_base_for_vector_ref.
20606 (vect_create_cond_for_align_checks): Likewise.
20607 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20608 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20609 (vect_recog_mask_conversion_pattern): Likewise.
20610 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20611 (new_stmt_vec_info): Remove redundant zeroing.
20613 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20615 * common/config/arm/arm-common.c (arm_be8_option): New function.
20616 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20617 (ISA_ARMv6): Add isa_bit_be8.
20618 * config/arm/arm.h (arm_be8_option): Add prototype.
20619 (BE8_SPEC_FUNCTION): New define.
20620 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20621 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20622 (mlittle-endian): Similarly.
20623 (mbe8, mbe32): New options.
20624 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20625 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20627 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20629 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20631 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20633 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20634 (cleanup_tree_cfg_bb): Use it.
20635 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20637 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20639 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20642 * loop-doloop.c (add_test): Update profile.
20644 2017-07-03 Martin Liska <mliska@suse.cz>
20647 * sanopt.c (rewrite_usage_of_param): New function.
20648 (sanitize_rewrite_addressable_params): Likewise.
20649 (pass_sanopt::execute): Call rewrite_usage_of_param.
20651 2017-07-03 Richard Biener <rguenther@suse.de>
20653 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20654 back to using VIEW_CONVERT_EXPR.
20656 2017-07-03 Martin Liska <mliska@suse.cz>
20659 * doc/extend.texi: Document when a resolver function is
20660 generated for target_clones.
20662 2017-07-03 Martin Liska <mliska@suse.cz>
20664 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20665 and shadow memory of auto variables which are subject of
20666 use-after-scope sanitization.
20667 (asan_expand_mark_ifn): Add do set only when is_poison.
20669 2016-07-03 Richard Biener <rguenther@suse.de>
20671 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20673 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20674 (vectorizable_reduction): Perform reduction PHI creation when
20675 visiting a reduction PHI and adjust and simplify code generation
20676 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20677 (vect_transform_loop): Visit reduction PHIs.
20678 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20679 defs into the SLP tree.
20680 (vect_build_slp_tree): Reduction defs terminate the recursion.
20681 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20683 (vect_get_vec_defs_for_stmt_copy): Export.
20684 (vect_get_vec_defs): Likewise.
20685 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20687 (vect_get_vec_defs_for_stmt_copy): Declare.
20688 (vect_get_vec_defs): Likewise.
20690 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20692 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20693 parameter with a "loop" parameter and use it instead of the
20694 loop containing DR_STMT. Don't check simple_iv when doing
20695 BB analysis. Describe the two analysis modes in the comment.
20697 2017-07-03 Tom de Vries <tom@codesourcery.com>
20699 PR tree-optimization/69468
20700 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20701 (find_same_succ_bb): Handle ignore_edge_flags.
20703 2017-07-03 Tom de Vries <tom@codesourcery.com>
20705 PR tree-optimization/81192
20706 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20708 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20710 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20712 2017-07-03 Tom de Vries <tom@codesourcery.com>
20714 PR tree-optimization/81192
20715 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20716 BB_SAME_SUCC (bb) == NULL.
20718 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20720 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20723 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20725 * dumpfile.c: Include profile-count.h
20726 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20728 (insert_cond_bb): Update profile.
20729 * tree-cfg.h (insert_cond_bb): Update prototype.
20730 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20731 * tree-dump.c: Do not include tree-cfg.
20733 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20735 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20737 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20739 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20744 * tree-complex.c (expand_complex_div_wide): update profile.
20746 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20747 Alan Hayward <alan.hayward@arm.com>
20748 David Sherwood <david.sherwood@arm.com>
20750 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20751 (CORETYPES_H): New define.
20752 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20753 (insn-modes-inline.h, s-modes-inline-h): New rules.
20754 (generated_files): Add insn-modes-inline.h.
20755 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20756 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20757 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20758 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20759 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20760 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20761 (build/gencodes.o, build/genconditions.o): Likewise.
20762 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20763 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20764 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20765 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20766 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20767 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20768 * coretypes.h: Include everything up to real.h for generators.
20769 Include insn-modes.h first. Include wide-int-print.h after
20770 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20771 * machmode.h: Don't include insn-modes.h here.
20772 * function-tests.c: Remove includes of signop.h, machmode.h,
20773 double-int.h and wide-int.h.
20775 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20777 * optc-save-gen.awk: Likewise.
20778 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20779 * godump.c: Remove include of wide-int-print.h.
20780 * pretty-print.h: Likewise.
20781 * wide-int-print.cc: Likewise.
20782 * wide-int.cc: Likewise.
20783 * hash-map-tests.c: Remove include of signop.h.
20784 * hash-set-tests.c: Likewise.
20785 * rtl-tests.c: Likewise.
20786 * mkconfig.sh: Remove include of machmode.h.
20787 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20789 (emit_insn_modes_inline_h): ...this new function. Emit the code
20790 into an insn-modes-inline.h header file, adding appropriate
20791 include guards and end comments.
20792 (emit_insn_modes_c_header): Remove include of machmode.h.
20793 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20795 (main): Handle -i flag and call emit_insn_modes_inline_h when
20798 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20800 * tree-ssa-strlen.c (strinfo): Rename the length field to
20801 nonzero_chars. Add a full_string_p field.
20802 (compare_nonzero_chars, zero_length_string_p): New functions.
20803 (get_addr_stridx): Add an offset_out parameter.
20804 Use compare_nonzero_chars.
20805 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20806 (new_strinfo): Update after above changes to strinfo.
20807 (set_endptr_and_length): Set full_string_p.
20808 (get_string_length): Update after above changes to strinfo.
20809 (unshare_strinfo): Update call to new_strinfo.
20810 (maybe_invalidate): Likewise.
20811 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20812 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20813 as a uhwi instead of an shwi. Update after above changes to
20814 strinfo and new_strinfo.
20815 (zero_length_string): Assert that chainsi contains full strings.
20816 Use zero_length_string_p. Update call to new_strinfo.
20817 (adjust_related_strinfos): Update after above changes to strinfo.
20818 Copy full_string_p from origsi.
20819 (adjust_last_stmt): Use zero_length_string_p.
20820 (handle_builtin_strlen): Update after above changes to strinfo and
20821 new_strinfo. Install the lhs as the string length if the previous
20822 entry didn't describe a full string.
20823 (handle_builtin_strchr): Update after above changes to strinfo
20825 (handle_builtin_strcpy): Likewise.
20826 (handle_builtin_strcat): Likewise.
20827 (handle_builtin_malloc): Likewise.
20828 (handle_pointer_plus): Likewise.
20829 (handle_builtin_memcpy): Likewise. Track nonzero characters
20830 that aren't necessarily followed by a nul terminator.
20831 (handle_char_store): Likewise.
20833 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20835 PR tree-optimization/80769
20836 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20837 for malloc and calloc. Document the new invariant that all related
20838 strinfos have delayed lengths or none do.
20839 (verify_related_strinfos): Move earlier in file.
20840 (set_endptr_and_length): New function, split out from...
20841 (get_string_length): ...here. Also set the lengths of related
20843 (zero_length_string): Assert that chainsi has known (rather than
20845 (adjust_related_strinfos): Likewise.
20847 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20849 PR tree-optimization/81136
20850 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20851 assert that two references with the same misalignment have the same
20852 compile-time misalignment if those compile-time misalignments
20855 2017-07-01 Andi Kleen <ak@linux.intel.com>
20857 * print-tree.c (print_node): Print all attributes.
20859 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20861 * cfg.c (scale_bbs_frequencies): New function.
20862 * cfg.h (scale_bbs_frequencies): Declare it.
20863 * cfgloopanal.c (single_likely_exit): Cleanup.
20864 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20866 (scale_loop_profile): Likewise.
20867 (loop_version): Likewise.
20868 (create_empty_loop_on_edge): Update.
20869 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20870 scale_loop_frequencies, scale_loop_profile, loopify,
20871 loop_version): Update prototypes.
20872 * modulo-sched.c (sms_schedule): Update.
20873 * predict.c (unlikely_executed_edge_p): Also check probability.
20874 (probably_never_executed_edge_p): Fix typo.
20875 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20876 * tree-parloops.c (gen_parallel_loop): Update.
20877 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20878 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20879 * tree-ssa-loop-split.c (split_loop): Update.
20880 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20881 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20882 (vect_loop_versioning): Update.
20883 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20885 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20887 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20889 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20891 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20892 to keep profile consistent.
20894 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20896 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20897 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20898 * profile-count.h (max_safe_multiplier): Make unsigned.
20899 (profile_count::guessed_zero): New.
20901 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20903 * bb-reorder.c (fix_up_crossing_landing_pad,
20904 fix_crossing_conditional_branches): Use make_single_succ_edge
20905 to keep profile consistent.
20907 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20909 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20912 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20915 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20916 the right scopes, make sure cond_jump isn't preserved between multiple
20917 iterations. Search for fallthru edge whenever there are 3+ edges and
20918 use find_fallthru_edge for it.
20920 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20922 Patch by Alexander Monakov <amonakov@ispras.ru>
20923 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20924 probabilities consistently.
20926 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20928 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20929 emit_cmp_and_jump_insns.
20931 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20934 * tree-inline.c (expand_call_inline): Combine profile statuses.
20936 2017-06-30 Andrew Pinski <apinski@cavium.com>
20938 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20939 fold_stmt returned true.
20941 2017-06-30 Nathan Sidwell <nathan@acm.org>
20943 * ggc.h (empty_string): Delete.
20944 * cfgexpand.c (expand_asm_stmt): Use plain "".
20945 * optabs.c (expand_asm_memory_barrier): Likewise.
20946 * stringpool.c (empty_string): Delete.
20947 (digit_vector, digit_string): Delete.
20948 (ggc_alloc_string): Use plain "", don't optimize single digit
20949 strings. Use ggc_alloc_atomic.
20951 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20953 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20954 comparison set and one other set, use the cost of the non-comparison
20957 2017-06-30 Nathan Sidwell <nathan@acm.org>
20959 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20962 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20964 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20965 loops. Remove now unneeded calls to gimple_switch_set_label() that
20966 just set removed labels to NULL_TREE.
20968 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20970 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20971 (set_range_info): ...here. Only call set_range_info_raw if domain
20973 (set_nonzero_bits): Call set_range_info_raw.
20974 * tree-ssanames.h (set_range_info_raw): New.
20976 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20979 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20980 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20981 of nonimmediate_operand and <store_mask_constraint> instead of m
20982 for the input operand. For V8FI iterator, always split if input
20983 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20984 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20985 <store_mask_predicate> instead of register_operand and
20986 <store_mask_constraint> instead of v for the input operand. Make
20987 sure both operands aren't MEMs for if not <mask_applied>.
20989 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20991 * lto-wrapper.c (copy_file) Close both file descriptors before
20994 2017-06-30 Martin Liska <mliska@suse.cz>
20997 * multiple_target.c (create_dispatcher_calls): Make ifunc
20998 also for function that don't have calls or are not referenced.
21000 2017-06-30 Richard Biener <rguenther@suse.de>
21002 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21003 analyze the first scalar stmt. Move vector type computation
21004 for the BB case here from ...
21005 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21006 live operation processing in the SLP case properly.
21008 2017-06-30 Richard Biener <rguenther@suse.de>
21010 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21012 2017-06-30 Martin Liska <mliska@suse.cz>
21015 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21016 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21018 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21020 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21023 2017-06-30 Richard Biener <rguenther@suse.de>
21025 PR tree-optimization/81249
21026 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21027 condition reduction result to original scalar type.
21029 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21031 * profile-count.h (enum profile_quality): Fix typos and whitespace
21034 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21036 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21037 type for branch probabilities.
21039 2017-06-29 Julian Brown <julian@codesourcery.com>
21040 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21042 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21043 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21044 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21045 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21047 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21049 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21050 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21051 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21052 CC usage from generic code to here.
21053 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21054 CC usage into the target macros.
21056 2017-06-29 Maya Rashish <coypu@sdf.org>
21058 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21061 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21063 * arm/arm-builtins.c: Include profile-count.h
21064 * except.c (sjlj_emit_function_enter): Use
21065 profile_probability::unlikely.
21067 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21069 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21070 and tocrel_offset be pointer args rather than implicitly using
21072 (legitimate_constant_pool_address_p, rs6000_emit_move,
21073 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21074 tocrel_offset and use in toc_relative_expr_p call.
21075 (print_operand, print_operand_address): Use static tocrel_base_oac
21076 and tocrel_offset_oac.
21077 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21080 2017-06-29 Maya Rashish <coypu@sdf.org>
21082 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21084 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21086 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21087 objects, take into account only the alignment of 'op0' and 'mode1' if
21090 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21092 * ccmp.c (ccmp_tree_comparison_p): New function.
21093 (ccmp_candidate_p): Update to use above function.
21094 (get_compare_parts): New function.
21095 (expand_ccmp_next): Update to use new functions.
21096 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21098 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21099 take mode as argument.
21100 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21101 * expr.c (expand_expr_real_1): Pass mode as argument.
21103 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21105 * combine.c (combine_instructions): Print insns to dump_file, together
21108 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21110 * asan.c (asan_emit_stack_protection): Update.
21111 (create_cond_insert_point): Update.
21112 * auto-profile.c (afdo_propagate_circuit): Update.
21113 * basic-block.h (struct edge_def): Turn probability to
21114 profile_probability.
21115 (EDGE_FREQUENCY): Update.
21116 * bb-reorder.c (find_traces_1_round): Update.
21117 (better_edge_p): Update.
21118 (sanitize_hot_paths): Update.
21119 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21120 (make_single_succ_edge): Update.
21121 (check_bb_profile): Update.
21122 (dump_edge_info): Update.
21123 (update_bb_profile_for_threading): Update.
21124 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21125 probabilitycount to 0.
21126 * cfgbuild.c (compute_outgoing_frequencies): Update.
21127 * cfgcleanup.c (try_forward_edges): Update.
21128 (outgoing_edges_match): Update.
21129 (try_crossjump_to_edge): Update.
21130 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21131 (expand_gimple_tailcall): Update.
21132 (construct_init_block): Use make_single_succ_edge.
21133 (construct_exit_block): Use make_single_succ_edge.
21134 * cfghooks.c (verify_flow_info): Update.
21135 (redirect_edge_succ_nodup): Update.
21136 (split_edge): Update.
21137 (account_profile_record): Update.
21138 * cfgloopanal.c (single_likely_exit): Update.
21139 * cfgloopmanip.c (scale_loop_profile): Update.
21140 (set_zero_probability): Remove.
21141 (duplicate_loop_to_header_edge): Update.
21142 * cfgloopmanip.h (loop_version): Update prototype.
21143 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21144 (force_nonfallthru_and_redirect): Update.
21145 (update_br_prob_note): Update.
21146 (rtl_verify_edges): Update.
21147 (purge_dead_edges): Update.
21148 (rtl_lv_add_condition_to_bb): Update.
21149 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21150 * cgraphunit.c (init_lowered_empty_function): Update.
21151 (cgraph_node::expand_thunk): Update.
21152 * cilk-common.c: Include profile-count.h
21153 * dojump.c (inv): Remove.
21154 (jumpifnot): Update.
21155 (jumpifnot_1): Update.
21156 (do_jump_1): Update.
21158 (do_jump_by_parts_greater_rtx): Update.
21159 (do_compare_rtx_and_jump): Update.
21160 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21161 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21162 * dwarf2cfi.c: Include profile-count.h
21163 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21164 (sjlj_emit_dispatch_table): Likewise.
21165 * explow.c: Include profile-count.h
21166 * expmed.c (emit_store_flag_force): Update.
21167 (do_cmp_and_jump): Update.
21168 * expr.c (compare_by_pieces_d::generate): Update.
21169 (compare_by_pieces_d::finish_mode): Update.
21170 (emit_block_move_via_loop): Update.
21171 (store_expr_with_bounds): Update.
21172 (store_constructor): Update.
21173 (expand_expr_real_2): Update.
21174 (expand_expr_real_1): Update.
21175 * expr.h (try_casesi, try_tablejump): Update prototypes.
21176 * gimple-pretty-print.c (dump_probability): Update.
21177 (dump_profile): New.
21178 (dump_gimple_label): Update.
21179 (dump_gimple_bb_header): Update.
21180 * graph.c (draw_cfg_node_succ_edges): Update.
21181 * hsa-gen.c (convert_switch_statements): Update.
21182 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21183 (find_if_case_1): Update.
21184 (find_if_case_2): Update.
21185 * internal-fn.c (expand_arith_overflow_result_store): Update.
21186 (expand_addsub_overflow): Update.
21187 (expand_neg_overflow): Update.
21188 (expand_mul_overflow): Update.
21189 (expand_vector_ubsan_overflow): Update.
21190 * ipa-cp.c (good_cloning_opportunity_p): Update.
21191 * ipa-split.c (split_function): Use make_single_succ_edge.
21192 * ipa-utils.c (ipa_merge_profiles): Update.
21193 * loop-doloop.c (add_test): Update.
21194 (doloop_modify): Update.
21195 * loop-unroll.c (compare_and_jump_seq): Update.
21196 (unroll_loop_runtime_iterations): Update.
21197 * lra-constraints.c (lra_inheritance): Update.
21198 * lto-streamer-in.c (input_cfg): Update.
21199 * lto-streamer-out.c (output_cfg): Update.
21200 * mcf.c (adjust_cfg_counts): Update.
21201 * modulo-sched.c (sms_schedule): Update.
21202 * omp-expand.c (expand_omp_for_init_counts): Update.
21203 (extract_omp_for_update_vars): Update.
21204 (expand_omp_ordered_sink): Update.
21205 (expand_omp_for_ordered_loops): Update.
21206 (expand_omp_for_generic): Update.
21207 (expand_omp_for_static_nochunk): Update.
21208 (expand_omp_for_static_chunk): Update.
21209 (expand_cilk_for): Update.
21210 (expand_omp_simd): Update.
21211 (expand_omp_taskloop_for_outer): Update.
21212 (expand_omp_taskloop_for_inner): Update.
21213 * omp-simd-clone.c (simd_clone_adjust): Update.
21214 * optabs.c (expand_doubleword_shift): Update.
21215 (expand_abs): Update.
21216 (emit_cmp_and_jump_insn_1): Update.
21217 (expand_compare_and_swap_loop): Update.
21218 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21219 * predict.c (predictable_edge_p): Update.
21220 (edge_probability_reliable_p): Update.
21221 (set_even_probabilities): Update.
21222 (combine_predictions_for_insn): Update.
21223 (combine_predictions_for_bb): Update.
21224 (propagate_freq): Update.
21225 (estimate_bb_frequencies): Update.
21226 (force_edge_cold): Update.
21227 * profile-count.c (profile_count::dump): Add missing space into dump.
21228 (profile_count::debug): Add newline.
21229 (profile_count::differs_from_p): Explicitly convert to unsigned.
21230 (profile_count::stream_in): Update.
21231 (profile_probability::dump): New member function.
21232 (profile_probability::debug): New member function.
21233 (profile_probability::differs_from_p): New member function.
21234 (profile_probability::differs_lot_from_p): New member function.
21235 (profile_probability::stream_in): New member function.
21236 (profile_probability::stream_out): New member function.
21237 * profile-count.h (profile_count_quality): Rename to ...
21238 (profile_quality): ... this one.
21239 (profile_probability): New.
21240 (profile_count): Update.
21241 * profile.c (compute_branch_probabilities): Update.
21242 * recog.c (peep2_attempt): Update.
21243 * sched-ebb.c (schedule_ebbs): Update.
21244 * sched-rgn.c (find_single_block_region): Update.
21245 (compute_dom_prob_ps): Update.
21246 (schedule_region): Update.
21247 * sel-sched-ir.c (compute_succs_info): Update.
21248 * stmt.c (struct case_node): Update.
21249 (do_jump_if_equal): Update.
21250 (get_outgoing_edge_probs): Update.
21251 (conditional_probability): Update.
21252 (emit_case_dispatch_table): Update.
21253 (expand_case): Update.
21254 (expand_sjlj_dispatch_table): Update.
21255 (emit_case_nodes): Update.
21256 * targhooks.c: Update.
21257 * tracer.c (better_p): Update.
21258 (find_best_successor): Update.
21259 * trans-mem.c (expand_transaction): Update.
21260 * tree-call-cdce.c: Update.
21261 * tree-cfg.c (gimple_split_edge): Upate.
21262 (move_sese_region_to_fn): Upate.
21263 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21264 * tree-eh.c (lower_resx): Upate.
21265 (cleanup_empty_eh_move_lp): Upate.
21266 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21267 * tree-inline.c (copy_edges_for_bb): Update.
21268 (copy_cfg_body): Update.
21269 * tree-parloops.c (gen_parallel_loop): Update.
21270 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21271 (gimple_gen_time_profiler): Update.
21272 * tree-ssa-dce.c (remove_dead_stmt): Update.
21273 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21274 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21275 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21276 (unloop_loops): Update.
21277 (try_peel_loop): Update.
21278 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21279 * tree-ssa-loop-split.c (connect_loops): Update.
21280 (split_loop): Update.
21281 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21282 (hoist_guard): Update.
21283 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21284 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21285 (value_replacement): Update.
21286 * tree-ssa-reassoc.c (branch_fixup): Update.
21287 * tree-ssa-tail-merge.c (replace_block_by): Update.
21288 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21289 (create_edge_and_update_destination_phis): Update.
21290 (compute_path_counts): Update.
21291 (recompute_probabilities): Update.
21292 (update_joiner_offpath_counts): Update.
21293 (freqs_to_counts_path): Update.
21294 (duplicate_thread_path): Update.
21295 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21296 (struct switch_conv_info): Update.
21297 (gen_inbound_check): Update.
21298 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21299 (vect_do_peeling): Update.
21300 (vect_loop_versioning): Update.
21301 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21302 (optimize_mask_stores): Update.
21303 * ubsan.c (ubsan_expand_null_ifn): Update.
21304 * value-prof.c (gimple_divmod_fixed_value): Update.
21305 (gimple_divmod_fixed_value_transform): Update.
21306 (gimple_mod_pow2): Update.
21307 (gimple_mod_pow2_value_transform): Update.
21308 (gimple_mod_subtract): Update.
21309 (gimple_mod_subtract_transform): Update.
21310 (gimple_ic): Update.
21311 (gimple_stringop_fixed_value): Update.
21312 (gimple_stringops_transform): Update.
21313 * value-prof.h: Update.
21315 2017-06-29 Carl Love <cel@us.ibm.com>
21317 * config/rs6000/rs6000-c.c: Add support for built-in functions
21318 vector signed int vec_signed (vector float);
21319 vector signed long long vec_signed (vector double);
21320 vector signed int vec_signed2 (vector double, vector double);
21321 vector signed int vec_signede (vector double);
21322 vector signed int vec_signedo (vector double);
21323 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21324 instruction generator.
21325 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21326 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21327 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21329 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21330 vunsignede_v2df): Add define_expands.
21331 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21332 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21333 VEC_UNSIGNEDO): Add definitions.
21334 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21335 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21336 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21337 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21338 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21339 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21340 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21341 vec_unsignede, vec_unsignedo): Add builtin defines.
21342 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21344 * doc/extend.texi: Update the built-in documentation file for the
21345 new built-in functions.
21347 2017-06-29 Richard Biener <rguenther@suse.de>
21349 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21350 reduction chains to LOOP_VINFO_REDUCTIONS.
21351 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21352 SLP reductions after processing reduction chains.
21354 2017-06-29 Nathan Sidwell <nathan@acm.org>
21356 * builtins.c (fold_builtin_FUNCTION): Use
21357 lang_hooks.decl_printable_name.
21359 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21361 PR middle-end/81194
21362 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21363 with only one label.
21364 * stmt.c (expand_case): Assert NCASES is greater than one.
21366 2017-06-29 Richard Biener <rguenther@suse.de>
21368 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21370 (group_case_labels): Likewise.
21371 (find_taken_edge): Push sanity checking on val to workers...
21372 (find_taken_edge_cond_expr): ... here
21373 (find_taken_edge_switch_expr): ... and here, handle cases
21374 with just a default label.
21375 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21376 (group_case_labels): Likewise.
21377 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21378 group_case_labels does anything cleanup the CFG again.
21380 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21382 PR tree-optimization/81196
21383 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21384 exit condition comparing two IVs.
21386 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21388 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21389 profile to the dummy entry at the end of the list of architectures.
21390 * config/arm/arm-cpu-cdata.h: Regenerated.
21392 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21393 Michael Collison <michael.collison@arm.com>
21396 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21398 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21399 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21400 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21401 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21402 of shift when the shift amount is masked with constant equal to
21403 the size of the mode.
21404 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21407 2017-06-29 Martin Liska <mliska@suse.cz>
21409 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21411 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21414 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21417 (edge_growth_cache_entry::edge_growth_cache_entry): New
21419 (reset_edge_growth_cache): Update to use constructor.
21421 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21423 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21424 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21425 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21427 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21429 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21430 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21432 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21434 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21435 (*-linux-uclibc*): Add t-uclibc tmake_file.
21436 * config/t-musl: New.
21437 * config/t-uclibc: New.
21439 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21441 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21443 (gen_comm_data): Emit architectural setting of arch_prof.
21444 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21446 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21447 (armv8-m.base, armv8-m.main): Likewise.
21448 * arm-protos.h (arm_build_target): Add profile field.
21449 (arch_option): Likewise.
21450 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21452 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21453 arm_active_target.profile.
21455 2017-06-28 Richard Biener <rguenther@suse.de>
21457 PR middle-end/81227
21458 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21459 TYPE_OVERFLOW_WRAPS.
21460 * match.pd (negate_expr_p): Likewise.
21461 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21462 fold_build2, not fold_binary.
21464 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21466 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21467 Convert memory address to Pmode.
21468 (aarch64_print_operand): Assert MEM operands are always Pmode.
21470 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21473 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21474 Remove redundant if.
21475 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21476 * config/arm/aarch-common-protos.h
21477 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21478 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21480 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21483 * multiple_target.c (create_dispatcher_calls): Set the default
21484 clone to be static, not public.
21486 2017-06-28 Richard Biener <rguenther@suse.de>
21488 * tree-vect-loop.c (vectorizable_reduction): Move special
21489 cond reduction IV var creation ...
21490 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21491 parameter. Use STMT_VINFO_VECTYPE.
21492 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21495 2017-06-28 Martin Liska <mliska@suse.cz>
21498 * ipa-visibility.c (non_local_p): Handle visibility.
21500 2017-06-28 Martin Liska <mliska@suse.cz>
21503 * common.opt: Add IntegerRange to various options.
21504 * opt-functions.awk (integer_range_info): New function.
21505 * optc-gen.awk: Add integer_range_info to cl_options struct.
21506 * opts-common.c (decode_cmdline_option): Handle
21507 CL_ERR_INT_RANGE_ARG.
21508 (cmdline_handle_error): Likewise.
21509 * opts.c (print_filtered_help): Show valid interval in
21510 when --help is provided.
21511 * opts.h (struct cl_option): Add range_min and range_max fields.
21512 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21514 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21516 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21517 (x * C EQ/NE y * C): New transformation.
21519 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21521 * genmultilib (combination_space): Accept '+' in option names.
21523 2017-06-28 Martin Liska <mliska@suse.cz>
21526 * asan.c (instrument_derefs): Bail out inner references
21527 that are hard register variables.
21529 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21532 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21533 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21535 2017-06-28 Richard Biener <rguenther@suse.de>
21537 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21538 (vect_get_slp_defs): Adjust.
21539 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21541 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21543 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21544 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21545 (vectorizable_reduction): Adjust.
21546 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21548 (vect_get_slp_defs): Likewise.
21549 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21550 (vectorizable_bswap): Adjust.
21551 (vectorizable_call): Likewise.
21552 (vectorizable_conversion): Likewise.
21553 (vectorizable_assignment): Likewise.
21554 (vectorizable_shift): Likewise.
21555 (vectorizable_operation): Likewise.
21556 (vectorizable_store): Likewise.
21557 (vectorizable_condition): Likewise.
21558 (vectorizable_comparison): Likewise.
21560 2017-06-28 Michael Collison <michael.collison@arm.com>
21563 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21565 (arm_gen_movmemqi): Removed unused variable 'i'.
21566 Convert 'for' loop into 'while' loop.
21567 (arm_expand_prologue): Remove last unnecessary set of insn.
21568 (thumb_pop): Remove unused variable 'pushed_words'.
21569 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21571 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21573 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21574 * config/s390/s390-protos.h: Add prototype of
21575 s390_rel_address_ok_p.
21576 * config/s390/s390.c (s390_got_symbol): New function.
21577 (s390_rel_address_ok_p): New function.
21578 (legitimize_pic_address): Use s390_rel_address_ok_p.
21579 (s390_load_got): Use s390_got_symbol.
21580 (s390_option_override): Issue error if
21581 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21582 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21584 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21586 2017-06-27 Andrew Pinski <apinski@cavium.com>
21588 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21589 (X * copysign (1.0, X)): New pattern.
21590 (X * copysign (1.0, -X)): New pattern.
21591 (copysign (-1.0, CST)): New pattern.
21593 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21595 * genmultilib (combination_space): Remove variable.
21596 Validate reuse rules against regular expression for any sequence
21597 of multilib options in any order.
21599 2017-06-27 Michael Collison <michael.collison@arm.com>
21601 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21602 call aarch64_split_simd_combine.
21603 * (aarch64_combine_internal<mode>): Delete pattern.
21604 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21605 Allow register and subreg operands.
21607 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21609 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21610 specific need, just fallback on defaults.
21611 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21613 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21614 Olivier Hainque <hainque@adacore.com>
21616 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21618 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21619 targets. Pick a default if no particular attempt applied.
21620 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21623 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21625 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21626 (x86_64-wrs-vxworks7): Likewise.
21628 2017-06-27 Marek Polacek <polacek@redhat.com>
21631 * ubsan.c (instrument_null): Check get_base_address's result for null.
21633 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21635 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21637 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21639 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21640 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21641 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21642 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21643 New function types.
21644 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21645 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21646 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21647 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21648 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21649 BUILT_IN_FEUPDATEENV): New builtins.
21650 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21651 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21652 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21653 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21655 (builtin_structptr_types): Adjust size.
21656 * tree.c (builtin_structptr_types): Add four entries.
21658 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21659 Olivier Hainque <hainque@adacore.com>
21661 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21662 (TLS_SYM): New local macro, forcing reference to __tls__ on
21663 link command lines for VxWorks 7 RTPs, triggering initialization
21665 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21666 OS features TLS support, true for RTPs on VxWorks 7.
21667 * config/vxworks.c (vxworks_override_options): Setup emutls
21670 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21672 * predict.c (test_prediction_value_range): Use -1U instead of -1
21673 to avoid narrowing conversion warning.
21674 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21675 to avoid narrowing conversion warning.
21676 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21678 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21680 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21682 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21683 64bit configurations.
21684 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21685 (SIZE_TYPE): Likewise.
21686 * config/vxworks.c (vxworks_emutls_var_fields): Use
21687 long_unsigned_type_node instead of unsigned_type_node as the offset
21688 field type, which is "pointer" mode in emutls.c.
21690 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21693 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21695 PR middle-end/81207
21696 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21697 gimple_vuse copying separately from gimple_vdef copying.
21699 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21701 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21704 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21705 Olivier Hainque <hainque@adacore.com>
21707 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21708 for all vxworks7 targets.
21709 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21710 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21711 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21712 variations for VX6/VX7 and 32/64bits later on in ...
21713 (VXWORKS_LIB_SPEC): Leverage new macros.
21714 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21715 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21717 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21719 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21720 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21722 2017-06-26 Carl Love <cel@us.ibm.com>
21724 * config/rs6000/rs6000-c.c: Add support for built-in functions
21725 vector bool char vec_reve (vector bool char);
21726 vector signed char vec_reve (vector signed char);
21727 vector unsigned char vec_reve (vector unsigned char);
21728 vector bool int vec_reve (vector bool int);
21729 vector signed int vec_reve (vector signed int);
21730 vector unsigned int vec_reve (vector unsigned int);
21731 vector bool long long vec_reve (vector bool long long);
21732 vector signed long long vec_reve (vector signed long long);
21733 vector unsigned long long vec_reve (vector unsigned long long);
21734 vector bool short vec_reve (vector bool short);
21735 vector signed short vec_reve (vector signed short);
21736 vector double vec_reve (vector double);
21737 vector float vec_reve (vector float);
21738 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21739 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21740 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21741 (altivec_vreve): New pattern.
21742 * config/rs6000/altivec.h (vec_reve): New define.
21743 * doc/extend.texi (vec_rev): Update the built-in documentation file
21744 for the new built-in functions.
21746 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21748 PR tree-optimization/71815
21749 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21751 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21753 (slsr_process_phi): Likewise.
21754 (replace_uncond_cands_and_profitable_phis): Don't replace a
21755 multiply candidate with a stride of 1 (copy or cast).
21756 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21758 (lowest_cost_path): Likewise.
21759 (total_savings): Likewise.
21761 2017-06-26 Richard Biener <rguenther@suse.de>
21764 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21765 Use def_builtin_pure for all gather builtins.
21767 2017-06-26 Richard Biener <rguenther@suse.de>
21769 PR tree-optimization/81203
21770 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21771 non-dominating BBs.
21773 2017-06-26 Marek Polacek <polacek@redhat.com>
21776 * doc/invoke.texi: Document -Wmultistatement-macros.
21778 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21780 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21781 arm_neon_ok_no_float_abi effective target.
21783 2017-06-26 Richard Biener <rguenther@suse.de>
21785 PR tree-optimization/80928
21786 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21787 (copy_bbs): Set BB_DUPLICATED flag early.
21788 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21790 (execute_on_shrinking_pred): Likewise.
21791 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21792 BB_DUPLICATED blocks.
21793 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21794 iterate over all PHIs considering removal of *gsi.
21796 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21798 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21801 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21803 * config/rs6000/rs6000-string.c: (expand_block_clear,
21804 do_load_for_compare, select_block_compare_mode,
21805 compute_current_alignment, expand_block_compare,
21806 expand_strncmp_align_check, expand_strn_compare,
21807 expand_block_move, rs6000_output_load_multiple)
21808 Move functions related to string/block move/compare
21809 to a separate file.
21810 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21811 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21812 for this function which is now used in two files.
21813 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21814 * config.gcc: Add rs6000-string.o to extra_objs for
21815 targets powerpc*-*-* and rs6000*-*-*.
21817 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21820 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21821 32-bit, since indexed is not valid for DImode.
21822 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21823 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21824 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21825 (define_peephole2 for Altivec d-form store): Likewise.
21828 * multiple_target.c (create_dispatcher_calls): Only create the
21829 dispatcher call if the function is the default clone of a
21830 versioned function.
21832 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21834 PR middle-end/80902
21835 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21836 a call, force the call to not be a tail call.
21838 2017-06-23 Jeff Law <law@redhat.com>
21840 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21843 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21845 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21846 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21847 handling for early expansion of vector shifts (sl,sr,sra,rl).
21848 (builtin_function_type): Add vector shift right instructions
21849 to the unsigned argument list.
21851 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21853 rtl-optimizatoin/79286
21854 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21855 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21856 trap. PIC register plus a const unspec without offset can never trap.
21858 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21860 * tree.h (builtin_structptr_type): New type.
21861 (builtin_structptr_types): Declare new array.
21862 * tree.c (builtin_structptr_types): New array.
21863 (free_lang_data, build_common_tree_nodes): Use it.
21865 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21868 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21871 2017-06-22 Matt Turner <mattst88@gmail.com>
21873 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21874 Lake models to skylake case. Assume skylake for unknown
21875 models with clflushopt.
21877 2017-06-22 Jeff Law <law@redhat.com>
21879 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21880 frame sizes that do not satisfy aarch64_uimm12_shift.
21882 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21884 * profile-count.h (apply_probability,
21885 apply_scale, probability_in): Fix checks for zero.
21887 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21889 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21890 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21892 2016-06-22 Richard Biener <rguenther@suse.de>
21894 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21895 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21896 REDUC_MAX_EXPR support.
21897 (vectorizable_reduction): Likewise.
21898 (vect_create_epilog_for_reduction): Likewise.
21900 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21902 * match.pd (A / (1 << B) -> A >> B): New.
21903 * generic-match-head.c: Include optabs-tree.h.
21904 * gimple-match-head.c: Likewise.
21905 * optabs-tree.h (target_supports_op_p): New.
21906 * optabs-tree.c (target_supports_op_p): New.
21908 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21910 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21911 $gcc_cv_ld --help output.
21912 (gcc_cv_ld_demangle): Likewise.
21913 (gcc_cv_ld_eh_frame_hdr): Likewise.
21914 (gcc_cv_ld_pie): Likewise.
21915 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21916 (gcc_cv_ld_buildid): Likewise.
21917 (gcc_cv_ld_sysroot): Likewise.
21918 (ld_bndplt_support): Likewise.
21919 (ld_pushpopstate_support): Likewise.
21920 * configure: Regenerate.
21921 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21923 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21926 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21927 operands indexes to avoid gap between operands and match_dups.
21929 2017-06-21 Andrew Pinski <apinski@cavium.com>
21931 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21932 Increment Arith_shift and Arith_shift_reg by 1.
21933 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21935 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21936 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21937 (aarch64_strip_extend): Add new argument and test for it.
21938 (aarch64_cheap_mult_shift_p): New function.
21939 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21940 add a cost if it is true.
21941 Update calls to aarch64_strip_extend.
21942 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21944 2017-06-21 Andrew Pinski <apinski@cavium.com>
21946 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21948 (thunderxt88): Likewise.
21949 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21950 (thunderx_prefetch_tune): New variable.
21951 (thunderx2t99_prefetch_tune): Update for the correct values.
21952 (thunderxt88_tunings): New variable.
21953 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21954 generic_prefetch_tune.
21955 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21957 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21959 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21960 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21961 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21962 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21963 (aarch64_atomic_cas<mode>, GPI): Likewise.
21965 2017-06-21 Martin Liska <mliska@suse.cz>
21967 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21968 statements on cold and hot labels.
21969 * predict.c (tree_estimate_probability_bb): Remove the
21970 prediction from this place.
21972 2017-06-21 Martin Liska <mliska@suse.cz>
21974 PR tree-optimization/79489
21975 * gimplify.c (maybe_add_early_return_predict_stmt): New
21977 (gimplify_return_expr): Call the function.
21978 * predict.c (tree_estimate_probability_bb): Remove handling
21980 * predict.def: Update comment about early return predictor.
21981 * gimple-predict.h (is_gimple_predict): New function.
21982 * predict.def: Change default value of early return to 66.
21983 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21985 * passes.def: Put pass_strip_predict_hints to the beginning of
21988 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21990 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21991 FUNCTION_DECL declarations.
21992 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21994 (dwaf2out_decl): Likewise.
21995 * godump.c (go_early_global_decl): Skip call to the real debug hook
21996 for FUNCTION_DECL declarations.
21997 * passes.c (rest_of_decl_compilation): Skip call to the
21998 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21999 -fdump-go-spec is passed.
22001 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22003 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22004 Reorder for compactness.
22005 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22006 (def_builtin_pure, def_builtin_pure2): New functions.
22007 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22009 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22011 * match.pd (nop_convert): New predicate.
22012 ((A +- CST1) +- CST2): Allow some NOP conversions.
22014 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22017 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22018 with ctors/dtors if GOVD_SHARED is set.
22020 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22022 * config/aarch64/aarch64.md (movti_aarch64):
22023 Emit mov rather than orr.
22024 (movtf_aarch64): Likewise.
22025 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22026 Emit mov rather than orr.
22028 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22030 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22031 Swap alternatives, make integer dup more expensive.
22033 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22035 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22036 Return true for non-tls symbols.
22038 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22040 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22041 (cortex-a75): Likewise.
22042 (cortex-a75.cortex-a55): Likewise.
22043 * config/aarch64/aarch64-tune.md: Regenerate.
22044 * doc/invoke.texi (-mtune): Document new values for -mtune.
22046 2017-06-21 Tom de Vries <tom@codesourcery.com>
22048 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22049 stack_size feature.
22050 (Effective-Target Keywords, Other attributes): Suggest using
22051 dg-add-options stack_size feature to get stack limit in stack_size
22052 effective target documentation.
22054 2017-06-21 Julian Brown <julian@codesourcery.com>
22055 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22057 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22058 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22059 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22061 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22062 attribute type list for neon_multiply.
22063 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22064 attribute type list for neon_multiply.
22065 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22066 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22067 attribute type list for neon_multiply.
22068 * config/arm/types.md (crypto_pmull): Add.
22069 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22070 attribute type list.
22072 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22074 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22077 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22079 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22080 to make sure not to dereference a NULL cost_classes_ptr pointer.
22082 2017-06-20 Carl Love <cel@us.ibm.com>
22084 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22085 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22086 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22087 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22088 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22089 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22090 VMULOSW): New enum "unspec" values.
22091 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22092 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22093 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22094 altivec_vmulosw): New patterns.
22095 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22096 VMULOSW): Add definitions.
22098 2017-06-20 Julia Koval <julia.koval@intel.com>
22100 * config/i386/i386.c: Fix rounding expand for new pattern.
22101 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22103 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22105 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22106 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22108 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22110 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22113 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22115 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22116 architecture, then by implementer ID.
22117 * config/aarch64/aarch64-tune.md: Regenerate.
22119 2017-06-20 Richard Biener <rguenther@suse.de>
22121 PR middle-end/81097
22122 * fold-const.c (split_tree): Fold to type before negating.
22124 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22126 * diagnostic-show-locus.c
22127 (selftest::test_fixit_deletion_affecting_newline): New function.
22128 (selftest::diagnostic_show_locus_c_tests): Call it.
22130 2017-06-20 Andreas Schwab <schwab@suse.de>
22133 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22136 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22138 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22139 __ARM_FEATURE_COPROC according to support.
22141 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22143 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22144 Rewritten to avoid overflow for > 32-bit pointers.
22147 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22148 by removing enum keyword.
22149 (ubsan_type_descriptor): Likewise. Formatting fix.
22152 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22153 splitter): Require TARGET_SSE2 in the condition.
22155 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22158 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22159 for doing vector set of SFmode on ISA 3.0.
22160 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22161 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22163 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22164 SFmode value into a V4SF variable that was extracted from another
22165 V4SF variable without converting the element to double precision
22166 and back to single precision vector format.
22167 (vsx_insert_extract_v4sf_p9_2): Likewise.
22169 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22171 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22172 in UWHI to avoid undefined overflow.
22175 * ubsan.h (enum ubsan_encode_value_phase): New.
22176 (ubsan_encode_value): Change second argument to
22177 enum ubsan_encode_value_phase with default value of
22178 UBSAN_ENCODE_VALUE_GENERIC.
22179 * ubsan.c (ubsan_encode_value): Change second argument to
22180 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22181 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22182 create_tmp_var_raw instead of create_tmp_var and use a
22184 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22185 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22186 ubsan_encode_value callers.
22189 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22190 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22191 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22193 2017-06-19 Richard Biener <rguenther@suse.de>
22195 PR middle-end/81118
22196 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22197 estimates if we changed anything.
22199 2017-06-19 Richard Biener <rguenther@suse.de>
22201 PR tree-optimization/80887
22202 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22203 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22204 simplified lookups, then reset mprts_hook.
22205 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22207 (try_to_simplify): Likewise.
22209 2017-06-19 Martin Liska <mliska@suse.cz>
22212 * gimplify.c (gimplify_switch_expr):
22213 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22215 2017-06-19 Martin Liska <mliska@suse.cz>
22217 * doc/install.texi: Document that PGO runs in 4 stages.
22219 2017-06-19 Martin Liska <mliska@suse.cz>
22222 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22223 to dispatcher function name.
22224 * multiple_target.c (replace_function_decl): New function.
22225 (create_dispatcher_calls): Redirect both edges and references.
22227 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22229 * profile-count.c (profile_count::dump): Dump quality.
22230 (profile_count::differs_from_p): Update for unsigned val.
22231 * profile-count.h (profile_count_quality): New enum.
22232 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22234 2017-06-19 Richard Biener <rguenther@suse.de>
22236 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22237 struct function as arg.
22238 (estimate_numbers_of_iterations): Export overload with loop arg.
22239 (free_numbers_of_iterations_estimates_loop): Use an overload of
22240 free_numbers_of_iterations_estimates instead.
22241 * tree-cfg.c (remove_bb): Adjust.
22242 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22243 * tree-parloops.c (gen_parallel_loop): Likewise.
22244 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22246 (tree_unroll_loops_completely): Likewise.
22247 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22248 Use an overload instead and export.
22249 (estimated_loop_iterations): Adjust.
22250 (max_loop_iterations): Likewise.
22251 (likely_max_loop_iterations): Likewise.
22252 (estimate_numbers_of_iterations): Take struct function as arg
22254 (loop_exits_before_overflow): Adjust.
22255 (free_numbers_of_iterations_estimates_loop): Use an overload.
22256 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22257 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22259 2017-06-19 Richard Biener <rguenther@suse.de>
22262 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22263 RANGE_EXPR conservatively.
22265 2017-06-16 Carl Love <cel@us.ibm.com>
22267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22268 definitions for vec_float, vec_float2, vec_floato,
22269 vec_floate built-ins.
22270 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22271 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22273 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22274 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22275 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22276 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22277 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22278 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22279 vec_floato): Add builtin defines.
22280 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22281 Update the built-in documentation file for the new built-in
22284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22286 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22287 (mthumb): Mark as the negative of -marm.
22289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22291 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22293 (ARM Options, -mtune): Document that this accepts the same
22294 extension options as -mcpu.
22295 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22299 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22300 permitted extensions.
22302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22304 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22305 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22306 (armv8-m.main): Add option +nodsp.
22307 * config/arm/arm-cpu-cdata.h: Regenerated.
22309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22311 * config/arm/t-fuchsia: New file.
22312 * config.gcc (arm*-*-fuchsia*): Use it.
22314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22316 * config/arm/t-symbian: Rewrite for new option infrastructure.
22318 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22320 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22321 (MULTILIB_REQUIRED): Likewise.
22323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22325 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22326 (MULTILIB_RESUE): Likewise.
22327 (MULTILIB_MATCHES): Likewise.
22328 (MULTLIB_REQUIRED): Likewise.
22330 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22332 * config/arm/t-rtems: Rewrite for new option framework.
22334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22336 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22337 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22338 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22339 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22340 * config/arm/t-multilib: ... here.
22341 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22342 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22343 armv7-a and armv8*-a when A-profile libraries have not been built.
22344 * config/arm/t-rmprofile: Rewrite.
22346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22348 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22349 with a backslash. Remove the backslash after substituting unescaped
22351 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22355 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22356 the driver to pass through the default setting of -mfloat-abi.
22357 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22359 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22360 (all_feat_combs): New rule.
22361 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22363 * config/arm/t-aprofile: Rewrite.
22365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22367 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22368 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22369 fpu is not specified by the user/command-line.
22370 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22371 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22372 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22373 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22374 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22375 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22379 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22380 * config/arm/t-arm-elf: Rewritten.
22382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22384 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22385 have some floating-point instructions.
22386 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22387 (TARGET_MAYBE_HARD_FLOAT): New macro.
22388 * config/arm/arm-builtins.c (arm_init_builtins): Use
22389 TARGET_MAYBE_HARD_FLOAT.
22390 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22394 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22395 (configargs.h): Include it.
22396 (arm_print_hint_for_fpu_option): New function.
22397 (arm_parse_fpu_option): New function.
22398 (candidate_extension): New class.
22399 (arm_canon_for_multilib): New function.
22400 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22401 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22402 (ARCH_CANONICAL_SPECS): New macro.
22403 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22407 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22408 are set after handling multilib fragments. Set target_cpu_default2
22411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22413 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22415 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22416 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22418 (check_arch): Likewise.
22419 * config/arm/arm.c (arm_configure_build_target): Handle
22420 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22421 options in the default.
22423 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22425 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22426 when an option is an alias of another.
22427 * config/arm/parsecpu.awk (optalias): New parser token.
22428 (gen_comm_data): Mark non-alias options as such. Emit entries
22429 for extension aliases.
22430 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22431 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22432 (armv6kz, armv6zk, armv6t2): Likewise.
22433 (armv7): Make vfpv3-d16 an alias.
22434 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22436 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22437 Sort in canonical order.
22438 (armv8-a): Sort in canonical order.
22439 (armv8.1-a, armv8.2-a): Likewise.
22440 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22442 (cortex-a9): Sort in canonical order.
22443 * config/arm/arm.c (selftests.h): Include it.
22444 (arm_test_cpu_arch_data): New function.
22445 (arm_run_self_tests): New function.
22446 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22447 (targetm): Move declaration to the end of the file.
22448 * arm-cpu-cdata.h: Regenerated.
22450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22452 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22453 call to target_mode_check describing the type of option passed.
22454 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22455 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22456 arm_parse_cpu_option_name to match parameters against list of
22458 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22459 arm_arch_core_flags data structure.
22460 * config/arm/arm-cpu_cdata.h: Regenerated.
22462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22464 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22466 (arm_print_hint_for_cpu_option): Likewise.
22467 (arm_print_hint_for_arch_option): Likewise.
22468 (arm_parse_cpu_option_name): Likewise.
22469 (arm_parse_arch_option_name): Likewise.
22470 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22471 of entries in the all_fpus list.
22472 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22473 (arm_parse_cpu_option_name): Declare.
22474 (arm_parse_arch_option_name): Declare.
22475 (arm_parse_option_features): Declare.
22476 (arm_intialize_isa): Declare.
22477 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22479 (gen_comm_data): ... here. Make definitions non-static.
22480 * config/arm/arm-cpu-data.h: Regenerated.
22481 * config/arm/arm-cpu-cdata.h: Regenerated.
22483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22485 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22486 (cpu_arch_extension): New structure.
22487 (cpu_arch_option, arch_option, cpu_option): New structures.
22488 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22489 architecture types.
22490 (gen_data): Generate new format data tables.
22491 * config/arm/arm.c (cpu_tune): New structure.
22492 (cpu_option, processors): Delete.
22493 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22494 (arm_print_hint_for_cpu_option): ... this and ...
22495 (arm_print_hint_for_arch_option): ... this.
22496 (arm_parse_arch_cpu_name): Delete. Replace with ...
22497 (arm_parse_cpu_option_name): ... this and ...
22498 (arm_parse_arch_option_name): ... this.
22499 (arm_unrecognized_feature): Change type of target parameter to
22501 (arm_parse_arch_cpu_features): Delete. Replace with ...
22502 (arm_parse_option_features): ... this.
22503 (arm_configure_build_target): Rework to use new configuration data
22505 (arm_print_tune_info): Rework for new configuration data tables.
22506 * config/arm/arm-cpu-data.h: Regenerated.
22507 * config/arm/arm-cpu.h: Regenerated.
22509 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22511 * Makefile.in (OBJS): Move sbitmap.o from here ...
22512 (OBJS-libcommon): ... to here.
22514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22516 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22517 (ISA_ALL_CRYPTO): New macro.
22518 (ISA_ALL_SIMD): New macro
22519 (ISA_ALL_FP): New macro.
22520 * config/arm/arm.c (fpu_bitlist): Update initializer.
22521 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22523 (arm9e): Add fpu. Add option for nofp
22524 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22525 (arm926ej-s, arm1026ej-s): Likewise.
22526 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22527 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22528 neon-fp16, neon-vfpv4, nofp and nosimd.
22529 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22530 (cortex-a8): Add fpu. Add option for nofp.
22531 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22532 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22533 (cortex-r4f): Add fpu.
22534 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22535 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22537 (cortex-r8): Likewise.
22538 (cortex-m4): Add fpu. Add option for nofp.
22539 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22540 (cortex-a17.cortex-a7): Likewise.
22541 (cortex-a32): Add fpu. Add options for crypto and nofp.
22542 (cortex-a35, cortex-a53): Likewise.
22543 (cortex-a57): Add fpu. Add option for crypto.
22544 (cortex-a72, cortex-a73): Likewise.
22545 (exynos-m1): Likewise.
22546 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22547 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22548 (cortex-m33): Add fpu. Add option for nofp.
22549 * config/arm/arm-cpu-cdata.h: Regenerated
22550 * config/arm/arm-cpu-data.h: Regenerated.
22552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22554 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22555 (armv5te, armv5tej): Likewise.
22556 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22557 (armv7): Add options fp and vfpv3-d16.
22558 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22559 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22561 (armv7ve): Likewise.
22562 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22563 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22564 (armv8-a): Add nocrypto option.
22565 (armv8.1-a, armv8.2-a): Likewise.
22566 (armv8-m.main): add options fp, fp.dp and nofp.
22568 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22570 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22572 (armv8-a+crc): Delete.
22573 (armv8.1-a): Add options simd, crypto and nofp.
22574 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22575 (armv8.2-a+fp16): Delete.
22576 (armv8-m.main): Add option dsp.
22577 (armv8-m.main+dsp): Delete.
22578 (cortex-a8): Add fpu. Add nofp option.
22579 (cortex-a9): Add fpu. Add nofp and nosimd options.
22580 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22581 link to main cpu and architecture data structures.
22582 (gen_comm_data): Only put isa attributes from the main architecture
22584 (option): New statement for architecture and CPU entries.
22585 * arm.c (struct cpu_option): New structure.
22586 (struct processors): Add entry for options.
22587 (arm_unrecognized_feature): New function.
22588 (arm_parse_arch_cpu_name): Ignore any characters after the first
22590 (arm_parse_arch_cpu_feature): New function.
22591 (arm_configure_build_target): Separate out any CPU and architecture
22592 features and parse separately. Don't error out if -mfpu=auto is
22593 used with only an architecture string.
22594 (arm_print_asm_arch_directives): New function.
22595 (arm_file_start): Call it.
22596 * config/arm/arm-cpu-cdata.h: Regenerated.
22597 * config/arm/arm-cpu-data.h: Likewise.
22598 * config/arm/arm-tables.opt: Likewise.
22600 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22602 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22603 assembler when it is not -mfpu=auto.
22605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22607 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22608 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22609 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22610 (ASM_CPU_SPEC): Rewrite.
22611 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22612 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22613 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22614 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22615 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22616 copied string is NUL-terminated. Also strip any characters prefixed
22618 (arm_rewrite_selected_arch): New function.
22619 (arm_rewrite_march): New function.
22621 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22623 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22624 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22625 (march, mcpu, mtune): Convert to string-based options.
22626 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22627 (arm_parse_arch_cpu_name): New function.
22628 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22629 identify selected architecture or CPU.
22630 (arm_option_save): New function.
22631 (TARGET_OPTION_SAVE): Redefine.
22632 (arm_option_restore): Restore string options.
22633 (arm_option_print): Print string options.
22635 2017-06-16 Martin Sebor <msebor@redhat.com>
22637 PR tree-optimization/80933
22638 PR tree-optimization/80934
22639 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22640 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22641 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22642 (gimple_fold_builtin): Call them.
22644 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22646 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22647 as unlikely; update profile.
22649 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22651 * predict.c (force_edge_cold): Handle declaring edges impossible
22654 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22656 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22658 (try_unroll_loop_completely): Fix reporting.
22660 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22662 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22664 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22667 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22668 if given a non-constant argument for an intrinsic which requires a
22671 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22673 * profile.c (compare_freqs): New function.
22674 (branch_prob): Sort edge list.
22675 (find_spanning_tree): Assume that the list is priority sorted.
22677 2017-06-16 Richard Biener <rguenther@suse.de>
22679 PR tree-optimization/81090
22680 * passes.def (pass_record_bounds): Remove.
22681 * tree-pass.h (make_pass_record_bounds): Likewise.
22682 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22683 make_pass_record_bounds): Likewise.
22684 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22685 not free niter estimates at the beginning but at the end.
22686 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22688 2017-06-16 Richard Biener <rguenther@suse.de>
22690 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22691 initializer to workaround ICE in host GCC 4.8.
22693 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22695 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22697 (clone_inlined_nodes): Update.
22699 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22701 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22702 prefetch settings, and enable prefetching by default at -O3.
22704 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22706 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22707 Set flag_prefetch_loop_arrays according to tuning data.
22709 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22711 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22712 New tune structure.
22713 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22714 [Unrelated to main purpose of the patch] Place the pointer field last
22715 to enable type checking errors when tune structure are wrongly merged.
22716 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22717 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22718 (thunderx2t99_prefetch_tune): New tune constants.
22719 (tune_params *_tunings): Update all tunings (no functional change).
22720 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22721 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22722 from tunings structures.
22724 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22727 * ubsan.c (instrument_null): Add T argument, use it instead
22728 of computing it based on IS_LHS.
22729 (instrument_object_size): Likewise.
22730 (pass_ubsan::execute): Adjust instrument_null and
22731 instrument_object_size callers to pass gimple_get_lhs or
22732 gimple_assign_rhs1 result to it. Use instrument_null instead of
22733 calling get_base_address and instrument_mem_ref. Handle
22734 aggregate call arguments for object-size sanitization.
22736 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22738 PR tree-optimization/81089
22739 * tree-vrp.c (is_masked_range_test): Validate operands of
22742 2017-06-15 Martin Sebor <msebor@redhat.com>
22745 * dumpfile.c (dump_register): Avoid calling memset to initialize
22746 a class with a default ctor.
22747 * gcc.c (struct compiler): Remove const qualification.
22748 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22749 * hash-table.h: Ditto.
22750 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22752 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22753 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22755 * params.h (struct param_info): Make struct members non-const.
22756 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22757 with default initialization.
22758 * vec.h (vec_copy_construct, vec_default_construct): New helper
22760 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22761 with vec_copy_construct.
22762 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22763 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22764 * doc/invoke.texi (-Wclass-memaccess): Document.
22766 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22768 * emit-rtl.h (is_leaf): Update comment about local
22769 register allocator.
22771 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22774 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22775 for a variable to have a section before checking if the section has a
22777 Set section to.persistent if persistent attribute is set.
22778 Warn if .persistent attribute is used on an automatic variable.
22780 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22782 PR rtl-optimization/80474
22783 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22785 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22787 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22790 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22792 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22793 (casesi): Emit bounds checking as RTL.
22794 (casesi_internal_mips16_<mode>): Remove bounds checking.
22796 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22798 * config/xtensa/xtensa.c (xtensa_option_override): Append
22799 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22800 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22801 xtensa_doloop_hooks): Define unconditionally.
22802 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22804 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22805 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22806 for it in xtensa_option_override.
22807 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22808 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22810 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22812 * doc/cppopts.texi: Document '-' special value to -MF.
22814 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22816 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22817 (cortex_a53_fconst): Likewise.
22818 (cortex_a53_fpmul): Likewise.
22819 (cortex_a53_f_load_64): Likewise.
22820 (cortex_a53_f_load_many): Likewise.
22821 (cortex_a53_advsimd_alu): Likewise.
22822 (cortex_a53_advsimd_alu_q): Likewise.
22823 (cortex_a53_advsimd_mul): Likewise.
22824 (cortex_a53_advsimd_mul_q): Likewise.
22825 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22826 Add missing fmul, r2f_cvt and fconst cases.
22828 2017-06-14 Richard Biener <rguenther@suse.de>
22830 PR middle-end/81088
22831 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22833 (fold_binary_loc): When associating do not treat pre-existing
22834 TREE_OVERFLOW on literal constants as a reason to allow
22835 TREE_OVERFLOW on associated literal constants.
22837 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22839 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22840 (MASK_FEATURES): New macro.
22841 * config/sparc/sparc.c (sparc_option_override): Remove the special
22842 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22844 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22846 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22847 a division of 0 if non-call exceptions are enabled.
22849 2017-06-14 Andrew Pinski <apinski@cavium.com>
22850 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22853 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22854 Improve vector initialization code gen for only variable case.
22856 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22858 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22860 2017-06-14 Richard Biener <rguenther@suse.de>
22862 PR tree-optimization/81083
22863 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22866 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22868 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22869 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22870 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22871 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22872 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22873 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22875 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22877 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22878 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22882 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22884 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22886 * config/rs6000/t-rtems: Don't handle SPE.
22888 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22890 * config/rs6000/t-linux: Don't handle SPE.
22892 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22894 * config/rs6000/eabispe.h: Delete file.
22896 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22898 * config/rs6000/t-spe: Delete file.
22900 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22902 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22903 (rs6000_legitimate_offset_address_p): Return false for anything in
22904 V2SImode or V2SFmode.
22906 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22908 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22909 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22910 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22912 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22913 (rs6000_legitimate_offset_address_p): Ditto.
22914 (rs6000_emit_move): Ditto.
22915 (rs6000_init_builtins): Remove V4HI_type_node.
22917 2017-06-13 Martin Liska <mliska@suse.cz>
22920 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22921 (gate_asan): Likewise.
22922 * asan.h (asan_no_sanitize_address_p): Remove the function.
22923 (sanitize_flags_p): New function.
22924 * builtins.def: Fix coding style.
22925 * common.opt: Use renamed enum value.
22926 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22927 * doc/extend.texi: Document no_sanitize attribute.
22928 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22929 to SANITIZE_UNDEFINED_NONDEFAULT.
22930 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22931 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22932 Use sanitize_flags_p.
22933 * gimplify.c (gimplify_function_tree): Likewise.
22934 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22935 * opts.c (parse_no_sanitize_attribute): New function.
22936 (common_handle_option): Use renamed enum value.
22937 * opts.h (parse_no_sanitize_attribute): Declare.
22938 * tree.c (sanitize_flags_p): New function.
22939 * tree.h: Declared here.
22940 * tsan.c: Use sanitize_flags_p.
22941 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22942 (instrument_mem_ref): Likewise.
22943 (instrument_bool_enum_load): Likewise.
22944 (do_ubsan_in_current_function): Remove the function.
22945 (pass_ubsan::execute): Use sanitize_flags_p.
22946 * ubsan.h: Remove do_ubsan_in_current_function
22947 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22948 (dump_function_to_file): Use it here.
22950 2017-06-13 Martin Jambor <mjambor@suse.cz>
22952 PR tree-optimization/80803
22953 PR tree-optimization/81063
22954 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22955 (propagate_subaccesses_across_link): Enqueue subtree whenever
22956 necessary instead of relying on the caller.
22958 2017-06-13 Martin Jambor <mjambor@suse.cz>
22960 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22961 that have a first_link.
22962 (sort_and_splice_var_accesses): Do not check first_link before
22964 (subtree_mark_written_and_enqueue): Likewise.
22965 (propagate_all_subaccesses): Likewise and do not stop at first
22966 parent with a first_link.
22968 2017-06-13 Martin Jambor <mjambor@suse.cz>
22970 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22973 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22975 * match.pd: New pattern.
22977 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22979 * tree-vrp.c (is_masked_range_test): New function.
22980 (register_edge_assert_for): Determine ranges for
22983 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22985 PR tree-optimization/67328
22986 * fold-const.c (maskable_range_p): New function.
22987 (build_range_check): Generate bittests if possible.
22989 2017-06-13 Martin Liska <mliska@suse.cz>
22991 * gimple-pretty-print.c (dump_probability): Add new argument.
22992 (dump_edge_probability): Dump both probability and count.
22993 (dump_gimple_label): Likewise.
22994 (dump_gimple_bb_header): Likewise.
22996 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22999 * config/avr/avr-devices.c: Fix indentation.
23000 * config/avr/gen-avr-mmcu-specs.c: Dito.
23002 2017-06-13 Richard Biener <rguenther@suse.de>
23004 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23005 instead get vector type from stmt_info.
23006 (vectorizable_reduction): Adjust. Remove dead code.
23008 2017-06-13 Richard Biener <rguenther@suse.de>
23010 PR middle-end/81065
23011 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23012 case of C * (x * C2 + C3).
23013 (fold_addr_of_array_ref_difference): Properly fold index difference.
23015 2017-06-12 David S. Miller <davem@davemloft.net>
23018 * config/sparc/sparc.md (return expander): Emit frame blockage if
23019 function uses alloca.
23021 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23023 * combine.c (make_field_assignment): Check len rather than the mode
23024 precision when calling force_to_mode.
23026 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23028 Support multilibs and devices that see flash in RAM address range.
23031 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23032 (avr_mcu_t) <flash_pm_offset>: New field.
23033 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23034 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23035 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23036 (AVR_TINY_PM_OFFSET): Remove macro.
23037 * config/avr/avr.opt (-mshort-calls): New option.
23038 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23039 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23040 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23041 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23042 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23043 instead of avr_arch->have_jmp_call.
23044 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23045 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23046 avr_arch->flash_pm_offset to define.
23047 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23048 new field flash_pm_offset. Add entry for avrxmega3.
23049 (avr_texinfo): Add entry for avrxmega3.
23050 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23051 attiny212, attiny214,
23052 attiny412, attiny414, attiny416, attiny417,
23053 attiny814, attiny816, attiny817,
23054 attiny1614, attiny1616, attiny1617,
23055 attiny3214, attiny3216, attiny3217.
23056 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23057 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23058 (avr_print_operand_address) [AVR_TINY]: Same.
23059 (avr_asm_init_sections) <readonly_data_section>: Only patch
23060 callback if avr_arch->flash_pm_offset = 0.
23061 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23062 for rodata if avr_arch->flash_pm_offset != 0.
23063 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23064 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23065 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23066 (m_options): Append opt_rcall.
23067 (m_dirnames): Append dir_rcall.
23068 * config/avr/t-multilib: Regenerate.
23070 * configure.ac [target=avr]: Check whether avrxmega3 default
23071 linker description file works as needed.
23072 * configure: Regenerate.
23073 * doc/avr-mmcu.texi: Regenerate.
23074 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23075 <__AVR_ARCH__>: Document avrxmega3 and 103.
23076 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23077 <__AVR_SHORT_CALLS__>: Document it.
23078 <__AVR_PM_BASE_ADDRESS__>: Document it.
23079 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23080 (AVR Variable Attributes) <progmem>: Document this is
23081 not needed for avrxmega3.
23082 (AVR Named Address Spaces) <__flash>: Dito.
23084 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23086 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23088 2017-06-12 Doug Rupp <rupp@adacore.com>
23090 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23091 Append vxworks-stdint.h to the tm_file list.
23092 * config/vxworks-stdint.h: New file.
23094 2017-06-12 Martin Liska <mliska@suse.cz>
23096 PR tree-optimization/81041
23097 * tree-profile.c (gimple_gen_ic_func_profiler):
23098 Create an extra BB in profile-generate
23099 (gimple_gen_time_profiler): Likewise.
23101 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23103 PR tree-optimization/81003
23104 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23105 (update_range_test): Use it instead of force_gimple_operand_gsi.
23107 2017-06-12 Richard Biener <rguenther@suse.de>
23109 PR tree-optimization/81053
23110 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23111 with backedge value not defined in loop. Simplify def stmt
23114 2017-06-11 Tom de Vries <tom@codesourcery.com>
23117 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23119 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23120 nvptx_cannot_force_const_mem.
23122 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23124 * opts.c (finish_options): Move test for flag_split_stack after
23125 it has been initialized.
23127 2017-06-11 Jason Merrill <jason@redhat.com>
23129 * tree.h (id_equal): New.
23130 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23131 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23132 instead of strcmp of IDENTIFIER_POINTER.
23134 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23136 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23137 (mark_all_inlined_calls_cdtor): Fix formating.
23138 (inline_transform): Rescale profile before inlining.
23140 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23142 * cgraph.h (cgraph_edge::clone): Update prototype.
23143 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23144 (cgraph_node::create_clone): Update.
23145 (cgraph_node::create_version_clone): Update.
23146 * tree-inline.c (copy_bb): Update.
23147 (expand_call_inline): Update.
23149 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23151 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23152 factored out from ...
23153 (rs6000_emit_prologue): ... here.
23155 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23157 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23158 factored out from ...
23159 (rs6000_emit_prologue): ... here.
23161 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23163 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23165 (handle_missing_profiles): Fix computation of tp_first_run.
23166 (counts_to_freqs): Do not touch freqs when count is 0.
23168 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23170 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23173 2017-06-10 Tom de Vries <tom@codesourcery.com>
23175 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23176 attributes): Document signal effective target.
23178 2017-06-10 Tom de Vries <tom@codesourcery.com>
23180 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23181 Document effective target stack_size.
23183 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23185 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23186 to the edit_context if they can be auto-applied.
23188 2017-06-9 Ian Lance Taylor <iant@golang.org>
23190 * opts.c (finish_options): If -fsplit-stack, disable implicit
23191 -forder-blocks-and-partition.
23192 * doc/invoke.texi (Optimize Options): Document that when using
23193 -fsplit-stack -forder-blocks-and-partition is not implicitly
23196 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23198 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23199 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23200 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23201 * builtins.def (abort, trap, unreachable): Declare cold.
23202 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23203 * tree-core.h (ECF_COLD): New.
23204 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23205 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23207 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23209 * predict.c (unlikely_executed_stmt_p): Cleanup.
23211 2017-06-09 Richard Biener <rguenther@suse.de>
23213 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23214 model if the ref is always written to.
23216 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23218 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23220 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23222 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23225 2017-06-09 Tom de Vries <tom@codesourcery.com>
23228 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23229 "target cannot support label values" when encountering LABEL_REF.
23231 2017-06-09 Martin Liska <mliska@suse.cz>
23233 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23234 (gimple_gen_ic_func_profiler): Emit direct comparison
23235 of __gcov_indirect_call_callee with NULL.
23236 (gimple_gen_time_profiler): Change probability from
23237 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23239 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23241 * profile.c (edge_gcov_counts): Turn to pointer.
23242 (compute_branch_probabilities, compute_branch_probabilities): Update.
23243 (branch_prob): Do not clear edge_gcov_count.
23244 * profile.h (edge_gcov_counts): Turn to pointer.
23245 (edge_gcov_count): Update.
23247 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23249 * gimple.h (gimple_check_failed): Mark cold.
23251 2017-06-09 Richard Biener <rguenther@suse.de>
23253 PR tree-optimization/66623
23254 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23255 refactor check_reduction into two parts, properly computing
23256 whether we have to check reduction validity for outer loop
23259 2017-06-09 Richard Biener <rguenther@suse.de>
23261 PR tree-optimization/79483
23262 * graphite-scop-detection.c (order): New global.
23263 (get_order): Compute bb to order mapping that satisfies code
23264 generation constraints.
23265 (cmp_pbbs): New helper.
23266 (build_scops): Start domwalk at entry block, sort generated
23269 2017-06-09 Richard Biener <rguenther@suse.de>
23271 PR middle-end/81007
23272 * ipa-polymorphic-call.c
23273 (ipa_polymorphic_call_context::restrict_to_inner_class):
23274 Skip FIELD_DECLs with error_mark_node type.
23275 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23278 2017-06-09 Martin Liska <mliska@suse.cz>
23280 * predict.c (struct branch_predictor): New struct.
23281 (test_prediction_value_range): New test.
23282 (predict_c_tests): New function.
23283 * selftest-run-tests.c (selftest::run_tests): Run the function.
23284 * selftest.h: Declare new tests.
23286 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23289 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23290 gen_add3_insn did not fail.
23291 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23292 r0, construct that number in a temporary reg and add that reg to r0.
23293 If asked to put the result in r0 as well, fail.
23295 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23297 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23298 for early expansion of vec_eqv.
23300 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23302 PR middle-end/81005
23303 * ubsan.c (instrument_null): Avoid pointless code temporary.
23304 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23306 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23310 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23313 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23316 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23318 * predict.c (unlikely_executed_edge_p): Move ahead.
23319 (probably_never_executed_edge_p): Use it.
23321 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23323 PR middle-end/79988
23324 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23325 gimple_call_builtin_p call.
23327 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23329 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23330 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23331 rtl_check_failed_type2, rtl_check_failed_code1,
23332 rtl_check_failed_code2, rtl_check_failed_code_mode,
23333 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23334 rtvec_check_failed_bounds, rtl_check_failed_flag,
23335 _fatal_insn_not_found, _fatal_insn): Likewise.
23336 * tree.h (tree_contains_struct_check_failed,
23337 tree_check_failed, tree_not_check_failed,
23338 tree_class_check_failed, tree_range_check_failed,
23339 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23340 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23341 tree_operand_check_failed, omp_clause_check_failed,
23342 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23345 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23347 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23348 flag_branch_probabilities.
23349 * ipa-inline.c (edge_badness): Likewise.
23350 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23351 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23352 * predict.c (maybe_hot_frequency_p): Likewise.
23353 (probably_never_executed): Likewise.
23354 * sched-ebb.c (schedule_ebbs): Likewise.
23355 * sched-rgn.c (find_single_block_region): Likewise.
23356 * tracer.c (tail_duplicate): Likewise.
23358 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23360 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23361 longer requires x_flag_profile_use.
23363 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23365 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23366 instead of flag_reorder_blocks_and_partition.
23367 * dbxout.c (dbxout_function_end): Likewise.
23368 * dwarf2out.c (gen_subprogram_die): Likewise.
23369 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23370 * hw-doloop.c (reorg_loops): Likewise.
23371 * varasm.c (assemble_start_function,
23372 assemble_end_function): Likewise.
23373 (decide_function_section): Do not check for
23374 flag_reorder_blocks_and_partition.
23376 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23378 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23380 (chkp_get_hard_register_fake_addr_expr): Ditto.
23381 (chkp_build_addr_expr): Add check for hard reg case.
23382 (chkp_parse_array_and_component_ref): Ditto.
23383 (chkp_find_bounds_1): Ditto.
23384 (chkp_process_stmt): Don't generate bounds store for
23387 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23389 * predict.c (maybe_hot_bb_p): Do not check profile status.
23390 (maybe_hot_edge_p): Likewise.
23391 (probably_never_executed): Check for zero counts even if profile
23393 (unlikely_executed_edge_p): New function.
23394 (unlikely_executed_stmt_p): New function.
23395 (unlikely_executed_bb_p): New function.
23396 (set_even_probabilities): Use unlikely predicates.
23397 (combine_predictions_for_bb): Likewise.
23398 (predict_paths_for_bb): Likewise.
23399 (predict_paths_leading_to_edge): Likewise.
23400 (determine_unlikely_bbs): New function.
23401 (estimate_bb_frequencies): Use it.
23402 (compute_function_frequency): Use zero counts even if profile is
23404 * profile-count.h: Fix typo.
23406 2017-08-08 Julia Koval <julia.koval@intel.com>
23408 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23409 _mm512_mask_cvtsepi16_storeu_epi8,
23410 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23411 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23412 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23413 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23414 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23415 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23416 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23417 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23418 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23419 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23420 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23421 __builtin_ia32_pmovuswb256mem_mask,
23422 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23423 __builtin_ia32_pmovwb512mem_mask): New builtins.
23425 2017-08-08 Julia Koval <julia.koval@intel.com>
23427 PR target/73350,80862
23428 * config/i386/subst.md (round): Fix round pattern.
23429 * config/i386/i386.c (ix86_erase_embedded_rounding):
23430 Fix erasing rounding for the fixed pattern.
23432 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23434 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23436 2017-06-08 Martin Liska <mliska@suse.cz>
23438 PR gcov-profile/80911
23439 * gcov.c (block_info::block_info): New constructor.
23441 2017-06-07 Carl Love <cel@us.ibm.com>
23443 * config/rs6000/rs6000-c: The return type of the following
23444 built-in functions was implemented as int not long long. Fix sign
23445 of return value for the unsigned version of vec_mulo and vec_mule.
23446 vector unsigned long long vec_bperm (vector unsigned long long,
23447 vector unsigned char)
23448 vector signed long long vec_mule (vector signed int,
23450 vector unsigned long long vec_mule (vector unsigned int,
23451 vector unsigned int)
23452 vector signed long long vec_mulo (vector signed int,
23454 vector unsigned long long vec_mulo (vector unsigned int,
23455 vector unsigned int)
23456 * doc/extend.texi: Fix the documentation for the built-in
23459 2017-06-07 Carl Love <cel@us.ibm.com>
23462 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23465 2017-06-07 Carl Love <cel@us.ibm.com>
23467 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23468 support, Generate doublehv for signed int/float for BE case only.
23470 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23472 * doc/invoke.texi (mcx16): Rewrite.
23474 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23476 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23477 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23478 *mov<mode>_softfloat, and an anonymous splitter): Use
23479 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23481 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23483 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23485 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23486 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23487 (rs6000_debug_reg_global): Adjust.
23488 (rs6000_init_hard_regno_mode_ok): Adjust.
23489 (rs6000_dbx_register_number): Adjust.
23490 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23491 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23492 Remove SPE_ACC and SPEFSCR.
23493 (REG_ALLOC_ORDER): Ditto.
23494 (FRAME_POINTER_REGNUM): Change to 111.
23495 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23496 (REG_CLASS_NAMES): Ditto.
23497 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23498 (REGISTER_NAMES): Ditto.
23499 (ADDITIONAL_REG_NAMES): Ditto.
23500 (rs6000_reg_names): Ditto.
23501 * config/rs6000/rs6000.md: Renumber some register number
23504 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23506 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23508 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23509 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23511 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23512 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23513 Delete the SPE high registers.
23514 (REG_ALLOC_ORDER): Ditto.
23515 (enum reg_class): Remove SPE_HIGH_REGS.
23516 (REG_CLASS_NAMES): Ditto.
23517 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23518 (REGISTER_NAMES): Ditto.
23519 (rs6000_reg_names): Ditto.
23520 * doc/tm.texi.in: Remove SPE as example.
23521 * doc/tm.texi: Regenerate.
23523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23525 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23526 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23527 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23528 * config/rs6000/rs6000.md (type): Remove "brinc".
23530 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23532 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23533 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23534 * config/rs6000/linuxspe.h: Delete file.
23535 * config/rs6000/rs6000.md: Don't include spe.md.
23536 * config/rs6000/spe.h: Delete file.
23537 * config/rs6000/spe.md: Delete file.
23538 * config/rs6000/t-rs6000: Remove spe.md.
23540 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23542 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23543 (reg_or_none500mem_operand): Delete.
23544 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23545 instead of reg_or_none500mem_operand.
23547 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23550 handling of SPE flags.
23551 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23553 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23555 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23557 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23558 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23559 paired_divv2sf3): Similar.
23560 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23561 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23562 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23564 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23565 Rename the paired_* instruction patterns.
23566 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23568 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23569 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23570 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23571 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23572 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23573 PAIRED_VECTOR_MODE.
23574 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23575 (spe_func_has_64bit_regs_p): Delete.
23576 (spe_expand_predicate_builtin): Delete.
23577 (spe_expand_evsel_builtin): Delete.
23578 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23579 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23580 (invalid_e500_subreg): Delete.
23581 (rs6000_legitimize_address): Always force_reg op2 as well, for
23582 paired single memory accesses.
23583 (rs6000_member_type_forces_blk): Delete.
23584 (rs6000_spe_function_arg): Delete.
23585 (rs6000_expand_unop_builtin): Delete SPE handling.
23586 (rs6000_expand_binop_builtin): Ditto.
23587 (spe_expand_stv_builtin): Delete.
23588 (bdesc_2arg_spe): Delete.
23589 (spe_expand_builtin): Delete.
23590 (spe_expand_predicate_builtin): Delete.
23591 (spe_expand_evsel_builtin): Delete.
23592 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23593 (spe_init_builtins): Delete.
23594 (spe_func_has_64bit_regs_p): Delete.
23595 (savres_routine_name): Delete "info" parameter. Adjust callers.
23596 (rs6000_emit_stack_reset): Ditto.
23597 (rs6000_dwarf_register_span): Delete.
23598 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23599 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23600 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23602 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23604 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23605 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23606 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23607 mulv2sf3, divv2sf3): Delete expanders.
23609 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23611 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23615 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23616 * config/rs6000/rs6000.c: Ditto.
23618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23620 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23621 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23622 comparison_operator.
23624 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23626 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23627 * config/rs6000/rs6000.opt: Ditto.
23628 * config/rs6000/t-rtems: Ditto.
23630 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23632 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23633 TARGET_E500_SINGLE by 0, simplify.
23634 * config/rs6000/rs6000.c: Ditto.
23635 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23636 (spe_build_register_parallel): Delete.
23637 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23638 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23639 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23640 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23641 (E500_CONVERT): Delete.
23642 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23644 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23646 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23647 * config/rs6000/dfp.md: Ditto.
23648 (negdd2, *negdd2_fpr): Merge.
23649 (absdd2, *absdd2_fpr): Merge.
23650 (negtd2, *negtd2_fpr): Merge.
23651 (abstd2, *abstd2_fpr): Merge.
23652 * config/rs6000/e500.h: Delete file.
23653 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23654 TARGET_FPRS by 1 and simplify.
23655 * config/rs6000/rs6000-c.c: Ditto.
23656 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23657 TARGET_DF_SPE by 0.
23658 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23660 * config/rs6000/rs6000.md: Ditto.
23661 (floatdidf2, *floatdidf2_fpr): Merge.
23662 (move_from_CR_gt_bit): Delete.
23663 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23664 (E500_CR_IOR_COMPARE): Delete.
23665 (All patterns that require !TARGET_FPRS): Delete.
23666 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23668 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23670 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23672 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23674 * graphds.c (add_edge): Intitialize edge's attached data.
23675 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23676 pointer parameter. Call pointed function on each edge during
23677 graph traversing. Skip traversing the edge when the function
23679 (graphds_dfs, graphds_scc): Ditto.
23680 (for_each_edge): New parameter. Pass the new parameter to callback
23682 * graphds.h (skip_edge_callback): New function pointer type.
23683 (graphds_dfs, graphds_scc): New function pointer parameter.
23684 (graphds_edge_callback, for_each_edge): New parameter.
23686 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23688 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23689 out code checking if runtime alias check is possible to below ...
23690 Call the new function.
23691 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23692 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23694 2017-06-07 Marek Polacek <polacek@redhat.com>
23697 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23698 TYPE_OVERFLOW_WRAPS check.
23700 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23702 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23703 if versioning is required.
23704 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23705 peeling with the check for versioning.
23707 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23709 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23710 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23711 Set true to new parameter if new ssa variable is defined.
23712 (vect_gen_vector_loop_niters): Refactor. Set range information
23713 for the new vector loop bound variable.
23714 (vect_do_peeling): Ditto.
23716 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23718 * tree-affine.c (ssa.h): Include header file.
23719 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23720 has wrapping overflow behavior.
23722 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23724 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23726 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23728 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23729 (tree_to_aff_combination): ... here.
23731 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23733 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23734 reg_pressure model function.
23735 (ivopts_global_cost_for_size): Delete.
23736 (determine_set_costs, iv_ca_recount_cost): Call new model function
23737 ivopts_estimate_reg_pressure.
23739 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23741 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23742 expensive than udiv. Remove floating point cases from mod.
23744 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23746 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23747 Increase idiv cost.
23749 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23751 * config/aarch64/aarch64.md
23752 (copysignsf3): Fix mask generation.
23754 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23756 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23758 (class dump_manager): Add register_dumps method.
23759 * dumpfile.c: Include langhooks.h.
23760 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23761 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23762 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23763 (dump_manager::dump_register): Start with 512 entries instead of 32.
23764 (dump_manager::register_dumps): New method.
23765 * toplev.c (general_init): Instead of invoking register_dumps
23766 langhook, invoke register_dumps method on the dump manager.
23767 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23770 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23772 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23773 Say that # requires an associated define_split to exist, and that
23774 the define_split must be suitable for use after register allocation.
23776 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23778 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23779 (compute_outgoing_frequencies): Also initialize zero counts.
23780 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23781 around loops; preserve more of profile when nothing changes.
23783 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23785 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23787 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23788 * config/arm/arm-cpu-cdata.h: Regenerate.
23789 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23790 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23791 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23792 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23793 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23795 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23797 * config/arm/t-rmprofile: Likewise.
23798 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23800 2017-06-06 David S. Miller <davem@davemloft.net>
23803 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23804 blockage if function uses alloca.
23806 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23808 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23809 New "uid" fields to hold pretty-print IDs of group and ref.
23810 Memory references are now identified as <group_id>:<ref_id>
23811 instead of using [random] addresses.
23812 (dump_mem_details): Simplify, no functional change.
23813 (dump_mem_ref): Simplify and make output more concise.
23814 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23815 (find_or_create_group): Initialize group uid.
23816 (record_ref): Initialize ref uid. Improve debug output.
23817 (prune_group_by_reuse, should_issue_prefetch_p,)
23818 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23819 (mark_nontemporal_store, determine_loop_nest_reuse):
23820 Improve debug output.
23822 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23824 * dbgcnt.def (prefetch): New debug counter.
23825 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23826 (schedule_prefetches): Stop issueing prefetches if debug counter
23829 2017-06-06 Tom de Vries <tom@codesourcery.com>
23831 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23832 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23834 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23836 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23837 Use aarch64_reg_or_zero predicate for operand 4.
23838 (aarch64_compare_and_swap<mode> define_insn_and_split):
23839 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23840 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23842 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23844 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23845 (arm_compute_save_core_reg_mask): This.
23846 (thumb1_compute_save_reg_mask): Rename into ...
23847 (thumb1_compute_save_core_reg_mask): This.
23848 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23849 (arm_compute_frame_layout): Likewise.
23851 2017-06-06 Richard Biener <rguenther@suse.de>
23853 PR tree-optimization/80974
23854 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23855 keep or clear leaders SSA info.
23857 2017-06-06 Tom de Vries <tom@codesourcery.com>
23859 * config/nvptx/nvptx.c (split_mode_p): New function.
23860 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23862 2017-06-06 Tom de Vries <tom@codesourcery.com>
23864 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23866 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23869 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23872 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23874 * shrink-wrap.c (handle_simple_exit): Update profile.
23875 (try_shrink_wrapping): Upate profile.
23877 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23879 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23880 (tree_guess_outgoing_edge_probabilities): New.
23881 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23882 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23884 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23886 * ipa-split.c (split_function): Initialize return bb profile.
23888 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23890 * profile.c (compute_branch_probabilities): Also initialize
23891 EXIT_BLOCK profile.
23893 2017-06-06 Richard Biener <rguenther@suse.de>
23895 PR tree-optimization/80928
23896 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23897 (vect_analyze_loop_operations): Properly guard analysis for
23899 (vect_transform_loop): Likewise.
23900 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23901 (vect_model_induction_cost): Do not cost for pure SLP.
23902 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23903 of induction in inner loop vectorization.
23904 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23905 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23906 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23908 (vect_analyze_slp_cost_1): Cost induction.
23909 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23910 (vect_get_slp_vect_defs): Likewise.
23911 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23912 (vect_transform_stmt): Handle SLP reductions.
23913 * tree-vectorizer.h (vectorizable_induction): Adjust.
23915 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23917 * config/rs6000/rs6000.c (make_resolver_func): Update
23918 init_lowered_empty_function call.
23920 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23922 * doc/invoke.texi: Document the -fprofile-abs-path option.
23923 * common.opt (fprofile-abs-path): New option.
23924 * gcov-io.h (gcov_write_filename): Declare.
23925 * gcov-io.c (gcov_write_filename): New function.
23926 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23927 * profile.c (output_location): Likewise.
23929 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23931 * shring-wrap.c: Revert accidental commit.
23933 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23935 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23937 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23939 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23941 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23942 profile in callgraph edge.
23943 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23944 (apply_scale): Likewise.
23945 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23946 Also scale profile when inlining function with zero profile.
23947 (initialize_cfun): Update exit block profile even when it is zero.
23948 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23949 when profile is read.
23951 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23953 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23954 (CLONE_*): New constants to define the processors we can generate
23955 code for with the target_clone attribute.
23956 (rs6000_clone_map): New array to identify which clone processors
23957 the current program is running on.
23958 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23959 target_clone attribute.
23960 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23961 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23962 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23963 (cpu_expand_builtin): Add support for target_clone attribute.
23964 (rs6000_valid_attribute_p): Allow "default" attribute.
23965 (get_decl_name): New debug function to simplify printing the
23966 current function name in debugging statements.
23967 (rs6000_clone_priority): New functions to support the target_clone
23968 attribute, and be able to generate code to switch between ISA 2.05
23969 through ISA 3.0 (power6 through power9).
23970 (rs6000_compare_version_priority): Likewise.
23971 (rs6000_get_function_versions_dispatcher): Likewise.
23972 (make_resolver_func): Likewise.
23973 (add_condition_to_bb): Likewise.
23974 (dispatch_function_versions): Likewise.
23975 (rs6000_generate_version_dispatcher_body): Likewise.
23976 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23977 (fusion_gpr_load_p): Fix a spacing issue.
23978 * doc/extend.texi (Common Function Attributes): Document that the
23979 PowerPC supports the target_clone attribute.
23981 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23983 * config/arm/arm.h: explain F symbol found in description of ARM
23984 register allocation in its legend.
23986 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23988 * config/mips/frame-header-opt.c: Include profile-count.h.
23989 * config/riscv/riscv.c: Include profile-count.h
23991 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23993 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23995 (sm_set_flag_if_changed): Add bbs field.
23996 (execute_sm_if_changed_flag_set): Pass BBS.
23997 (execute_sm): Update.
23999 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24001 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24004 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24006 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24007 (peephole2): New peephole2 to emit the above.
24008 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24010 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24012 * config/aarch64/aarch64.c (define_peephole2 above
24013 *sub_<shift>_<mode>): New peephole.
24015 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24017 * config/i386/i386.c (make_resolver_func): Update.
24018 * Makefile.in: Add profile-count.h and profile-count.o
24019 * auto-profile.c (afdo_indirect_call): Update to new API.
24020 (afdo_set_bb_count): Update.
24021 (afdo_propagate_edge): Update.
24022 (afdo_propagate_circuit): Update.
24023 (afdo_calculate_branch_prob): Update.
24024 (afdo_annotate_cfg): Update.
24025 * basic-block.h: Include profile-count.h
24026 (struct edge_def): Turn count to profile_count.
24027 (struct basic_block_def): Likewie.
24028 (REG_BR_PROB_BASE): Move to profile-count.h
24029 (RDIV): Move to profile-count.h
24030 * bb-reorder.c (max_entry_count): Turn to profile_count.
24031 (find_traces): Update.
24032 (rotate_loop):Update.
24033 (connect_traces):Update.
24034 (sanitize_hot_paths):Update.
24035 * bt-load.c (migrate_btr_defs): Update.
24036 * cfg.c (RDIV): Remove.
24037 (init_flow): Use alloc_block.
24038 (alloc_block): Uninitialize count.
24039 (unchecked_make_edge): Uninitialize count.
24040 (check_bb_profile): Update.
24041 (dump_edge_info): Update.
24042 (dump_bb_info): Update.
24043 (update_bb_profile_for_threading): Update.
24044 (scale_bbs_frequencies_int): Update.
24045 (scale_bbs_frequencies_gcov_type): Update.
24046 (scale_bbs_frequencies_profile_count): New.
24047 * cfg.h (update_bb_profile_for_threading): Update.
24048 (scale_bbs_frequencies_profile_count): Declare.
24049 * cfgbuild.c (compute_outgoing_frequencies): Update.
24050 (find_many_sub_basic_blocks): Update.
24051 * cfgcleanup.c (try_forward_edges): Update.
24052 (try_crossjump_to_edge): Update.
24053 * cfgexpand.c (expand_gimple_tailcall): Update.
24054 (construct_exit_block): Update.
24055 * cfghooks.c (verify_flow_info): Update.
24056 (dump_bb_for_graph): Update.
24057 (split_edge): Update.
24058 (make_forwarder_block): Update.
24059 (duplicate_block): Update.
24060 (account_profile_record): Update.
24061 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24062 (get_estimated_loop_iterations): Update.
24063 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24064 (single_likely_exit): Update.
24065 * cfgloopmanip.c (scale_loop_profile): Update.
24067 (set_zero_probability): Update.
24068 (lv_adjust_loop_entry_edge): Update.
24069 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24070 (purge_dead_edges): Update.
24071 (rtl_account_profile_record): Update.
24072 * cgraph.c (cgraph_node::create): Uninitialize count.
24073 (symbol_table::create_edge): Uninitialize count.
24074 (cgraph_update_edges_for_call_stmt_node): Update.
24075 (cgraph_edge::dump_edge_flags): Update.
24076 (cgraph_node::dump): Update.
24077 (cgraph_edge::maybe_hot_p): Update.
24078 * cgraph.h: Include profile-count.h
24079 (create_clone), create_edge, create_indirect_edge): Update.
24080 (cgraph_node): Turn count to profile_count.
24081 (cgraph_edge0: Likewise.
24082 (make_speculative, clone): Update.
24083 (create_edge): Update.
24084 (init_lowered_empty_function): Update.
24085 * cgraphclones.c (cgraph_edge::clone): Update.
24086 (duplicate_thunk_for_node): Update.
24087 (cgraph_node::create_clone): Update.
24088 * cgraphunit.c (cgraph_node::analyze): Update.
24089 (cgraph_node::expand_thunk): Update.
24090 * final.c (dump_basic_block_info): Update.
24091 * gimple-streamer-in.c (input_bb): Update.
24092 * gimple-streamer-out.c (output_bb): Update.
24093 * graphite.c (print_global_statistics): Update.
24094 (print_graphite_scop_statistics): Update.
24095 * hsa-brig.c: Include basic-block.h.
24096 * hsa-dump.c: Include basic-block.h.
24097 * hsa-gen.c (T sum_slice): Update.
24098 (convert_switch_statements):Update.
24099 * hsa-regalloc.c: Include basic-block.h.
24100 * ipa-chkp.c (chkp_produce_thunks): Update.
24101 * ipa-cp.c (struct caller_statistics): Update.
24102 (init_caller_stats): Update.
24103 (gather_caller_stats): Update.
24104 (ipcp_cloning_candidate_p): Update.
24105 (good_cloning_opportunity_p): Update.
24106 (get_info_about_necessary_edges): Update.
24107 (dump_profile_updates): Update.
24108 (update_profiling_info): Update.
24109 (update_specialized_profile): Update.
24110 (perhaps_add_new_callers): Update.
24111 (decide_about_value): Update.
24112 (ipa_cp_c_finalize): Update.
24113 * ipa-devirt.c (struct odr_type_warn_count): Update.
24114 (struct decl_warn_count): Update.
24115 (struct final_warning_record): Update.
24116 (possible_polymorphic_call_targets): Update.
24117 (ipa_devirt): Update.
24118 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24119 * ipa-icf.c (sem_function::merge): Update.
24120 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24121 * ipa-inline.c (compute_uninlined_call_time): Update.
24122 (compute_inlined_call_time): Update.
24123 (want_inline_small_function_p): Update.
24124 (want_inline_self_recursive_call_p): Update.
24125 (edge_badness): Update.
24126 (lookup_recursive_calls): Update.
24127 (recursive_inlining): Update.
24128 (inline_small_functions): Update.
24129 (dump_overall_stats): Update.
24130 (dump_inline_stats): Update.
24131 * ipa-profile.c (ipa_profile_generate_summary): Update.
24132 (ipa_propagate_frequency): Update.
24133 (ipa_profile): Update.
24134 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24135 * ipa-utils.c (ipa_merge_profiles): Update.
24136 * loop-doloop.c (doloop_modify): Update.
24137 * loop-unroll.c (report_unroll): Update.
24138 (unroll_loop_runtime_iterations): Update.
24139 * lto-cgraph.c (lto_output_edge): Update.
24140 (lto_output_node): Update.
24141 (input_node): Update.
24142 (input_edge): Update.
24143 (merge_profile_summaries): Update.
24144 * lto-streamer-in.c (input_cfg): Update.
24145 * lto-streamer-out.c (output_cfg): Update.
24146 * mcf.c (create_fixup_graph): Update.
24147 (adjust_cfg_counts): Update.
24148 (sum_edge_counts): Update.
24149 * modulo-sched.c (sms_schedule): Update.
24150 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24151 * predict.c (maybe_hot_count_p): Update.
24152 (probably_never_executed): Update.
24153 (dump_prediction): Update.
24154 (combine_predictions_for_bb): Update.
24155 (propagate_freq): Update.
24156 (handle_missing_profiles): Update.
24157 (counts_to_freqs): Update.
24158 (rebuild_frequencies): Update.
24159 (force_edge_cold): Update.
24160 * predict.h: Include profile-count.h
24161 (maybe_hot_count_p, counts_to_freqs): UPdate.
24162 * print-rtl-function.c: Do not include cfg.h
24163 * print-rtl.c: Include basic-block.h
24164 * profile-count.c: New file.
24165 * profile-count.h: New file.
24166 * profile.c (is_edge_inconsistent): Update.
24167 (correct_negative_edge_counts): Update.
24168 (is_inconsistent): Update.
24169 (set_bb_counts): Update.
24170 (read_profile_edge_counts): Update.
24171 (compute_frequency_overlap): Update.
24172 (compute_branch_probabilities): Update; Initialize and deinitialize
24174 (branch_prob): Update.
24175 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24176 (edge_gcov_count): New.
24177 (bb_gcov_count): New.
24178 * shrink-wrap.c (try_shrink_wrapping): Update.
24179 * tracer.c (better_p): Update.
24180 * trans-mem.c (expand_transaction): Update.
24181 (ipa_tm_insert_irr_call): Update.
24182 (ipa_tm_insert_gettmclone_call): Update.
24183 * tree-call-cdce.c: Update.
24184 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24185 (gimple_duplicate_sese_tail): Update.
24186 (gimple_account_profile_record): Update.
24187 (execute_fixup_cfg): Update.
24188 * tree-inline.c (copy_bb): Update.
24189 (copy_edges_for_bb): Update.
24190 (initialize_cfun): Update.
24191 (freqs_to_counts): Update.
24192 (copy_cfg_body): Update.
24193 (expand_call_inline): Update.
24194 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24195 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24196 (try_unroll_loop_completely): Update.
24197 (try_peel_loop): Update.
24198 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24199 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24200 * tree-ssa-loop-split.c (connect_loops): Update.
24201 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24202 * tree-ssa-reassoc.c (branch_fixup): Update.
24203 * tree-ssa-tail-merge.c (replace_block_by): Update.
24204 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24205 (compute_path_counts): Update.
24206 (update_profile): Update.
24207 (recompute_probabilities): Update.
24208 (update_joiner_offpath_counts): Update.
24209 (estimated_freqs_path): Update.
24210 (freqs_to_counts_path): Update.
24211 (clear_counts_path): Update.
24212 (ssa_fix_duplicate_block_edges): Update.
24213 (duplicate_thread_path): Update.
24214 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24215 (struct switch_conv_info): Update.
24216 * tree-tailcall.c (decrease_profile): Update.
24217 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24218 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24219 * value-prof.c (check_counter): Update.
24220 (gimple_divmod_fixed_value): Update.
24221 (gimple_mod_pow2): Update.
24222 (gimple_mod_subtract): Update.
24223 (gimple_ic_transform): Update.
24224 (gimple_stringop_fixed_value): Update.
24225 * value-prof.h (gimple_ic): Update.
24227 2017-06-02 Carl Love <cel@us.ibm.com>
24229 * config/rs6000/rs6000-c: Add support for built-in functions
24230 vector double vec_doublee (vector signed int);
24231 vector double vec_doublee (vector unsigned int);
24232 vector double vec_doublee (vector float);
24233 vector double vec_doubleh (vector signed int);
24234 vector double vec_doubleh (vector unsigned int);
24235 vector double vec_doubleh (vector float);
24236 vector double vec_doublel (vector signed int);
24237 vector double vec_doublel (vector unsigned int);
24238 vector double vec_doublel (vector float);
24239 vector double vec_doubleo (vector signed int);
24240 vector double vec_doubleo (vector unsigned int);
24241 vector double vec_doubleo (vector float);.
24242 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24243 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24245 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24246 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24247 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24249 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24250 vec_doublel, vec_doubleh.
24251 * doc/extend.texi: Update the built-in documentation file for the
24252 new built-in functions.
24254 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24257 * ipa-inline-analysis.c (free_growth_caches): Set
24258 edge_removal_hook_holder to NULL after removing it.
24260 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24262 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24263 comparision with zero.
24265 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24266 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24267 for early expansion of vec_min and vec_max builtins.
24268 (builtin_function_type): Add min/max unsigned variants to those
24269 identified as having unsigned arguments.
24271 2017-06-02 Olivier Hainque <hainque@adacore.com>
24273 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24275 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24277 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24278 Use VALL_F16 iterator rather than VALL.
24280 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24282 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24283 Emit CBNZ inside loop when doing a strong exchange and comparing
24284 against zero. Generate the CC flags after the loop.
24286 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24288 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24289 (dl_section_ref): New.
24290 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24291 On AIX, append an expression to subtract the size of the
24292 section length to dl_section_ref.
24294 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24296 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24297 for early expansion of vector absolute builtins.
24299 2017-06-02 Richard Biener <rguenther@suse.de>
24301 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24302 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24304 2017-06-02 Richard Biener <rguenther@suse.de>
24306 PR tree-optimization/80948
24307 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24308 stmt order as well.
24310 2017-06-02 Richard Biener <rguenther@suse.de>
24312 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24314 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24315 for inductions as relevant.
24317 2017-06-02 Richard Biener <rguenther@suse.de>
24319 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24320 (vectorizable_induction): ... this. Remove dead code.
24322 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24324 * builtins. (expand_builtin_alloca): Remove second parameter and
24325 infer its value from the first parameter instead.
24326 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24328 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24330 PR rtl-optimization/80903
24331 * loop-doloop.c (add_test): Unshare sequence.
24333 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24335 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24337 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24339 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24341 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24342 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24343 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24344 (xlogue_layout::get_stub_rtx): Make static.
24345 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24346 (xlogue_layout::compute_stub_managed_regs): Rename to...
24347 (xlogue_layout::count_stub_managed_regs): ...this.
24348 (xlogue_layout::is_stub_managed_reg): New function.
24349 (xlogue_layout::m_stub_names): Rename to...
24350 (xlogue_layout::s_stub_names): ...this, make static.
24351 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24352 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24353 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24354 xlogue_layout::s_stub_names): Instantiate statics.
24355 (stub_managed_regs): Remove.
24356 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24357 (disable_call_ms2sysv_xlogues): Rename to...
24358 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24359 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24361 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24362 change after reload_completed.
24363 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24365 (ix86_expand_prologue): Likewise.
24366 (ix86_expand_epilogue): Likewise.
24367 (ix86_expand_split_stack_prologue): Likewise.
24368 (ix86_compute_frame_layout): Remove frame parameter ...
24369 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24370 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24372 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24373 (ix86_frame): Move from here ...
24374 * config/i386/i386.h (ix86_frame): ... to here.
24375 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24376 complete ix86_frame data structure instead. Remove some_ld_name.
24378 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24380 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24381 symbols that hold a DECL_VALUE_EXPR.
24383 2017-06-01 Martin Jambor <mjambor@suse.cz>
24385 PR tree-optimization/80898
24386 * tree-sra.c (process_subtree_disqualification): Removed.
24387 (disqualify_candidate): Do not acll
24388 process_subtree_disqualification.
24389 (subtree_mark_written_and_enqueue): New function.
24390 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24391 RHS has been disqualified and re-queue LHS if necessary. Apart
24392 from that, ignore disqualified RHS.
24394 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24396 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24397 address fetch for z10 or later.
24399 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24401 * config/arc/arc.md (tst_movb): Add guard when splitting.
24403 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24405 * config/arc/arc.c (arc_can_eliminate): Test against
24406 arc_frame_pointer_needed.
24408 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24410 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24411 to prevent store reordering.
24412 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24413 (type): Add block type.
24414 (stack_tie): Define special instruction to be used in
24417 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24419 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24420 constraint. It is not valid for the pattern.
24421 (noncommutative_binary_comparison): Likewise.
24423 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24425 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24428 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24430 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24431 be used by the reg-alloc.
24433 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24435 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24436 reg-alloc when having mul64 or mul32x16 instructions.
24437 (mulsidi3): Likewise.
24438 (umulsidi3): Likewise.
24439 (mulsi32x16): New pattern.
24440 (mulsi64): Likewise.
24441 (mulsidi64): Likewise.
24442 (umulsidi64): Likewise.
24443 (MUL32x16_REG): Define.
24444 (mul64_600): Use MUL32x16_REG.
24445 (mac64_600): Likewise.
24446 (umul64_600): Likewise.
24447 (umac64_600): Likewise.
24449 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24451 * config/arc/arc.md (mulsi3_700): Make it commutative.
24453 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24455 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24457 (*sign_extendsidi2_insn): Likewise for movstosw.
24459 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24461 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24462 the type of the input discriminant value. Convert the
24463 discriminant value of signedness vary.
24465 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24467 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24468 Add to -Wall section.
24470 2017-06-01 Richard Biener <rguenther@suse.de>
24472 PR middle-end/66313
24473 * fold-const.c (fold_plusminus_mult_expr): If the factored
24474 factor may be zero use a wrapping type for the inner operation.
24475 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24476 and handle moved defs.
24477 (process_assignment): Properly guard the unary op case. Return a
24478 tri-state indicating that moving the stmt before the call may allow
24479 to continue. Pass through to_move.
24480 (find_tail_calls): Handle moving unrelated defs before
24483 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24486 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24487 splitter result in the canonical way.
24489 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24491 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24492 also for 32bit target. Update insn attributes.
24493 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24495 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24497 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24498 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24499 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24500 (_mm_maskz_min_ss): New intrinsics.
24502 2017-05-31 Martin Liska <mliska@suse.cz>
24504 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24505 Change comment style to one we normally use.
24506 (vectorizable_reduction): Likewise.
24507 (vectorizable_induction): Likewise.
24508 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24509 (vectorizable_call): Likewise.
24510 (vectorizable_simd_clone_call): Likewise.
24511 (vectorizable_conversion): Likewise.
24512 (vectorizable_assignment): Likewise.
24513 (vectorizable_shift): Likewise.
24514 (vectorizable_operation): Likewise.
24515 (vectorizable_store): Likewise.
24516 (vectorizable_load): Likewise.
24517 * tree-vectorizer.h: Likewise.
24519 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24521 * passes.c (emergency_dump_function): New.
24522 * tree-pass.h (emergency_dump_function): Declare.
24523 * plugin.c (plugins_internal_error_function): Remove.
24524 * plugin.h (plugins_internal_error_function): Remove declaration.
24525 * toplev.c (internal_error_function): New static function. Use it...
24526 (general_init): ...here.
24528 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24530 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24531 (arc_rtx_costs): Add costs for new patterns.
24532 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24533 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24535 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24537 * tree-ssa-strlen.c (get_next_strinfo): New function.
24538 (get_stridx_plus_constant): Use it.
24539 (zero_length_string): Likewise.
24540 (adjust_related_strinfos): Likewise.
24541 (adjust_last_stmt): Likewise.
24543 2017-05-31 Richard Biener <rguenther@suse.de>
24546 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24547 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24549 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24551 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24552 loop_vinfo argument and use of dependence distance vectors.
24553 Check instead whether the two references differ only in their
24554 initial value and assume that they have the same alignment if the
24555 difference is a multiple of the vector alignment.
24556 (vect_analyze_data_refs_alignment): Update call accordingly.
24558 2017-05-31 Martin Liska <mliska@suse.cz>
24561 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24563 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24565 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24566 (create_intersect_range_checks): Move from ...
24567 * tree-data-ref.c (create_intersect_range_checks_index)
24568 (create_intersect_range_checks): ... to here.
24569 (create_runtime_alias_checks): New function factored from ...
24570 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24571 here. Call above function.
24572 * tree-data-ref.h (create_runtime_alias_checks): New function.
24574 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24576 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24577 segment length for dr_b and compute it in wide_int.
24579 2017-05-31 Richard Biener <rguenther@suse.de>
24581 PR tree-optimization/80906
24582 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24583 and pass through iv_map.
24584 (copy_bb_and_scalar_dependences): Adjust.
24585 (translate_pending_phi_nodes): Likewise.
24586 (copy_loop_close_phi_args): Handle code-generating IVs instead
24589 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24591 * diagnostic-color.c (color_dict): Add "type-diff".
24592 (parse_gcc_colors): Update comment.
24593 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24594 -fdiagnostics-show-template-tree and -fno-elide-type.
24595 (GCC_COLORS): Add type-diff to example.
24597 (-fdiagnostics-show-template-tree): New.
24598 (-fno-elide-type): New.
24599 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24600 the pp_format_decoder callback. Call any m_format_postprocessor's
24602 (pretty_printer::pretty_printer): Initialize
24603 m_format_postprocessor.
24604 (pretty_printer::~pretty_printer): Delete any
24605 m_format_postprocessor.
24606 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24607 (class format_postprocessor): New class.
24608 (struct pretty_printer::format_decoder): Document the new parameters.
24609 (struct pretty_printer::m_format_postprocessor): New field.
24610 * tree-diagnostic.c (default_tree_printer): Update for new
24611 bool and const char ** params.
24612 * tree-diagnostic.h (default_tree_printer): Likewise.
24614 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24616 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24617 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24619 rs6000_gen_cell_microcode code.
24620 (rs6000_final_prescan_insn): Delete.
24621 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24622 "warn-cell-microcode" entries.
24623 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24624 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24625 throughout. Change cc_reg_not_micro_cr0_operand to
24626 cc_reg_not_cr0_operand throughout.
24627 (*extendhi<mode>2_noload): Delete.
24628 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24629 (mwarn-cell-microcode): Delete.
24630 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24631 -mgen-cell-microcode and -mwarn-cell-microcode.
24633 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24636 * config/i386/constraints.md (Yd): New constraint.
24638 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24639 and (?Yd, r) alternatives. Update insn attributes.
24640 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24641 and (?*Yd, r) alternatives. Update insn attributes.
24642 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24644 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24646 * gimplify.c (gimplify_modify_expr): Don't create a
24647 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24650 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24652 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24654 2017-05-30 Richard Biener <rguenther@suse.de>
24656 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24657 and reduc_def fields.
24658 (STMT_VINFO_REDUC_TYPE): New define.
24659 (STMT_VINFO_REDUC_DEF): Likewise.
24660 (vect_force_simple_reduction): Adjust prototype.
24661 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24662 (vect_is_simple_reduction): Remove check_reduction argument.
24663 (vect_force_simple_reduction): Adjust and set
24664 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24665 (vectorizable_reduction): Do not re-do reduction analysis
24666 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24667 * tree-parloops.c (gather_scalar_reductions): Adjust.
24669 2017-05-30 Richard Biener <rguenther@suse.de>
24671 PR middle-end/80901
24672 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24675 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24677 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24678 Introduce unknown_misalignment parameter and remove vf.
24679 (vect_peeling_hash_get_lowest_cost):
24680 Pass unknown_misalignment parameter.
24681 (vect_enhance_data_refs_alignment):
24682 Fix unsupportable data ref treatment.
24684 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24686 * tree-vect-data-refs.c (vect_get_data_access_cost):
24687 Workaround for SLP handling.
24688 (vect_enhance_data_refs_alignment):
24689 Compute costs for doing no peeling at all, compare to the best
24690 peeling costs so far and avoid peeling if cheaper.
24692 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24694 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24695 Return peeling info and set costs to zero for unlimited cost
24697 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24698 with unknown misalignment. Compute and costs for unknown
24699 misalignment, compare them to the costs for known misalignment
24700 and choose the cheapest for peeling.
24702 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24704 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24705 (vect_get_peeling_costs_all_drs): Create function.
24706 (vect_peeling_hash_get_lowest_cost):
24707 Use vect_get_peeling_costs_all_drs.
24708 (vect_peeling_supportable): Create function.
24709 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24711 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24713 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24714 DR_HAS_NEGATIVE_STEP.
24715 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24716 (vect_enhance_data_refs_alignment): Use.
24717 (vect_duplicate_ssa_name_ptr_info): Use.
24718 * tree-vectorizer.h (dr_misalignment): Use.
24719 (known_alignment_for_access_p): Use.
24721 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24724 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24726 (has_section_name): New function.
24728 2017-05-30 Martin Liska <mliska@suse.cz>
24731 * auto-profile.c (get_function_decl_from_block): Fix
24734 2017-05-30 Richard Biener <rguenther@suse.de>
24736 PR middle-end/80876
24737 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24739 2017-05-30 Martin Liska <mliska@suse.cz>
24741 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24742 * dumpfile.h (struct dump_file_info): Remove ctors.
24744 2017-05-30 Martin Liska <mliska@suse.cz>
24746 * predict.def: Fix GNU coding style.
24748 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24750 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24751 Mark 'to' argument with ATTRIBUTE_UNUSED.
24753 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24755 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24756 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24757 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24760 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24762 * doc/install.texi (Options specification): Restore entry of
24763 --enable-sjlj-exceptions.
24765 2017-05-27 Michael Eager <eager@eagercon.com>
24768 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24770 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24772 * config/microblaze/microblaze.h
24773 (FIXED_REGISTERS): Update in macro.
24774 (CALL_USED_REGISTERS): Update in macro.
24776 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24778 * doc/install.texi: Add links to macOS binary distributions.
24780 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24784 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24786 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24788 2017-05-26 Martin Liska <mliska@suse.cz>
24790 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24792 2017-05-26 Martin Liska <mliska@suse.cz>
24794 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24795 always leading ';; '.
24796 (dump_bb_info): Likewise.
24797 (brief_dump_cfg): Likewise.
24798 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24799 * dumpfile.c: Remove usage of TDF_VERBOSE.
24800 * dumpfile.h (enum dump_kind): Likewise.
24801 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24802 * print-tree.c (debug_verbose): Remove.
24803 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24804 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24805 * tree-diagnostic.c (default_tree_printer): Replace
24806 TDF_DIAGNOSTIC with TDF_SLIM.
24808 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24810 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24811 in parameter loop, rather than loop_vinfo.
24812 (create_intersect_range_checks): Ditto.
24813 (vect_create_cond_for_alias_checks): Update call to above functions.
24815 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24817 PR tree-optimization/80815
24818 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24819 for merging runtime alias checks. Handle negative DR_STEPs.
24821 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24823 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24825 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24826 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24827 out code pruning runtime alias checks.
24828 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24829 factored out from above.
24830 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24832 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24834 (prune_runtime_alias_test_list): New decalaration.
24836 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24838 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24839 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24840 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24841 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24842 (operator==, comp_dr_with_seg_len_pair): Ditto.
24843 (vect_prune_runtime_alias_test_list): Ditto.
24845 2017-05-26 Martin Liska <mliska@suse.cz>
24848 * params.def: Bound partial-inlining-entry-probability param.
24850 2017-05-26 Marek Polacek <polacek@redhat.com>
24853 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24856 2017-05-26 Richard Biener <rguenther@suse.de>
24858 PR tree-optimization/80842
24859 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24862 2017-05-26 Richard Biener <rguenther@suse.de>
24864 PR tree-optimization/80844
24865 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24867 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24869 * doc/md.texi (Machine Constraints): Update x86 family
24870 machine constraints section to match 'config/i386/constraints.md'.
24872 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24874 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24876 2017-05-25 Nathan Sidwell <nathan@acm.org>
24878 * doc/invoke.texi (--enable-languages): Update documentation.
24880 2017-05-25 Martin Liska <mliska@suse.cz>
24882 * dumpfile.c: Add TDF_FOLDING.
24883 * dumpfile.h (enum dump_kind): Likewise.
24884 * genmatch.c (dt_simplify::gen_1): Use it.
24886 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24888 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24890 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24892 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24893 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24895 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24897 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24898 * match.pd (X == C): Rewrite it here.
24899 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24900 with_certain_nonzero_bits2): New predicates.
24901 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24903 2017-05-24 Nathan Sidwell <nathan@acm.org>
24905 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24908 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24911 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24913 * config/powerpcspe: New port. Files are copied from the rs6000
24914 port, with "rs6000" in filenames replaced by "powerpcspe".
24916 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24918 PR rtl-optimization/80754
24919 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24921 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24923 * config/sparc/sparc.md (length): Return the correct value for -mflat
24924 sibcalls to match output_sibcall.
24926 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24930 * config/rs6000/rs6000.c (struct machine_function): Add new field
24932 (rs6000_get_separate_components): Init that field, use it.
24933 (rs6000_components_for_bb): Use the field.
24935 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24937 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24939 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24941 PR middle-end/80823
24942 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24944 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24947 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24948 address against address_operand predicate.
24949 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24951 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24953 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24954 parameters passed indirectly.
24956 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24958 * config/i386/i386.md (*movdi_internal): Remove SSE4
24959 alternative 18 (?r, *v). Update insn attributes.
24960 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24961 Update insn attributes.
24962 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24963 Update insn attributes.
24964 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24965 alternative 1 (r, v). Remove isa attribute.
24966 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24967 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24968 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24970 2017-05-23 Tom de Vries <tom@codesourcery.com>
24972 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24975 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24977 * cgraphunit.c (symbol_table::process_new_functions): Update.
24978 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24979 (inline_generate_summary): Rename to ...
24980 (ipa_fn_summary_generate): ... this one.
24981 (inline_read_summary): Rename to ...
24982 (ipa_fn_summary_read): ... this one.
24983 (inline_write_summary): Rename to ...
24984 (ipa_fn_summary_write): ... this one.
24985 (inline_free_summary): Rename to ...
24986 (ipa_free_fn_summary): ... this one.
24987 (pass_data_local_fn_summary, pass_local_fn_summary,
24988 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24989 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24990 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24991 make_pass_ipa_fn_summary): New.
24992 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24993 inline_write_summary, inline_free_summary): Remove.
24994 (ipa_free_fn_summary) : New.
24995 * ipa-inline.c (ipa_inline): Update.
24996 (pass_ipa_inline): Do not generate summaries.
24997 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24999 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25000 and add pass_ipa_fn_summary.
25001 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25003 (make_pass_inline_parameters): Remove.
25005 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25007 * omp-low.c (struct omp_context): Remove "default_kind" member.
25010 * omp-offload.c (execute_oacc_device_lower): Remove the
25011 parallelism dimensions function attributes for unparallelized
25012 OpenACC kernels constructs.
25014 2017-05-23 Martin Liska <mliska@suse.cz>
25016 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25018 (cgraph_edge::make_speculative): Likewise.
25019 (cgraph_edge::resolve_speculation): Likewise.
25020 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25021 (cgraph_node::dump): Likewise.
25022 * cgraph.h: Likewise.
25023 * cgraphunit.c (analyze_functions): Likewise.
25024 (symbol_table::compile): Likewise.
25025 * ipa-cp.c (print_all_lattices): Likewise.
25026 (determine_versionability): Likewise.
25027 (initialize_node_lattices): Likewise.
25028 (ipcp_verify_propagated_values): Likewise.
25029 (estimate_local_effects): Likewise.
25030 (update_profiling_info): Likewise.
25031 (create_specialized_node): Likewise.
25032 (perhaps_add_new_callers): Likewise.
25033 (decide_about_value): Likewise.
25034 (decide_whether_version_node): Likewise.
25035 (identify_dead_nodes): Likewise.
25036 (ipcp_store_bits_results): Likewise.
25037 * ipa-devirt.c (dump_targets): Likewise.
25038 (ipa_devirt): Likewise.
25039 * ipa-icf.c (sem_item::dump): Likewise.
25040 (sem_function::equals): Likewise.
25041 (sem_variable::equals): Likewise.
25042 (sem_item_optimizer::read_section): Likewise.
25043 (sem_item_optimizer::execute): Likewise.
25044 (congruence_class::dump): Likewise.
25045 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25046 (dump_inline_summary): Likewise.
25047 (estimate_node_size_and_time): Likewise.
25048 (inline_analyze_function): Likewise.
25049 * ipa-inline-transform.c (inline_call): Likewise.
25050 * ipa-inline.c (report_inline_failed_reason): Likewise.
25051 (want_early_inline_function_p): Likewise.
25052 (edge_badness): Likewise.
25053 (update_edge_key): Likewise.
25054 (inline_small_functions): Likewise.
25055 * ipa-profile.c (ipa_profile): Likewise.
25056 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25057 (ipa_make_edge_direct_to_target): Likewise.
25058 (remove_described_reference): Likewise.
25059 (ipa_impossible_devirt_target): Likewise.
25060 (propagate_controlled_uses): Likewise.
25061 (ipa_print_node_params): Likewise.
25062 (ipcp_transform_function): Likewise.
25063 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25064 (propagate_pure_const): Likewise.
25065 * ipa-reference.c (generate_summary): Likewise.
25066 (read_write_all_from_decl): Likewise.
25067 (propagate): Likewise.
25068 (ipa_reference_read_optimization_summary): Likewise.
25069 * ipa-utils.c (ipa_merge_profiles): Likewise.
25070 * ipa.c (walk_polymorphic_call_targets): Likewise.
25071 (symbol_table::remove_unreachable_nodes): Likewise.
25072 (ipa_single_use): Likewise.
25073 * passes.c (execute_todo): Likewise.
25074 * predict.c (drop_profile): Likewise.
25075 * symtab.c (symtab_node::get_dump_name): New function.
25076 (symtab_node::dump_name): Likewise.
25077 (symtab_node::dump_asm_name): Likewise.
25078 (symtab_node::dump_references): Likewise.
25079 (symtab_node::dump_referring): Likewise.
25080 (symtab_node::dump_base): Likewise.
25081 (symtab_node::debug_symtab): Likewise.
25082 * tree-sra.c (convert_callers_for_node): Likewise.
25083 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25084 * value-prof.c (init_node_map): Likewise.
25086 2017-05-23 Martin Liska <mliska@suse.cz>
25088 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25089 and symtab_node::debug_symtab to symbol_table::debug.
25090 * cgraphunit.c (analyze_functions): Use the renamed function.
25091 (symbol_table::compile): Likewise.
25092 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25093 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25094 * passes.c (execute_todo): Likewise.
25095 * symtab.c (symbol_table::dump): New function.
25096 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25098 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25100 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25101 that nonconst implies exec.
25103 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25105 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25106 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25107 (inline_edge_summary_vec): Turn into ...
25108 (ipa_call_summaries): ... this one.
25109 (redirect_to_unreachable, edge_set_predicate,
25110 evaluate_properties_for_edge, inline_summary_alloc,
25111 reset_ipa_call_summary, reset_inline_summary,
25112 inline_summary_t::duplicate): Update.
25113 (inline_edge_duplication_hook): Turn to ...
25114 (ipa_call_summary_t::duplicate): ... this one.
25115 (inline_edge_removal_hook): Turn to ...
25116 (ipa_call_summary_t::remove): ... this one.
25117 (dump_inline_edge_summary): Turn to ...
25118 (dump_ipa_call_summary): ... this one.
25119 (estimate_function_body_sizes): Update.
25120 (inline_update_callee_summaries): Update.
25121 (remap_edge_change_prob): Update.
25122 (remap_edge_summaries): Update.
25123 (inline_merge_summary): Update.
25124 (do_estimate_edge_time): Update.
25125 (inline_generate_summary): Update.
25126 (inline_read_section): Update.
25127 (inline_read_summary): Update.
25128 (inline_free_summary): Update.
25129 * ipa-inline.c (can_inline_edge_p): Update.
25130 (compute_inlined_call_time): Update.
25131 (want_inline_small_function_p): Update.
25132 (edge_badness): Update.
25133 (early_inliner): Update.
25134 * ipa-inline.h (inline_edge_summary): Turn to ...
25135 (ipa_call_summary): ... this one.
25136 (ipa_call_summary_t): New class.
25137 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25138 (ipa_call_summaries): New.
25139 (inline_edge_summary): Remove.
25140 (estimate_edge_growth): Update.
25141 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25142 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25143 * ipa-split.c (execute_split_functions): Update.
25144 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25146 2017-05-23 Tom de Vries <tom@codesourcery.com>
25148 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25149 attributes): Document rdrand effective target.
25151 2017-05-23 Tom de Vries <tom@codesourcery.com>
25153 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25154 attributes): Sort alphabetically.
25156 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25158 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25160 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25163 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25164 V2DF/V2DI splat into two separate patterns, one that handles
25165 registers, and the other that only handles memory. Drop support
25166 for splatting from a GPR on ISA 2.07 and then splitting the
25167 splat into direct move and splat.
25168 (vsx_splat_<mode>_reg): Likewise.
25169 (vsx_splat_<mode>_mem): Likewise.
25171 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25173 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25175 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25177 PR middle-end/80809
25178 * omp-low.c (finish_taskreg_remap): New function.
25179 (finish_taskreg_scan): If unit size of ctx->record_type
25180 is non-constant, unshare the size expression and replace
25181 decls in it with possible outer var refs.
25183 PR middle-end/80809
25184 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25185 GOVD_SHARED rather than GOVD_PRIVATE with it.
25186 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25187 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25189 PR middle-end/80853
25190 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25191 as last argument to build_outer_var_ref for pointer bases of array
25192 section reductions.
25194 2017-05-19 Martin Sebor <msebor@redhat.com>
25196 * print-tree.c (print_node): Print DECL_READ_P flag.
25198 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25200 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25201 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25202 * cgraph.c: Likewise.
25203 * cgraphunit.c: Likewise.
25204 * gengtype.c: Likewise.
25205 * ipa-cp.c: Likewise.
25206 * ipa-devirt.c: Likewise.
25207 * ipa-icf.c: Likewise.
25208 * ipa-predicate.c: Likewise.
25209 * ipa-profile.c: Likewise.
25210 * ipa-prop.c: Likewise.
25211 * ipa-split.c: Likewise.
25213 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25214 edge_predicate_pool, dump_inline_hints,
25215 inline_summary::account_size_time, redirect_to_unreachable,
25216 edge_set_predicate, set_hint_predicate,
25217 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25218 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25219 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25220 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25221 ipa_call_summary_t::remove, initialize_growth_caches,
25222 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25223 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25224 mark_modified, unmodified_parm_1, unmodified_parm,
25225 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25226 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25227 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25228 will_be_nonconstant_predicate, record_modified_bb_info,
25229 get_minimal_bb, record_modified, param_change_prob,
25230 phi_result_unknown_predicate, predicate_for_phi_result,
25231 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25232 estimate_function_body_sizes, compute_inline_parameters,
25233 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25234 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25235 inline_update_callee_summaries, remap_edge_change_prob,
25236 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25237 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25238 inline_analyze_function, inline_summary_t::insert,
25239 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25240 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25241 inline_free_summary): Move to ipa-fnsummary.h
25242 (predicate_t): Remove.
25243 * ipa-fnsummary.c: New file.
25244 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25245 (enum inline_hints_vals, inline_hints, agg_position_info,
25246 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25247 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25248 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25249 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25250 inline_read_summary, inline_write_summary, inline_free_summary,
25251 inline_analyze_function, initialize_inline_failed,
25252 inline_merge_summary, inline_update_overall_summary,
25253 compute_inline_parameters): Move to ipa-fnsummary.h
25254 * ipa-fnsummary.h: New file.
25255 * ipa-inline-transform.h: Include ipa-inline.h.
25256 * ipa-inline.c: LIkewise.
25258 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25260 * ipa-inline.c (edge_badness): Use inlined_time instead of
25261 inline_summaries->get.
25263 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25265 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25267 2017-05-22 Nathan Sidwell <nathan@acm.org>
25269 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25270 (fdump-lang): Document 'raw' option.
25271 * dumpfile.h (TDI_tu): Delete.
25272 * dumpfile.c (dump_files): Remove translation-unit.
25273 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25275 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25277 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25278 command option from $(AWK) call.
25279 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25280 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25281 [FORMAT]: Remove handling of variable.
25282 * config/avr/t-multilib: Regenerate.
25284 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25286 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25288 (dump_inline_summary): Do not print self_time.
25289 (estimate_function_body_sizes): Do not set self_time.
25290 (compute_inline_parameters): Likewise.
25291 (inline_read_section, inline_write_summary): Do not stream self_time.
25292 * ipa-inline.h (inline_summary): Drop self_time.
25294 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25296 * ipa-inline-analysis.c (account_size_time): Rename to ...
25297 (inline_summary::account_size_time): ... this one.
25298 (reset_ipa_call_summary): Turn to ...
25299 (ipa_call_summary::reset): ... this one.
25300 (reset_inline_summary): Turn to ...
25301 (inline_summary::reset): ... this one.
25302 (inline_summary_t::remove): Update.
25303 (inline_summary_t::duplicate): Update.
25304 (ipa_call_summary_t::remove): Update.
25305 (dump_inline_summary): Update.
25306 (estimate_function_body_sizes): Update.
25307 (compute_inline_parameters): Update.
25308 (estimate_node_size_and_time): Update.
25309 (inline_merge_summary): Update.
25310 (inline_update_overall_summary): Update.
25311 (inline_read_section): Update.
25312 (inline_write_summary): Update.
25313 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25314 add account_size_time and reset member functions.
25315 (ipa_call_summary): Add reset function.
25316 * ipa-predicate.h (predicate::operator &): Constify.
25318 2017-05-22 Richard Biener <rguenther@suse.de>
25320 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25322 2017-05-19 Jason Merrill <jason@redhat.com>
25324 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25326 2017-05-19 Marek Polacek <polacek@redhat.com>
25329 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25330 TYPE_OVERFLOW_WRAPS checks.
25332 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25334 * tree-core.h (enum omp_clause_default_kind): Add
25335 "OMP_CLAUSE_DEFAULT_PRESENT".
25336 * tree-pretty-print.c (dump_omp_clause): Handle it.
25337 * gimplify.c (enum gimplify_omp_var_data): Add
25338 "GOVD_MAP_FORCE_PRESENT".
25339 (gimplify_adjust_omp_clauses_1): Map it to
25340 "GOMP_MAP_FORCE_PRESENT".
25341 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25343 * gimplify.c (oacc_default_clause): Clarify.
25345 2017-05-19 Nathan Sidwell <nathan@acm.org>
25347 LANG_HOOK_REGISTER_DUMPS
25348 * toplev.c (general_init): Call register dump lang hook.
25349 * doc/invoke.texi: Document -fdump-lang option family.
25350 * dumpfile.c (dump_files): Remove class dump here.
25351 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25352 * dumpfile.h (tree_dump_index): Remove TDI_class.
25353 * langhooks-def.h (lhd_register_dumps): Declare.
25354 (LANG_HOOKS_REGISTER_DUMPS): Define.
25355 (LANG_HOOKS_INITIALIZER): Add it.
25356 * langhooks.c (lhd_register_dumps): Define.
25357 * langhooks.h (struct lang_hooks): Add register_dumps.
25359 2017-05-19 Nathan Sidwell <nathan@acm.org>
25361 * context.h (context::set_passes): New.
25362 * context.c (context::context): Do not create pass manager.
25363 * toplev.c (general_init): Create pass manager here.
25365 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25367 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25368 use this splitter if two add or or instructions would also work for
25369 the constant we want to generate.
25371 2017-05-19 Richard Biener <rguenther@suse.de>
25374 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25375 predicate evaluation.
25377 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25379 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25381 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25384 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25386 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25387 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25388 (inline_edge_summary_vec): Turn into ...
25389 (ipa_call_summaries): ... this one.
25390 (redirect_to_unreachable, edge_set_predicate,
25391 evaluate_properties_for_edge, inline_summary_alloc,
25392 reset_ipa_call_summary, reset_inline_summary,
25393 inline_summary_t::duplicate): Update.
25394 (inline_edge_duplication_hook): Turn to ...
25395 (ipa_call_summary_t::duplicate): ... this one.
25396 (inline_edge_removal_hook): Turn to ...
25397 (ipa_call_summary_t::remove): ... this one.
25398 (dump_inline_edge_summary): Turn to ...
25399 (dump_ipa_call_summary): ... this one.
25400 (estimate_function_body_sizes): Update.
25401 (inline_update_callee_summaries): Update.
25402 (remap_edge_change_prob): Update.
25403 (remap_edge_summaries): Update.
25404 (inline_merge_summary): Update.
25405 (do_estimate_edge_time): Update.
25406 (inline_generate_summary): Update.
25407 (inline_read_section): Update.
25408 (inline_read_summary): Update.
25409 (inline_free_summary): Update.
25410 * ipa-inline.c (can_inline_edge_p): Update.
25411 (compute_inlined_call_time): Update.
25412 (want_inline_small_function_p): Update.
25413 (edge_badness): Update.
25414 (early_inliner): Update.
25415 * ipa-inline.h (inline_edge_summary): Turn to ...
25416 (ipa_call_summary): ... this one.
25417 (ipa_call_summary_t): New class.
25418 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25419 (ipa_call_summaries): New.
25420 (inline_edge_summary): Remove.
25421 (estimate_edge_growth): Update.
25422 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25423 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25424 * ipa-split.c (execute_split_functions): Update.
25425 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25427 2017-05-19 Richard Biener <rguenther@suse.de>
25429 PR middle-end/80764
25430 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25432 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25434 * config/rs6000/rs6000.c (struct machine_function): Add field
25435 fpr_is_wrapped_separately.
25436 (rs6000_get_separate_components): Use 64 components. Handle the
25437 new FPR components.
25438 (rs6000_components_for_bb): Handle the FPR components.
25439 (rs6000_emit_prologue_components): Handle the FPR components.
25440 (rs6000_emit_epilogue_components): Handle the FPR components.
25441 (rs6000_set_handled_components): Handle the FPR components.
25442 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25443 that are already separately shrink-wrapped.
25444 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25445 that are already separately shrink-wrapped.
25447 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25450 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25453 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25454 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25455 cases where the register allocator uses a move and an offsettable
25456 memory operation to/from a FPR register on ISA 2.06/2.07.
25457 (define_peephole2 for Altivec d-form store): Likewise.
25459 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25462 * config/i386/mmx.md (*mov<mode>_internal): Enable
25463 alternatives 11, 12, 13 and 14 also for 32bit targets.
25464 Remove alternatives 15, 16, 17 and 18.
25465 * config/i386/sse.md (vec_concatv2di): Change
25466 alternative (!x, *y) to (x, ?!*Yn).
25468 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25470 * dumpfile.h (enum dump_kind): Remove stray comma.
25472 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25474 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25475 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25476 predicate::num_conditions
25477 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25478 (CHANGED): turn into predicate::changed.
25479 (agg_position_info): Move to ipa-predicate.h
25480 (add_condition, predicate::add_clause, predicate::operator &=,
25481 predicate::or_with, predicate::evaluate, predicate::probability,
25482 dump_condition, dump_clause, predicate::dump,
25483 predicate::remap_after_duplication, predicate::remap_after_inlining,
25484 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25485 (evaluate_conditions_for_known_args): Update.
25486 (set_cond_stmt_execution_predicate): Update.
25487 * ipa-inline.h: Include ipa-predicate.h
25488 (condition, inline_param_summary, conditions, agg_position_info,
25489 predicate): Move to ipa-predicate.h
25490 * ipa-predicate.c: New file.
25491 * ipa-predicate.h: New file.
25493 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25495 * final.c (leaf_function_p): Check we are not in a sequence.
25497 2017-05-18 Martin Liska <mliska@suse.cz>
25499 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25500 * dumpfile.c (dump_register): Use new enum dump_kind.
25501 (get_dump_file_name): Likewise.
25502 (dump_enable_all): Likewise.
25503 (dump_switch_p_1): Likewise.
25504 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25505 * dumpfile.h (enum dump_kind): New enum type.
25506 (struct dump_file_info): Create constructor and
25507 format fields and comments.
25508 * passes.c (pass_manager::register_one_dump_file):
25510 * statistics.c (statistics_early_init): Likewise.
25511 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25512 TDF_TREE with TDF_SLIM.
25513 (gather_memory_references_ref): Likewise.
25515 2017-05-18 Martin Liska <mliska@suse.cz>
25517 * vec.h (struct vnull): Use it.
25519 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25521 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25522 (true_predicate, false_predicate, true_predicate_p,
25523 false_predicate_p): Remove.
25524 (single_cond_predicate, not_inlined_predicate): Turn to member function
25526 (add_condition): Update.
25527 (add_clause): Turn to...
25528 (predicate::add_clause): ... this one; update; allow passing NULL
25530 (and_predicates): Turn to ...
25531 (predicate::operator &=): ... this one.
25532 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25533 (or_predicates): Turn to ...
25534 (predicate::or_with): ... this one.
25535 (evaluate_predicate): Turn to ...
25536 (predicate::evaluate): ... this one.
25537 (predicate_probability): Turn to ...
25538 (predicate::probability): ... this one.
25539 (dump_condition): Update.
25540 (dump_predicate): Turn to ...
25541 (predicate::dump): ... this one.
25542 (account_size_time): Update.
25543 (edge_set_predicate): Update.
25544 (set_hint_predicate): UPdate.
25545 (evaluate_conditions_for_known_args): Update.
25546 (evaluate_properties_for_edge): Update.
25547 (remap_predicate_after_duplication): Turn to...
25548 (predicate::remap_after_duplication): ... this one.
25549 (remap_hint_predicate_after_duplication): Update.
25550 (inline_summary_t::duplicate): UPdate.
25551 (dump_inline_edge_summary): Update.
25552 (dump_inline_summary): Update.
25553 (set_cond_stmt_execution_predicate): Update.
25554 (set_switch_stmt_execution_predicate): Update.
25555 (compute_bb_predicates): Update.
25556 (will_be_nonconstant_expr_predicate): Update.
25557 (will_be_nonconstant_predicate): Update.
25558 (phi_result_unknown_predicate): Update.
25559 (predicate_for_phi_result): Update.
25560 (array_index_predicate): Update.
25561 (estimate_function_body_sizes): Update.
25562 (estimate_node_size_and_time): Update.
25563 (estimate_ipcp_clone_size_and_time): Update.
25564 (remap_predicate): Rename to ...
25565 (predicate::remap_after_inlining): ... this one.
25566 (remap_hint_predicate): Update.
25567 (inline_merge_summary): Update.
25568 (inline_update_overall_summary): Update.
25569 (estimate_size_after_inlining): Update.
25570 (read_predicate): Rename to ...
25571 (predicate::stream_in): ... this one.
25572 (read_inline_edge_summary): Update.
25573 (write_predicate): Rename to ...
25574 (predicate::stream_out): ... this one.
25575 (write_inline_edge_summary): Update.
25576 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25577 (clause_t): Turn to uint32_t
25578 (predicate): Turn to class; implement constructor and operators
25580 (size_time_entry): Update.
25581 (inline_summary): Update.
25582 (inline_edge_summary): Update.
25584 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25586 * fold-const.c (fold_binary_loc): Move transformation...
25587 * match.pd (C - X CMP X): ... here.
25589 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25591 * config/sparc/sparc.c (sparc_option_override): Set function
25592 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25593 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25595 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25597 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25599 2017-05-18 Marek Polacek <polacek@redhat.com>
25602 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25603 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25604 gimple_assign_load_p.
25606 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25608 PR middle-end/80692
25609 * real.c (do_compare): Give decimal_do_compare preference over
25610 comparing just the signs.
25612 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25614 * doc/md.texi (Canonicalization of Instructions): Describe the
25615 canonical form of instructions that inherently set a condition
25618 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25620 PR middle-end/80775
25621 * tree-cfg.c: Move deletion of unreachable case statements to after
25622 the merging of consecutive case labels.
25624 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25626 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25627 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25628 restoring of callee-saved registers.
25630 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25632 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25633 * config/visium/visium.c (single_set_and_flags): Likewise.
25634 * config/visium/visium.md (Substitutions): Likewise.
25636 2017-05-17 Martin Liska <mliska@suse.cz>
25638 * cfg.c: Introduce dump_flags_t type and
25639 use it instead of int type.
25641 * cfghooks.c: Likewise.
25642 * cfghooks.h (struct cfg_hooks): Likewise.
25643 * cfgrtl.c: Likewise.
25644 * cfgrtl.h: Likewise.
25645 * cgraph.c (cgraph_node::get_body): Likewise.
25646 * coretypes.h: Likewise.
25647 * domwalk.c: Likewise.
25648 * domwalk.h: Likewise.
25649 * dumpfile.c (struct dump_option_value_info): Likewise.
25650 (dump_enable_all): Likewise.
25651 (dump_switch_p_1): Likewise.
25652 (opt_info_switch_p): Likewise.
25653 * dumpfile.h (enum tree_dump_index): Likewise.
25654 (struct dump_file_info): Likewise.
25655 * genemit.c: Likewise.
25656 * generic-match-head.c: Likewise.
25657 * gengtype.c (open_base_files): Likewise.
25658 * gimple-pretty-print.c: Likewise.
25659 * gimple-pretty-print.h: Likewise.
25660 * graph.c (print_graph_cfg): Likewise.
25661 * graphite-scop-detection.c (dot_all_sese): Likewise.
25662 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25663 * loop-unroll.c (report_unroll): Likewise.
25664 * passes.c (pass_manager::register_one_dump_file): Likewise.
25665 * print-tree.c: Likewise.
25666 * statistics.c: Likewise.
25667 * tree-cfg.c: Likewise.
25668 * tree-cfg.h: Likewise.
25669 * tree-dfa.c: Likewise.
25670 * tree-dfa.h: Likewise.
25671 * tree-dump.c (dump_function): Likewise.
25672 * tree-dump.h (struct dump_info): Likewise.
25673 * tree-pretty-print.c: Likewise.
25674 * tree-pretty-print.h: Likewise.
25675 * tree-ssa-live.c: Likewise.
25676 * tree-ssa-live.h: Likewise.
25677 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25678 * tree-vect-loop.c: Likewise.
25679 * tree-vect-slp.c: Likewise.
25681 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25682 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25684 PR tree-optimization/80457
25685 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25686 of all arguments to a statement as scalar_to_vec operations.
25687 (vectorizable_call): Adjust call to vect_model_simple_cost for
25689 (vectorizable_conversion): Likewise.
25690 (vectorizable_assignment): Likewise.
25691 (vectorizable_shift): Likewise.
25692 (vectorizable_operation): Likewise.
25693 (vectorizable_comparison): Likewise.
25694 (vect_is_simple_cond): Record the def types for operands.
25695 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25696 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25697 for statement argument count.
25699 2017-05-16 Carl Love <cel@us.ibm.com>
25701 * config/rs6000/rs6000-c: Add support for built-in functions
25702 vector unsigned long long vec_bperm (vector unsigned long long,
25703 vector unsigned char)
25704 vector signed long long vec_mule (vector signed int,
25706 vector unsigned long long vec_mule (vector unsigned int,
25707 vector unsigned int)
25708 vector signed long long vec_mulo (vector signed int,
25710 vector unsigned long long vec_mulo (vector unsigned int,
25711 vector unsigned int)
25712 vector signed char vec_sldw (vector signed char,
25713 vector signed char,
25715 vector unsigned char vec_sldw (vector unsigned char,
25716 vector unsigned char,
25718 vector signed short vec_sldw (vector signed short,
25719 vector signed short,
25721 vector unsigned short vec_sldw (vector unsigned short,
25722 vector unsigned short,
25724 vector signed int vec_sldw (vector signed int,
25727 vector unsigned int vec_sldw (vector unsigned int,
25728 vector unsigned int,
25730 vector signed long long vec_sldw (vector signed long long,
25731 vector signed long long,
25733 vector unsigned long long vec_sldw (vector unsigned long long,
25734 vector unsigned long long,
25736 * config/rs6000/rs6000-c: Add support for built-in functions
25737 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25738 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25739 * doc/extend.texi: Update the built-in documentation for the
25740 new built-in functions.
25742 2017-05-16 Marek Polacek <polacek@redhat.com>
25746 * tree.c (save_expr): Don't fold the expression.
25748 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25750 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25751 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25752 and (?*y,m). Update insn attributes.
25754 2017-05-16 Martin Liska <mliska@suse.cz>
25756 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25757 flags argument of print_gimple_stmt, print_gimple_expr,
25758 print_generic_stmt and print_generic_expr.
25759 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25760 * coretypes.h: Likewise.
25761 * except.c (dump_eh_tree): Likewise.
25762 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25763 * gimple-pretty-print.h: Likewise.
25764 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25765 (backprop::push_to_worklist): Likewise.
25766 (backprop::pop_from_worklist): Likewise.
25767 (backprop::process_use): Likewise.
25768 (backprop::intersect_uses): Likewise.
25769 (note_replacement): Likewise.
25770 * gimple-ssa-store-merging.c
25771 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25772 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25773 (pass_store_merging::execute): Likewise.
25774 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25775 (ssa_base_cand_dump_callback): Likewise.
25776 (dump_incr_vec): Likewise.
25777 (replace_refs): Likewise.
25778 (replace_mult_candidate): Likewise.
25779 (create_add_on_incoming_edge): Likewise.
25780 (create_phi_basis): Likewise.
25781 (insert_initializers): Likewise.
25782 (all_phi_incrs_profitable): Likewise.
25783 (introduce_cast_before_cand): Likewise.
25784 (replace_one_candidate): Likewise.
25785 * gimplify.c (gimplify_expr): Likewise.
25786 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25787 (set_rename): Likewise.
25788 (rename_uses): Likewise.
25789 (copy_loop_phi_nodes): Likewise.
25790 (add_close_phis_to_merge_points): Likewise.
25791 (copy_loop_close_phi_args): Likewise.
25792 (copy_cond_phi_args): Likewise.
25793 (graphite_copy_stmts_from_block): Likewise.
25794 (translate_pending_phi_nodes): Likewise.
25795 * graphite-poly.c (print_pdr): Likewise.
25796 (dump_gbb_cases): Likewise.
25797 (dump_gbb_conditions): Likewise.
25798 (print_scop_params): Likewise.
25799 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25800 (build_cross_bb_scalars_use): Likewise.
25801 (gather_bbs::before_dom_children): Likewise.
25802 * hsa-dump.c (dump_hsa_immed): Likewise.
25803 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25804 (get_replacement_map): Likewise.
25805 * ipa-inline-analysis.c (dump_condition): Likewise.
25806 (estimate_function_body_sizes): Likewise.
25807 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25808 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25809 * ipa-prop.c (ipa_dump_param): Likewise.
25810 (ipa_print_node_jump_functions_for_edge): Likewise.
25811 (ipa_modify_call_arguments): Likewise.
25812 (ipa_modify_expr): Likewise.
25813 (ipa_dump_param_adjustments): Likewise.
25814 (ipa_dump_agg_replacement_values): Likewise.
25815 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25816 * ipa-pure-const.c (check_stmt): Likewise.
25817 (pass_nothrow::execute): Likewise.
25818 * ipa-split.c (execute_split_functions): Likewise.
25819 * omp-offload.c (dump_oacc_loop_part): Likewise.
25820 (dump_oacc_loop): Likewise.
25821 * trans-mem.c (tm_log_emit): Likewise.
25822 (tm_memopt_accumulate_memops): Likewise.
25823 (dump_tm_memopt_set): Likewise.
25824 (dump_tm_memopt_transform): Likewise.
25825 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25826 (print_loop): Likewise.
25827 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25828 (chkp_gather_checks_info): Likewise.
25829 (chkp_get_check_result): Likewise.
25830 (chkp_remove_check_if_pass): Likewise.
25831 (chkp_use_outer_bounds_if_possible): Likewise.
25832 (chkp_reduce_bounds_lifetime): Likewise.
25833 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25834 (chkp_mark_completed_bounds): Likewise.
25835 (chkp_register_incomplete_bounds): Likewise.
25836 (chkp_mark_invalid_bounds): Likewise.
25837 (chkp_maybe_copy_and_register_bounds): Likewise.
25838 (chkp_build_returned_bound): Likewise.
25839 (chkp_get_bound_for_parm): Likewise.
25840 (chkp_build_bndldx): Likewise.
25841 (chkp_get_bounds_by_definition): Likewise.
25842 (chkp_generate_extern_var_bounds): Likewise.
25843 (chkp_get_bounds_for_decl_addr): Likewise.
25844 * tree-chrec.c (chrec_apply): Likewise.
25845 * tree-data-ref.c (dump_data_reference): Likewise.
25846 (dump_subscript): Likewise.
25847 (dump_data_dependence_relation): Likewise.
25848 (analyze_overlapping_iterations): Likewise.
25849 * tree-inline.c (expand_call_inline): Likewise.
25850 (tree_function_versioning): Likewise.
25851 * tree-into-ssa.c (dump_defs_stack): Likewise.
25852 (dump_currdefs): Likewise.
25853 (dump_names_replaced_by): Likewise.
25854 (dump_update_ssa): Likewise.
25855 (update_ssa): Likewise.
25856 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25857 * tree-parloops.c (build_new_reduction): Likewise.
25858 (try_create_reduction_list): Likewise.
25859 (ref_conflicts_with_region): Likewise.
25860 (oacc_entry_exit_ok_1): Likewise.
25861 (oacc_entry_exit_single_gang): Likewise.
25862 * tree-pretty-print.h: Likewise.
25863 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25864 (get_scalar_evolution): Likewise.
25865 (add_to_evolution): Likewise.
25866 (get_loop_exit_condition): Likewise.
25867 (analyze_evolution_in_loop): Likewise.
25868 (analyze_initial_condition): Likewise.
25869 (analyze_scalar_evolution): Likewise.
25870 (instantiate_scev): Likewise.
25871 (number_of_latch_executions): Likewise.
25872 (gather_chrec_stats): Likewise.
25873 (final_value_replacement_loop): Likewise.
25874 (scev_const_prop): Likewise.
25875 * tree-sra.c (dump_access): Likewise.
25876 (disqualify_candidate): Likewise.
25877 (create_access): Likewise.
25878 (reject): Likewise.
25879 (maybe_add_sra_candidate): Likewise.
25880 (create_access_replacement): Likewise.
25881 (analyze_access_subtree): Likewise.
25882 (analyze_all_variable_accesses): Likewise.
25883 (sra_modify_assign): Likewise.
25884 (initialize_constant_pool_replacements): Likewise.
25885 (find_param_candidates): Likewise.
25886 (decide_one_param_reduction): Likewise.
25887 (replace_removed_params_ssa_names): Likewise.
25888 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25889 * tree-ssa-copy.c (dump_copy_of): Likewise.
25890 (copy_prop_visit_cond_stmt): Likewise.
25891 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25892 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25893 (record_equivalences_from_stmt): Likewise.
25894 * tree-ssa-dse.c (compute_trims): Likewise.
25895 (delete_dead_call): Likewise.
25896 (delete_dead_assignment): Likewise.
25897 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25898 (forward_propagate_into_cond): Likewise.
25899 (pass_forwprop::execute): Likewise.
25900 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25901 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25903 (move_computations_worker): Likewise.
25904 (execute_sm): Likewise.
25905 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25906 (remove_exits_and_undefined_stmts): Likewise.
25907 (remove_redundant_iv_tests): Likewise.
25908 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25909 (adjust_iv_update_pos): Likewise.
25910 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25911 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25912 (value_replacement): Likewise.
25913 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25914 * tree-ssa-pre.c (print_pre_expr): Likewise.
25915 (get_representative_for): Likewise.
25916 (create_expression_by_pieces): Likewise.
25917 (insert_into_preds_of_block): Likewise.
25918 (eliminate_insert): Likewise.
25919 (eliminate_dom_walker::before_dom_children): Likewise.
25920 (eliminate): Likewise.
25921 (remove_dead_inserted_code): Likewise.
25922 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25923 * tree-ssa-reassoc.c (get_rank): Likewise.
25924 (eliminate_duplicate_pair): Likewise.
25925 (eliminate_plus_minus_pair): Likewise.
25926 (eliminate_not_pairs): Likewise.
25927 (undistribute_ops_list): Likewise.
25928 (eliminate_redundant_comparison): Likewise.
25929 (update_range_test): Likewise.
25930 (optimize_range_tests_var_bound): Likewise.
25931 (optimize_vec_cond_expr): Likewise.
25932 (rewrite_expr_tree): Likewise.
25933 (rewrite_expr_tree_parallel): Likewise.
25934 (linearize_expr): Likewise.
25935 (break_up_subtract): Likewise.
25936 (linearize_expr_tree): Likewise.
25937 (attempt_builtin_powi): Likewise.
25938 (attempt_builtin_copysign): Likewise.
25939 (transform_stmt_to_copy): Likewise.
25940 (transform_stmt_to_multiply): Likewise.
25941 (dump_ops_vector): Likewise.
25942 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25943 (print_scc): Likewise.
25944 (set_ssa_val_to): Likewise.
25945 (visit_reference_op_store): Likewise.
25946 (visit_use): Likewise.
25947 (sccvn_dom_walker::before_dom_children): Likewise.
25948 (run_scc_vn): Likewise.
25949 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25951 (expr_hash_elt::print): Likewise.
25952 (const_and_copies::pop_to_marker): Likewise.
25953 (const_and_copies::record_const_or_copy_raw): Likewise.
25954 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25955 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25956 (dump_predicates): Likewise.
25957 (find_uninit_use): Likewise.
25958 (warn_uninitialized_phi): Likewise.
25959 (pass_late_warn_uninitialized::execute): Likewise.
25960 * tree-ssa.c (verify_vssa): Likewise.
25961 (verify_ssa): Likewise.
25962 (maybe_optimize_var): Likewise.
25963 * tree-vrp.c (dump_value_range): Likewise.
25964 (dump_all_value_ranges): Likewise.
25965 (dump_asserts_for): Likewise.
25966 (register_edge_assert_for_2): Likewise.
25967 (vrp_visit_cond_stmt): Likewise.
25968 (vrp_visit_switch_stmt): Likewise.
25969 (vrp_visit_stmt): Likewise.
25970 (vrp_visit_phi_node): Likewise.
25971 (simplify_cond_using_ranges_1): Likewise.
25972 (fold_predicate_in): Likewise.
25973 (evrp_dom_walker::before_dom_children): Likewise.
25974 (evrp_dom_walker::push_value_range): Likewise.
25975 (evrp_dom_walker::pop_value_range): Likewise.
25976 (execute_early_vrp): Likewise.
25978 2017-05-16 Richard Biener <rguenther@suse.de>
25980 * dwarf2out.c (loc_list_from_tree_1): Do not create
25981 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25983 2017-05-16 Richard Biener <rguenther@suse.de>
25985 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25987 (note_variable_value_in_expr): If we resolved the decl ref
25988 do not push to the stack.
25990 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25992 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25993 operations in fast-math mode.
25994 (vaddq_f16): Likewise.
25995 (vmul_f16): Likewise.
25996 (vmulq_f16): Likewise.
25997 (vsub_f16): Likewise.
25998 (vsubq_f16): Likewise.
25999 * config/arm/neon.md (add<mode>3): New.
26001 (fma:<VH:mode>3): New. Also remove outdated comment.
26004 2017-05-16 Martin Liska <mliska@suse.cz>
26008 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26009 (odr_types_equivalent_p): Likewise.
26011 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26013 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26015 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26018 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26019 non-interunit SSE move alternatives with '?'.
26020 (zero-extendsidi peephole2): New peephole to skip intermediate
26021 general register in SSE zero-extend sequence.
26023 2017-05-15 Jeff Law <law@redhat.com>
26025 * reorg.c (relax_delay_slots): Create a new variable to hold
26026 the temporary target rather than clobbering TARGET_LABEL.
26028 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26029 missing argument to extract_bit_field call.
26030 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26032 2017-05-15 Martin Liska <mliska@suse.cz>
26035 * gcc.c (process_command): Do not allow empty argument of -o option.
26037 2017-05-15 Renlin Li <renlin.li@arm.com>
26039 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26040 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26041 * config/aarch64/constraints.md (Usf): Add long call check.
26042 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26043 (call_value): Likewise.
26044 (sibcall): Likewise.
26045 (sibcall_value): Likewise.
26047 (call_value_insn): New.
26048 (sibcall_insn): Update rtx pattern.
26049 (sibcall_value_insn): Likewise.
26050 (call_internal): Remove.
26051 (call_value_internal): Likewise.
26052 (sibcall_internal): Likewise.
26053 (sibcall_value_internal): Likewise.
26054 (call_reg): Likewise.
26055 (call_symbol): Likewise.
26056 (call_value_reg): Likewise.
26057 (call_value_symbol): Likewise.
26059 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26062 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26064 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26066 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26067 compatible with CCGOCmode and with CCZmode.
26069 2017-05-14 Martin Sebor <msebor@redhat.com>
26071 PR middle-end/77671
26072 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26073 (gimple_fold_builtin_snprintf): Same.
26074 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26075 (gimple_fold_builtin_snprintf): Same.
26076 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26077 of character types.
26078 (is_call_safe): New function.
26079 (try_substitute_return_value): Call it.
26080 (try_simplify_call): New function.
26081 (pass_sprintf_length::handle_gimple_call): Call it.
26083 2017-05-14 Martin Sebor <msebor@redhat.com>
26085 PR middle-end/80669
26086 * builtins.c (expand_builtin_stpncpy): Simplify.
26088 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26090 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26091 * config/i386/i386.h
26092 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26093 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26094 (struct machine_function): Add new members call_ms2sysv,
26095 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26096 (struct machine_frame_state): New fields sp_realigned and
26097 sp_realigned_offset.
26098 * config/i386/i386.c
26099 (enum xlogue_stub): New enum.
26100 (enum xlogue_stub_sets): New enum.
26101 (class xlogue_layout): New class.
26102 (struct ix86_frame): New fields stack_realign_allocate_offset,
26103 stack_realign_offset and outlined_save_offset. Modify comments to
26104 detail stack layout when using out-of-line stubs.
26105 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26106 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26107 -mcall-ms2sysv-xlogues.
26108 (stub_managed_regs): New static variable.
26109 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26110 registers managed by out-of-line stub.
26111 (disable_call_ms2sysv_xlogues): New function.
26112 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26113 m->call_ms2sysv when appropriate and compute frame layout for
26115 (sp_valid_at, fp_valid_at): New inline functions.
26116 (choose_basereg): New function.
26117 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26119 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26120 Use align parameter of choose_baseaddr to generated aligned SSE movs
26122 (pro_epilogue_adjust_stack): Modify to track
26123 machine_frame_state::sp_realigned.
26124 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26125 (ix86_nsaved_sseregs): Likewise.
26126 (ix86_emit_save_regs): Likewise.
26127 (ix86_emit_save_regs_using_mov): Likewise.
26128 (ix86_emit_save_sse_regs_using_mov): Likewise.
26129 (get_scratch_register_on_entry): Likewise.
26130 (gen_frame_set): New function.
26131 (gen_frame_load): Likewise.
26132 (gen_frame_store): Likewise.
26133 (emit_outlined_ms2sysv_save): Likewise.
26134 (emit_outlined_ms2sysv_restore): Likewise.
26135 (ix86_expand_prologue): Modify stack re-alignment code and call
26136 emit_outlined_ms2sysv_save when appropriate.
26137 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26138 parameter rtx_insn *insn, which allows the function to be used to only
26139 generate the notes.
26140 (ix86_expand_epilogue): Modify validity checks of frame and stack
26141 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26142 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26143 * config/i386/predicates.md
26144 (save_multiple): New predicate.
26145 (restore_multiple): Likewise.
26146 * config/i386/sse.md
26147 (save_multiple<mode>): New pattern.
26148 (save_multiple_realign<mode>): Likewise.
26149 (restore_multiple<mode>): Likewise.
26150 (restore_multiple_and_return<mode>): Likewise.
26151 (restore_multiple_leave_return<mode>): Likewise.
26152 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26154 2017-05-14 Julia Koval <julia.koval@intel.com>
26156 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26157 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26158 (__builtin_ia32_xsetbv): New builtins.
26159 * config/i386/i386.c (ix86_expand_special_args_builtin):
26161 (ix86_expand_builtin): Special expand for new intrinsics.
26162 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26163 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26164 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26166 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26168 * cfganal.c (inverted_post_order_compute): Change argument type
26170 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26171 * df-core.c (rest_of_handle_df_initialize): Adjust.
26172 (rest_of_handle_df_finish): Likewise.
26173 (df_analyze_1): Likewise.
26174 (df_analyze): Likewise.
26175 (loop_inverted_post_order_compute): Change argument to be a vec *.
26176 (df_analyze_loop): Adjust.
26177 (df_get_n_blocks): Likewise.
26178 (df_get_postorder): Likewise.
26179 * df.h (struct df_d): Change field to be a vec.
26180 * lcm.c (compute_laterin): Adjust.
26181 (compute_available): Likewise.
26182 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26183 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26184 * tree-ssa-pre.c (compute_antic): Likewise.
26186 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26188 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26189 (depth_first_search::depth_first_search): Change structure init
26190 function to this constructor.
26191 (depth_first_search::add_bb): Rename function to this member.
26192 (depth_first_search::execute): Likewise.
26193 (flow_dfs_compute_reverse_finish): Adjust.
26195 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26197 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26198 (longest_simple_path): Likewise.
26199 * shrink-wrap.c (spread_components): Likewise.
26200 (disqualify_problematic_components): Likewise.
26201 (emit_common_heads_for_components): Likewise.
26202 (emit_common_tails_for_components): Likewise.
26203 (insert_prologue_epilogue_for_components): Likewise.
26205 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26207 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26210 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26212 * df-core.c (df_set_blocks): Start using auto_bitmap.
26213 (df_compact_blocks): Likewise.
26214 * df-problems.c (df_rd_confluence_n): Likewise.
26215 * df-scan.c (df_insn_rescan_all): Likewise.
26216 (df_process_deferred_rescans): Likewise.
26217 (df_update_entry_block_defs): Likewise.
26218 (df_update_exit_block_uses): Likewise.
26219 (df_entry_block_bitmap_verify): Likewise.
26220 (df_exit_block_bitmap_verify): Likewise.
26221 (df_scan_verify): Likewise.
26222 * lra-constraints.c (lra_constraints): Likewise.
26223 (undo_optional_reloads): Likewise.
26224 (lra_undo_inheritance): Likewise.
26225 * lra-remat.c (calculate_gen_cands): Likewise.
26226 (do_remat): Likewise.
26227 * lra-spills.c (assign_spill_hard_regs): Likewise.
26228 (spill_pseudos): Likewise.
26229 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26230 (bitmap_set_subtract_values): Likewise.
26232 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26234 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26235 management with auto_bitmap.
26236 (fix_inter_tick): Likewise.
26237 (fix_recovery_deps): Likewise.
26238 * ira.c (add_store_equivs): Likewise.
26239 (find_moveable_pseudos): Likewise.
26240 (split_live_ranges_for_shrink_wrap): Likewise.
26241 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26242 (rtx_reuse_manager::seen_def_p): Likewise.
26243 (rtx_reuse_manager::set_seen_def): Likewise.
26244 * print-rtl.h (class rtx_reuse_manager): Likewise.
26246 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26248 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26250 (migrate_btr_def): Likewise.
26251 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26252 * df-core.c (loop_post_order_compute): Likewise.
26253 (loop_inverted_post_order_compute): Likewise.
26254 * hsa-common.h: Likewise.
26255 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26256 * init-regs.c (initialize_uninitialized_regs): Likewise.
26257 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26258 (inline_small_functions): Likewise.
26259 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26260 * ira.c (combine_and_move_insns): Likewise.
26261 (build_insn_chain): Likewise.
26262 * loop-invariant.c (find_invariants): Likewise.
26263 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26264 * predict.c (tree_predict_by_opcode): Likewise.
26265 (predict_paths_leading_to): Likewise.
26266 (predict_paths_leading_to_edge): Likewise.
26267 (estimate_loops_at_level): Likewise.
26268 (estimate_loops): Likewise.
26269 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26270 (spread_components): Likewise.
26271 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26272 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26273 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26274 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26275 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26276 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26277 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26278 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26279 (mark_threaded_blocks): Likewise.
26280 (thread_through_all_blocks): Likewise.
26281 * tree-ssa.c (verify_ssa): Likewise.
26282 (execute_update_addresses_taken): Likewise.
26283 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26285 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26287 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26289 (post_order_compute): Likewise.
26290 (inverted_post_order_compute): Likewise.
26291 (pre_and_rev_post_order_compute_fn): Likewise.
26293 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26295 * genrecog.c (int_set::int_set): Explicitly construct our
26296 auto_vec base class.
26297 * vec.h (auto_vec::auto_vec): New constructor.
26299 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26301 * bitmap.h (class auto_bitmap): New constructor taking
26302 bitmap_obstack * argument.
26304 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26306 * bitmap.h (class auto_bitmap): Change type of m_bits to
26307 bitmap_head, and adjust ctor / dtor and member operators.
26309 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26311 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26312 when returned register mode doesn't match original mode.
26314 2017-05-12 Jeff Law <law@redhat.com>
26315 Jakub Jelinek <jakub@redhat.com>
26317 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26318 we look for cc setter after the compare-elim changes.
26319 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26320 within the vector to match what compare-elim now expects.
26321 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26322 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26324 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26325 after the compare-elim changes.
26326 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26327 the vector to match what compare-elim now expects.
26328 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26329 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26330 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26331 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26332 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26334 * config/visium/visium.c (single_set_and_flags): Fix where
26335 we look for cc setter after the compare-elim changes.
26336 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26337 with the vector to match what compare-elim now expects.
26338 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26339 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26340 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26341 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26342 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26343 (neg<mode>2_insn_set_overflow): Likewise.
26345 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26347 PR middle-end/79794
26348 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26349 maybe_expand_insn call, set ops[0].target. If still set after call,
26350 set alt_rtl. Add extra arg to recursive calls.
26351 (extract_bit_field): Add alt_rtl argument. Pass to
26353 * expmed.h (extract_bit_field): Fix prototype.
26354 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26355 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26356 to extract_bit_field_calls.
26357 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26358 Pass alt_rtl to extract_bit_field calls.
26359 * calls.c (store_unaligned_arguments_into_psuedos)
26360 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26361 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26363 * optabs.h (struct expand_operand): Add target bitfield.
26365 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26367 * compare-elim.c (try_eliminate_compare): Canonicalize
26368 operation with embedded compare to
26369 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26370 (set (reg) (operation)].
26372 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26374 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26377 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26378 cost of adding a carry flag for ADC instruction.
26379 [case MINUS]: Ignore the cost of subtracting a carry flag
26380 for SBB instruction.
26382 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26384 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26386 * config/rs6000/bmiintrin.h: New file.
26387 * config/rs6000/bmi2intrin.h: New file.
26388 * config/rs6000/x86intrin.h: New file.
26390 2017-05-12 Jeff Law <law@redhat.com>
26392 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26395 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26397 PR middle-end/80707
26398 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26400 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26402 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26403 early expansion of vector divide builtins.
26404 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26405 builtins identified as having unsigned arguments.
26407 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26409 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26410 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26411 expansion of vector logical operations (and, andc, or, xor,
26414 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26416 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26417 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26419 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26422 early GIMPLE expansion of vector multiplies.
26424 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26426 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26427 TARGET_HAVE_MOVT conditional.
26428 (movt splitter): Likewise.
26430 2017-05-12 Richard Biener <rguenther@suse.de>
26432 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26433 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26434 Fold all stmts not inplace.
26436 2017-05-12 Richard Biener <rguenther@suse.de>
26438 PR tree-optimization/80713
26439 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26440 inserted_exprs bit for not removed stmts.
26442 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26444 PR middle-end/69921
26445 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26446 parallelized" attribute for parallelized OpenACC kernels.
26447 * omp-offload.c (execute_oacc_device_lower): Use it.
26449 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26450 Set "oacc kernels" attribute.
26451 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26452 parameter. Adjust all users.
26453 (oacc_fn_attrib_kernels_p): Remove function.
26454 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26455 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26456 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26457 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26458 assert "oacc kernels" attribute is set.
26460 2017-05-11 Carl Love <cel@us.ibm.com>
26462 * config/rs6000/rs6000-c: Add support for built-in functions
26463 vector unsigned char vec_popcnt (vector signed char)
26464 vector unsigned char vec_popcnt (vector unsigned char)
26465 vector unsigned short vec_popcnt (vector signed short)
26466 vector unsigned short vec_popcnt (vector unsigned short)
26467 vector unsigned int vec_popcnt (vector signed int)
26468 vector unsigned int vec_popcnt (vector unsigned int)
26469 vector unsigned long long vec_popcnt (vector signed long long)
26470 vector unsigned long long vec_popcnt (vector unsigned long long)
26471 vector signed long long vec_slo (vector signed long long,
26472 vector signed char)
26473 vector signed long long vec_slo (vector signed long long,
26474 vector unsigned char)
26475 vector unsigned long long vec_slo (vector unsigned long long,
26476 vector signed char)
26477 vector unsigned long long vec_slo (vector unsigned long long,
26478 vector unsigned char)
26479 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26480 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26481 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26482 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26483 * doc/extend.texi: Update the built-in documentation file for the
26484 new built-in functions.
26486 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26488 * attribs.h (sorted_attr_string): Move machine independent
26489 functions for target clone support from the i386 port to common
26490 code. Rename ix86_function_versions to common_function_versions.
26491 Rename make_name to make_unique_name.
26492 (common_function_versions): Likewise.
26493 (make_unique_name): Likewise.
26494 (make_dispatcher_decl): Likewise.
26495 (is_function_default_version): Likewise.
26496 * attribs.c (attr_strcmp): Likewise.
26497 (sorted_attr_string): Likewise.
26498 (common_function_versions): Likewise.
26499 (make_unique_name): Likewise.
26500 (make_dispatcher_decl): Likewise.
26501 (is_function_default_version): Likewise.
26502 * config/i386/i386.c (attr_strcmp): Likewise.
26503 (sorted_attr_string): Likewise.
26504 (ix86_function_versions): Likewise.
26505 (make_name): Likewise.
26506 (make_dispatcher_decl): Likewise.
26507 (is_function_default_version): Likewise.
26508 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26510 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26513 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26514 Account for direct move costs for vec_construct of integer
26517 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26520 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26521 (UNSPEC_STX_ATOMIC): Ditto.
26522 (loaddi_via_sse): New insn.
26523 (storedi_via_sse): Ditto.
26524 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26525 Update corresponding peephole2 patterns.
26526 (atomic_storedi_fpu): Ditto.
26528 2017-05-11 Julia Koval <julia.koval@intel.com>
26530 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26531 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26533 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26534 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26535 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26537 2017-05-11 Nathan Sidwell <nathan@acm.org>
26539 * graphite-poly.c: Include dumpfile.h.
26541 * dumpfle.h (dump_function): Declare here ...
26542 * tree-dump.h (dump_function): ... not here.
26543 * dumpfile.c: #include tree-cfg.h.
26544 (dump_function): Move here from ...
26545 * tree-dump.c (dump_function): ... here.
26546 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26547 * graphite-poly.c: Don't include tree-dump.h.
26548 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26549 * print-tree.c: Likewise.
26550 * stor-layout.c: Likewise.
26551 * tree-nested.c: Likewise.
26553 * dumpfile.c (dump_start): Use TDF_FLAGS.
26554 (dump_enable_all): Fix TDF_KIND check thinko.
26556 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26558 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26559 array entries to represent two legal parameterizations of the
26560 overloaded __builtin_cmpb function, as represented by the
26561 P6_OV_BUILTIN_CMPB constant.
26562 (altivec_resolve_overloaded_builtin): Add special case handling
26563 for the __builtin_cmpb function, as represented by the
26564 P6_OV_BUILTIN_CMPB constant.
26565 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26566 (BU_P6_64BIT_2): New macro.
26567 (BU_P6_OVERLOAD_2): New macro
26568 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26569 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26570 (CMPB): Add overload support to represent both 32-bit and 64-bit
26571 compare-bytes function.
26572 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26573 support for TARGET_CMPB.
26574 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26575 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26576 documentation of the __builtin_cmpb overloaded built-in function.
26578 2017-05-11 Richard Biener <rguenther@suse.de>
26580 PR tree-optimization/80705
26581 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26582 bases are not vectorizable.
26584 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26586 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26587 when counting register pressure.
26589 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26591 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26592 (iv_ca_more_deps): Renamed to ...
26593 (iv_ca_compare_deps): ... this.
26594 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26596 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26598 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26600 (determine_group_iv_costs): ... here.
26601 (find_inv_vars_cb): Record inv var if it's not recorded before.
26603 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26605 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26606 (get_shiftadd_cost): Ditto.
26608 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26610 * tree-ssa-address.c: Include header file.
26611 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26613 (add_to_parts): Refactor.
26614 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26615 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26618 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26620 PR tree-optimization/53090
26621 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26623 (extract_cond_operands): Detect condition with IV on both sides
26624 and return COMP_IV_EXPR_2.
26625 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26626 (rewrite_use_compare): Simplify by removing call to function
26627 extract_cond_operands.
26629 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26631 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26632 (extract_cond_operands): Detect condition comparing against non-
26633 invariant bound and return appropriate enum value.
26634 (find_interesting_uses_cond): Update use of extract_cond_operands.
26635 Handle its return value accordingly.
26636 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26638 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26640 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26641 nonlinear iv_use computation in loop invariant sensitive way.
26643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26645 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26646 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26650 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26651 (dump_cand): Support iv_cand.inv_exprs.
26652 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26654 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26659 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26661 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26662 as local function. Include necessary header files.
26663 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26665 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26667 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26669 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26671 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26672 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26673 RSHIFT_EXPR and BIT_NOT_EXPR.
26675 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26677 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26678 (adjust_setup_cost): New parameter supporting round up adjustment.
26679 (struct address_cost_data): Delete.
26680 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26681 (split_address_cost, ptr_difference_cost): Delete.
26682 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26683 (struct ainc_cost_data): New struct.
26684 (get_address_cost_ainc): New function.
26685 (get_address_cost, get_computation_cost): Reimplement.
26686 (determine_group_iv_cost_address): Record inv_expr for all uses of
26688 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26689 (iv_ca_has_deps): Reimplemented to ...
26690 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26692 (iv_ca_extend): Call iv_ca_more_deps.
26694 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26696 * tree-ssa-address.c (struct mem_address): Move to header file.
26697 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26698 * tree-ssa-address.h (struct mem_address): Move from C file.
26699 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26703 * tree-affine.h (aff_combination_type): New interface.
26704 (aff_combination_zero_p): Remove static.
26705 (aff_combination_const_p): New interface.
26706 (aff_combination_singleton_var_p): New interfaces.
26708 2017-05-11 Richard Biener <rguenther@suse.de>
26710 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26711 Skip unreachable blocks and destinations.
26712 (eliminate): Move stmt removal and fixup ...
26713 (fini_eliminate): ... here. Skip inserted exprs.
26714 (pass_pre::execute): Move fini_pre after fini_eliminate.
26715 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26716 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26717 PRE to get rid of dead code that has invalid SSA form and
26718 split critical edges again.
26720 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26722 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26724 2017-05-11 Richard Biener <rguenther@suse.de>
26726 * passes.c (execute_function_todo): Verify loops if they are
26727 said to be up-to-date.
26728 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26729 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26731 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26734 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26735 handle calling assemble_external ourself.
26738 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26739 modes with zero size. Enhance comment.
26741 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26743 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26744 built-ins for vec_xl and vec_xst with short and char pointer
26747 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26749 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26750 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26751 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26752 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26753 (_mm_maskz_min_round_ss): New intrinsics.
26754 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26755 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26756 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26757 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26758 (__builtin_ia32_minss_mask_round): New builtins.
26759 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26760 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26761 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26763 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26764 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26766 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26769 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26771 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26772 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26773 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26774 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26775 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26776 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26777 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26778 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26779 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26780 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26781 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26782 (__builtin_ia32_mulss_mask_round): New builtins.
26783 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26784 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26785 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26787 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26788 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26790 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26793 2017-05-10 Julia Koval <julia.koval@intel.com>
26795 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26796 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26797 (_mm256_setr_m128i): New intrinsics.
26799 2017-05-10 Julia Koval <julia.koval@intel.com>
26801 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26802 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26803 (_mm_maskz_rcp14_ss): New intrinsics.
26804 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26805 (__builtin_ia32_rcp14ss_mask): New builtins.
26806 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26808 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26810 PR tree-optimization/51513
26811 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26812 (assert_unreachable_fallthru_edge_p): Use it.
26813 (group_case_labels_stmt): Likewise.
26814 * tree-cfg.h: Prototype it.
26815 * stmt.c: Include cfghooks.h and tree-cfg.h.
26816 (emit_case_dispatch_table) <gap_label>: New local variable.
26817 Use it to fill dispatch table gaps.
26818 Test for default_label before updating probabilities.
26819 (expand_case) <default_label>: Remove unneeded initialization.
26820 Test for unreachable default case statement and remove its edge.
26821 Set default_label accordingly.
26822 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26824 2017-05-10 Carl Love <cel@us.ibm.com>
26826 * config/rs6000/rs6000-c: Add support for built-in functions
26827 vector signed char vec_neg (vector signed char)
26828 vector signed short int vec_neg (vector short int)
26829 vector signed int vec_neg (vector signed int)
26830 vector signed long long vec_neg (vector signed long long)
26831 vector float vec_neg (vector float)
26832 vector double vec_neg (vector double)
26833 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26835 * config/rs6000/altivec.h: Add define for vec_neg
26836 * doc/extend.texi: Update the built-in documentation for the
26837 new built-in functions.
26839 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26841 PR tree-optimization/77644
26842 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26844 2017-05-10 Nathan Sidwell <nathan@acm.org>
26846 * dumpfile.h (TDI_lang_all): New.
26847 (TDF_KIND): New. Renumber others
26848 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26850 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26852 (get_dump_file_name): Adjust suffix generation.
26853 (dump_enable_all): Use TDF_KIND.
26854 * doc/invoke.texi (-fdump-lang-all): Document.
26856 * dumpfile.h: Tabify.
26858 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26861 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26862 Move member access before delete.
26864 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26866 * tree-inline.c (expand_call_inline): Split block at stmt
26869 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26872 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26873 are now unused after splitting mov{sf,sd}_hardfloat.
26874 (f32_lr2): Likewise.
26875 (f32_lm): Likewise.
26876 (f32_lm2): Likewise.
26877 (f32_li): Likewise.
26878 (f32_li2): Likewise.
26879 (f32_lv): Likewise.
26880 (f32_sr): Likewise.
26881 (f32_sr2): Likewise.
26882 (f32_sm): Likewise.
26883 (f32_sm2): Likewise.
26884 (f32_si): Likewise.
26885 (f32_si2): Likewise.
26886 (f32_sv): Likewise.
26887 (f32_dm): Likewise.
26888 (f32_vsx): Likewise.
26889 (f32_av): Likewise.
26890 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26891 For movsf, order stores so the VSX stores occur before the GPR
26892 store which encourages the register allocator to use a traditional
26893 FPR instead of a GPR. For movsd, order the stores so that the GPR
26894 store comes before the VSX stores to allow the power6 to work.
26895 This is due to the power6 not having a 32-bit integer store
26896 instruction from a FPR.
26897 (movsf_hardfloat): Likewise.
26898 (movsd_hardfloat): Likewise.
26900 2017-05-09 Martin Sebor <msebor@redhat.com>
26902 PR translation/80280
26903 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26906 PR translation/80280
26907 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26908 data member added in r247778.
26909 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26911 2017-05-09 Nathan Sidwell <nathan@acm.org>
26913 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26915 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26918 2017-05-09 Marek Polacek <polacek@redhat.com>
26920 * doc/invoke.texi: Fix typo.
26922 2017-05-09 Richard Biener <rguenther@suse.de>
26924 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26925 (vrp_val_is_min): Likewise.
26926 (set_value_range_to_value): Likewise.
26927 (set_value_range_to_nonnegative): Likewise.
26928 (gimple_assign_nonzero_p): Likewise.
26929 (gimple_stmt_nonzero_p): Likewise.
26930 (vrp_int_const_binop): Likewise. Remove unreachable case.
26931 (adjust_range_with_scev): Adjust comments.
26932 (compare_range_with_value): Likewise.
26933 (extract_range_from_phi_node): Likewise.
26934 (test_for_singularity): Likewise.
26936 2017-05-09 Richard Biener <rguenther@suse.de>
26938 * tree-vrp.c (get_single_symbol): Add assert that we don't
26939 get overflowed constants as invariant part.
26940 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26941 checks. Use wi::cmp instead of recursing for integer constants.
26942 (compare_values): Just ignore whether we assumed undefined
26943 overflow instead of failing the compare.
26944 (extract_range_for_var_from_comparison_expr): Add comment before the
26945 TREE_NO_WARNING sets.
26946 (test_for_singularity): Likewise.
26947 (extract_range_from_comparison): Do not disable optimization
26948 when we assumed undefined overflow.
26949 (extract_range_basic): Remove init of unused var.
26951 2017-05-09 Richard Biener <rguenther@suse.de>
26953 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26954 (extract_range_from_multiplicative_op_1): Adjust.
26955 (extract_range_from_binary_expr_1): Use int_const_binop.
26957 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26960 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26961 rs6000_store_data_bypass_p in seven define_bypass directives and
26962 in several comments.
26963 * config/rs6000/rs6000-protos.h: Add prototype for
26964 rs6000_store_data_bypass_p function.
26965 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26966 function implements slightly different (rs6000-specific) semantics
26967 than store_data_bypass_p, returning false rather than aborting
26968 with assertion error when arguments do not satisfy the
26969 requirements of store data bypass.
26970 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26971 rs6000_store_data_bypass_p.
26973 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26975 * config/xtensa/xtensa-protos.h
26976 (xtensa_initial_elimination_offset): New declaration.
26977 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26978 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26979 macro definition, add case for FRAME_POINTER_REGNUM when
26980 FRAME_GROWS_DOWNWARD.
26981 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26982 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26983 xtensa_initial_elimination_offset.
26985 2017-05-08 Nathan Sidwell <nathan@acm.org>
26987 * doc/invoke.texi: Alphabetize -fdump options.
26989 2017-05-08 Martin Sebor <msebor@redhat.com>
26991 PR translation/80280
26992 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26994 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26996 * target.def (compute_frame_layout): New optional target hook.
26997 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26998 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26999 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27001 * reload1.c (verify_initial_elim_offsets): Likewise.
27002 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27003 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27004 (arm_get_frame_offsets): Split up into this ...
27005 (arm_compute_frame_layout): ... and this function.
27007 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27009 * config/aarch64/constraints.md (Usa): New constraint.
27010 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27012 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27014 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27015 with_multilib_list after it has been checked.
27017 2017-05-08 Richard Biener <rguenther@suse.de>
27019 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27020 (bitmap_set_subtract_values): Likewise.
27022 2017-05-08 Richard Biener <rguenther@suse.de>
27024 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27025 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27027 (gimple_stmt_nonzero_warnv_p): Rename to ...
27028 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27030 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27031 (extract_range_basic): Adjust, do not disable propagation on
27032 strict overflow sensitive simplification.
27033 (vrp_visit_cond_stmt): Likewise.
27035 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27037 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27038 body size unconditionally.
27040 2017-05-07 Jeff Law <law@redhat.com>
27043 2017-05-06 Jeff Law <law@redhat.com>
27044 PR tree-optimization/78496
27045 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27048 PR tree-optimization/78496
27049 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27050 (simplify_stmt_using_ranges): Call it.
27051 (vrp_dom_walker::before_dom_children): Extract equivalences
27052 from an ASSERT_EXPR with an equality comparison against a
27055 2017-05-06 Jeff Law <law@redhat.com>
27057 PR tree-optimization/78496
27058 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27061 PR tree-optimization/78496
27062 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27063 (simplify_stmt_using_ranges): Call it.
27064 (vrp_dom_walker::before_dom_children): Extract equivalences
27065 from an ASSERT_EXPR with an equality comparison against a
27068 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27070 * lra-constraints.c (lra_copy_reg_equiv): New function.
27071 (split_reg): Use it to copy equivalence information from the
27072 original register to the spill register.
27074 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27076 PR rtl-optimization/75964
27077 * simplify-rtx.c (simplify_const_relational_operation): Remove
27078 invalid handling of comparisons of integer ABS.
27080 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27082 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27083 initialize to zero.
27084 (init_regs): Remove declaration.
27085 (function_arg_advance_32): Initialize error_p as boolean variable.
27087 2017-05-05 Nathan Sidwell <nathan@acm.org>
27089 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27090 lines. Use for (;;).
27092 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27094 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27095 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27096 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27097 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27098 VF=2 that require versioning.
27100 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27102 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27105 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27107 * diagnostic.h (diagnostic_override_option_index): Convert from
27108 macro to inline function.
27110 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27112 * diagnostic.c (last_module_changed_p): New function.
27113 (set_last_module): New function.
27114 (diagnostic_report_current_module): Convert macro usage to
27115 the above functions.
27116 * diagnostic.h (diagnostic_context::last_module): Strengthen
27117 from const line_map * to const line_map_ordinary *.
27118 (diagnostic_last_module_changed): Delete macro.
27119 (diagnostic_set_last_module): Delete macro.
27121 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27123 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27124 with diagnostic_report_diagnostic.
27125 (diagnostic_n_impl_richloc): Likewise.
27126 * diagnostic.h (report_diagnostic): Delete macro.
27127 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27128 with diagnostic_report_diagnostic.
27129 * substring-locations.c (format_warning_va): Likewise.
27131 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27133 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27134 save/restor of format_spec. Move option-printing code to...
27135 (print_option_information): ...this new function, and
27136 reimplement by simply printing to the pretty_printer,
27137 rather than appending to the format string.
27139 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27141 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27142 handling logic into...
27143 (update_effective_level_from_pragmas): ...this new function.
27145 2017-05-04 Andrew Waterman <andrew@sifive.com>
27147 * config/riscv/riscv.opt (mstrict-align): New option.
27148 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27149 (SLOW_UNALIGNED_ACCESS): Define.
27150 (riscv_slow_unaligned_access): Declare.
27151 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27153 (riscv_slow_unaligned_access): New variable.
27154 (rocket_tune_info): Set slow_unaligned_access to true.
27155 (optimize_size_tune_info): Set slow_unaligned_access to false.
27156 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27157 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27158 (riscv_option_override): Set riscv_slow_unaligned_access.
27159 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27161 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27163 * config/riscv/riscv.md: Unify indentation.
27165 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27170 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27172 (extendsi<mode>2): Add support for doing sign extension via
27173 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27174 don't have ISA 3.0 instructions.
27175 (extendsi<mode>2 splitter): Likewise.
27176 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27177 generate the normal insns since SImode can now go in vector
27178 registers. Disallow the special UNSPECs needed for previous
27179 machines to hide SImode being used. Add new insns
27180 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27181 (fix_trunc<mode>si2_stfiwx): Likewise.
27182 (fix_trunc<mode>si2_internal): Likewise.
27183 (fixuns_trunc<mode>si2): Likewise.
27184 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27185 (fctiw<u>z_<mode>_smallint): Likewise.
27186 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27187 of floating point to 32-bit integer from doing a direct move to
27188 the GPR registers to do a store.
27189 (fctiwz_<mode>): Break long line.
27191 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27193 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27194 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27195 (addr_list, addr_offset_valid_p): New.
27196 (split_address_groups): Check offset validity with above function.
27197 (gt-tree-ssa-loop-ivopts.h): Include header file.
27199 2017-05-05 Nathan Sidwell <nathan@acm.org>
27201 * config.gcc (arm*-*-*): Add missing 'fi'.
27203 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27205 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27206 included in -fopt-info does not matter.
27207 * doc/optinfo.texi (-fopt-info): Fix description of default
27208 behavour. Explicitly say order of options included in -fopt-info
27211 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27213 * config.gcc: Allow combinations of aprofile and rmprofile values for
27214 --with-multilib-list.
27215 * config/arm/t-multilib: New file.
27216 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27217 variables. Remove setting of ISA and floating-point ABI in
27218 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27219 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27220 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27221 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27223 * config/arm/t-rmprofile: Likewise except for the matches changes.
27224 * doc/install.texi (--with-multilib-list): Document the combination of
27225 aprofile and rmprofile values and warn about pitfalls in doing that.
27227 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27229 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27230 (movdi_aarch64): Likewise.
27232 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27234 PR tree-optimization/80632
27235 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27237 (build_arrays): Initialize it for virtual phis.
27238 (fix_phi_nodes): Use it for virtual phis.
27240 PR tree-optimization/80558
27241 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27242 [x, y] op z into [x op, y op z] for op & or | if conditions
27245 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27246 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27249 * config/arm/arm.md (use_literal_pool): Remove.
27250 (64-bit immediate split): No longer takes cost into consideration
27251 if arm_disable_literal_pool is enabled.
27252 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27253 used when arm_disable_literal_pool is enabled.
27254 (arm_max_const_double_inline_cost): Remove use of
27255 arm_disable_literal_pool.
27256 (push_minipool_fix): Add assert.
27257 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27258 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27259 (no_literal_pool_sf_immediate): New.
27261 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27263 PR tree-optimization/80613
27264 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27265 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27267 2017-05-05 Richard Biener <rguenther@suse.de>
27269 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27271 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27273 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27274 of this flag from insn conditions due to removal from r247495.
27276 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27278 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27280 (arm_early_store_addr_dep_ptr): Likewise.
27281 * config/arm/aarch-common-protos.h
27282 (arm_early_load_addr_dep_ptr): Add prototype.
27283 (arm_early_store_addr_dep_ptr): Likewise.
27284 * config/arm/cortex-a53.md: Add new bypasses.
27286 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27288 * tree.c (next_type_uid): Change type to unsigned.
27289 (type_hash_canon): Decrement back next_type_uid if
27290 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27291 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27294 2017-05-04 Martin Sebor <msebor@redhat.com>
27296 * builtins.c: Fix a trivial typo in a comment.
27298 PR middle-end/79234
27299 * builtins.c (check_sizes): Adjust to handle reading past the end.
27300 Avoid printing excessive upper bound of ranges. Use %E to print
27301 tree nodes instead of converting them to %wu.
27302 (expand_builtin_memchr): New function.
27303 (compute_dest_size): Rename...
27304 (compute_objsize): ...to this.
27305 (expand_builtin_memcpy): Adjust.
27306 (expand_builtin_mempcpy): Adjust.
27307 (expand_builtin_strcat): Adjust.
27308 (expand_builtin_strcpy): Adjust.
27309 (check_strncat_sizes): Adjust.
27310 (expand_builtin_strncat): Adjust.
27311 (expand_builtin_strncpy): Adjust and simplify.
27312 (expand_builtin_memset): Adjust.
27313 (expand_builtin_bzero): Adjust.
27314 (expand_builtin_memcmp): Adjust.
27315 (expand_builtin): Handle memcmp.
27316 (maybe_emit_chk_warning): Check strncat just once.
27318 2017-05-04 Martin Sebor <msebor@redhat.com>
27320 PR preprocessor/79214
27321 PR middle-end/79222
27322 PR middle-end/79223
27323 * builtins.c (check_sizes): Add inlining context and issue
27324 warnings even when -Wno-system-headers is set.
27325 (check_strncat_sizes): Same.
27326 (expand_builtin_strncat): Same.
27327 (expand_builtin_memmove): New function.
27328 (expand_builtin_stpncpy): Same.
27329 (expand_builtin): Handle memmove and stpncpy.
27331 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27333 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27334 which is not used any more.
27336 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27338 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27340 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27342 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27343 (cortexa53_tunings): Likewise.
27344 (cortexa57_tunings): Likewise.
27345 (cortexa72_tunings): Likewise.
27346 (cortexa73_tunings): Likewise.
27348 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27350 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27351 Set loop alignment to 8.
27353 2017-05-04 Martin Sebor <msebor@redhat.com>
27355 PR translation/80280
27356 * builtins.c (expand_builtin_object_size): Add missing quoting to
27357 %D and like directives.
27358 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27359 (hsa_type_for_tree_type): Same.
27360 (verify_function_arguments): Same.
27361 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27362 * varasm.c (get_section): Same.
27365 2017-05-04 Martin Sebor <msebor@redhat.com>
27367 PR translation/80280
27368 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27370 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27372 * config/aarch64/aarch64.c (generic_addrcost_table):
27373 Change HI/TI mode setting.
27375 2017-05-04 Martin Jambor <mjambor@suse.cz>
27377 PR tree-optimization/80622
27378 * tree-sra.c (comes_initialized_p): New function.
27379 (build_accesses_from_assign): Only set write lazily when
27380 comes_initialized_p is false.
27381 (analyze_access_subtree): Use comes_initialized_p.
27382 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27383 instead of testing for PARM_DECL.
27385 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27387 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27388 constraint on operand 0 to allow more general addressing modes.
27389 Adjust output template.
27390 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27392 * config/aarch64/aarch64-protos.h
27393 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27394 * config/aarch64/constraints.md (Dp): New address constraint.
27395 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27398 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27400 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27401 update use of estimate_ipcp_clone_size_and_time.
27402 (estimate_local_effects): Update use of
27403 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27404 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27405 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27406 Return nonspecialized time.
27408 2017-05-04 Richard Biener <rguenther@suse.de>
27410 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27411 for the last VUSE which def dominates the PHI. Directly call
27413 (get_continuation_for_phi_1): Remove.
27415 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27417 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27418 to explain the use of truncating division. Cap the number of
27419 iterations to the maximum given by nb_iterations_upper_bound,
27422 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27424 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27425 * configure: Regenerate.
27426 * config.in: Regenerate.
27427 * config/i386/driver-mingw32.c: new file.
27428 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27429 * config.host: Link driver-mingw32.o on MinGW host.
27430 * doc/install.texi: Document new --enable-mingw-wildcard configure
27433 2017-05-04 Marek Polacek <polacek@redhat.com>
27435 PR tree-optimization/80612
27436 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27438 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27439 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27441 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27442 (movt splitter): Likewise.
27443 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27444 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27445 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27446 block for Thumb-1 with MOVT.
27447 (thumb2_legitimate_address_p): Move code block ...
27448 (can_avoid_literal_pool_for_label_p): ... into this new function.
27449 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27451 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27452 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27453 "M-profile targets with the MOVT instruction".
27455 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27457 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27458 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27459 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27461 2017-05-04 Martin Liska <mliska@suse.cz>
27463 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27464 variable cond_code.
27466 2017-05-04 Richard Biener <rguenther@suse.de>
27468 * tree.c (array_at_struct_end_p): Handle arrays at struct
27469 end with flexarrays more conservatively. Refactor and treat
27470 arrays of arrays or aggregates more strict. Fix
27471 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27472 * tree.c (array_at_struct_end_p): Adjust prototype.
27473 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27474 * gimple-fold.c (get_range_strlen): Likewise.
27475 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27477 2017-05-04 Richard Biener <rguenther@suse.de>
27479 PR tree-optimization/31130
27480 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27482 (supports_overflow_infinity): Likewise.
27483 (is_negative_overflow_infinity): Likewise.
27484 (is_positive_overflow_infinity): Likewise.
27485 (is_overflow_infinity): Likewise.
27486 (stmt_overflow_infinity): Likewise.
27487 (overflow_infinity_range_p): Likewise.
27488 (usable_range_p): Remove as always returning true.
27489 (make_overflow_infinity): Remove.
27490 (negative_overflow_infinity): Likewise.
27491 (positive_overflow_infinity): Likewise.
27492 (avoid_overflow_infinity): Likewise.
27493 (set_value_range): Adjust accordingly.
27494 (set_value_range_to_nonnegative): Likewise, remove now unused
27495 overflow_infinity arg.
27496 (vrp_operand_equal_p): Adjust.
27497 (update_value_range): Likewise.
27498 (range_int_cst_singleton_p): Likewise.
27499 (operand_less_p): Likewise.
27500 (compare_values_warnv): Likewise.
27501 (extract_range_for_var_from_comparison_expr): Likewise.
27502 (vrp_int_const_binop): Likewise.
27503 (zero_nonzero_bits_from_vr): Likewise.
27504 (extract_range_from_multiplicative_op_1): Likewise.
27505 (extract_range_from_binary_expr_1): Likewise.
27506 (extract_range_from_unary_expr): Likewise.
27507 (extract_range_from_comparison): Likewise.
27508 (extract_range_basic): Likewise.
27509 (adjust_range_with_scev): Likewise.
27510 (compare_ranges): Likewise.
27511 (compare_range_with_value): Likewise.
27512 (dump_value_range): Likewise.
27513 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27515 (simplify_cond_using_ranges): Adjust.
27517 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27519 * brig-builtins.def: Added a builtin for class_f64.
27520 * builtin-types.def: Added a builtin type needed by class_f64.
27522 2017-05-03 Jason Merrill <jason@redhat.com>
27524 * timevar.def: Add TV_CONSTEXPR.
27526 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27528 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27530 2017-05-03 Martin Jambor <mjambor@suse.cz>
27532 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27533 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27534 * ipa-cp.c (ipcp_propagate_stage): Do not call
27535 ipa_update_after_lto_read.
27536 * ipa-inline.c (ipa_inline): Likewise.
27538 2017-05-03 Martin Jambor <mjambor@suse.cz>
27540 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27541 tag. Added a default constructor and a destructor.
27542 (ipa_edge_args_sum_t): New class;
27543 (ipa_edge_args_sum): Declare.
27544 (ipa_edge_args_vector): Remove declaration.
27545 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27546 (ipa_free_edge_args_substructures): Remove declaration.
27547 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27548 (ipa_edge_args_info_available_for_edge_p): Likewise.
27549 * ipa-prop.c (ipa_edge_args_vector): Removed.
27550 (edge_removal_hook_holder): Likewise.
27551 (edge_duplication_hook_holder): Likewise.
27552 (ipa_edge_args_sum): New variable.
27553 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27554 ipa_edge_args_vector.
27555 (ipa_free_edge_args_substructures): Likewise.
27556 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27557 ipa_edge_args_vector.
27558 (ipa_edge_removal_hook): Turned into method
27559 ipa_edge_args_sum_t::remove.
27560 (ipa_edge_duplication_hook): Turned into method
27561 ipa_edge_args_sum_t::duplicate.
27562 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27563 registering edge hooks.
27564 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27565 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27566 ipa_edge_args_sum instead of ipa_edge_args_vector.
27567 * ipa-profile.c (ipa_profile): Likewise.
27569 2017-05-03 Martin Jambor <mjambor@suse.cz>
27571 * symbol-summary.h (function_summary): New method exists.
27572 (function_summary::symtab_removal): Deallocate through release.
27573 (call_summary): New class.
27574 (gt_ggc_mx): New overload.
27575 (gt_pch_nx): Likewise.
27576 (gt_pch_nx): Likewise.
27578 2017-05-03 Jeff Law <law@redhat.com>
27580 PR tree-optimization/78496
27581 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27582 from simplify_cond_using_ranges. Split off code to walk
27583 backwards through casts into ...
27584 (simplify_cond_using_ranges_2): New function.
27585 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27586 (execute_vrp): After identifying jump threads, call
27587 simplify_cond_using_ranges_2.
27589 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27592 * ipa-inline.h (inline_summary): Add ctor.
27593 (create_ggc): Do not use ggc_cleared_alloc.
27595 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27596 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27598 * gcc.c (handle_braces): Support escaping in switch matching
27600 * doc/invoke.texi (Spec Files): Document it.
27601 Remove superfluous @code markup in items.
27603 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27605 * diagnostic-show-locus.c (struct column_range): New struct.
27606 (get_affected_columns): New function.
27607 (get_printed_columns): New function.
27608 (struct correction): New struct.
27609 (correction::ensure_capacity): New function.
27610 (correction::ensure_terminated): New function.
27611 (struct line_corrections): New struct.
27612 (line_corrections::~line_corrections): New dtor.
27613 (line_corrections::add_hint): New function.
27614 (layout::print_trailing_fixits): Reimplement in terms of the new
27616 (selftest::test_overlapped_fixit_printing): New function.
27617 (selftest::diagnostic_show_locus_c_tests): Call it.
27619 2017-05-03 Nathan Sidwell <nathan@acm.org>
27621 Canonicalize canonical type hashing
27622 * tree.h (type_hash_canon_hash): Declare.
27623 * tree.c (type_hash_list, attribute_hash_list): Move into
27624 type_hash_canon_hash.
27625 (build_type_attribute_qual_variant): Break out hash code calc into
27626 type_hash_canon_hash.
27627 (type_hash_canon_hash): New. Generic type hash computation.
27628 (build_range_type_1, build_array_type_1, build_function_type,
27629 build_method_type_directly, build_offset_type, build_complex_type,
27630 make_vector_type): Call it.
27632 2017-05-03 Richard Biener <rguenther@suse.de>
27634 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27635 When all DRs have unknown misaligned do not always peel
27636 when there is a store but apply the same costing model as if
27637 there were only loads.
27639 2017-05-03 Richard Biener <rguenther@suse.de>
27642 PR tree-optimization/80492
27643 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27644 compare_base_decls returning dont-know properly.
27646 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27648 * config/arm/iterators.md (CCSI): New mode iterator.
27649 (arch): New mode attribute.
27650 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27651 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27652 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27653 code iterator for success result mode.
27654 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27655 the corresponding new insn generators.
27657 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27660 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27661 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27663 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27665 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27666 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27667 (DDR_A): Wrap DDR argument in brackets.
27668 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27669 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27670 (DDR_REVERSED_P): Likewise.
27672 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27674 PR tree-optimization/79472
27675 * tree-switch-conversion.c (struct switch_conv_info): Add
27676 contiguous_range and default_case_nonstandard fields.
27677 (collect_switch_conv_info): Compute contiguous_range and
27678 default_case_nonstandard fields, don't clear final_bb if
27679 contiguous_range and only the default case doesn't have the required
27681 (check_all_empty_except_final): Set default_case_nonstandard instead
27682 of failing if contiguous_range and the default case doesn't have empty
27684 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27685 and only the default case doesn't have the required constants. Skip
27687 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27688 if default_case_nonstandard.
27689 (build_constructors): Build constant 1 just once. Assert that default
27690 values aren't inserted in between cases if contiguous_range. Skip
27692 (build_arrays): Skip virtual phis.
27693 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27694 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27695 Handle virtual phis.
27696 (gen_inbound_check): Handle default_case_nonstandard case.
27697 (process_switch): Adjust check_final_bb caller. Call
27698 gather_default_values with the first non-default case instead of
27699 default case if default_case_nonstandard.
27701 2017-05-02 Nathan Sidwell <nathan@acm.org>
27703 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27704 check. Fix formatting.
27706 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27708 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27709 errors when comparing specialized and unspecialized times.
27711 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27713 * diagnostic-show-locus.c
27714 (layout::should_print_annotation_line_p): Make private.
27715 (layout::print_annotation_line): Make private.
27716 (layout::annotation_line_showed_range_p): Make private.
27717 (layout::show_ruler): Make private.
27718 (layout::print_source_line): Make private. Pass in line and
27719 line_width, rather than calling location_get_source_line. Drop
27721 (layout::print_leading_fixits): New method.
27722 (layout::print_any_fixits): Rename to...
27723 (layout::print_trailing_fixits): ...this, and make private.
27724 Don't print newline fixits.
27725 (diagnostic_show_locus): Move logic for printing one row into...
27726 (layout::print_line): ...this new function. Move the
27727 location_get_source_line call and error-handling from
27728 print_source_line to here. Call print_leading_fixits, and rename
27729 print_any_fixits to print_trailing_fixits.
27730 (selftest::test_fixit_insert_containing_newline): Update now that
27731 newlines are partially supported.
27732 (selftest::test_fixit_insert_containing_newline_2): New test.
27733 (selftest::test_fixit_replace_containing_newline): Update comments.
27734 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27735 * edit-context.c (class added_line): New class.
27736 (class edited_line): Describe newline handling in comment.
27737 (edited_line::actually_edited_p): New method.
27738 (edited_line::print_content): Delete redundant decl.
27739 (edited_line::m_predecessors): New field.
27740 (edited_file::print_content): Call edited_line::print_content.
27741 (edited_file::print_diff): Update to support newlines.
27742 (edited_file::print_diff_hunk): Likewise.
27743 (edited_file::print_run_of_changed_lines): New function.
27744 (edited_file::print_diff_line): Convert to...
27745 (print_diff_line): ...this.
27746 (edited_file::get_effective_line_count): New function.
27747 (edited_line::edited_line): Initialize new field m_predecessors.
27748 (edited_line::~edited_line): Clean up m_predecessors.
27749 (edited_line::apply_fixit): Handle newlines.
27750 (edited_line::get_effective_line_count): New function.
27751 (edited_line::print_content): New function.
27752 (edited_line::print_diff_lines): New function.
27753 (selftest::test_applying_fixits_insert_containing_newline): New
27755 (selftest::test_applying_fixits_replace_containing_newline): New
27757 (selftest::insert_line): New function.
27758 (selftest::test_applying_fixits_multiple_lines): Add example of
27760 (selftest::edit_context_c_tests): Call the new tests.
27762 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27764 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27765 parameter cand. Update dump information.
27766 (get_computation_cost): Update uses.
27768 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27770 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27771 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27772 (get_computation_at, rewrite_use_address): Update use of
27773 get_computation_aff.
27775 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27777 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27778 (get_computation): Delete.
27779 (get_computation_cost): Implement like get_computation_cost_at.
27780 Use get_computation_at.
27781 (get_computation_cost_at): Delete.
27782 (rewrite_use_nonlinear_expr): Use get_computation_at.
27783 (rewrite_use_compare, remove_unused_ivs): Ditto.
27785 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27787 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27789 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27791 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27792 (ivopts_global_cost_for_size): Rename parameter and update uses.
27793 (iv_ca_recount_cost): Update uses.
27794 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27795 candidates seperately in n_invs and n_cands.
27796 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27798 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27800 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27801 (find_inv_vars_cb): New.
27802 (find_depends): Renamed to ...
27803 (find_inv_vars): ... this.
27804 (add_candidate_1, force_var_cost): Call find_inv_vars.
27805 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27807 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27809 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27810 inv_vars. Add inv_exprs.
27811 (struct iv_cand): Rename depends_on to inv_vars.
27812 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27813 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27814 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27815 (dump_cand): Dump inv_vars.
27816 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27817 (record_invariant, find_depends, add_candidate_1): Ditto.
27818 (set_group_iv_cost, force_var_cost): Ditto.
27819 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27820 (get_computation_cost_at, get_computation_cost): Ditto.
27821 (determine_group_iv_cost_generic): Ditto.
27822 (determine_group_iv_cost_address): Ditto.
27823 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27824 (determine_group_iv_costs): Ditto.
27825 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27826 (iv_ca_set_remove_invariants): Renamed to ...
27827 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27828 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27829 (iv_ca_set_add_invariants): Renamed to ...
27830 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27831 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27832 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27833 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27834 (create_new_ivs): Remove useless dump.
27836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27838 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27840 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27841 (iv_ca_set_no_cp, create_new_iv): Ditto.
27843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27845 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27847 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27849 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27850 function tree_check2.
27852 2017-05-02 Martin Liska <mliska@suse.cz>
27854 * doc/gcov.texi: Add missing preposition.
27855 * gcov.c (function_info::function_info): Properly fill up
27856 all member variables.
27858 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27860 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27862 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27864 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27866 2017-05-02 Martin Liska <mliska@suse.cz>
27869 * lto-streamer-in.c (lto_read_tree_1): Remove
27870 LTO_STREAMER_DEBUG.
27871 * lto-streamer.c (struct tree_hash_entry): Likewise.
27872 (struct tree_entry_hasher): Likewise.
27873 (tree_entry_hasher::hash): Likewise.
27874 (tree_entry_hasher::equal): Likewise.
27875 (lto_streamer_init): Likewise.
27876 (lto_orig_address_map): Likewise.
27877 (lto_orig_address_get): Likewise.
27878 (lto_orig_address_remove): Likewise.
27879 * lto-streamer.h: Likewise.
27880 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27881 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27883 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27885 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27886 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27887 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27888 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27889 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27890 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27891 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27892 (mm_maskz_sub_ss): New intrinsics.
27893 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27894 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27895 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27896 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27897 (__builtin_ia32_subss_mask_round): New builtins.
27898 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27899 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27900 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27902 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27903 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27905 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27908 2017-05-02 Martin Jambor <mjambor@suse.cz>
27910 PR tree-optimization/78687
27911 * tree-sra.c (access): New field parent.
27912 (process_subtree_disqualification): New function.
27913 (disqualify_candidate): Call it.
27914 (build_accesses_from_assign): Reset write flag if creating an
27916 (build_access_subtree): Fill in parent field and also prpagate
27917 down grp_write flag.
27918 (create_artificial_child_access): New parameter set_grp_write, set
27919 grp_write to its value.
27920 (propagate_subaccesses_across_link): Also propagate grp_write flag
27922 (propagate_all_subaccesses): Push the closest parent back to work
27923 queue if add_access_to_work_queue returned true.
27925 2017-05-02 Richard Biener <rguenther@suse.de>
27927 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27928 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27929 -fstrict-overflow documentation.
27930 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27931 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27932 flag_strict_overflow.
27933 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27934 * lto-opts.c (lto_write_options): Do not stream it.
27935 * lto-wrapper.c (merge_and_complain): Do not handle it.
27936 * opts.c (default_options_table): Do not set -fstrict-overflow.
27937 (finish_options): Likewise do not clear it when sanitizing.
27938 * simplify-rtx.c (simplify_const_relational_operation): Do not
27939 test flag_strict_overflow.
27941 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27943 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27944 using enabled attribute.
27945 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27946 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27947 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27948 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27949 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27950 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27951 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27952 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27953 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27954 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27956 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27958 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27960 2017-05-02 Richard Biener <rguenther@suse.de>
27962 PR tree-optimization/80591
27964 2017-04-10 Richard Biener <rguenther@suse.de>
27966 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27969 2017-05-02 Richard Biener <rguenther@suse.de>
27971 PR tree-optimization/80549
27972 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27973 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27974 headers if they do not have a preheader.
27976 2017-05-02 Martin Liska <mliska@suse.cz>
27979 * common.opt: Fix typo.
27980 * doc/invoke.texi: Likewise.
27982 2017-05-01 Jan Beulich <jbeulich@suse.com>
27984 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27985 swapping, add (x,x,m,x,n) alternative.
27987 2017-05-01 Nathan Sidwell <nathan@acm.org>
27989 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27990 unnecessary unadjusted_alignment check.
27992 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27995 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27997 * gimplify.c (gimplify_cilk_detach): New function.
27998 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27999 * tree-core.h: Document EXPR_CILK_SPAWN.
28000 * tree.h (EXPR_CILK_SPAWN): Define.
28002 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28004 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28005 to use new fixit_hint representation, using the "replace" logic.
28006 (get_line_span_for_fixit_hint): Likewise.
28007 (layout::print_any_fixits): Likewise.
28008 (selftest::test_one_liner_many_fixits): Rename to...
28009 (selftest::test_one_liner_many_fixits_1): ...this, and update
28010 comment and expected output to reflect that the multiple fix-it
28011 hints are now consolidated into one insertion.
28012 (selftest::test_one_liner_many_fixits_2): New test.
28013 (selftest::test_diagnostic_show_locus_one_liner): Update for
28015 (selftest::test_fixit_consolidation): Update for fix-it API
28017 * diagnostic.c (print_parseable_fixits): Likewise.
28018 * edit-context.c (edited_line::m_line_events): Convert from
28019 auto_vec <line_event *> to auto_vec <line_event>.
28020 (class line_event): Convert from abstract base class to a concrete
28021 class, taking over the role of replace_event.
28022 (class insert_event): Delete.
28023 (class replace_event): Rename to class line_event. Convert to
28025 (edit_context::add_fixits): Reimplement.
28026 (edit_context::apply_insert): Delete.
28027 (edit_context::apply_replace): Rename to...
28028 (edit_context::apply_fixit): ...this. Convert to half-open range.
28029 (edited_file::apply_insert): Delete.
28030 (edited_file::apply_replace): Rename to...
28031 (edited_file::apply_fixit): ...this.
28032 (edited_line::~edited_line): Drop deletion of events.
28033 (edited_line::apply_insert): Delete.
28034 (edited_line::apply_replace): Rename to...
28035 (edited_line::apply_fixit): ...this. Convert to half-open range.
28036 Update for change to type of m_line_events.
28037 * edit-context.h (edit_context::apply_insert): Delete.
28038 (edit_context::apply_replace): Rename to...
28039 (edit_context::apply_fixit): ...this.
28041 2017-05-01 Martin Sebor <msebor@redhat.com>
28043 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28046 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28049 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28050 __get_cpuid_max returns 0.
28051 (__get_cpuid_count): Ditto.
28053 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28055 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28056 replacement expression is another instance of one of its arguments.
28058 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28061 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28062 check for stack push/pop autoinc.
28063 * config/i386/i386.c (ix86_agi_dependent): Return false
28064 if the only reason why modified_in_p returned true is that
28065 addr is SP based and set_insn is a push or pop.
28067 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28069 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28072 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28075 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28076 (account_size_time): Use two predicates - exec_pred and
28078 (evaluate_conditions_for_known_args): Compute both clause and
28080 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28081 (inline_summary_t::duplicate): Update.
28082 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28084 (compute_inline_parameters): Likewise.
28085 (estimate_edge_size_and_time): Update caluclation of time.
28086 (estimate_node_size_and_time): Compute both time and nonspecialized
28088 (estimate_ipcp_clone_size_and_time): Update.
28089 (inline_merge_summary): Update.
28090 (do_estimate_edge_time): Update.
28091 (do_estimate_edge_size): Update.
28092 (do_estimate_edge_hints): Update.
28093 (inline_read_section, inline_write_summary): Stream both new predicates.
28094 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28096 (compute_inlined_call_time): Cleanup.
28097 (big_speedup_p): Update.
28098 (edge_badness): Update.
28099 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28100 (size_time_entry): Replace predicate by exec_predicate and
28101 nonconst_predicate.
28102 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28103 (estimate_edge_time): Return also nonspec_time.
28104 (reset_edge_growth_cache): Update.
28106 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28108 PR rtl-optimization/80491
28109 * ifcvt.c (noce_process_if_block): When looking for x setter
28110 with missing else_bb, don't check only the insn right before
28111 cond_earliest, but look for the last insn that x is modified in
28112 within the same bb.
28114 PR rtl-optimization/80491
28115 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28117 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28119 PR tree-optimization/80487
28120 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28122 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28124 PR tree-optimization/79697
28125 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28126 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28127 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28129 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28130 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28132 2017-04-28 Martin Sebor <msebor@redhat.com>
28134 PR tree-optimization/80523
28135 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28136 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28138 (maybe_warn, format_directive, parse_directive): Use new functions.
28139 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28141 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28143 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28145 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28147 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28148 target_header_dir): Set correctly.
28149 * configure: Regenerated.
28150 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28151 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28152 instead of SYSTEM_HEADER_DIR.
28154 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28156 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28157 (estimate_local_effects): Likewise.
28158 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28159 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28160 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28161 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28162 do_estimate_edge_time, estimate_edge_time): Likewise.
28163 * ipa-inline-analysis.c (estimate_node_size_and_time,
28164 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28165 (estimate_time_after_inlining): Remove.
28167 2017-04-28 Martin Liska <mliska@suse.cz>
28169 * doc/gcov.texi: Enhance documentation of gcov.
28171 2017-04-28 Martin Liska <mliska@suse.cz>
28173 * doc/gcov.texi: Sort options in alphabetic order.
28174 * doc/gcov-dump.texi: Likewise.
28175 * doc/gcov-tool.texi: Likewise.
28176 * gcov.c (print_usage): Likewise.
28177 * gcov-dump.c (print_usage): Likewise.
28178 * gcov-tool.c (print_merge_usage_message): Likewise.
28179 (print_rewrite_usage_message): Likewise.
28180 (print_overlap_usage_message): Likewise.
28182 2017-04-28 Martin Liska <mliska@suse.cz>
28184 PR gcov-profile/53915
28185 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28187 2017-04-28 Martin Liska <mliska@suse.cz>
28189 PR gcov-profile/79891
28190 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28191 is marked by compiler as living on a line.
28192 (get_cycles_count): Remove usage of the union.
28193 (output_intermediate_file): Likewise.
28194 (find_source): Fix GNU coding style.
28195 (accumulate_line_counts): Remove old non-all block mode.
28196 (output_lines): Remove usage of the union.
28197 * profile.c (output_location): Include all BBs, even if
28198 belonging to a same line (and file) as a previous BB.
28200 2017-04-28 Martin Liska <mliska@suse.cz>
28202 * gcov.c (process_args): Handle new argument 'w'.
28203 (read_graph_file): Assign ID to BBs.
28204 (output_branch_count): Display BB # if verbose flag is set.
28205 (output_lines): Likewise for arcs.
28206 (print_usage): Add '--verbose' option help.
28207 * doc/gcov.texi: Document --verbose (-w) option.
28209 2017-04-28 Martin Liska <mliska@suse.cz>
28211 * gcov.c (struct block_location_info): New struct.
28212 (process_file): Fill up the new structure.
28213 (read_graph_file): Replace usage of encoding by the newly added
28215 (add_line_counts): Likewise.
28216 (accumulate_line_counts): Remove usage of the union.
28217 (function_info::function_info): New function.
28218 (function_info::~function_info): Likewise.
28219 (process_file): Call delete instead of release_function.
28220 (release_function): Release the function.
28221 (release_structures): Call delete instead of release_function.
28222 (solve_flow_graph): Replace usage of num_blocks.
28223 (find_exception_blocks): Likewise.
28224 (output_lines): Fix GNU coding style.
28226 2017-04-28 Martin Liska <mliska@suse.cz>
28229 * coverage.c (coverage_remove_note_file): New function.
28230 * coverage.h: Declare the function.
28231 * toplev.c (finalize): Clean if an error has been seen.
28233 2017-04-28 Martin Liska <mliska@suse.cz>
28235 PR gcov-profile/80031
28236 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28237 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28238 * gcov.c (read_graph_file): Read just number of blocks.
28239 * profile.c (branch_prob): Do not stream 0 flags per a basic
28242 2017-04-28 Martin Liska <mliska@suse.cz>
28244 * gcov-dump.c (tag_*): Add new argument to declarations.
28245 (dump_gcov_file): Likewise.
28246 (tag_blocks): Add and use new argument depth.
28247 (tag_arcs): Likewise.
28248 (tag_lines): Likewise.
28249 (tag_counters): Likewise.
28250 (tag_summary): Likewise.
28251 (dump_working_sets): Use depth to do a proper indentation.
28253 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28256 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28257 * symtab.c (symtab_node::debug_symtab): Move definition here.
28259 2017-04-28 Richard Biener <rguenther@suse.de>
28261 * lto-streamer.h (LTO_major_version): Bump to 7.
28263 2017-04-28 Richard Biener <rguenther@suse.de>
28265 * tree-vrp.c (assert_info): New struct.
28266 (add_assert_info): New helper.
28267 (register_edge_assert_for_2): Refactor to add asserts to a vector
28269 (register_edge_assert_for_1): Likewise.
28270 (register_edge_assert_for): Likewise.
28271 (finish_register_edge_assert_for): New helper actually registering
28272 asserts where live on edge.
28273 (find_conditional_asserts): Adjust.
28274 (find_switch_asserts): Likewise.
28275 (evrp_dom_walker::try_find_new_range): Generalize.
28276 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28278 2017-04-27 Marek Polacek <polacek@redhat.com>
28281 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28282 arg10 and arg11 to itype.
28284 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28286 * doc/extend.texi (Object Size Checking): Improve grammar.
28288 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28291 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28292 that the logic for permitting reciprocal estimates matches that
28295 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28298 * tree.c (type_cache_hasher::equal): Only compare
28299 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28300 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28301 non-aggregate element types.
28302 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28303 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28305 2017-04-27 Richard Biener <rguenther@suse.de>
28307 PR middle-end/80533
28308 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28309 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28310 keeping a reference to a trailing array.
28312 2017-04-27 Richard Biener <rguenther@suse.de>
28314 PR middle-end/80539
28315 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28316 being in loop-closed SSA form conservatively.
28317 (chrec_fold_multiply_poly_poly): Likewise.
28319 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28321 PR middle-end/79665
28322 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28323 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28325 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28328 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28329 (aarch64_function_arg_alignment): Return unsigned int again, but still
28330 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28331 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28332 Don't emit -Wpsabi note.
28333 (aarch64_function_arg_boundary): Likewise.
28334 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28337 2017-04-26 Nathan Sidwell <nathan@acm.org>
28339 * tree.h (crc32_unsigned_n): Declare.
28340 (crc32_unsigned, crc32_unsigned): Make inline.
28341 * tree.c (crc32_unsigned_bits): Replace with ...
28342 (crc32_unsigned_n): ... this.
28343 (crc32_unsigned, crc32_byte): Remove.
28344 (crc32_string): Remove unnecessary braces.
28346 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28348 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28349 * ipa-inline-analysis.c (MAX_TIME): Remove.
28350 (account_size_time): Use sreal for time.
28351 (dump_inline_summary): Update.
28352 (estimate_function_body_sizes): Update.
28353 (estimate_edge_size_and_time): Update.
28354 (estimate_calls_size_and_time): Update.
28355 (estimate_node_size_and_time): Update.
28356 (inline_merge_summary): Update.
28357 (inline_update_overall_summary): Update.
28358 (estimate_time_after_inlining): Update.
28359 (inline_read_section): Update.
28360 (inline_write_summary): Update.
28361 * ipa-inline.c (compute_uninlined_call_time): Update.
28362 (compute_inlined_call_time): Update.
28363 (recursive_inlining): Update.
28364 (inline_small_functions): Update.
28365 (dump_overall_stats): Update.
28366 * ipa-inline.h: Include sreal.h.
28367 (size_time_entry): Turn time to sreal.
28368 (inline_summary): Turn self_time nad time to sreal.
28370 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28372 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28374 (sreal::stream_out, sreal::stream_in): New.
28375 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28377 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28379 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28382 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28385 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28387 (dimode_scalar_chain::compute_convert_gain): Ditto.
28388 (dimode_scalar_chain::make_vector_copies): Ditto.
28389 (dimode_scalar_chain::convert_reg): Ditto.
28390 (dimode_scalar_chain::convert_insn): Ditto.
28391 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28392 (VI248_AVX512BW_1): New mode iterator.
28393 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28394 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28397 2017-04-25 Martin Sebor <msebor@redhat.com>
28399 PR tree-optimization/80497
28400 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28401 constants are representable in HOST_WIDE_INT.
28402 (parse_directive): Ditto.
28404 2017-04-25 Martin Sebor <msebor@redhat.com>
28407 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28408 (new_zero_array): Adjust signature.
28409 (dom_info::dom_init): Used unsigned rather that size_t.
28410 (dom_info::dom_info): Same.
28412 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28413 Jakub Jelinek <jakub@redhat.com>
28416 * config/arm/arm.c: Include gimple.h.
28417 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28418 returns negative, increment ncrn only if it returned positive.
28419 (arm_needs_doubleword_align): Return int instead of bool,
28420 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28421 members, but if there is any such non-FIELD_DECL
28422 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28423 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28424 returns negative, increment nregs only if it returned positive.
28425 (arm_setup_incoming_varargs): Likewise.
28426 (arm_function_arg_boundary): Emit -Wpsabi note if
28427 arm_needs_doubleword_align returns negative, return
28428 DOUBLEWORD_ALIGNMENT only if it returned positive.
28430 2017-04-25 Marek Polacek <polacek@redhat.com>
28433 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28434 first argument to type.
28436 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28439 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28440 type checks to test for compatibility instead of equality.
28442 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28443 Jakub Jelinek <jakub@redhat.com>
28446 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28448 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28449 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28450 the alignment computation, but return their maximum in warn_alignment.
28451 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28452 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28454 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28455 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28458 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28460 * config/arc/simdext.md (dmpyh): Fix typo.
28462 2017-04-25 Richard Biener <rguenther@suse.de>
28464 PR tree-optimization/80492
28465 * alias.c (compare_base_decls): Handle registers with asm
28466 specification conservatively.
28467 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28468 compare_base_decls returning dont-know properly.
28470 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28472 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28473 (legitimate_offset_address_p): New function.
28474 (arc_legitimate_address_p): Use above function.
28476 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28478 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28480 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28482 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28483 ACCH registers whenever they are available.
28485 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28487 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28488 double regs fix when not used.
28490 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28492 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28494 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28495 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28497 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28499 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28500 when emitting short ADD instructions.
28502 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28504 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28506 (cmpsi_cc_c_insn): Likewise.
28507 (cbranchsi4_scratch): Compute proper instruction length using
28508 compact_hreg_operand.
28509 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28511 2017-04-25 Richard Biener <rguenther@suse.de>
28513 PR middle-end/80509
28514 * passes.c (pass_manager::pass_manager): Initialize
28515 m_name_to_pass_map.
28517 2017-04-25 Richard Biener <rguenther@suse.de>
28519 PR tree-optimization/79201
28520 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28522 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28525 * config/s390/vector.md: Split MEM->GPR vector moves for
28526 non-s_operand addresses.
28528 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28531 * config/s390/predicates.md (reload_const_wide_int_operand): New
28533 * config/s390/s390.md ("movti"): Remove d/P alternative.
28534 ("movti_bigconst"): New pattern definition.
28536 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28539 * s390-protos.h (s390_expand_cs_hqi): Removed.
28540 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28541 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28542 modes as well as CCZ1mode and CCZmode.
28543 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28544 signature of s390_emit_compare_and_swap.
28545 (s390_expand_cs_hqi): Likewise, make static.
28546 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28547 compare-and-swap, in some cases.
28548 (s390_expand_cs): Wrapper function.
28549 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28551 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28552 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28553 patterns for small and large integers. Forbid symref memory operands.
28554 Move expander to s390.c. Require cc register.
28555 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28556 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28557 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28558 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28559 symref memory operands. Remove CC mode and call s390_match_ccmode
28561 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28563 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28565 * config/s390/s390.md (define_peephole2): New peephole to help
28566 combining the load-and-test pattern with volatile memory.
28568 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28570 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28571 with CCZmode for TARGET_Z196.
28573 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28575 PR rtl-optimization/80501
28576 * combine.c (make_compound_operation_int): Set subreg_code to SET
28577 even for AND with mask of the sign bit of mode.
28579 PR rtl-optimization/80500
28580 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28581 sum's initial value.
28583 2017-04-25 Julian Brown <julian@codesourcery.com>
28584 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28586 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28588 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28590 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28592 2017-04-25 Julian Brown <julian@codesourcery.com>
28593 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28595 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28596 (thunderx2t99_sha): New Reservation.
28598 2017-04-25 Julian Brown <julian@codesourcery.com>
28599 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28601 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28602 type for 1-element load.
28604 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28606 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28608 2017-04-24 Martin Jambor <mjambor@suse.cz>
28610 PR tree-optimization/80293
28611 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28612 char arrays not totally scalarizable if it is false.
28613 (analyze_all_variable_accesses): Pass correct value in the new
28614 parameter. Add a statistics counter.
28616 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28618 PR middle-end/79931
28619 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28621 2017-04-24 Richard Biener <rguenther@suse.de>
28623 PR tree-optimization/80494
28624 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28625 out for complex types.
28627 2017-04-24 Richard Biener <rguenther@suse.de>
28629 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28630 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28631 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28632 (DFS): Adjust and never fail.
28633 (sccvn_dom_walker::fail): Remove.
28634 (sccvn_dom_walker::before_dom_children): Adjust.
28635 (run_scc_vn): Likewise and never fail.
28636 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28637 (pass_fre::execute): Likewise.
28639 2017-04-24 Richard Biener <rguenther@suse.de>
28641 PR tree-optimization/79725
28642 * tree-ssa-sink.c (statement_sink_location): Return whether
28643 failure reason was zero uses. Move that check later.
28644 (sink_code_in_bb): Deal with zero uses by removing the stmt
28647 2017-04-24 Richard Biener <rguenther@suse.de>
28650 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28651 pointer-based references.
28653 2017-04-24 Richard Biener <rguenther@suse.de>
28656 * pass_manager.h (pass_manager::operator new): Remove.
28657 (pass_manager::operator delete): Likewise.
28658 * passes.c (pass_manager::operator new): Remove.
28659 (pass_manager::operator delete): Likewise.
28660 (pass_manager::pass_manager): Zero individual pass members.
28662 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28665 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28666 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28667 Check "XEXP (src, 1)" operand here.
28668 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28669 Check "XEXP (src, 1)" operand here.
28670 (dimode_scalar_chain::make_vector_copies): Detect count register
28671 of a shift instruction. Zero extend count register from QImode
28672 to DImode to satisfy vector shift pattern count operand predicate.
28673 Substitute vector shift count operand with a DImode copy.
28674 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28677 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28679 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28680 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28681 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28682 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28683 (UNSPEC_NOREX_MEM): Remove definition.
28685 2017-04-21 Richard Biener <rguenther@suse.de>
28687 PR tree-optimization/79547
28688 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28689 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28690 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28691 without any constraints.
28693 2017-04-21 Richard Biener <rguenther@suse.de>
28695 PR tree-optimization/78847
28696 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28698 2017-04-21 Richard Biener <rguenther@suse.de>
28700 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28701 (build_distinct_type_copy): Likewise.
28702 (build_variant_type_copy): Likewise.
28703 * tree.c (build_qualified_type): Pass down mem-stat info.
28704 (build_distinct_type_copy): Likewise.
28705 (build_variant_type_copy): Likewise.
28707 2017-04-21 Richard Biener <rguenther@suse.de>
28709 PR tree-optimization/80237
28710 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28712 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28713 for a simplified result.
28715 2016-04-21 Richard Biener <rguenther@suse.de>
28717 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28718 sth as strict as a simple_iv but a chrec without symbols and an
28719 operand defined in the loop we are peeling (and not some subloop).
28720 (propagate_constants_for_unrolling): Propagate all constants.
28722 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28725 * config/i386/i386.c (print_reg): Remove assert for disalowed
28726 regno values, call output_operand_lossage instead.
28728 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28731 * config/i386/constraints.md (Yc): New register constraint.
28732 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28733 Use Yc constraint for alternative 2 of operand 0. Remove
28734 preferred_for_speed attribute.
28736 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28738 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28739 lastprivate clauses in SIMT case.
28741 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28743 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28745 2017-04-20 Richard Biener <rguenther@suse.de>
28747 PR tree-optimization/57796
28748 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28749 as N scalar stores.
28750 (vect_model_load_cost): Cost gathers as N scalar loads.
28752 2017-04-20 Richard Biener <rguenther@suse.de>
28754 * ggc-page.c (ggc_allocated_p): Rename to ...
28755 (safe_lookup_page_table_entry): ... this and return the lookup
28757 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28759 2017-04-20 Richard Biener <rguenther@suse.de>
28761 PR tree-optimization/80453
28762 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28763 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28764 from the conditions.
28765 (vn_phi_eq): Pass them down.
28766 (vn_phi_lookup): Record them.
28767 (vn_phi_insert): Likewise.
28769 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28771 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28772 uninitialized variable warning to avoid buffer overrun.
28774 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28777 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28778 is suppressed for '{ 0 }' in C.
28780 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28782 * BASE-VER: Set to 8.0.0.
28784 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28786 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28787 priority .init_array and .fini_array section with SECTION_NOTYPE
28790 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28792 PR middle-end/80423
28793 * tree.h (build_array_type): Add typeless_storage default argument.
28794 * tree.c (type_cache_hasher::equal): Also compare
28795 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28796 (build_array_type): Add typeless_storage argument, set
28797 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28799 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28800 (build_array_type): Likewise. Add typeless_storage argument.
28802 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28803 Jakub Jelinek <jakub@redhat.com>
28805 PR tree-optimization/80426
28806 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28807 operation on symbolic operands, also compute the overflow for the
28808 invariant part when the operation degenerates into a negation.
28810 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28813 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28814 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28817 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28819 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28822 * config/avr/avr.c (tree.h): Include it.
28823 (cgraph.h): Include it.
28824 (avr_encode_section_info): Don't warn for uninitialized progmem
28825 variable if it's just an alias.
28827 2017-04-19 Richard Biener <rguenther@suse.de>
28830 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28831 when needed by AutoPGO.
28833 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28836 * doc/lto.texi: Remove an extra 'that'.
28838 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28840 PR rtl-optimization/80429
28841 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28842 are only used in debug insns.
28844 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28845 Vladimir Makarov <vmakarov@redhat.com>
28847 * config/sparc/predicates.md (input_operand): Add comment. Return
28848 true for any memory operand when LRA is in progress.
28849 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28851 2017-04-18 Jeff Law <law@redhat.com>
28854 * mips.md ({return,simple_return}_internal): Do not overwrite
28857 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28859 PR tree-optimization/80443
28860 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28861 instead of adding 1, subtract -1 and similarly instead of subtracting
28864 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28866 PR rtl-optimization/80357
28867 * haifa-sched.c (tmp_bitmap): New variable.
28868 (model_recompute): Handle duplicate use records.
28869 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28870 (free_global_sched_pressure_data): Free it.
28872 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28875 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28876 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28877 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28878 instead of SYSTEM_HEADER_DIR.
28880 2017-04-18 Jeff Law <law@redhat.com>
28882 PR middle-end/80422
28883 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28884 predecessors after walking up the insn chain.
28886 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28889 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28890 sizetype type into debug info.
28892 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28895 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28896 unneeded test for TARGET_UPPER_REGS_SF.
28897 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28899 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28902 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28903 instead of gsi_after_labels.
28905 2017-04-18 Jeff Law <law@redhat.com>
28907 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28911 2017-04-13 Jeff Law <law@redhat.com>
28912 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28913 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28915 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28918 * config/avr/avr.c (intl.h): Include it.
28919 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28921 2017-04-18 Martin Liska <mliska@suse.cz>
28923 PR gcov-profile/78783
28924 * gcov-tool.c (gcov_output_files): Validate that destination
28925 file is either removed by the tool or by a user.
28927 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28928 Guy Benyei <guybe@mellanox.com>
28930 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28931 block, and do not negate it, the stored id is already negative.
28933 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28935 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28937 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28940 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28941 masks of options that should be turned off if the VSX vector
28942 options are turned off.
28943 (OTHER_P8_VECTOR_MASKS): Likewise.
28944 (OTHER_VSX_VECTOR_MASKS): Likewise.
28945 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28946 rs6000_disable_incompatible_switches to validate no type switches
28948 (rs6000_incompatible_switch): New function to disallow turning on
28949 other vector options if -mno-vsx, -mno-power8-vector, or
28950 -mno-power9-vector are specified.
28952 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28954 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28956 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28958 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28959 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28960 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28961 (ARG_POINTER_CFA_OFFSET): Likewise.
28963 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28965 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28966 conditions to take advantage of various optimizations.
28968 2017-04-13 Jeff Law <law@redhat.com>
28970 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28971 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28972 (zero_extendsidi2_dext): Likewise.
28974 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28977 * fold-const.c (fold_ternary_loc): Revert
28978 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28981 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28983 PR rtl-optimization/80343
28984 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28985 new scratch pseudo.
28987 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28990 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28991 to ubsan_encode_value.
28993 2017-04-13 Jeff Law <law@redhat.com>
28995 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28996 appearing in DEBUG_INSNs.
28998 2017-04-13 Martin Liska <mliska@suse.cz>
29000 PR gcov-profile/80413
29001 * gcov-io.c (gcov_write_string): Copy to buffer just when
29002 allocated size is greater than zero.
29004 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29007 * dwarf2out.c (decls_for_scope): Ignore declarations of
29008 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29010 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29013 * ipa-visibility.c (non_local_p): Fix typos.
29014 (localize_node): When localizing symbol in same comdat group,
29015 dissolve the group only when we know external symbols are going
29017 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29019 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29021 PR tree-optimization/79390
29022 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29023 order does not result in usable sequence, retry with reversed operand
29029 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29030 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29031 op0 instead of fold_convert_loc (loc, type, arg0).
29033 2017-04-12 Jeff Law <law@redhat.com>
29035 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29036 has a delay slot in the generated code.
29038 * config/cris/cris.md (cris_preferred_reload_class): Return
29039 GENNONACR_REGS rather than GENERAL_REGS.
29041 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29044 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29045 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29046 signedness of the result type.
29048 2017-04-12 Richard Biener <rguenther@suse.de>
29049 Jeff Law <law@redhat.com>
29051 PR tree-optimization/80359
29052 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29053 trim stores to TARGET_MEM_REFs.
29055 2017-04-12 Richard Biener <rguenther@suse.de>
29057 PR tree-optimization/79390
29058 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29059 threading case even more.
29061 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29064 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29065 for quad_address_p for TImode, instead of just not indexed_address.
29067 2017-04-12 Richard Biener <rguenther@suse.de>
29068 Bernd Edlinger <bernd.edlinger@hotmail.de>
29070 PR middle-end/79671
29071 * alias.c (component_uses_parent_alias_set_from): Handle
29072 TYPE_TYPELESS_STORAGE.
29073 (get_alias_set): Likewise.
29074 * tree-core.h (tree_type_common): Add typeless_storage flag.
29075 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29076 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29077 for types containing members with TYPE_TYPELESS_STORAGE.
29078 (place_field): Likewise.
29079 (layout_type): Likewise for ARRAY_TYPE.
29080 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29081 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29082 TYPE_TYPELESS_STORAGE.
29083 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29085 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29088 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29089 first argument to type.
29091 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29095 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29096 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29097 (rs6000_expand_binop_builtin): Likewise.
29098 (rs6000_expand_ternop_builtin): Likewise; also add missing
29099 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29100 vshasigma built-ins.
29101 * doc/extend.texi: Document that vec_xxpermdi's third argument
29102 must be a constant.
29104 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29106 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29107 Use shift_const cost parameter when calculating gain of STV shifts.
29109 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29111 PR rtl-optimization/70478
29112 * lra-constraints.c (process_alt_operands): Check memory for
29113 disfavoring memory insn operand.
29115 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29117 PR middle-end/80100
29118 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29119 left shift in unsigned HOST_WIDE_INT type.
29121 PR rtl-optimization/80385
29122 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29123 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29126 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29127 if they have any depend clauses.
29129 2017-04-11 Martin Liska <mliska@suse.cz>
29132 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29133 * ipa-split.c (split_function): Create a local comdat symbol
29134 if caller is in a comdat group.
29136 2017-04-11 Martin Liska <mliska@suse.cz>
29139 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29142 2017-04-11 Martin Sebor <msebor@redhat.com>
29144 PR middle-end/80364
29145 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29146 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29148 (directive::set_width, directive::set_precision, format_character):
29150 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29153 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29156 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29157 conflict, set target->arch_name instead of target->cpu_name.
29159 2017-04-11 Richard Biener <rguenther@suse.de>
29161 PR tree-optimization/80374
29162 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29163 build_zero_cst, remove fold_convertible_p check again.
29165 2017-04-11 Martin Liska <mliska@suse.cz>
29168 * ubsan.c (instrument_object_size): Do not instrument register
29171 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29174 * config/i386/i386-builtin-types.def
29175 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29176 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29177 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29178 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29179 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29180 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29181 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29182 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29183 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29184 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29185 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29186 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29187 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29188 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29189 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29190 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29191 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29192 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29193 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29194 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29195 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29196 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29197 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29198 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29199 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29200 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29201 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29202 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29203 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29204 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29205 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29206 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29207 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29208 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29209 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29210 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29211 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29212 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29213 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29214 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29215 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29216 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29217 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29218 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29219 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29220 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29222 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29223 flag to second_arg_count, handle 4 argument function type _COUNT
29224 aliases, handle second_arg_count on second argument rather than last.
29226 2017-04-10 Jeff Law <law@redhat.com>
29228 PR tree-optimization/80374
29229 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29230 record anything if we can not convert integer_zero_node to the
29233 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29236 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29237 Enhance special handling given to the TARGET_P9_MINMAX option in
29238 relation to certain other options.
29240 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29242 PR tree-optimization/80153
29243 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29244 remove POINTER_PLUS_EXPR's base part directly, rather than through
29247 2017-04-10 Richard Biener <rguenther@suse.de>
29248 Bin Cheng <bin.cheng@arm.com>
29250 PR tree-optimization/80153
29251 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29252 the first element of pointer type aff_tree. Build result expr in
29254 (add_elt_to_tree): Convert to type unconditionally. Remove other
29255 fold_convert calls.
29256 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29257 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29259 2017-04-10 Richard Biener <rguenther@suse.de>
29261 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29264 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29266 PR rtl-optimization/70478
29267 * lra-constraints.c (curr_small_class_check): New.
29268 (update_and_check_small_class_inputs): New.
29269 (process_alt_operands): Update curr_small_class_check. Disfavor
29270 alternative insn memory operands. Check available regs for small
29273 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29276 * config/mips/mips.opt (-mvirt): Update description.
29277 * doc/invoke.texi (-mvirt): Likewise.
29279 2017-04-10 Richard Biener <rguenther@suse.de>
29281 PR middle-end/80362
29282 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29283 looking for NEGATE_EXPR in -A / -B to A / B folding.
29285 2017-04-10 Martin Liska <mliska@suse.cz>
29287 PR gcov-profile/80224
29288 * gcov.c (print_usage): Fix usage string.
29289 (get_gcov_intermediate_filename): Remove.
29290 (output_gcov_file): Use both for normal and intermediate format.
29291 (generate_results): Do not initialize special file for
29292 intermediate format.
29294 2017-04-10 Richard Biener <rguenther@suse.de>
29296 PR tree-optimization/80304
29297 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29300 2017-04-10 Nathan Sidwell <nathan@acm.org>
29303 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29304 (rs6000_init_builtins): Use it.
29306 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29308 * config/arm/arm.md (<mrc>): Add mode to SET source.
29309 (<mrrc>): Likewise.
29311 2017-04-10 Richard Biener <rguenther@suse.de>
29313 PR middle-end/80344
29314 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29316 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29319 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29320 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29321 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29322 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29323 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29324 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29325 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29326 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29327 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29328 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29329 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29330 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29331 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29332 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29333 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29334 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29335 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29336 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29337 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29338 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29339 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29340 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29341 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29343 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29345 PR rtl-optimization/70478
29346 * lra-constraints.c: Reverse the last patch.
29348 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29350 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29351 Add comment for WCHAR_T.
29353 2017-04-08 Martin Liska <mliska@suse.cz>
29356 2017-04-07 Martin Liska <mliska@suse.cz>
29359 * ipa-split.c (split_function): Add function part to a same comdat
29362 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29365 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29367 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29369 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29370 (vec_pairing): ...to this.
29371 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29372 (rs6000_sched_init): Adjust for name change.
29373 (struct rs6000_sched_context): Likewise.
29374 (rs6000_init_sched_context): Likewise.
29375 (rs6000_set_sched_context): Likewise.
29377 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29383 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29385 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29386 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29387 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29389 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29391 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29393 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29395 PR rtl-optimization/70703
29396 * ira-color.c (update_conflict_hard_regno_costs): Use
29397 int64_t instead of HOST_WIDE_INT.
29399 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29401 PR rtl-optimization/70478
29402 * lra-constraints.c (process_alt_operands): Disfavor alternative
29403 insn memory operands.
29405 2017-04-07 Jeff Law <law@redhat.com>
29407 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29408 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29410 2017-04-07 Martin Liska <mliska@suse.cz>
29413 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29414 Show error message instead of an ICE.
29416 2017-04-07 Martin Liska <mliska@suse.cz>
29419 * ipa-split.c (split_function): Add function part to a same comdat
29422 2017-04-07 Richard Biener <rguenther@suse.de>
29424 PR middle-end/80341
29425 * tree.c (get_unwidened): Also handle ! for_type case for
29427 * convert.c (do_narrow): Split out from ...
29428 (convert_to_integer_1): ... here. Do not pass final truncation
29429 type to get_unwidened for TRUNC_DIV_EXPR.
29431 2017-04-07 Richard Biener <rguenther@suse.de>
29433 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29435 (aff_combination_const): Adjust.
29436 (aff_combination_scale): Likewise.
29437 (aff_combination_add_elt): Likewise.
29438 (aff_combination_add_cst): Likewise.
29439 (aff_combination_convert): Likewise.
29440 (add_elt_to_tree): Likewise. Remove unused argument.
29441 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29443 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29445 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29447 * config/arm/arm.c (arm_default_short_enums): Use
29448 ARM_DEFAULT_SHORT_ENUMS.
29449 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29451 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29454 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29455 members with redundant out-of-class redeclaration.
29457 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29460 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29461 * config/i386/i386.md (*zero_extendsidi2):
29462 Add (?*x,*x) and (?*v,*v) alternatives.
29464 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29467 * config/i386/i386.c (ix86_expand_builtin)
29468 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29469 mode from insn data. Convert operands to insn operand mode.
29470 Copy operands that don't satisfy insn predicate to a register.
29472 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29474 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29477 2017-04-06 Richard Biener <rguenther@suse.de>
29479 PR tree-optimization/80334
29480 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29481 preserve alignment of accesses.
29483 2017-04-06 Richard Biener <rguenther@suse.de>
29485 PR tree-optimization/80262
29486 * tree-sra.c (build_ref_for_offset): Preserve address-space
29488 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29489 Drop useless address-space information on MEM_REF offsets.
29491 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29493 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29495 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29497 PR rtl-optimization/70703
29498 * ira-color.c (update_conflict_hard_regno_costs): Use
29499 HOST_WIDE_INT instead of long.
29501 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29504 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29505 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29507 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29508 for x86_64 target. Handle -m3dnowa option.
29510 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29512 PR rtl-optimization/70703
29513 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29514 (update_conflict_hard_regno_costs): Use long instead of unsigned
29515 arithmetic for cost calculation.
29517 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29518 Bernd Edlinger <bernd.edlinger@hotmail.de>
29521 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29524 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29527 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29528 ptr_mode with Pmode throughout.
29529 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29530 into probe_stack_range and use DImode.
29532 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29535 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29536 call_eh_return is true.
29538 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29540 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29541 Initialize last_match_fntype_index.
29543 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29546 * tree-nvr.c: Include internal-fn.h.
29547 (pass_return_slot::execute): Ignore internal calls without
29550 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29551 Richard Biener <rguenther@suse.de>
29554 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29555 captures used multiple times, except for the last use.
29556 * generic-match-head.c: Include gimplify.h.
29558 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29560 PR tree-optimization/79390
29561 * target.h (struct noce_if_info): Declare.
29562 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29563 * target.def (noce_conversion_profitable_p): New target hook.
29564 * ifcvt.h (struct noce_if_info): New type, moved from ...
29565 * ifcvt.c (struct noce_if_info): ... here.
29566 (noce_conversion_profitable_p): Renamed to ...
29567 (default_noce_conversion_profitable_p): ... this. No longer
29569 (noce_try_store_flag_constants, noce_try_addcc,
29570 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29571 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29572 instead of noce_conversion_profitable_p.
29573 * config/i386/i386.c: Include ifcvt.h.
29574 (ix86_option_override_internal): Don't override
29575 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29576 (ix86_noce_conversion_profitable_p): New function.
29577 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29578 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29579 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29580 * doc/tm.texi: Regenerated.
29582 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29584 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29587 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29590 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29591 instructions for small multiply cores.
29593 2017-04-04 Jeff Law <law@redhat.com>
29595 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29597 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29598 that are not set by the loop over the elements.
29600 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29603 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29604 int mode, convert_modes it to mode as unsigned, otherwise use
29605 lowpart_subreg to mode rather than SImode.
29606 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29607 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29608 Use DImode instead of SImode for the shift count operand.
29609 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29612 2017-04-04 Richard Biener <rguenther@suse.de>
29614 PR middle-end/80281
29615 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29616 arithmetic done for the negate or the plus. Simplify.
29617 (A - (-B) -> A + B): Likewise.
29618 * fold-const.c (split_tree): Make sure to not negate pointers.
29620 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29622 PR rtl-optimization/60818
29623 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29624 a compare of comparisons with the thing compared if this results
29625 in a different machine mode.
29627 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29629 * alias.c (base_alias_check): Fix typo in comment.
29630 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29631 * cgraphunit.c (symbol_table::compile): Likewise.
29632 * collect2.c (maybe_run_lto_and_relink): Likewise.
29633 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29634 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29635 * config/avr/avr.c (avr_map_op_t): Likewise.
29636 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29637 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29638 * config/epiphany/epiphany.md (movcc): Likewise.
29639 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29640 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29642 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29643 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29644 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29645 * config/sh/sh.c (sh_rtx_costs): Likewise.
29646 * fold-const.c (fold_truth_andor): Likewise.
29647 * genautomata.c (collapse_flag): Likewise.
29648 * gengtype.h (struct type::u::s): Likewise.
29649 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29650 * input.c (FORMAT_AMOUNT): Likewise.
29651 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29652 (known_aggs_to_agg_replacement_list): Likewise.
29653 * ipa-inline-analysis.c: Likewise.
29654 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29655 * ipa-polymorphic-call.c
29656 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29657 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29658 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29660 * modulo-sched.c (apply_reg_moves): Likewise.
29661 * omp-expand.c (build_omp_regions_1): Likewise.
29662 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29663 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29664 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29665 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29666 * value-prof.c: Likewise.
29667 * var-tracking.c (val_reset): Likewise.
29669 2017-04-03 Richard Biener <rguenther@suse.de>
29671 PR tree-optimization/80275
29672 * fold-const.c (split_address_to_core_and_offset): Handle
29675 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29677 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29678 descriptors is at least equal to that of functions.
29680 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29682 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29684 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29687 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29688 (mov<IMOD4:mode>): New expander.
29689 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29691 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29693 PR rtl-optimization/79405
29694 * fwprop.c (propagations_left): New variable.
29695 (forward_propagate_into): Decrement it.
29696 (fwprop_init): Initialize it.
29697 (fw_prop): If the variable has reached zero, stop propagating.
29698 (fwprop_addr): Ditto.
29700 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29703 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29704 a FUNCTION_DECL, pass it as decl instead of origin to
29707 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29709 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29712 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29715 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29716 TARGET_VSX_SMALL_INTEGER.
29718 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29720 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29721 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29723 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29725 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29726 extraction from odd-numbered MSA register.
29728 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29730 PR middle-end/80173
29731 * expmed.c (store_bit_field_1): Don't attempt to create
29732 a word subreg out of hard registers wider than word if they
29733 have HARD_REGNO_NREGS of 1 for their mode.
29735 PR middle-end/80163
29736 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29737 conversions to integer types wider than word and pointer.
29740 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29741 (rtx_equal_for_cselib_p): Pass 0 to it.
29742 * cselib.c (cselib_hasher::equal): Likewise.
29743 (rtx_equal_for_cselib_1): Add depth argument. If depth
29744 is 128, don't look up VALUE locs and punt. Increment
29745 depth in recursive calls when walking VALUE locs.
29747 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29749 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29750 (make_gcov_file_name): Use the canonical path name for generating
29752 (read_line): Fix handling of files with ascii null bytes.
29754 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29756 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29757 to initialise a vector register instead
29758 of using a const_int.
29760 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29762 PR translation/80189
29763 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29764 diagnostic messages.
29766 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29769 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29770 (dfp_diex_<mode>): Update mode of operand 1.
29771 * doc/extend.texi (dxex, dxexq): Document change to return type.
29772 (diex, diexq): Document change to argument type.
29774 2017-03-30 Martin Jambor <mjambor@suse.cz>
29777 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29778 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29779 it reflects the signature changes performed at the callee side.
29780 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29781 to cgraph_build_function_type_skip_args.
29782 (build_function_decl_skip_args): Adjust call to the above function.
29784 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29787 * config/i386/sse.md
29788 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29789 register as dest whenever it is a MEM not rtx_equal_p to the
29790 corresponding dup operand, and when forcing into reg move the
29791 reg into the memory afterwards.
29792 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29793 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29794 for the force_reg mode.
29795 (avx512vl_vextractf128<mode>): Use register as dest either
29796 always when a MEM, or when it is a MEM not rtx_equal_p to the
29797 corresponding dup operand, or even not when it is a CONST_VECTOR
29798 depending on the mode and lo vs. hi.
29799 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29801 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29802 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29803 Likewise. Require that operands[2] is even.
29804 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29805 Remove extraneous parens. Require that operands[2] is a multiple
29807 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29808 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29809 disallow memory then.
29811 2017-03-30 Richard Biener <rguenther@suse.de>
29813 PR tree-optimization/77498
29814 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29815 to non-constants over backedges.
29817 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29819 PR rtl-optimization/80233
29820 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29821 as last_combined_insn. Do not test for BARRIER_P separately.
29823 2017-03-29 Andreas Schwab <schwab@suse.de>
29826 * calls.c (prepare_call_address): Convert funexp to Pmode before
29827 copying to temp reg.
29829 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29831 PR tree-optimization/80158
29832 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29833 Handle possible future case of more than one alternate
29835 (replace_rhs_if_not_dup): Likewise.
29836 (replace_one_candidate): Likewise.
29838 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29840 PR rtl-optimization/80193
29841 * ira.c (ira): Do not check allocation for LRA.
29843 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29845 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29846 (nvptx_output_simt_exit): Declare.
29847 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29848 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29849 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29850 (nvptx_declare_function_name): ...here. Emit declaration of local
29851 memory space buffer for omp_simt_enter insn.
29852 (nvptx_output_unisimt_switch): New.
29853 (nvptx_output_softstack_switch): New.
29854 (nvptx_output_simt_enter): New.
29855 (nvptx_output_simt_exit): New.
29856 * config/nvptx/nvptx.h (struct machine_function): New fields
29857 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29858 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29859 (UNSPECV_SIMT_EXIT): Ditto.
29860 (omp_simt_enter_insn): New insn.
29861 (omp_simt_enter): New expansion.
29862 (omp_simt_exit): New insn.
29863 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29865 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29866 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29867 (expand_GOMP_SIMT_EXIT): New.
29868 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29869 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29870 (GOMP_SIMT_EXIT): Ditto.
29871 * target-insns.def (omp_simt_enter): New insn.
29872 (omp_simt_exit): Ditto.
29873 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29875 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29876 (lower_rec_input_clauses): Likewise.
29877 (lower_lastprivate_clauses): Handle SIMT privatization.
29879 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29880 (ompdevlow_adjust_simt_enter): New.
29881 (find_simtpriv_var_op): New.
29882 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29883 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29885 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29886 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29887 (copy_decl_for_dup_finish): Ditto.
29889 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29891 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29894 * config/i386/i386.c (ix86_option_override_internal): Always
29895 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29897 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29899 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29901 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29903 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29904 mark new edge's irreducible flag accordign to it.
29905 (vect_do_peeling): Check loop preheader edge's irreducible flag
29906 and pass it to function slpeel_add_loop_guard.
29908 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29910 PR tree-optimization/80218
29911 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29912 Update block frequencies and counts.
29914 2017-03-28 Richard Biener <rguenther@suse.de>
29916 PR tree-optimization/78644
29917 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29918 of a simplification result we may not use it at all.
29920 2017-03-28 Richard Biener <rguenther@suse.de>
29923 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29924 without arguments, generate default definition of a SSA name.
29926 2017-03-28 Richard Biener <rguenther@suse.de>
29928 PR middle-end/80222
29929 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29930 TYPE_REF_CAN_ALIAS_ALL references.
29931 * fold-const.c (fold_indirect_ref_1): Likewise.
29933 2017-03-28 Martin Liska <mliska@suse.cz>
29936 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29937 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29939 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29940 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29942 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29943 (EXTRA_SPECS): Define.
29944 (SUBTARGET_EXTRA_SPECS): Likewise.
29945 (SUBTARGET_CPP_SPEC): Likewise.
29946 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29947 SUBTARGET_EXTRA_SPECS.
29948 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29950 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29952 * config/arc/simdext.md (vst64_insn): Update pattern.
29953 (vld32wh_insn): Likewise.
29954 (vld32wl_insn): Likewise.
29955 (vld64_insn): Likewise.
29956 (vld32_insn): Likewise.
29958 2017-03-28 Marek Polacek <polacek@redhat.com>
29961 * fold-const.c (fold_comparison): Use protected_set_expr_location
29962 instead of SET_EXPR_LOCATION.
29964 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29966 * tree.c (add_expr): Avoid name lookup warning.
29968 2017-03-27 Jeff Law <law@redhat.com>
29970 PR tree-optimization/80216
29971 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29972 function name. Limit recursion depth.
29973 (record_temporary_equivalences): Corresponding changes.
29975 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29977 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29980 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29983 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29985 * cfgcleanup.c (reg_note_cfa_p): New array.
29986 (insns_have_identical_cfa_notes): New function.
29987 (old_insns_match_p): Don't cross-jump in between /f
29988 and non-/f instructions. If both i1 and i2 are frame related,
29989 verify all CFA notes, their order and content.
29991 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29994 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29995 HImode and SImode with zero extend to DImode to one insn.
29996 (bswap<mode>2_extenddi): Likewise.
29997 (bswapsi2_extenddi): Likewise.
29998 (bswaphi2_extendsi): Likewise.
29999 (bswaphi2): Combine bswap HImode and SImode into one insn.
30000 Separate memory insns from swapping register.
30001 (bswapsi2): Likewise.
30002 (bswap<mode>2): Likewise.
30003 (bswaphi2_internal): Delete, no longer used.
30004 (bswapsi2_internal): Likewise.
30005 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30006 store, and gpr<-gpr swap insns.
30007 (bswap<mode>2_store): Likewise.
30008 (bswaphi2_reg): Register only splitter, combine with the splitter.
30009 (bswaphi2 splitter): Likewise.
30010 (bswapsi2_reg): Likewise.
30011 (bswapsi2 splitter): Likewise.
30012 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30013 the insns into load, store, and register/register insns.
30014 (bswapdi2_ldbrx): Likewise.
30015 (bswapdi2_load): Likewise.
30016 (bswapdi2_store): Likewise.
30017 (bswapdi2_reg): Likewise.
30019 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30021 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30022 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30024 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30027 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30030 special handling for target option conflicts between dform
30031 options (-mpower9-dform, -mpower9-dform-vector,
30032 -mpower9-dform-scalar) and -mno-direct-move.
30034 2017-03-27 Richard Biener <rguenther@suse.de>
30036 PR tree-optimization/80181
30037 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30039 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30041 * config/arc/predicates.md (move_double_src_operand): Replace the
30042 call to move_double_src_operand with a call to address_operand.
30044 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30046 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30047 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30048 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30050 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30052 * config/arc/predicates.md (long_immediate_loadstore_operand):
30053 Consider scaled addresses cases.
30055 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30057 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30058 restored when in interrupt.
30059 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30060 doesn't have delay slot.
30062 2017-03-27 Richard Biener <rguenther@suse.de>
30065 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30066 inlined thunk clones.
30068 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30071 * asan.c (instrument_derefs): Copy over last operand from
30072 original COMPONENT_REF to the new COMPONENT_REF with
30073 DECL_BIT_FIELD_REPRESENTATIVE.
30074 * ubsan.c (instrument_object_size): Likewise.
30076 2017-03-27 Richard Biener <rguenther@suse.de>
30078 PR tree-optimization/80170
30079 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30080 sure DR/SCEV didnt fold in constants we do not see when looking
30081 at the reference base alignment.
30083 2017-03-27 Richard Biener <rguenther@suse.de>
30085 PR middle-end/80171
30086 * gimple-fold.c (fold_ctor_reference): Properly guard against
30087 NULL return value from canonicalize_constructor_val.
30089 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30092 * config/i386/i386.c (ix86_expand_builtin)
30093 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30094 flags reg setting and flags reg using instructions.
30095 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30096 clobbering instructions to zero extend op2.
30098 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30100 * doc/install.texi (Configuration) <--with-aix-soname>:
30101 Update link to AIX ld.
30103 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30105 PR rtl-optimization/80160
30106 PR rtl-optimization/80159
30107 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30108 reg_alternate_class into account.
30110 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30113 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30114 to consider in curr_insn_transform.
30116 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30118 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30119 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30120 and emit_mode_inner.
30122 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30124 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30125 argument to the overloaded builtin variants. Use the new flag to
30126 deprecate certain builtin variants.
30127 * config/s390/s390-builtin-types.def: Add new builtin types.
30128 * config/s390/s390-builtins.h: Support new flags field for
30129 overloaded builtins.
30130 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30131 (s390_macro_to_expand): Enable vector float data type.
30132 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30133 builtins by incrementing the __VEC__ version number.
30134 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30136 (s390_resolve_overloaded_builtin): Emit error messages depending
30137 on the builtin flags.
30138 * config/s390/s390.c (s390_expand_builtin): Support additional
30139 flags argument. Change error message to match the messages
30140 emitted in s390-c.c.
30141 * config/s390/s390.md: New UNSPEC_* constants.
30142 (op_type): Add new instruction types.
30143 * config/s390/vecintrin.h: Add new builtins and test data class
30145 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30146 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30147 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30148 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30149 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30150 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30152 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30153 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30154 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30155 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30157 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30158 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30159 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30160 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30162 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30163 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30164 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30165 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30166 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30167 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30168 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30170 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30171 ("vec_scatter_element<V_HW_4:mode>_DI")
30172 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30173 ("vec_fpint<mode>", "vflls")
30174 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30175 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30176 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30177 ("*vec_cmphe<mode>_cc"): ... these.
30179 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30180 mode constant instead of magic value.
30182 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30184 * config/s390/s390.c (s390_expand_vec_compare): Support other
30185 vector floating point modes than just V2DF.
30186 (s390_expand_vcond): Likewise.
30187 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30188 (s390_cannot_change_mode_class): Prevent mode changes between TF
30189 and V1TF in vector registers.
30190 * config/s390/s390.md (DF, SF): New mode attributes.
30191 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30192 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30193 SFmode support for VRs.
30194 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30196 (VFT, VF_HW): New mode iterators.
30197 (vw, sdx): New mode attributes.
30198 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30199 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30200 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30201 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30202 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30203 also the new vector floating point modes. Renaming to ...
30205 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30206 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30207 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30208 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30209 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30210 ("vec_unordered<mode>"): ... these.
30212 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30213 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30214 ("*vec_extendv2df"): New insn definitions.
30216 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30218 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30219 ("mulditi3_2", "*muldi3_sign"): New patterns.
30220 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30221 rename the pattern definition.
30223 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30225 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30227 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30231 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30232 instruction if possible.
30233 * config/s390/vector.md (vec_halfnumelts): New mode
30235 ("*vec_vllezlf<mode>"): New pattern.
30237 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30239 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30240 ("popcountv4si2", "popcountv2di2"): Rename to ...
30241 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30242 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30244 ("popcount<mode>2_vxe"): New pattern.
30246 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30248 * common/config/s390/s390-common.c (processor_flags_table): Add
30250 * config.gcc: Add arch12.
30251 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30252 Default to arch12 for unknown CPU model numbers.
30253 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30254 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30255 PROCESSOR_max sanity check.
30256 * config/s390/s390-opts.h (enum processor_type): Add
30258 * config/s390/s390.c (processor_table): Add arch12.
30259 (s390_expand_builtin): Add check for B_VXE flag.
30260 (s390_issue_rate): Add PROCESSOR_ARCH12.
30261 (s390_get_sched_attrmask): Likewise.
30262 (s390_get_unit_mask): Likewise.
30263 (s390_sched_score): Enable z13 scheduling for arch12.
30264 (s390_sched_reorder): Likewise.
30265 (s390_sched_variable_issue): Likewise.
30266 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30268 (s390_tune_attr): Use z13 scheduling also for arch12.
30269 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30270 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30271 (TARGET_VXE_P): New macros.
30272 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30273 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30274 * config/s390/s390.opt: Add arch12 as processor_type.
30276 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30278 * config/s390/s390.md
30279 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30280 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30281 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30283 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30284 Rename expanders to ...
30286 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30287 ("fixuns_truncdddi2_emu"): ... these.
30289 ("fixuns_trunc<mode>si2_emu"): New expander.
30291 ("*fixuns_truncdfdi2_z13"): Rename to ...
30292 ("*fixuns_truncdfdi2_vx"): ... this.
30294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30296 * config/s390/2964.md: Remove the single element vector compare
30297 instructions which are no longer used.
30298 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30300 (s390_canonicalize_comparison): Remove handling of DFmode
30302 (s390_expand_vec_compare_scalar): Remove function.
30303 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30304 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30306 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30308 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30310 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30312 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30313 will anyway by matched by mov<mode>_64dfp.
30315 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30317 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30318 vlef/vstef. Add missing operand to vleif.
30320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30322 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30323 pair for all vector types with 64 bit elements.
30324 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30325 * config/s390/vector.md (V_HW_64): ... here.
30326 (V_128_NOSINGLE): New mode iterator.
30327 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30328 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30329 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30330 ("*vec_load_pairv2di"): Change to ...
30331 ("*vec_load_pair<mode>"): ... this one.
30333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30335 * config/s390/constraints.md: Add comments.
30336 (jKK): Reject element sizes > 8 bytes.
30337 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30339 * config/s390/s390.md: Add the s_operand checks formerly in
30340 s390_split_ok_p to various splitters where they are still
30342 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30343 for 128 bit vectors. Plus two splitters.
30345 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30347 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30350 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30353 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30354 error if the boundary argument is not constant.
30356 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30358 PR rtl-optimization/80112
30359 * loop-doloop.c (doloop_condition_get): Don't check condition
30360 if cmp isn't SET with IF_THEN_ELSE src.
30362 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30364 PR tree-optimization/80158
30365 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30366 replacing a candidate statement, also replace it for the
30367 candidate's alternate interpretation.
30368 (replace_rhs_if_not_dup): Likewise.
30369 (replace_one_candidate): Likewise.
30371 2017-03-24 Richard Biener <rguenther@suse.de>
30373 PR tree-optimization/80167
30374 * graphite-isl-ast-to-gimple.c
30375 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30377 (translate_isl_ast_to_gimple::get_rename): Likewise.
30379 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30382 handling of certain combinations of target options, including the
30383 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30384 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30386 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30389 * config/arm/arm.md (*load_multiple): Add reload_completed to
30390 matching condition.
30392 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30393 Richard Biener <rguenth@suse.de>
30395 PR tree-optimization/79908
30396 PR tree-optimization/80136
30397 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30398 been cast away, gimplify_and_add suffices.
30400 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30402 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30404 2017-03-23 Richard Biener <rguenther@suse.de>
30406 PR tree-optimization/80032
30407 * gimplify.c (gimple_push_cleanup): Forced unconditional
30408 cleanups still have to go to the conditional_cleanups
30411 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30413 PR tree-optimization/80072
30414 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30416 (next_operand_entry_id): Change type to unsigned int.
30417 (sort_by_operand_rank): Make sure to return the right return value
30418 even if unsigned fields are bigger than INT_MAX.
30419 (struct oecount): Change cnt and id type to unsigned int.
30420 (oecount_hasher::equal): Formatting fix.
30421 (oecount_cmp): Make sure to return the right return value
30422 even if unsigned fields are bigger than INT_MAX.
30423 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30426 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30427 TREE_READONLY on result if writing it more than once.
30430 * doc/invoke.texi (-fsanitize=thread): Document that with
30431 -fnon-call-exceptions atomics are not able to throw
30435 * tsan.c: Include tree-eh.h.
30436 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30437 maybe_clean_or_replace_eh_stmt where needed.
30438 (instrument_memory_accesses): Add cfg_changed argument.
30439 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30440 if it returned true.
30441 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30443 PR rtl-optimization/63191
30444 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30445 wrapper function, moved the whole old content into ...
30446 (ix86_delegitimize_address_1): ... this. New inline function.
30447 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30448 true as last argument instead of ix86_delegitimize_address.
30450 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30452 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30453 cortexa57_branch_cost.
30455 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30457 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30459 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30462 * doc/md.texi (Constraints): Document wA constraint.
30463 * config/rs6000/constraints.md (wA): New.
30464 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30465 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30466 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30467 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30469 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30472 * gimplify.c (is_oacc_declared): New function.
30473 (oacc_default_clause): Use it to set default flags for acc declared
30474 variables inside parallel regions.
30475 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30476 declared variables.
30477 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30478 declare attribute to any decl as necessary.
30480 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30483 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30484 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30485 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30486 (arm_arch_lpae): This.
30487 * config/arm/arm.c (arm_arch7ve): Rename into ...
30488 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30489 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30492 2017-03-22 Martin Liska <mliska@suse.cz>
30495 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30496 error message instead of an ICE.
30498 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30500 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30502 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30504 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30509 2017-03-21 Martin Sebor <msebor@redhat.com>
30511 * doc/extend.texi: Use "cannot" instead of "can't."
30512 * doc/hostconfig.texi: Same.
30513 * doc/install.texi: Same.
30514 * doc/invoke.texi: Same.
30515 * doc/loop.texi: Same.
30516 * doc/md.texi: Same.
30517 * doc/objc.texi: Same.
30518 * doc/rtl.texi: Same.
30519 * doc/tm.texi: Same.
30520 * doc/tm.texi.in: Same.
30521 * doc/trouble.texi: Same.
30523 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30526 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30527 (collect_checksum_attributes): Set it.
30528 (die_checksum_ordered): Use it.
30530 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30532 PR tree-optimization/79908
30533 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30534 change: For a VA_ARG whose LHS has been cast away, use
30535 force_gimple_operand to construct the side effects.
30537 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30539 PR translation/80001
30540 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30541 more amenable to translation.
30542 (oacc_loop_auto_partitions): Likewise.
30544 2017-03-21 Marek Polacek <polacek@redhat.com>
30545 Martin Sebor <msebor@redhat.com>
30547 PR tree-optimization/80109
30548 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30549 on INTEGRAL_TYPE_P.
30551 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30552 Segher Boessenkool <segher@kernel.crashing.org>
30555 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30556 check reg_used_between_p between insn and one of succ or succ2
30557 depending on if succ is artificial insn not inserted into insn
30560 2017-03-21 Martin Liska <mliska@suse.cz>
30562 PR gcov-profile/80081
30563 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30564 * doc/gcc.texi: Include gcov-dump stuff.
30565 * doc/gcov-dump.texi: New file.
30567 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30569 PR rtl-optimization/79150
30570 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30571 conditional jump, if the jump is the last insn of the loop.
30573 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30574 Richard Biener <rguenth@suse.de>
30576 PR tree-optimization/79908
30577 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30578 been cast away, use force_gimple_operand to construct the side
30581 2017-03-21 Martin Liska <mliska@suse.cz>
30583 PR libfortran/79956
30584 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30587 2017-03-21 Brad Spengler <spender@grsecurity.net>
30590 * plugin.c (htab_hash_plugin): New function.
30591 (add_new_plugin): Use it and adjust.
30592 (parse_plugin_arg_opt): Adjust.
30593 (init_one_plugin): Likewise.
30595 2017-03-21 Richard Biener <rguenther@suse.de>
30597 PR tree-optimization/80032
30598 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30599 if set force the cleanup to happen unconditionally.
30600 (gimplify_target_expr): Push inserted clobbers with force_uncond
30601 to avoid them being removed by control-dependent DCE.
30603 2017-03-21 Richard Biener <rguenther@suse.de>
30605 PR tree-optimization/80122
30606 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30607 va_arg_pack_len when the inlined call stmt requires pack
30609 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30611 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30614 * tsan.c (instrument_builtin_call): If the memory model argument
30615 is not a constant, assume it is valid.
30618 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30621 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30623 PR rtl-optimization/79910
30624 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30625 if its dest is used by an insn before I2 (other than the combined
30626 insns themselves, which are properly handled already).
30628 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30631 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30633 * combine.c (record_used_regs): New static function.
30634 (try_combine): Handle situations where there is an additional
30635 instruction between I2 and I3 which needs to have a LOG_LINK
30639 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30641 * combine.c (try_combine): Delete redundant i1 test. Call
30642 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30644 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30647 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30648 alternatives 13/14.
30650 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30652 PR tree-optimization/80054
30653 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30654 the optimization if a PHI or any of its arguments is not dominated
30655 by the candidate's basis. Use gphi* rather than gimple* as
30657 (replace_profitable_candidates): Clean up a gimple* variable that
30658 should be a gphi* variable.
30660 2017-03-20 Martin Sebor <msebor@redhat.com>
30663 * doc/extend.texi (attribute constructor): Document present limitation.
30665 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30668 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30669 __POWER9_VECTOR__ #ifdef control, change template definition to
30670 use Power9-specific built-in function.
30671 (vec_any_eq): Likewise.
30672 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30673 to control outcomes from this test.
30674 (vector_ae_<mode>p): For VEC_F modes, likewise.
30676 2017-03-20 Ian Lance Taylor <iant@google.com>
30678 * config/i386/i386.c (ix86_function_regparm): Save an extra
30679 register for -fsplit-stack with DECL_STATIC_CHAIN.
30681 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30684 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30685 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30687 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30689 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30691 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30694 2017-03-20 Marek Polacek <polacek@redhat.com>
30697 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30699 2017-03-20 Richard Biener <rguenther@suse.de>
30701 PR tree-optimization/80113
30702 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30703 allocate extra SSA name for PHI def.
30704 (add_close_phis_to_outer_loops): Likewise.
30705 (add_close_phis_to_merge_points): Likewise.
30706 (copy_loop_close_phi_args): Likewise.
30707 (copy_cond_phi_nodes): Likewise.
30709 2017-03-20 Martin Liska <mliska@suse.cz>
30711 PR middle-end/79753
30712 * tree-chkp.c (chkp_build_returned_bound): Do not build
30713 returned bounds for a LHS that's not a BOUNDED_P type.
30715 2017-03-20 Martin Liska <mliska@suse.cz>
30719 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30720 COMPLEX_CST and VECTOR_CST.
30722 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30725 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30726 target operand. A new splitter adds the clobber statement in case
30727 the target operand is dead anyway.
30729 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30731 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30732 to age-old versions of binutils and glibc.
30734 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30736 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30738 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30740 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30742 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30744 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30745 requirement for binutils 2.13.
30747 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30749 * combine.c (try_combine): Delete redundant i1 test. Call
30750 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30752 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30754 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30755 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30757 <riscv64-*-elf>: Re-arrange section
30758 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30759 <riscv32-*-linux>: Likewise.
30760 <riscv64-*-elf>: Likewise
30761 <riscv64-*-linux>: Likewise.
30763 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30766 * aarch64.opt(verbose-cost-dump): Fix typo.
30768 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30771 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30772 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30774 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30776 * reload.c (find_reloads): When reloading a nonoffsettable address,
30777 use RELOAD_OTHER for it and its address reloads.
30779 PR rtl-optimization/79910
30780 * combine.c (record_used_regs): New static function.
30781 (try_combine): Handle situations where there is an additional
30782 instruction between I2 and I3 which needs to have a LOG_LINK
30785 2017-03-17 Jeff Law <law@redhat.com>
30787 PR tree-optimization/71437
30788 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30789 conditional in the hash table first.
30790 (vrp_dom_walker::before_dom_children): Extract condition from
30791 ASSERT_EXPR. Record condition, its inverion and any implied
30792 conditions as well.
30794 2017-03-17 Marek Polacek <polacek@redhat.com>
30795 Markus Trippelsdorf <markus@trippelsdorf.de>
30797 PR tree-optimization/80079
30798 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30801 2017-03-17 Richard Biener <rguenther@suse.de>
30803 PR middle-end/80075
30804 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30805 Properly verify the LHS before the RHS possibly claims to be
30807 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30810 2017-03-17 Martin Jambor <mjambor@suse.cz>
30812 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30813 (List of -O2 options): Likewise.
30814 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30817 2017-03-17 Tom de Vries <tom@codesourcery.com>
30819 * gcov-dump.c (print_usage): Print bug_report_url.
30821 2017-03-17 Richard Biener <rguenther@suse.de>
30823 PR middle-end/80050
30824 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30825 (parser::peek): Likewise.
30827 2017-03-17 Richard Biener <rguenther@suse.de>
30829 PR tree-optimization/80048
30830 * sese.c (free_sese_info): Properly release rename_map and
30831 copied_bb_map elements.
30833 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30835 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30836 Add linked-list forward and backlinks. Insert on
30837 construction, remove on destruction.
30838 (class pass_store_merging): Add m_stores_head field.
30839 (pass_store_merging::terminate_and_process_all_chains):
30840 Iterate over m_stores_head list.
30841 (pass_store_merging::terminate_all_aliasing_chains):
30843 (pass_store_merging::execute): Check for debug stmts first.
30844 Push new chains onto the m_stores_head stack.
30846 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30849 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30850 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30851 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30853 2017-03-16 Jeff Law <law@redhat.com>
30855 PR tree-optimization/71437
30856 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30857 member function. Implementation moved into after_dom_children
30858 member function and into the threader's thread_outgoing_edges
30860 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30861 some code into new thread_outgoing_edges.
30862 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30863 definition. Simplify marker handling (do it here). Assume we always
30864 have the available expression and the const/copies tables.
30865 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30867 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30868 * tree-vrp.c (equiv_stack): No longer file scoped.
30869 (vrp_dom_walker): New class.
30870 (vrp_dom_walker::before_dom_children): New member function.
30871 (vrp_dom_walker::after_dom_children): Likewise.
30872 (identify_jump_threads): Setup domwalker. Use it rather than
30873 walking edges in a random order by hand. Simplify setup/finalization.
30874 (finalize_jump_threads): Remove.
30875 (vrp_finalize): Do not call identify_jump_threads here.
30876 (execute_vrp): Do it here instead and call thread_through_all_blocks
30879 PR tree-optimization/71437
30880 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30882 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30884 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30885 (dom_opt_dom_walker::thread_across_edge): Remove
30886 handle_dominating_asserts argument. All callers changed.
30887 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30888 changes. Remove calls to lhs_of_dominating_assert. Other
30889 uses of handle_dominating_asserts turn into unconditional code
30890 (simplify_control_stmt_condition_1): Likewise.
30891 (simplify_control_stmt_condition): Likewise.
30892 (thread_through_normal_block, thread_across_edge): Likewise.
30893 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30894 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30895 object if it is not an SSA_NAME.
30896 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30897 before calling into the VRP specific simplifiers.
30898 (identify_jump_threads): Remove handle_dominating_asserts
30901 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30904 * tree-diagnostic.c (default_tree_printer): No longer static.
30905 * tree-diagnostic.h (default_tree_printer): New prototype.
30907 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30909 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30910 Change ins into fmov.
30912 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30914 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30915 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30916 Use h_con constraint for operand 1.
30917 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30918 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30920 2017-03-15 Jeff Law <law@redhat.com>
30922 PR tree-optimization/71437
30923 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30924 (record_temporary_equivalences): Use it.
30926 PR tree-optimization/71437
30927 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30928 tree-ssa-scopedtables.
30929 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30930 (record_conditions, record_cond, vuse_eq): Likewise.
30931 (record_edge_info): Adjust to API tweak of record_conditions.
30932 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30933 (record_temporary_equivalences, optimize_stmt): Likewise.
30934 (eliminate_redundant_computations): Likewise.
30935 (record_equivalences_from_stmt): Likewise.
30936 * tree-ssa-scopedtables.c: Include options.h and params.h.
30937 (vuse_eq): New function, moved from tree-ssa-dom.c
30938 (build_and_record_new_cond): Likewise.
30939 (record_conditions): Likewise. Accept vector of conditions rather
30940 than edge_equivalence structure for first argument.
30941 for the first argument.
30942 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30943 from tree-ssa-dom.c.
30944 (avail_exprs_stack::record_cond): Likewise.
30945 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30946 from tree-ssa-dom.c.
30947 (avail_exprs_stack): Add new member functions lookup_avail_expr
30949 (record_conditions): Declare.
30951 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30954 * lra-constraints.c (process_alt_operands): Increase reject for
30955 reloading an input/output operand.
30957 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30960 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30961 insns to convert from signed/unsigned char/short to IEEE 128-bit
30963 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30965 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30968 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30969 subreg of inner mode for values already in registers.
30971 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30973 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30974 iteration reg is used after the loop.
30976 2017-03-14 Martin Sebor <msebor@redhat.com>
30978 PR tree-optimization/79800
30979 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30980 precision in negative-positive range.
30981 (format_floating): Call non-const overload with adjusted precision.
30983 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30986 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30989 2017-03-14 Martin Sebor <msebor@redhat.com>
30991 PR middle-end/80020
30992 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30993 * builtins.def (aligned_alloc): Use it.
30996 * Makefile.in (GTFILES): Add calls.c.
30997 * calls.c: Include "gt-calls.h".
30999 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31001 PR rtl-optimization/79728
31002 * regs.h (struct target_regs): New field
31003 x_contains_allocatable_regs_of_mode.
31004 (contains_allocatable_regs_of_mode): New macro.
31005 * reginfo.c (init_reg_sets_1): Initialize it, and change
31006 contains_reg_of_mode so it includes global regs as well.
31007 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31008 rather than contains_regs_of_mode.
31010 2017-03-14 Martin Liska <mliska@suse.cz>
31012 * doc/invoke.texi: Document options that can't be combined with
31013 -fcheck-pointer-bounds.
31015 2017-03-14 Martin Liska <mliska@suse.cz>
31017 PR middle-end/79831
31018 * doc/invoke.texi (-Wchkp): Document the option.
31020 2017-03-14 Martin Liska <mliska@suse.cz>
31022 * Makefile.in: Install gcov-dump.
31024 2017-03-14 Martin Liska <mliska@suse.cz>
31026 * multiple_target.c (expand_target_clones): Bail out for
31027 an invalid attribute.
31029 2017-03-14 Richard Biener <rguenther@suse.de>
31031 * alias.c (struct alias_set_entry): Pack properly.
31032 * cfgloop.h (struct loop): Likewise.
31033 * cse.c (struct set): Likewise.
31034 * ipa-utils.c (struct searchc_env): Likewise.
31035 * loop-invariant.c (struct invariant): Likewise.
31036 * lra-remat.c (struct cand): Likewise.
31037 * recog.c (struct change_t): Likewise.
31038 * rtl.h (struct address_info): Likewise.
31039 * symbol-summary.h (function_summary): Likewise.
31040 * tree-loop-distribution.c (struct partition): Likewise.
31041 * tree-object-size.c (struct object_size_info): Likewise.
31042 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31043 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31044 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31045 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31046 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31047 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31048 (struct _stmt_vec_info): Likewise.
31050 2017-03-14 Martin Liska <mliska@suse.cz>
31053 * multiple_target.c (create_dispatcher_calls): Check that
31054 a target can create a function dispatcher.
31056 2017-03-14 Martin Liska <mliska@suse.cz>
31059 * multiple_target.c (expand_target_clones): Drop local.local
31060 flag for default implementation.
31062 2017-03-14 Richard Biener <rguenther@suse.de>
31064 PR tree-optimization/80030
31065 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31067 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31069 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31070 gcc_fallthrough() instead of __attribute__((fallthrough));
31072 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31074 * doc/gcc.texi: Remove "up" link to (DIR).
31075 * doc/gccint.texi: Ditto.
31077 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31079 * doc/install.texi (Specific) <avr>: Remove reference to
31082 2017-03-13 Jeff Law <law@redhat.com>
31084 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31085 attribute rather than comments.
31087 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31088 match_scratch operand is highest.
31090 2017-03-13 Martin Liska <mliska@suse.cz>
31092 PR middle-end/78339
31093 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31094 is a CHKP clone, use original declaration.
31096 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31098 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31099 (arc_conditional_register_usage): Use a different allocation order
31100 when optimizing for size.
31101 * common/config/arc/arc-common.c (arc_option_optimization_table):
31102 Section anchors default on when optimizing for size.
31104 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31106 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31108 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31110 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31111 * config/arc/arc.md (cpu_facility): Add cd variant.
31112 (*movqi_insn): Add code density variant.
31113 (*movhi_insn): Likewise.
31114 (*movqi_insn): Likewise.
31115 (*addsi3_mixed): Likewise.
31116 (subsi3_insn): Likewise.
31118 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31120 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31122 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31124 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31125 expressions with MINUS and UNARY ops.
31127 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31130 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31132 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31133 between vec_select and vector argument.
31134 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31135 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31136 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31137 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31138 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31139 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31141 2017-03-13 Richard Biener <rguenther@suse.de>
31144 * params.def (vect-max-peeling-for-alignment): Fix typo.
31146 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31148 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31149 issue that only occurred with binutils below 2.18.
31151 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31153 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31154 refer to binutils 2.11/2.12 minimum.
31156 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31158 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31159 ftp.kernel.org and simplify binutils requirement.
31161 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31163 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31165 (Optimize Options): Ditto. Also remove redundancy.
31167 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31169 PR translation/79848
31170 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31172 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31173 to G_ to avoid double translation.
31175 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31177 PR translation/79923
31178 * auto-profile.c (get_combined_location): Convert leading
31179 character of diagnostics to lower case and remove trailing period.
31180 (read_profile): Likewise for various diagnostics.
31181 * config/arm/arm.c (arm_option_override): Remove trailing period
31182 from various diagnostics.
31183 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31184 (msp430_expand_delay_cycles): Likewise.
31186 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31189 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31190 full command-line argument, rather than just "str".
31191 (aarch64_validate_march): Likewise.
31192 (aarch64_validate_mtune): Likewise.
31194 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31196 PR rtl-optimization/78911
31197 * lra-assigns.c (must_not_spill_p): New function.
31198 (spill_for): Use it.
31200 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31202 PR tree-optimization/79981
31203 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31204 ATOMIC_COMPARE_EXCHANGE ifn result.
31205 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31206 IFN_ATOMIC_COMPARE_EXCHANGE.
31208 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31211 * opts.c (parse_sanitizer_options): Add missing question mark to
31212 "did you mean" message.
31214 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31216 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31218 (VMULEUH_UNS): Likewise.
31219 (VMULOUB_UNS): Likewise.
31220 (VMULOUH_UNS): Likewise.
31221 * config/rs6000/rs6000.c (builtin_function_type): Remove
31222 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31224 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31227 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31228 x with result of extra_parsing_for_operand_code_0.
31229 (function_reader::extra_parsing_for_operand_code_0): Convert
31230 return type from void to rtx, returning x. When reading
31231 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31232 larger size containing struct block_symbol.
31234 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31236 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31237 -mfloat128-hardware without -m64.
31239 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31242 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31243 entries to the case statement that marks unsigned arguments to
31244 overloaded functions.
31246 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31249 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31251 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31254 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31255 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31257 2017-03-10 Martin Liska <mliska@suse.cz>
31261 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31262 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31265 2017-03-10 Olivier Hainque <hainque@adacore.com>
31267 * tree-switch-conversion (array_value_type): Start by resetting
31268 candidate type to it's main variant.
31270 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31272 PR rtl-optimization/79909
31273 * combine.c (try_combine): Use simplify_replace_rtx on individual
31274 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31275 of the whole CALL_INSN_FUNCTION_USAGE.
31277 PR tree-optimization/79972
31278 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31279 get_range_info on SSA_NAMEs. Formatting fixes.
31281 2017-03-10 Richard Biener <rguenther@suse.de>
31282 Jakub Jelinek <jakub@redhat.com>
31284 PR tree-optimization/77975
31285 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31286 edge to be constant.
31287 (get_val_for): For constant x return it. Formatting fix.
31288 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31289 would use the same bases as the current one.
31291 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31293 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31294 instead of vec_select for V1TImode.
31295 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31297 (VSX_LE_128): Add V1TI to this mode iterator.
31298 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31299 (*vsx_le_perm_store_<mode>): Likewise.
31300 (pre-reload splitter for VSX stores): Likewise.
31301 (post-reload splitter for VSX stores): Likewise.
31302 (*vsx_xxpermdi2_le_<mode>): Likewise.
31303 (*vsx_lxvd2x2_le_<mode>): Likewise.
31304 (*vsx_stxvd2x2_le_<mode>): Likewise.
31306 2017-03-09 Michael Eager <eager@eagercon.com>
31308 Correct failures with --enable-checking=yes,rtl.
31310 * config/microblaze/microblaze.c (microblaze_expand_shift):
31311 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31312 test for const0_rtx.
31313 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31314 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31316 2017-03-09 Richard Biener <rguenther@suse.de>
31318 PR tree-optimization/79977
31319 * graphite-scop-detection.c (scop_detection::merge_sese):
31320 Handle the case of extra exits to blocks dominating the entry.
31322 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31324 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31327 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31329 PR rtl-optimization/79949
31330 * lra-constraints.c (process_alt_operands): Check memory when
31331 trying to predict a cycle. Print about the overall increase.
31333 2017-03-09 Richard Biener <rguenther@suse.de>
31335 PR middle-end/79971
31336 * gimple-expr.c (useless_type_conversion_p): Preserve
31337 TYPE_SATURATING for fixed-point types.
31339 2017-03-09 Richard Biener <rguenther@suse.de>
31342 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31343 alignment of BLKmode params.
31345 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31348 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31349 (VALL_NO_V2Q): Likewise.
31351 * config/aarch64/aarch64-simd.md
31352 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31354 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31355 VALL_NO_V2Q mode iterator.
31356 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31358 2017-03-09 Martin Liska <mliska@suse.cz>
31360 PR tree-optimization/79631
31361 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31362 tree_int_cst_sign_bit just for INTEGER constants.
31364 2017-03-09 Martin Liska <mliska@suse.cz>
31368 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31371 2017-03-09 Marek Polacek <polacek@redhat.com>
31374 * tree.c (inchash::add_expr): Handle TREE_VEC.
31376 2017-03-09 Martin Liska <mliska@suse.cz>
31379 (chkp_narrow_size_and_offset): New function.
31380 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31381 (void chkp_parse_bit_field_ref): New function.
31382 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31383 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31385 2017-03-09 Martin Liska <mliska@suse.cz>
31388 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31389 (chkp_find_bounds_1): Remove gcc_unreachable.
31391 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31394 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31395 BUILT_IN_SYNC*, determine the access type from the size suffix and
31396 always build a MEM_REF with that type. Handle forgotten
31397 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31400 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31401 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31402 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31403 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31404 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31405 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31406 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31407 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31408 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31409 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31410 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31411 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31412 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31413 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31414 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31415 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31416 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31417 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31418 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31419 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31420 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31421 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31422 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31423 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31424 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31425 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31426 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31427 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31428 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31429 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31430 definitions outside of __OPTIMIZE__ guarded section.
31433 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31434 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31435 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31436 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31439 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31441 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31442 ("vfenez<mode>"): Add missing constraints.
31444 2017-03-08 Martin Sebor <msebor@redhat.com>
31447 * config/nds32/nds32.c (nds32_option_override):
31448 Fix misspelled diagnostic.
31450 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31453 * gimplify.c (gimplify_omp_for): Replace index var in outer
31454 taskloop statement with an artificial variable and add
31455 OMP_CLAUSE_PRIVATE clause for it.
31457 2017-03-08 Richard Biener <rguenther@suse.de>
31459 PR tree-optimization/79955
31460 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31461 for accesses that are completely outside of the variable.
31463 2017-03-08 Andrew Haley <aph@redhat.com>
31465 PR tree-optimization/79943
31466 * tree-ssa-loop-split.c (compute_new_first_bound): When
31467 calculating the new upper bound, (END-BEG) should be added, not
31470 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31472 * config/avr/avr.md (setmemhi): Make sure match_dup
31473 operand number comes before match_scratch.
31475 2017-03-08 Richard Biener <rguenther@suse.de>
31477 PR tree-optimization/79920
31478 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31479 with ncopies == 1 to ...
31480 (vect_transform_slp_perm_load): ... here. Properly compute
31481 all element loads by iterating VF times over the group. Do
31482 not handle ncopies (computed in a broken way) in
31483 vect_create_mask_and_perm.
31485 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31488 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31489 is a uniform vector, use uniform_vector_p return value instead of
31490 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31492 2017-03-07 Marek Polacek <polacek@redhat.com>
31494 PR middle-end/79809
31495 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31496 (alloca_call_type): Likewise.
31498 2017-03-07 Martin Liska <mliska@suse.cz>
31500 * gcov.c (process_args): Put comment to correct location.
31502 2017-03-07 Martin Liska <mliska@suse.cz>
31504 PR middle-end/68270
31505 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31506 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31507 (chkp_narrow_bounds_for_field): Likewise.
31508 (chkp_parse_array_and_component_ref): Pass one more argument to
31511 2017-03-07 Richard Biener <rguenther@suse.de>
31513 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31516 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31518 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31519 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31521 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31524 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31525 to end of description.
31526 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31528 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31530 PR rtl-optimization/79901
31531 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31533 (*avx512f_<code><mode>3<mask_name>): ... this.
31534 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31535 iterator instead of VI8_AVX2_AVX512BW.
31537 PR rtl-optimization/79901
31538 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31539 min/max expander, expand it using expand_vec_cond_expr.
31542 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31545 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31548 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31549 to void * for PCH reasons.
31550 * dwarf2out.c (output_loc_operands, output_die): Cast
31551 v.val_vec.array to unsigned char *.
31553 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31556 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31559 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31561 PR rtl-optimization/79571
31562 * lra-constraints.c (process_alt_operands): Calculate static
31563 reject and subtract it from overall when only addresses will be
31566 2017-03-06 Julia Koval <julia.koval@intel.com>
31569 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31570 incoming stack boundary to 128 for 64-bit targets.
31572 2017-03-06 Richard Biener <rguenther@suse.de>
31574 PR tree-optimization/79894
31575 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31576 to NULL after folding it.
31578 2017-03-06 Richard Biener <rguenther@suse.de>
31580 PR tree-optimization/79824
31581 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31582 check disabling peeling for gaps.
31584 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31586 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31587 attributes): Document gettimeofday.
31589 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31591 * config/s390/s390.c (s390_option_override_internal): Set
31592 PARAM_MIN_VECT_LOOP_BOUND
31594 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31596 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31597 * config/s390/s390.md: Likewise.
31599 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31602 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31603 (<avx2_avx512>_perm<mode>): Rename to ...
31604 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31606 (<avx512>_perm<mode>_mask): Rename to ...
31607 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31609 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31610 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31611 instead of VI8F_256_512.
31612 (avx512f_perm<mode>): New define_expand.
31613 (avx512f_perm<mode>_mask): Likewise.
31614 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31615 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31617 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31619 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31620 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31622 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31624 2017-03-06 Martin Liska <mliska@suse.cz>
31627 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31628 when having a SSA NAME w/o VAR_DECL assigned to it.
31630 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31632 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31633 msa_dpsub_<su>_d): Fix MODE for vec_select.
31635 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31637 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31639 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31641 2017-03-06 Richard Biener <rguenther@suse.de>
31643 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31644 * plugin.c (register_plugin_info): Likewise.
31645 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31647 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31649 * config/i386/sse.md (sse_storehps, sse_storelps,
31650 avx_<castmode><avxsizesuffix>_<castmode>,
31651 avx512f_<castmode><avxsizesuffix>_<castmode>,
31652 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31653 in condition that at least one operand is not a MEM.
31655 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31657 PR middle-end/79805
31658 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31659 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31661 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31662 gimple_call_nothrow_p flag based on whether original builtin can throw.
31663 If it can, emit following stmts on the fallthrough edge.
31664 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31665 don't create new bb if inserting just debug stmts on the edge, try to
31666 insert them on the fallthru bb or just reset debug stmts.
31668 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31671 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31672 restore recog_data (including the operand rtxes inside it) around
31673 the call to get_insn_template.
31675 2017-03-03 Martin Sebor <msebor@redhat.com>
31677 PR tree-optimization/79699
31678 * context.c (context::~context): Free MPFR caches to avoid
31679 a memory leak on program exit.
31681 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31683 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31684 Use wide_int::ulow () instead of .elt (0).
31686 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31688 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31689 (*pushxf): Limit oF constraint to 32bit targets and add oC
31690 constraint for 64bit targets.
31691 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31692 (*pushdf): Change rmF constraint to rmC.
31694 2017-03-03 Martin Liska <mliska@suse.cz>
31696 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31697 Remove unused variable.
31699 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31702 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31703 is a memory operand, increase num_memory.
31704 (ix86_expand_args_builtin): Likewise.
31706 2017-03-03 Jan Hubicka <jh@suse.cz>
31709 * ipa-devirt.c (maybe_record_node): Properly handle
31710 __cxa_pure_virtual visibility.
31712 2017-03-03 Martin Liska <mliska@suse.cz>
31714 PR tree-optimization/79803
31715 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31717 (pass_loop_prefetch::execute): Disabled optimization if an
31718 assumption about L1 cache size is not met.
31720 2017-03-03 Martin Liska <mliska@suse.cz>
31722 PR rtl-optimization/79574
31723 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31724 (hash_scan_set): Likewise.
31725 (dump_hash_table): Likewise.
31726 (hoist_code): Likewise.
31728 2017-03-03 Richard Biener <rguenther@suse.de>
31730 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31731 in place of uhigh ().
31732 (fixed_convert_from_real): Likewise.
31734 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31737 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31739 2017-03-03 Richard Biener <rguenther@suse.de>
31741 PR middle-end/79818
31742 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31743 TYPE_OVERFLOW_UNDEFINED check.
31745 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31747 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31749 (vector_ae_<mode>_p): Likewise.
31750 (vector_nez_<mode>_p): Likewise.
31751 (vector_ne_v2di_p): Likewise.
31752 (vector_ae_v2di_p): Likewise.
31753 (vector_ne_<mode>_p): Likewise.
31754 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31756 (vsx_tsqrt<mode>2_fe): Likewise.
31758 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31761 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31763 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31765 PR rtl-optimization/79780
31766 * cprop.c (one_cprop_pass): When second and further conditional trap
31767 in a single basic block is turned into an unconditional trap, turn it
31768 into a deleted note to avoid RTL verification failures.
31770 2017-03-02 Richard Biener <rguenther@suse.de>
31772 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31774 2017-03-02 Richard Biener <rguenther@suse.de>
31776 PR tree-optimization/79345
31778 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31779 param and abort the walk, returning -1 if it is hit.
31780 (walk_aliased_vdefs): Take a limit param and pass it on.
31781 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31782 defaulting to 0 and return a signed int.
31783 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31784 (check_defs): New helper.
31785 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31786 about uninitialized memory.
31787 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31788 bogus uninitialized warning.
31789 (fixed_convert_from_real): Likewise.
31791 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31793 PR tree-optimization/66768
31794 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31795 iv_use if base object can't be determined.
31797 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31799 PR tree-optimization/79345
31800 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31801 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31802 (get_pattern_stats): Initialize it.
31803 * genemit.c (gen_expand): Verify match_scratch numbers come after
31804 match_operand/match_dup numbers.
31805 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31806 match_scratch numbers.
31807 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31809 * config/s390/s390.md (trunctdsd2): Likewise.
31811 2017-03-02 Richard Biener <rguenther@suse.de>
31813 * wide-int.h (wide_int_storage::operator=): Implement in terms
31816 2017-03-02 Richard Biener <rguenther@suse.de>
31818 PR tree-optimization/79777
31819 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31820 the to insert expression to sth existing.
31822 2017-03-01 Martin Sebor <msebor@redhat.com>
31824 PR middle-end/79692
31825 * gimple-ssa-sprintf.c
31826 (directive::known_width_and_precision): New function.
31827 (format_integer): Use it.
31828 (get_mpfr_format_length): Consider the full range of precision
31829 when computing %g output with the # flag. Set the likely byte
31830 count to 3 rather than 1 when precision is indeterminate.
31831 (format_floating): Correct the lower bound of precision.
31833 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31835 * doc/invoke.texi: Document default code model for 64-bit Linux.
31837 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31840 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31841 udiv rather than div since input pattern is unsigned.
31843 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31845 * config/i386/i386.c (print_reg): Warn for values of
31846 unsupported size in integer register.
31848 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31851 * config/rs6000/predicates.md (current_file_function_operand): Do
31852 not allow self calls to be local if the function is replaceable.
31854 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31857 * config/rs6000/altivec.h (vec_ctz and others): Change the
31858 preprocessor macro that controls conditional compilation from
31859 _ARCH_PWR9 to __POWER9_VECTOR__.
31860 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31861 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31862 control (instead of _ARCH_PWR9 control) so that template
31863 definition uses power9-specific function.
31864 (vec_any_eq): Likewise.
31865 (vec_all_ne): Change macro definition to use a power9-specific
31866 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31867 _ARCH_PWR9 control).
31868 (vec_any_eq) Likewise.
31869 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31870 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31871 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31872 support for xvcmpnedp instruction.
31873 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31874 macro expansion so that Power9 implementation of vec_all_ne does
31875 not use the AltiVec predicate framework.
31876 (VCMPNEH_P): Likewise.
31877 (VCMPNEW_P): Likewise.
31878 (VCMPNED_P): Likewise.
31879 (VCMPNEFP_P): Likewise.
31880 (VCMPNEDP_P): Likewise.
31881 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31882 implementation of vec_any_eq to not use AltiVec predicate
31884 (VCMPAEH_P): Likewise.
31885 (VCMPAEW_P): Likewise.
31886 (VCMPAED_P): Likewise.
31887 (VCMPAEFP_P): Likewise.
31888 (VCMPAEDP_P): Likewise.
31889 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31890 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31891 not use the AltiVec predicate framework.
31892 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31893 of vec_any_eq to not use AltiVec predicate framework.
31894 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31895 support for predefined __POWER9_VECTOR__ macro to indicate that
31896 Power9 instruction selection is enabled.
31897 (altivec_overloaded_builtins): Remove extraneous
31898 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31899 function argument types RS6000_BTI_bool_V16QI and
31900 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31901 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31902 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31903 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31904 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31905 Power9 for implementations of vec_cmpne. Change the signature for
31906 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31907 (representing vec_all_ne) to remove the previously described first
31908 argument of type RS6000_BTI_INTSI, as this was an artifact of
31909 reliance on the AltiVec predicate framework, which is no longer
31910 used in the implementation of these functions. Add
31911 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31912 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31913 since, unlike the AltiVec predicate framework implementation, we
31914 do not share function descriptors between vec_alle and vec_anyeq.
31915 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31916 set of modes that receive special treatment even when
31917 TARGET_P9_VECTOR is true. The special treatment emits code that
31918 does not depend on Power9 instructions.
31919 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31920 define_expand to not rely on AltiVec predicate framework.
31921 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31923 (vector_ne_v2di_p): Change this define_expand to not rely on
31924 AltiVec predicate framework.
31925 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31927 (vector_ne_<mode>_p): Change this define_expand to not rely on
31928 AltiVec predicate framework.
31929 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31931 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31932 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31933 define_insn pattern.
31934 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31935 define_insn pattern because the xvcmpne<VSs>. instruction is not
31937 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31938 instruction is not supported.
31940 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31942 * config/nvptx/nvptx.c: Include intl.h.
31944 2017-03-01 Martin Jambor <mjambor@suse.cz>
31947 * ipa-prop.h (ipa_bits): Removed field known.
31948 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31949 to pointers. Adjusted their comments to warn about their sharing.
31950 (ipcp_transformation_summary): Change bits to a vector of pointers.
31951 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31952 (ipa_get_ipa_bits_for_value): Declare.
31953 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31954 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31955 (ipa_bits_hash_table): Likewise.
31956 (ipa_vr_ggc_hash_traits): Likewise.
31957 (ipa_vr_hash_table): Likewise.
31958 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31959 being pointers and vr_known being removed.
31960 (ipa_set_jf_unknown): Likewise.
31961 (ipa_get_ipa_bits_for_value): New function.
31962 (ipa_set_jfunc_bits): Likewise.
31963 (ipa_get_value_range): New overloaded functions.
31964 (ipa_set_jfunc_vr): Likewise.
31965 (ipa_compute_jump_functions_for_edge): Use the above functions to
31966 construct bits and vr parts of jump functions.
31967 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31968 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31970 (ipcp_grow_transformations_if_necessary): Also allocate
31971 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31973 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31974 them. Fix too long lines.
31975 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31976 vr_known being removed.
31977 (ipa_read_jump_function): Use new setter functions to construct bits
31978 and vr parts of jump functions or set them to NULL.
31979 (write_ipcp_transformation_info): Adjust for bits being pointers.
31980 (read_ipcp_transformation_info): Likewise.
31981 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31983 Include gt-ipa-prop.h.
31984 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31986 (ipcp_store_bits_results): Likewise.
31987 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31988 Do not write to existing jump functions but use a temporary instead.
31990 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31993 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31994 attempt to use its first operand as BIT_FIELD_REF base.
31996 2017-03-01 Richard Biener <rguenther@suse.de>
31998 PR middle-end/79721
31999 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32000 interpolating formula in wrapping arithmetic.
32001 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32003 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32005 PR tree-optimization/79734
32006 * tree-vect-generic.c (expand_vector_condition): Optimize
32007 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32008 Handle VEC_COND_EXPR where comparison has different inner width from
32009 type's inner width.
32011 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32013 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32014 markup, and similar issues. Remove @opindex entries for things
32015 that aren't options. Add missing -mmpy-option entries.
32017 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32019 PR tree-optimization/79737
32020 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32021 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32022 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32023 instead of byte_size. Formatting fix.
32024 (shift_bytes_in_array_right): Formatting fix.
32026 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32029 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32030 condition on optimize for the leaf function test.
32032 2017-02-28 Martin Liska <mliska@suse.cz>
32035 * read-rtl-function.c (function_reader::handle_unknown_directive):
32036 Bail out when one uses -flto.
32038 2017-02-28 Martin Liska <mliska@suse.cz>
32040 * common.opt: Replace space with tabular for options of <number>
32042 * config/i386/i386.opt: Show <number> value for
32043 -mlarge-data-threshold.
32044 * opts.c (print_filtered_help): Do not display number in hexadecimal
32047 2017-02-28 Martin Liska <mliska@suse.cz>
32049 * common.opt: Fix --help=option -Q for options which are of
32052 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32054 * config/i386/i386.c (print_reg): Error out for values
32055 of 8-bit size in invalid integer register.
32057 2017-02-28 Martin Sebor <msebor@redhat.com>
32059 PR tree-optimization/79691
32060 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32062 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32065 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32066 gcc_unreachable with output_operand_lossage.
32068 2017-02-28 Richard Biener <rguenther@suse.de>
32070 PR tree-optimization/79740
32071 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32073 (visit_nary_op): Insert the nary into the hashtable if we
32074 pattern-matched sth.
32075 * tree-ssa-pre.c (eliminate_insert): Robustify.
32077 2017-02-28 Richard Biener <rguenther@suse.de>
32079 PR middle-end/79731
32080 * fold-const.c (decode_field_reference): Reject out-of-bound
32083 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32085 * config/i386/i386.c: Include intl.h.
32086 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32087 instead of just cond ? "..." : "...".
32088 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32089 * coverage.c (read_counts_file): Likewise.
32090 * omp-offload.c: Include intl.h.
32091 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32092 of just cond ? "..." : "...".
32093 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32094 of just cond ? "..." : "...".
32096 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32099 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32101 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32102 'tune for' CPU name.
32103 * config/arm/arm-cpu-data.h: Regenerated.
32105 2017-02-28 Richard Biener <rguenther@suse.de>
32107 PR tree-optimization/79732
32108 * tree-inline.c (expand_call_inline): Do not shadow var.
32110 2017-02-28 Richard Biener <rguenther@suse.de>
32112 PR tree-optimization/79723
32113 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32114 address-space properly.
32116 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32118 * doc/optinfo.texi (Optimization groups): Fix option used for
32120 * doc/invoke.texi (-fopt-info): Document "omp".
32121 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32122 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32123 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32124 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32125 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32127 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32129 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32130 OPTGROUP_OMP with "omp".
32132 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32135 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32136 for arithmetic shift of unsigned V2DI.
32138 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32140 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32141 arc/linux.h headers.
32142 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32143 (LINK_SPEC): Likewise.
32144 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32145 (EXTRA_SPECS): Likewise.
32146 (STARTFILE_SPEC): Likewise.
32147 (ENDFILE_SPEC): Likewise.
32148 (LIB_SPEC): Likewise.
32149 (TARGET_SDATA_DEFAULT): Likewise.
32150 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32151 (MULTILIB_DEFAULTS): Likewise.
32152 (DWARF2_UNWIND_INFO): Likewise.
32153 * config/arc/big.h: New file.
32154 * config/arc/elf.h: Likewise.
32155 * config/arc/linux.h: Likewise.
32156 * config/arc/t-uClibc: Remove.
32158 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32160 PR tree-optimization/77536
32161 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32162 (tree_transform_and_unroll_loop): Use above function to compute the
32163 estimated niter of unrolled loop and use it when scaling profile.
32164 Also use count info rather than frequency if it's non-zero.
32165 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32166 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32167 (vect_transform_loop): Call above function.
32169 2017-02-27 Richard Biener <rguenther@suse.de>
32171 PR tree-optimization/45397
32172 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32173 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32174 (visit_nary_op): Add pattern matching for CSEing sign-changed
32175 or truncated operations with wider ones.
32177 2017-02-27 Richard Biener <rguenther@suse.de>
32179 PR tree-optimization/79690
32180 * tree-vect-stmts.c (vectorizable_store): Use vector type
32181 built from the DR with address-space.
32183 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32185 * doc/invoke.texi (Optimize Options): Refine the description
32186 of asan-use-after-return.
32188 2017-02-25 Alan Modra <amodra@gmail.com>
32190 PR rtl-optimization/79584
32191 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32192 base, not ad->base_term, the reg within base. Remove assertion
32193 that ad->base == ad->base_term. Replace gen_int_mode using
32194 bogus mode with const0_rtx.
32196 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32198 PR middle-end/79396
32199 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32200 FMA_EXPR like tcc_binary or tcc_unary.
32202 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32205 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32207 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32208 (output_loc_operands): Handle DW_OP_call_ref and
32209 DW_OP_GNU_variable_value.
32210 (struct variable_value_struct): New type.
32211 (struct variable_value_hasher): Likewise.
32212 (variable_value_hash): New variable.
32213 (string_types): Remove.
32214 (copy_loc_descr): New function.
32215 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32216 (prepend_loc_descr_to_each): New function.
32217 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32218 instead of add_loc_descr_to_each if the first argument is single
32219 location list and the second has multiple.
32220 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32221 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32222 when looking for variable value which doesn't have other location info.
32223 (loc_list_from_tree): Formatting fix.
32224 (gen_array_type_die): Simplify DW_AT_string_length handling.
32225 (adjust_string_types): Remove.
32226 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32227 string_types. Call resolve_variable_values.
32228 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32229 (resolve_addr_in_expr): Likewise. Add A argument.
32230 (copy_deref_exprloc): Remove deref argument. Adjust for the
32231 original expression being DW_OP_GNU_variable_value with optionally
32232 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32233 optionally after it.
32234 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32235 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32236 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32237 (variable_value_hasher::hash, variable_value_hasher::equal): New
32239 (resolve_variable_value_in_expr, resolve_variable_value,
32240 resolve_variable_values, note_variable_value_in_expr,
32241 note_variable_value): New functions.
32242 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32245 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32248 * opts.h (handle_generated_option): Add GENERATED_P argument.
32249 * opts-common.c (handle_option): Adjust function comment.
32250 (handle_generated_option): Add GENERATED_P argument, pass it to
32252 (control_warning_option): Pass false to handle_generated_option
32254 * opts.c (maybe_default_option): Pass true to handle_generated_option
32256 * optc-gen.awk: Likewise.
32258 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32260 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32261 a REG, look at the REG it is a SUBREG of.
32262 (splitter for cmpeqsi_t): Ditto.
32264 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32266 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32267 the special USEs with the pattern of the insn, not the insn itself.
32269 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32272 * doc/invoke.texi: Document -mload-store-pairs.
32274 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32275 Sandra Loosemore <sandra@codesourcery.com>
32277 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32278 argument isn't a CONST_INT.
32279 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32280 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32281 (nios2_validate_compare): Bypass alternate compare logic if *op2
32282 is not a CONST_INT.
32283 (ldstwm_operation_p): Return false if first_base is not a REG or
32284 if first_offset is not a CONST_INT.
32286 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32288 * config/cris/cris.md: Use correct operand in a define_peephole2.
32290 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32292 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32294 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32296 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32297 this_insn if it is an INSN or JUMP_INSN.
32298 (force_offsettable): Look at base, not at addr.
32299 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32300 on things that aren't necessarily CONST_INTs.
32302 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32304 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32305 -mfpmath=sse is the default also for x86-32 targets with SSE2
32306 instruction set when @option{-ffast-math} is enabled
32308 2017-02-24 Jeff Law <law@redhat.com>
32310 PR rtl-optimizatoin/79286
32311 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32314 2017-02-24 Richard Biener <rguenther@suse.de>
32316 PR tree-optimization/79389
32317 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32320 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32322 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32323 function comment to reflect reality.
32324 (loop_exits_before_overflow): Fix typo in function description.
32326 2017-02-24 Richard Biener <rguenther@suse.de>
32328 PR tree-optimization/79389
32329 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32330 properly that a threading opportunity exists. Detect conditional
32331 copy/constant propagation opportunities.
32333 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32335 * config/visium/visium.md (type): Add trap.
32336 (b): New mode attribute.
32337 (*btst): Rename into...
32338 (*btst<mode>): ...this and adjust.
32339 (*cbranchsi4_btst_insn): Rename into...
32340 (*cbranch<mode>4_btst_insn): ...this and adjust.
32341 (trap): New define_insn.
32343 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32345 PR tree-optimization/79389
32346 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32347 (noce_reversed_cond_code): New function.
32348 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32349 reversed_comparison_code. Formatting fix.
32350 (noce_try_store_flag): Test rev_cond != NULL in addition to
32351 reversed_comparison_code.
32352 (noce_try_store_flag_constants): Likewise.
32353 (noce_try_store_flag_mask): Likewise.
32354 (noce_try_addcc): Use rev_cond if non-NULL instead of
32355 reversed_comparison_code.
32356 (noce_try_cmove_arith): Likewise. Formatting fixes.
32357 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32358 (noce_find_if_block): Initialize rev_cond.
32359 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32360 instead of false as last argument never attempt to reverse it
32363 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32365 PR tree-optimization/79663
32366 * tree-predcom.c (combine_chains): Process refs in reverse order
32367 only for ZERO length chains, and add explaining comment.
32369 2017-02-23 Jeff Law <law@redhat.com>
32371 PR tree-optimization/79578
32372 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32373 in call to operand_equal_p.
32375 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32378 * config/i386/cpuid.h: Fix another undefined behavior.
32380 2017-02-23 Richard Biener <rguenther@suse.de>
32382 PR tree-optimization/79683
32383 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32384 vector types for data-refs.
32386 2017-02-23 Martin Liska <mliska@suse.cz>
32388 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32390 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32392 PR middle-end/79665
32393 * internal-fn.c (get_range_pos_neg): Moved to ...
32394 * tree.c (get_range_pos_neg): ... here. No longer static.
32395 * tree.h (get_range_pos_neg): New prototype.
32396 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32397 are known to be in between 0 and signed maximum inclusive, try to
32398 expand both unsigned and signed divmod and use the cheaper one from
32401 2017-02-22 Jeff Law <law@redhat.com>
32403 PR tree-optimization/79578
32404 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32405 to compare base operands.
32407 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32410 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32411 gpc_reg_operand instead of fpr_reg_operand.
32413 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32415 * config/mips/mips.c (mips_return_in_memory): Force FP
32416 vector types to be returned in memory for o32 ABI.
32418 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32420 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32421 instead of DW_TAG_member for static data member declarations and don't
32422 set no_linkage_name for static inline data members.
32423 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32426 2017-02-22 Martin Liska <mliska@suse.cz>
32428 * doc/invoke.texi: Replace inequality signs with square brackets
32431 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32434 * lra-constraints.c (simplify_operand_subreg): Handle
32435 WORD_REGISTER_OPERATIONS targets.
32437 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32440 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32441 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32442 elimination by swapping fld*.
32444 2017-02-22 Richard Biener <rguenther@suse.de>
32446 PR tree-optimization/79673
32447 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32448 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32449 irrelevant address-space qualifiers and avoiding a
32450 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32452 2017-02-22 Richard Biener <rguenther@suse.de>
32454 PR tree-optimization/79666
32455 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32456 to not symbolically negate if that may introduce undefined
32459 2017-02-22 Martin Liska <mliska@suse.cz>
32462 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32463 * data-streamer-out.c (streamer_write_gcov_count_stream):
32465 * value-prof.c (stream_out_histogram_value): Make assert more
32466 precise based on type of counter.
32468 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32471 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32472 Use nonimmediate_operand instead of memory_operand for operand 1.
32473 (float-extend standard_x87sse_constant_load splitter): Ditto.
32475 2017-02-21 Jeff Law <law@redhat.com>
32477 PR tree-optimization/79621
32478 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32479 blocks with edges to themselves.
32481 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32484 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32485 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32486 Use gimple_call_builtin_p.
32489 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32490 on temporarily removed DEBUG_INSNs.
32492 PR tree-optimization/79649
32493 * tree-loop-distribution.c (classify_partition): Give up on
32494 non-generic address space loads/stores.
32496 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32498 * doc/loop.texi (Loop manipulation): Remove nonexistent
32499 tree_ssa_loop_version from the documentation.
32500 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32502 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32505 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32506 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32507 * config/rs6000/rs6000.c: Include except.h.
32508 (rs6000_expand_split_stack_prologue): Call
32509 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32511 2017-02-21 Martin Jambor <mjambor@suse.cz>
32514 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32515 have been analyzed.
32517 2017-02-21 Martin Jambor <mjambor@suse.cz>
32519 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32520 for backward compatibility only.
32521 * doc/invoke.texi (Option Summary): Remove all references to
32522 -fipa-cp-alignment.
32524 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32528 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32530 * lra-constraints.c (curr_insn_transform): Handle
32531 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32533 2017-02-21 Martin Liska <mliska@suse.cz>
32535 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32537 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32539 PR translation/79638
32540 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32542 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32545 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32546 (arm_function_ok_for_sibcall): Return false for an indirect call by
32547 descriptor if all the argument registers are used.
32548 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32549 alignment of the function.
32551 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32553 PR tree-optimization/61441
32554 * simplify-rtx.c (simplify_const_unary_operation): For
32555 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32556 the sNaN unmodified.
32558 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32560 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32561 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32562 instead of SYSTEM_HEADER_DIR.
32564 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32565 Martin Liška <mliska@suse.cz>
32567 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32568 Fix typos and grammar, use active voice, and clarify.
32570 2017-02-20 Marek Polacek <polacek@redhat.com>
32572 PR middle-end/79537
32573 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32576 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32578 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32581 * config/i386/i386.c (ix86_expand_builtin): Handle
32582 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32583 ix86_builtins_isa[fcode].isa as a requirement of those
32584 flags and any other flag in the bitmask.
32585 (ix86_init_mmx_sse_builtins): Use 0 instead of
32586 ~OPTION_MASK_ISA_64BIT as mask.
32587 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32588 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32589 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32590 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32592 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32595 * lra-constraints.c (split_reg): Check requested split mode
32596 is supported by the register.
32598 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32600 * lra-constraints.c (simplify_operand_subreg): Remove early
32603 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32606 * lra-constraints.c (curr_insn_transform): Tighten condition
32607 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32609 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32612 * lra-constraints.c (curr_insn_transform): Handle
32613 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32615 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32618 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32620 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32622 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32625 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32628 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32630 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32633 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32635 * final.c (last_columnnum, override_columnnum): New variables.
32636 (final_start_function): Set last_columnnum, pass it to begin_prologue
32637 hook and pass 0 to dwarf2out_begin_prologue.
32638 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32639 to source_line debug hook.
32640 (notice_source_line): Compute last_columnnum and for debug_column_info
32641 return true on column changes.
32642 * debug.h (struct gcc_debug_hooks): Add column argument to
32643 source_line and begin_prologue hooks.
32644 (debug_nothing_int_charstar_int_bool): Remove prototype.
32645 (debug_nothing_int_int_charstar,
32646 debug_nothing_int_int_charstar_int_bool): New prototypes.
32647 (dwarf2out_begin_prologue): Add column argument.
32648 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32649 begin_prologue hooks.
32650 (debug_nothing_int_charstar_int_bool): Remove.
32651 (debug_nothing_int_int_charstar,
32652 debug_nothing_int_int_charstar_int_bool): New functions.
32653 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32654 through to dwarf2out_source_line.
32655 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32656 (dwarf2out_source_line): Add column argument, emit it if requested.
32657 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32659 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32660 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32661 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32662 through to dwarf2out_begin_prologue.
32663 (vmsdbgout_source_line): Add column argument, pass it through to
32664 dwarf2out_source_line.
32665 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32666 dbxout_source_line caller.
32667 (dbxout_source_line): Add column argument.
32669 * common.opt (gno-column-info, gcolumn-info): New options.
32670 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32671 (check_die): Also test for multiple DW_AT_decl_column attributes.
32672 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32673 DW_AT_decl_column if requested.
32674 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32676 (gen_variable_die): Likewise.
32677 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32678 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32681 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32682 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32683 (ix86_handle_option): Handle OPT_m3dnowa.
32684 * doc/invoke.texi (-m3dnowa): Document.
32685 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32686 -m3dnowa instead of -m3dnow -march=athlon.
32689 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32690 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32692 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32695 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32696 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32697 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32698 generator for vsx_xxpermdi_<mode>_be.
32699 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32700 force big-endian semantics.
32701 (vsx_xxpermdi_<mode>_be): New define_expand with same
32702 implementation as previous version of vsx_xxpermdi_<mode>.
32704 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32706 PR tree-optimization/79327
32707 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32708 variable, its initialization and use.
32710 2017-02-17 Julia Koval <julia.koval@intel.com>
32712 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32713 (OPTION_MASK_ISA_PKU_UNSET): New.
32714 (ix86_handle_option): Handle -mrdpid.
32715 * config/i386/cpuid.h (bit_RDPID): New.
32716 * config/i386/driver-i386.c (host_detect_local_cpu):
32717 Detect RDPID feature.
32718 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32719 * config/i386/i386-c.c (ix86_target_macros_internal):
32721 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32722 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32723 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32724 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32725 * config/i386/i386.md (define_insn "rdpid"): New.
32726 * config/i386/i386.opt Add -mrdpid.
32727 * config/i386/immintrin.h (_rdpid_u32): New.
32729 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32731 PR rtl-optimization/79541
32732 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32733 instead of transforming it into USE.
32735 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32737 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32738 If HONOR_SNANS (SFmode) force the input to a register.
32739 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32740 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32741 an frsp or similar insn.
32743 2017-02-17 Martin Liska <mliska@suse.cz>
32745 PR rtl-optimization/79577
32746 * params.def (selsched-max-sched-times): Increase minimum to 1.
32748 2017-02-17 Martin Liska <mliska@suse.cz>
32750 PR rtl-optimization/79574
32751 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32753 2017-02-17 Martin Liska <mliska@suse.cz>
32755 PR tree-optimization/79529
32756 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32757 ssa_defined_default_def_p to handle cases which are implicitly
32759 * tree-ssa.c (ssa_defined_default_def_p): New function.
32760 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32761 which are implicitly defined.
32762 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32764 2017-02-17 Richard Biener <rguenther@suse.de>
32766 PR middle-end/79576
32767 * params.def (max-ssa-name-query-depth): Limit to 10.
32769 2017-02-17 Richard Biener <rguenther@suse.de>
32771 PR tree-optimization/79552
32772 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32775 2017-02-17 Richard Biener <rguenther@suse.de>
32778 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32780 2017-02-17 Marek Polacek <polacek@redhat.com>
32782 PR middle-end/79536
32783 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32784 (fold_negate_expr): New wrapper.
32786 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32788 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32789 Correct terminology and de-emphasize pre-standard behavior.
32791 2017-02-16 Alan Modra <amodra@gmail.com>
32793 PR rtl-optimization/79286
32794 * ira.c (def_dominates_uses): New function.
32795 (update_equiv_regs): Don't create an equivalence for insns that
32796 may trap where the register def does not dominate the use.
32798 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32800 PR rtl-optimization/78127
32801 * lra.c (lra): Call lra_eliminate before finish the loop after
32804 2017-02-16 Richard Biener <rguenther@suse.de>
32806 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32808 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32809 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32810 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32811 (isl_val_int_from_wi): New function.
32812 (extract_affine_gmp): Rename to ...
32813 (extract_affine_wi): ... this, take a widest_int.
32814 (extract_affine_int): Just wrap extract_affine_wi.
32815 (add_param_constraints): Use isl_val_int_from_wi.
32816 (add_loop_constraints): Likewise, and extract_affine_wi.
32818 2017-02-15 Jeff Law <law@redhat.com>
32820 PR middle-end/79521
32821 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32822 ira_init_register_move_cost_if_necessary.
32824 2017-02-15 Martin Sebor <msebor@redhat.com>
32826 PR middle-end/32003
32827 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32828 removed in a prior commit.
32830 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32832 PR tree-optimization/79347
32833 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32834 counters during peeling.
32836 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32838 * Makefile.in (site.exp): Remove "set ISLVER".
32840 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32843 * real.c (real_from_integer): Call real_convert even for decimal.
32845 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32848 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32850 2017-02-14 Andrew Pinski <apinski@cavium.com>
32852 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32853 cores and change the partno/implementer to be correct.
32854 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32855 the 'B" as the implementer.
32856 * config/aarch64/aarch64-tune.md: Regenerate.
32858 2017-02-14 Carl Love <cel@us.ibm.com>
32860 * config/rs6000/rs6000.c: Add case statement entry to make the
32861 xvcvuxdsp built-in argument unsigned.
32862 * config/rs6000/vsx.md: Fix the source and return operand types so they
32863 match the instruction definitions from the ISA document. Fix typo
32864 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32867 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32870 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32871 member early_clobber_alts.
32872 * lra-lives.c (reg_early_clobber_p): New.
32873 (process_bb_lives): Use it.
32874 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32875 (debug_operand_data): Initialize early_clobber_alts.
32876 (setup_operand_alternative): Set up early_clobber_alts.
32877 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32878 alternatives to new_insn_reg.
32879 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32881 (lra_update_insn_regno_info): Pass the new arg.
32883 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32885 PR middle-end/79505
32886 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32887 (new_oacc_loop_raw): Don't clear already cleared fields.
32890 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32891 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32892 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32894 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32897 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32899 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32902 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32903 the extra instruction to the right place to store 128-bit constant
32906 2017-02-14 Martin Sebor <msebor@redhat.com>
32908 PR middle-end/79448
32909 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32910 warning for strings of unknown length.
32912 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32914 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32916 2017-02-14 Jeff Law <law@redhat.com>
32919 * ira-costs.c (scan_one_insn): Initialize register move costs
32920 for pseudos seen in USE/CLOBBER insns.
32922 PR tree-optimization/79095
32923 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32924 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32925 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32926 if the operands are known to be not equal, then the resulting range
32928 (intersect_ranges): If the new range is ~[0,0] and the old range is
32929 wide, then prefer ~[0,0].
32930 * tree-vrp.c (overflow_comparison_p_1): New function.
32931 (overflow_comparison_p): New function.
32932 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32933 if NAME is used in an overflow test.
32934 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32935 overflow check that can be expressed as an equality test, then adjust
32936 ops to be that equality test.
32938 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32940 * config/s390/s390-builtin-types.def: Remove flags argument.
32941 * config/s390/s390.c (s390_init_builtins): Likewise.
32943 2017-02-14 Martin Liska <mliska@suse.cz>
32945 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32946 vector. Fix trailing white spaces.
32948 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32950 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32953 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32955 PR rtl-optimization/68664
32956 * config/arm/arm.c (arm_sched_can_speculate_insn):
32957 New function. Declare prototype.
32958 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32960 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32962 PR rtl-optimization/68664
32963 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32965 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32967 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32969 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32970 max skip bytes for function, loop and jump.
32972 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32974 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32975 ABS_EXPR for gimple dump.
32977 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32980 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32982 PR tree-optimization/79408
32983 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32984 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32985 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32986 also if rhs1 is INTEGER_CST.
32988 2017-02-14 Richard Biener <rguenther@suse.de>
32990 PR middle-end/79432
32991 * tree-into-ssa.c (insert_phi_nodes): When the function can
32992 have abnormal edges rewrite SSA names with broken use-def
32993 dominance out of SSA and register them for PHI insertion.
32995 2017-02-13 Martin Sebor <msebor@redhat.com>
32997 PR middle-end/79496
32998 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32999 clearing info.nowrite flag when snprintf size argument is a range.
33001 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33003 * cprop.c (cprop_jump): Add missing space in string literal.
33004 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33005 (get_constraint_for_component_ref): Likewise.
33006 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33007 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33008 * lra-constraints.c (process_alt_operands): Likewise.
33009 * ipa-inline.c (inline_small_functions): Likewise.
33010 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33011 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33012 * trans-mem.c (diagnose_tm_1_op): Likewise.
33013 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33014 (grid_parallel_clauses_gridifiable): Likewise.
33016 * config/nvptx/mkoffload.c (process): Add space in between
33019 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33020 "MOD4_SSE_REGS" and "ALL_REGS".
33022 * spellcheck.c (test_data): Add , in between "foo" and "food".
33024 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33027 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33028 boundary crossing check and subsequent code generation agree.
33030 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33032 * config/aarch64/aarch64.c (has_memory_op): Delete.
33033 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33036 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33038 PR rtl-optimization/79388
33039 PR rtl-optimization/79450
33040 * combine.c (distribute_notes): When removing TEM_INSN for which
33041 corresponding dest has last value recorded, invalidate that last
33044 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33046 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33047 of explicit '@'. Add missing assembly comment marker on branch costs
33050 2017-02-13 Nathan Sidwell <nathan@acm.org>
33052 * gengtype-lex.l (<in_struct>): Add '/'.
33054 2017-02-13 Martin Liska <mliska@suse.cz>
33057 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33059 2017-02-13 Richard Biener <rguenther@suse.de>
33061 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33063 * configure: Re-generate.
33064 * config.in: Likewise.
33065 * graphite-dependences.c: Simplify as if
33066 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33067 * graphite-isl-ast-to-gimple.c: Likewise.
33068 * graphite-optimize-isl.c: Likewise.
33069 * graphite-poly.c: Likewise.
33070 * graphite-sese-to-poly.c: Likewise.
33071 * graphite.h: Likewise.
33072 * toplev.c: Include isl/version.h and use isl_version () for
33073 printing the ISL version.
33074 * doc/install.texi: Update ISL requirement.
33076 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33078 * doc/standards.texi (Standards): Update reference to
33081 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33083 * doc/extend.texi (Named Address Spaces): sourceware.org now
33085 * doc/install.texi (Binaries): Ditto.
33088 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33090 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33091 standard terminology "stringize"/"stringizing" throughout.
33092 * doc/cppinternals.texi: Likewise.
33094 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33096 * doc/extend.texi: Fix some spelling mistakes and typos.
33097 * doc/invoke.texi: Likewise.
33099 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33102 * params.def (inline-min-speedup) Change from 10 to 8.
33104 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33106 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33109 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33112 * ipa-inline-analysis.c (get_minimal_bb): New function.
33113 (record_modified): Use it.
33114 (remap_edge_change_prob): Handle also ancestor functions.
33116 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33118 * doc/contrib.texi (Contributors): Remove broken link into
33119 the Mauve CVS repository.
33121 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33123 PR middle-end/79454
33124 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33125 result computation whenever lhs doesn't have vector mode, not
33126 just when it has BLKmode.
33128 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33130 * doc/makefile.texi (profiledbootstrap): Refer to the
33131 installation instructions only in textual form.
33133 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33136 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33138 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33140 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33141 (Specific): Update mingw-w64 reference.
33143 (Specific): Remove broken link to Renesas RX processor.
33145 2017-02-10 Richard Biener <rguenther@suse.de>
33147 * toplev.c (process_options): Do not mention obsolete graphite
33148 options when printing sorry message about missing graphite support.
33149 Mention -floop-nest-optimize.
33151 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33153 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33154 (vtst_p16): Likewise.
33155 (vtstq_p8): Likewise.
33156 (vtstq_p16): Likewise.
33158 (vtstq_p64): Likewise.
33159 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33160 (vset_lane_p64): New.
33161 (vsetq_lane_p64): New.
33163 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33165 PR tree-optimization/79411
33166 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33167 stmt operands are SSA_NAMEs used in abnormal phis.
33168 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33171 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33174 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33177 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33179 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33181 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33183 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33187 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33188 not arbitrary TREE_CONSTANT.
33191 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33192 "omp declare target link" attribute unless is_global_var.
33193 * omp-offload.c (find_link_var_op): Likewise.
33195 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33196 Chung-Lin Tang <cltang@codesourcery.com>
33198 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33200 (gimplify_adjust_omp_clauses): Don't delete TILE.
33201 (gimplify_omp_for): Deal with TILE.
33202 * internal-fn.c (expand_GOACC_TILE): New function.
33203 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33205 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33206 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33208 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33209 avoid DIV for outermost collapse var.
33210 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33211 Remove out of date comments, fix whitespace.
33212 * omp-general.c (omp_extract_for_data): Deal with tiling.
33213 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33214 adjust OLF_DIM_BASE value.
33215 (struct omp_for_data): Add tiling field.
33216 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33217 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33218 for auto loops. Remove default auto determining, moved to
33219 oacc_loop_fixed_partitions.
33220 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33221 stmts, add e_mask field.
33222 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33223 (oacc_thread_numbers): Use oacc_dim_call.
33224 (oacc_xform_tile): New.
33225 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33226 (finish_oacc_loop): Adjust for ifns vector.
33227 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33228 add case for GOACC_TILE fns.
33229 (oacc_loop_xform_loop): Delete.
33230 (oacc_loop_process): Iterate over call list directly, and add
33231 handling for GOACC_TILE fns.
33232 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33234 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33235 vector partitioning to outer loops. Assign 2 partitions to loops
33236 when available. Add TILE handling.
33237 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33238 (execite_oacc_device_lower): Process GOACC_TILE fns,
33239 ignore unknown specs.
33240 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33241 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33242 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33244 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33246 * configure.ac (ACX_BUGURL): Update.
33247 * configure: Regenerate.
33249 2017-02-09 Richard Biener <rguenther@suse.de>
33251 PR tree-optimization/69823
33252 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33253 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33255 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33257 * config/arc/arc-c.def: Add __NPS400__ definition.
33258 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33259 (TARGET_NPS400): Define.
33261 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33263 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33265 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33266 pointer, arch_info.
33267 (arc_cpu_types): Fill the arch_info field with a pointer into the
33268 arc_arch_types table.
33269 (arc_selected_cpu): Declare.
33270 * config/arc/arc.c (arc_selected_cpu): Make global.
33271 (arc_selected_arch): Delete.
33272 (arc_base_cpu): Delete.
33273 (arc_override_options): Remove references to deleted variables,
33274 update access to arch information.
33275 (ARC_OPT): Update access to arch information.
33276 (ARC_OPTX): Likewise.
33277 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33278 (TARGET_ARC600): Update access to arch information.
33279 (TARGET_ARC601): Likewise.
33280 (TARGET_ARC700): Likewise.
33281 (TARGET_EM): Likewise.
33282 (TARGET_HS): Likewise.
33283 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33286 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33289 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33290 condition/operands for integer GE/LE/GEU/LEU operations.
33292 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33294 PR translation/79397
33295 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33298 2017-02-08 Martin Jambor <mjambor@suse.cz>
33301 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33302 whether allocation happened.
33303 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33304 nothing was allocated.
33306 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33308 PR tree-optimization/79408
33309 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33310 constant, but SSA_NAME with a known integer range, use the minimum
33311 of that range instead of op1 to determine if modulo can be replaced
33312 with its first operand.
33314 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33316 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33318 2017-02-08 Richard Biener <rguenther@suse.de>
33320 PR tree-optimization/71824
33321 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33322 Check all loops contained in the merged region.
33324 2017-02-07 Andrew Pinski <apinski@cavium.com>
33326 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33328 2017-02-07 Andrew Pinski <apinski@cavium.com>
33330 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33331 (thunderxt88): Likewise.
33332 (thunderxt81): Disable LSE and change v8.1 to v8.
33333 (thunderxt83): Likewise.
33335 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33336 Richard Biener <rguenther@suse.de>
33338 PR middle-end/79399
33339 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33340 type from int to size_t.
33341 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33343 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33345 PR rtl-optimization/79386
33346 * cprop.c (bypass_conditional_jumps): Initialize
33347 bypass_last_basic_block already before splitting bbs after
33348 unconditional traps...
33349 (bypass_conditional_jumps): ... rather than here.
33352 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33353 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33354 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33355 fix -masm=intel patterns.
33357 2017-02-07 Richard Biener <rguenther@suse.de>
33359 PR tree-optimization/79256
33360 PR middle-end/79278
33361 * builtins.c (get_object_alignment_2): Use min_align_of_type
33362 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33363 and ADJUST_FIELD_ALIGN.
33365 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33367 * doc/tm.texi: Regenerate.
33368 * stor-layout.c (layout_decl): Adjust.
33369 (update_alignment_for_field): Likewise.
33370 (place_field): Likewise.
33371 (min_align_of_type): Likewise.
33372 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33373 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33374 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33375 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33376 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33377 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33378 * config/i386/i386.c (x86_field_alignment): Likewise.
33379 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33380 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33381 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33382 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33383 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33384 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33388 2017-01-30 Richard Biener <rguenther@suse.de>
33390 PR tree-optimization/79256
33391 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33392 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33395 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33397 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33398 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33399 builtins to SImode and emit a zero-extend, if necessary.
33401 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33403 * docs/invoke.texi (RISC-V Options): Alphabetize.
33405 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33407 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33410 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33412 * config/riscv/riscv.c: New file.
33413 * common/config/riscv/riscv-common.c: Likewise.
33414 * config.gcc: Likewise.
33415 * config/riscv/constraints.md: Likewise.
33416 * config/riscv/elf.h: Likewise.
33417 * config/riscv/generic.md: Likewise.
33418 * config/riscv/linux.h: Likewise.
33419 * config/riscv/multilib-generator: Likewise.
33420 * config/riscv/peephole.md: Likewise.
33421 * config/riscv/pic.md: Likewise.
33422 * config/riscv/predicates.md: Likewise.
33423 * config/riscv/riscv-builtins.c: Likewise.
33424 * config/riscv/riscv-c.c: Likewise.
33425 * config/riscv/riscv-ftypes.def: Likewise.
33426 * config/riscv/riscv-modes.def: Likewise.
33427 * config/riscv/riscv-opts.h: Likewise.
33428 * config/riscv/riscv-protos.h: Likewise.
33429 * config/riscv/riscv.h: Likewise.
33430 * config/riscv/riscv.md: Likewise.
33431 * config/riscv/riscv.opt: Likewise.
33432 * config/riscv/sync.md: Likewise.
33433 * config/riscv/t-elf-multilib: Likewise.
33434 * config/riscv/t-linux: Likewise.
33435 * config/riscv/t-linux-multilib: Likewise.
33436 * config/riscv/t-riscv: Likewise.
33437 * configure.ac: Likewise.
33438 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33439 Waterman as RISC-V maintainers.
33440 * doc/install.texi: Add RISC-V entries.
33441 * doc/invoke.texi: Add RISC-V options section.
33442 * doc/md.texi: Add RISC-V constraints section.
33443 * configure: Regenerated.
33445 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33448 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33449 false values to be constant vectors with all 0 or all 1 bits set.
33450 (vcondu<mode><mode>): Likewise.
33451 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33453 (fpmask_comparison_operator): Update comment.
33454 (vecint_comparison_operator): New predicate.
33455 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33456 vector conditionals when the true and false values are constant
33457 vectors with all 0 bits or all 1 bits set.
33459 2017-02-06 Martin Sebor <msebor@redhat.com>
33461 PR tree-optimization/79376
33462 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33464 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33466 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33467 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33468 to simplify split condition.
33470 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33472 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33473 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33476 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33478 PR rtl-optimization/68664
33479 * target.def (can_speculate_insn): New hook.
33480 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33481 * doc/tm.texi: Regenerate.
33482 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33483 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33484 (rs6000_sched_can_speculate_insn): New function.
33486 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33488 PR tree-optimization/79284
33489 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33490 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33491 vectorizable_mask_load_store, vectorizable_operation,
33492 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33493 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33494 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33495 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33496 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33497 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33498 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33499 is_gimple_assign (stmt). Replace another such test with
33500 is_gimple_assign (stmt).
33502 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33505 * config/avr/avr.c (rtl-iter.h): Include it.
33506 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33507 (avr_legitimate_combined_insn): ...and implementation.
33509 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33511 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33512 * config/s390/s390.c (s390_const_operand_ok)
33513 (s390_canonicalize_comparison, s390_extract_part)
33514 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33515 (s390_contiguous_bitmask_p, s390_rtx_costs)
33516 (legitimize_pic_address): Likewise.
33517 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33518 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33519 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33520 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33521 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33523 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33525 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33526 REGNO($0) == REGNO($1).
33528 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33530 * config/s390/linux.h(SIZE_TYPE): Add comment.
33532 2017-02-06 Julian Brown <julian@codesourcery.com>
33533 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33534 Virendra Pathak <virendra.pathak@broadcom.com>
33536 * config/aarch64/aarch64-cores.def: Change the scheduler
33538 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33539 * config/aarch64/thunderx2t99.md: New file.
33541 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33543 * doc/standards.texi (Go Language): Update link to language
33546 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33548 * tree-eh.c (lower_resx): Sanitize profile.
33549 (cleanup_empty_eh_move_lp): Likewise.
33551 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33554 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33556 * cfgloopmanip.h (loop_version): Update prototype.
33557 * modulo-sched.c (sms_schedule): Update call of loop_version.
33558 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33559 * tree-parloops.c (gen_parallel_loop): Likewise.
33560 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33561 * tree-ssa-loop-split.c (split_loop): Likewise.
33562 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33563 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33565 2017-02-05 Martin Liska <mliska@suse.cz>
33568 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33570 (print_operand_address): Initialize a struct to zero.
33572 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33574 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33575 garbage collector only in textual form.
33577 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33579 * doc/extend.texi (x86 specific memory model extensions for
33580 transactional memory): Simplify a phrase.
33582 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33585 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33586 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33587 (atomic_storedi_1): Likewise.
33589 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33591 PR tree-optimization/79338
33592 * tree-parloops.c (gather_scalar_reductions): Don't call
33593 vect_analyze_loop_form for loop->inner before destroying loop's
33596 2017-02-03 Martin Sebor <msebor@redhat.com>
33598 PR tree-optimization/79327
33599 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33600 when precision has resulted in leading zeros.
33601 (format_integer): Adjust the likely counter to assume an unknown
33602 argument that may be zero is non-zero.
33604 2017-02-03 Jason Merrill <jason@redhat.com>
33607 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33608 avoid copying non-taken branch.
33610 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33612 PR tree-optimization/79340
33613 * tree-vect-loop.c (vectorizable_reduction): Release
33614 vec_defs elements after safe_splicing them into other vectors.
33617 PR tree-optimization/79327
33618 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33619 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33621 (format_integer): Use wide_int_to_tree instead of build_int_cst
33622 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33623 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33624 of shortest and longest sequence.
33626 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33628 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33629 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33631 2017-02-03 Walter Lee <walt@tilera.com>
33634 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33635 after initial stackframe link reg save.
33636 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33638 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33641 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33642 wu for stxssp alternative.
33644 2017-02-03 Martin Sebor <msebor@redhat.com>
33646 PR tree-optimization/79352
33647 * gimple-fold.c (get_range_strlen): Add argument.
33648 (get_range_strlen): Change return type to bool.
33649 (get_maxval_strlen): Pass in a dummy argument.
33650 * gimple-fold.h (get_range_strlen): Change return type to bool.
33651 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33652 * tree.h (array_at_struct_end_p): Add argument.
33653 * tree.c (array_at_struct_end_p): Handle it.
33655 2017-02-03 Martin Liska <mliska@suse.cz>
33658 * multiple_target.c (create_dispatcher_calls): Redirect edge
33659 from a caller of a dispatcher.
33660 (expand_target_clones): Make the clones local.
33661 (ipa_target_clone): Do both target clones and resolvers.
33662 (ipa_dispatcher_calls): Remove the pass.
33663 (pass_dispatcher_calls::gate): Likewise.
33664 (make_pass_dispatcher_calls): Likewise.
33665 * passes.def (pass_target_clone): Put as very first IPA early
33668 2017-02-03 Martin Liska <mliska@suse.cz>
33670 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33671 in case of a function with ifunc attribute.
33673 2017-02-03 Martin Liska <mliska@suse.cz>
33675 * cgraph.c (cgraph_node::dump): Dump function version info.
33676 * symtab.c (symtab_node::dump_base): Add missing new line.
33678 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33680 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33681 (ifcombine_ifandif): Use it.
33683 2017-02-03 Martin Liska <mliska@suse.cz>
33685 * doc/invoke.texi: Document default value for
33686 use-after-scope-direct-emission-threshold.
33688 2017-02-03 Martin Liska <mliska@suse.cz>
33690 PR tree-optimization/79339
33691 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33692 (format_floating): Likewise.
33694 2017-02-03 Martin Liska <mliska@suse.cz>
33697 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33699 (ipa_node_params_t::remove): Likewise.
33700 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33701 initialization from removed ipa_node_params_t::insert.
33702 (ipa_node_params::~ipa_node_params): Move from removed
33703 ipa_node_params_t::release.
33704 * symbol-summary.h (symbol_summary::m_released): New member.
33705 Do not release a summary twice. Do not allow to call finalizer
33706 for types of a summary that live in GGC memory.
33708 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33710 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33713 2017-02-02 Martin Sebor <msebor@redhat.com>
33715 PR middle-end/79275
33716 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33717 (format_string): Tighten up the range of output for non-constant
33718 strings and correct the expected range for wide non-constant strings.
33720 2017-02-02 Martin Sebor <msebor@redhat.com>
33722 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33724 PR middle-end/32003
33725 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33727 (-fdump-tree-@var): Add to index and document how to come up
33728 with pass-specific option and dump file names.
33729 (-fdump-passes): Clarify where to look for output.
33731 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33733 PR middle-end/77445
33734 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33735 statistics of the analyzed path; allow threading for speed when
33736 any of BBs along the path are optimized for speed.
33738 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33740 PR middle-end/78468
33741 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33742 settings of the virtual registers.
33745 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33747 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33748 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33751 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33753 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33754 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33756 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33758 * config/s390/s390.md: Add missing comments with the expanded
33760 * config/s390/vector.md: Likewise.
33761 * config/s390/vx-builtins.md: Likewise.
33763 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33766 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33767 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33768 conditions on a single line.
33770 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33772 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33773 __S390_VX__ to __VX__.
33775 2017-02-01 Andrew Pinski <apinski@cavium.com>
33777 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33778 stmt_info to record_stmt_cost.
33779 (vect_get_known_peeling_cost): Pass stmt_info if known to
33781 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33782 cpu_vector_cost field into
33783 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33784 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33785 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33786 splitting of scalar_stmt_cost and vec_stmt_cost.
33787 (thunderx_vector_cost): Likewise.
33788 (cortexa57_vector_cost): LIkewise.
33789 (exynosm1_vector_cost): Likewise.
33790 (xgene1_vector_cost): Likewise.
33791 (thunderx2t99_vector_cost): Improve after the splitting of the two
33793 (aarch64_builtin_vectorization_cost): Update for the splitting of
33794 scalar_stmt_cost and vec_stmt_cost.
33796 2017-02-01 Torvald Riegel <triegel@redhat.com>
33797 Richard Henderson <rth@redhat.com>
33799 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33800 conditional on existance of a fast atomic load.
33801 * optabs-query.c (can_atomic_load_p): New function.
33802 * optabs-query.h (can_atomic_load_p): Declare it.
33803 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33804 no fast atomic load is available for the particular size of access.
33805 (expand_atomic_compare_and_swap): Likewise.
33806 (expand_atomic_load): Likewise.
33807 (expand_atomic_store): Likewise.
33808 (expand_atomic_fetch_op): Likewise.
33809 * testsuite/lib/target-supports.exp
33810 (check_effective_target_sync_int_128): Remove x86 because it provides
33811 no fast atomic load.
33812 (check_effective_target_sync_int_128_runtime): Likewise.
33814 2017-02-01 Richard Biener <rguenther@suse.de>
33816 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33817 (graphite_transform_loops): Provide opt-info for optimized nests.
33818 * tree-parloop.c (parallelize_loops): Provide opt-info for
33819 parallelized loops.
33821 2017-02-01 Richard Biener <rguenther@suse.de>
33823 PR middle-end/79315
33824 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33825 was not set before.
33827 2017-02-01 Richard Biener <rguenther@suse.de>
33829 PR tree-optimization/71824
33830 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33831 Verify the loops are valid in the merged SESE region.
33832 (scop_detection::can_represent_loop_1): Check analyzing the
33833 evolution of the number of iterations in the region succeeds.
33835 2017-01-31 Ian Lance Taylor <iant@golang.org>
33837 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33838 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33840 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33842 PR preprocessor/79210
33843 * input.c (get_substring_ranges_for_loc): Replace line_width
33844 assertion with error-handling.
33846 2017-01-31 Richard Biener <rguenther@suse.de>
33848 PR tree-optimization/77318
33849 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33850 (create_pw_aff_from_tree): Take loop parameter.
33851 (add_condition_to_pbb): Pass loop of the condition to
33852 create_pw_aff_from_tree.
33854 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33856 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33857 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33859 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33863 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33865 (convert_int_to_float128): Likewise.
33866 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33867 (convert_int_to_float128): Likewise.
33868 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33869 (UNSPEC_IEEE128_CONVERT): Likewise.
33870 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33871 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33872 Use local variables for IBM extended format.
33873 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33874 (fix_trunc<mode>si2_fprs): Likewise.
33875 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33876 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33877 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33878 to know that we can now have integers of all sizes in vector
33880 (fix<uns>_<mode>di2_hw): Likewise.
33881 (float<uns>_<mode>si2_hw): Likewise.
33882 (fix_<mode>si2_hw): Likewise.
33883 (fixuns_<mode>si2_hw): Likewise.
33884 (float<uns>_<mode>di2_hw): Likewise.
33885 (float_<mode>di2_hw): Likewise.
33886 (float_<mode>si2_hw): Likewise.
33887 (floatuns_<mode>di2_hw): Likewise.
33888 (floatuns_<mode>si2_hw): Likewise.
33889 (xscvqp<su>wz_<mode>): Delete, no longer used.
33890 (xscvqp<su>dz_<mode>): Likewise.
33891 (xscv<su>dqp_<mode>): Likewise.
33892 (ieee128_mfvsrd_64bit): Likewise.
33893 (ieee128_mfvsrd_32bit): Likewise.
33894 (ieee128_mfvsrwz): Likewise.
33895 (ieee128_mtvsrw): Likewise.
33896 (ieee128_mtvsrd_64bit): Likewise.
33897 (ieee128_mtvsrd_32bit): Likewise.
33899 2017-01-31 Martin Liska <mliska@suse.cz>
33902 * ipa-prop.c (ipa_free_all_node_params): Call release method
33903 instead of ~sumbol_summary to not to trigger double times
33906 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33908 PR tree-optimization/71691
33909 * bitmap.h (class auto_bitmap): New.
33910 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33911 is_maybe_undefined instead of ssa_undefined_value_p.
33913 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33915 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33916 __S390_ARCH_LEVEL__ to __ARCH__.
33918 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33920 PR tree-optimization/79267
33921 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33922 if should_remove_lhs_p is true.
33924 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33927 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33928 (add_alignment_attribute): New.
33929 (base_type_die): Add alignment attribute.
33930 (subrange_type_die): Likewise.
33931 (modified_type_die): Likewise.
33932 (gen_array_type_die): Likewise.
33933 (gen_descr_array_type_die: Likewise.
33934 (gen_enumeration_type_die): Likewise.
33935 (gen_subprogram_die): Likewise.
33936 (gen_variable_die): Likewise.
33937 (gen_field_die): Likewise.
33938 (gen_ptr_to_mbr_type_die): Likewise.
33939 (gen_struct_or_union_type_die): Likewise.
33940 (gen_subroutine_type_die): Likewise.
33941 (gen_typedef_die): Likewise.
33942 (base_type_cmp): Compare alignment attribute.
33944 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33947 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33948 (setb_unsigned) New pattern for setb with CCUNS.
33949 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33950 subfc./subfe sequence to avoid overflow problems. Generate a
33951 shorter sequence with cmpld/setb for power9.
33952 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33953 for generating subfc. instruction.
33954 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33955 now uses this instruction.
33957 2017-01-30 Ian Lance Taylor <iant@google.com>
33960 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33961 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33963 2017-01-30 Martin Sebor <msebor@redhat.com>
33965 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33966 Move constant to the right of a relational operator.
33967 (get_mpfr_format_length, format_character, format_string): Ditto.
33968 (should_warn_p, maybe_warn): Same.
33970 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33972 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33975 * asan.c (get_translation_unit_decl): Remove function.
33976 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33978 2017-01-30 Martin Liska <mliska@suse.cz>
33980 PR gcov-profile/79259
33981 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33982 -fprofile-generate.
33984 2017-01-30 Martin Liska <mliska@suse.cz>
33987 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33988 Initialize variables with NULL value.
33990 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33993 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33995 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33997 2017-01-30 Richard Biener <rguenther@suse.de>
33999 PR tree-optimization/79276
34000 * tree-vrp.c (process_assert_insertions): Properly adjust common
34001 when removing a duplicate.
34003 2017-01-30 Richard Biener <rguenther@suse.de>
34005 PR tree-optimization/79256
34006 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34007 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34009 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34011 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34014 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34015 ("*r<noxa>sbg_<mode>_sll_bitmask")
34016 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34017 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34018 Use contiguous_bitmask_nowrap_operand.
34020 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34023 * config/rs6000/altivec.h (vec_xl): Revise #define.
34024 (vec_xst): Likewise.
34026 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34028 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34030 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34032 PR rtl-optimization/79194
34033 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34034 traps before call to bypass_conditional_jumps.
34036 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34038 PR tree-optimization/71374
34039 * lra-constraints.c (check_conflict_input_operands): New.
34040 (match_reload): Use it.
34042 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34045 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34046 account to calculate conflict_set.
34048 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34050 PR rtl-optimization/78559
34051 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34052 other_insn in combine.
34054 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34056 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34057 uint16_type_node for BT_UINT16.
34059 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34061 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34062 "RTL Tests" to menu.
34063 (GIMPLE Tests): New node.
34064 (RTL Tests): New node.
34066 2017-01-27 Richard Biener <rguenther@suse.de>
34068 PR tree-optimization/79245
34069 * tree-loop-distribution.c (distribute_loop): Apply cost
34070 modeling also to detected patterns.
34072 2017-01-27 Richard Biener <rguenther@suse.de>
34074 PR tree-optimization/71433
34075 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34076 (compare_assert_loc): New function.
34077 (process_assert_insertions): Sort and optimize assert locations
34078 to remove duplicates and push down identical assertions on
34079 edges to their destination block.
34081 2017-01-27 Richard Biener <rguenther@suse.de>
34083 PR tree-optimization/79244
34084 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34085 out SSA names even if abnormal.
34087 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34089 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34090 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34091 instead of MPFR_RNDN.
34093 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34096 * arm.c (arm_option_override): Don't call build_target_option_node
34097 until after doing all option overrides.
34098 (arm_valid_target_attribute_tree): Likewise.
34100 2017-01-27 Martin Liska <mliska@suse.cz>
34102 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34103 for {cd}tors and C++ {cd}tors.
34105 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34107 * config/s390/s390.md ("*setmem_long_and")
34108 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34110 2017-01-26 Martin Sebor <msebor@redhat.com>
34112 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34115 2017-01-26 Martin Sebor <msebor@redhat.com>
34117 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34118 HAVE_DFmode before using XFmode or DFmode.
34119 (parse_directive): Avoid using the z length modifier to avoid
34120 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34122 PR middle-end/78703
34123 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34124 to accept adjustment as an array.
34125 (get_int_range): New function.
34126 (struct directive): Make width and prec arrays.
34127 (directive::set_width, directive::set_precision): Call get_int_range.
34128 (format_integer, format_floating): Handle width and precision ranges.
34129 (format_string, parse_directive): Same.
34131 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34134 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34135 comdat_type_p set, just clone them, but keep the children in the
34139 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34140 which have direct callers with -fvar-tracking-assignments enabled
34142 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34143 inside of type units.
34145 2017-01-26 Martin Sebor <msebor@redhat.com>
34147 PR middle-end/78703
34148 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34150 (struct format_result): Replace number_chars, number_chars_min,
34151 and number_chars_max with a single member of struct result_range.
34153 (format_result::operator+=): Adjust.
34154 (struct fmtresult): Remove bounded. Handle likely and unlikely
34156 (fmtresult::adjust_for_width_or_precision): New function.
34157 (fmtresult:type_max_digits): New function.
34158 (bytes_remaining): Handle likely and unlikely counters.
34159 (min_bytes_remaining): Remove.
34160 (format_percent): Simplify.
34161 (format_integer, format_floating): Set likely and unlikely counters.
34162 (get_string_length, format_character, format_string): Same.
34163 (format_plain, should_warn_p): New function.
34164 (maybe_warn): Call should_warn_p. Update diagnostic messages
34165 and handle those for all directives, including plain strings.
34166 (format_directive): Handle likely and unlikely counters.
34167 Remove unnecessary quoting from diagnostics. Add an informational
34169 (add_bytes): Remove.
34170 (pass_sprintf_length::compute_format_length): Simplify.
34171 (try_substitute_return_value): Handle likely and unlikely counters.
34173 2017-01-26 Carl Love <cel@us.ibm.com>
34175 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34176 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34178 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34181 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34182 endianess for subregs into account.
34183 * lra-constraints.c (lra_constraints): Do risky transformations
34184 always on the first iteration.
34185 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34186 last_call_used_reg_set.
34187 (process_bb_lives): Define and use last_call_used_reg_set.
34188 * lra.c (lra): Always continue after lra_constraints on the first
34191 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34193 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34195 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34197 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34199 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34200 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34201 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34202 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34203 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34204 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34205 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34206 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34207 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34209 2017-01-26 Marek Polacek <polacek@redhat.com>
34212 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34213 for the third operand.
34215 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34217 PR middle-end/79236
34218 * omp-low.c (struct omp_context): Add simt_stmt field.
34219 (scan_omp_for): Return omp_context *.
34220 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34221 context to the _simt_ SIMD stmt.
34222 (lower_omp_for): For combined SIMD with sibling _simt_
34223 SIMD, make sure to use the same decls in _looptemp_
34224 clauses as in the sibling.
34226 2017-01-26 David Sherwood <david.sherwood@arm.com>
34228 PR middle-end/79212
34229 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34232 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34235 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34236 emit fld b; fld a; if possible.
34238 * brig-builtins.def: Update copyright years.
34239 * config/arm/arm_acle_builtins.def: Update copyright years.
34241 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34244 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34245 constraint instead of o for the stxsd instruction.
34247 2017-01-25 Carl Love <cel@us.ibm.com>
34249 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34250 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34252 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34254 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34256 2017-01-25 Richard Biener <rguenther@suse.de>
34258 PR tree-optimization/69264
34259 * target.def (vector_alignment_reachable): Improve documentation.
34260 * doc/tm.texi: Regenerate.
34261 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34263 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34264 earlier changes with respect to TYPE_USER_ALIGN.
34265 (vector_alignment_reachable_p): Likewise. Improve dumping.
34267 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34270 * config/arm/arm.md (xordi3): Force constant operand into a register
34273 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34275 * doc/invoke.texi (-fstore-merging): Correct default optimization
34276 levels at which it is enabled.
34277 (-O): Move -fstore-merging from list to...
34280 2017-01-25 Richard Biener <rguenther@suse.de>
34283 * omp-expand.c: Include debug.h.
34284 (expand_omp_taskreg): Make sure to generate early debug before
34285 outlining anything from a function.
34286 (expand_omp_target): Likewise.
34287 (grid_expand_target_grid_body): Likewise.
34289 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34292 * asan.c (get_translation_unit_decl): New function.
34293 (asan_add_global): Extract modules file name from globals
34294 TRANSLATION_UNIT_DECL name.
34296 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34299 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34300 for long calls with APCS frame and VFP.
34302 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34304 * cfg.c (original_copy_tables_initialized_p): New function.
34305 * cfg.h (original_copy_tables_initialized_p): New decl.
34306 * cfgrtl.c (relink_block_chain): Guard the call to
34307 free_original_copy_tables with a call to
34308 original_copy_tables_initialized_p.
34309 * cgraph.h (symtab_node::native_rtl_p): New decl.
34310 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34311 (symtab_node::needed_p): Don't assert for early assembly output
34312 for __RTL functions.
34313 (cgraph_node::finalize_function): Set "force_output" for __RTL
34315 (cgraph_node::analyze): Bail out early for __RTL functions.
34316 (analyze_functions): Update assertion to support __RTL functions.
34317 (cgraph_node::expand): Bail out early for __RTL functions.
34318 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34320 * function.h (struct function): Update comment for field
34322 * gimple-expr.c: Include "tree-pass.h".
34323 (gimple_has_body_p): Return false for __RTL functions.
34324 * Makefile.in (OBJS): Add run-rtl-passes.o.
34325 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34327 (gcc::pass_manager::get_clean_slate): New accessor.
34328 * passes.c: Include "insn-addr.h".
34329 (should_skip_pass_p): Add logging. Update logic for running
34330 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34331 property-provider override so it is only done for gimple passes.
34333 (skip_pass): New function.
34334 (execute_one_pass): Call skip_pass when skipping passes.
34335 * read-md.c (md_reader::read_char): Support filtering
34336 the input to a subset of line numbers.
34337 (md_reader::md_reader): Initialize fields
34338 m_first_line and m_last_line.
34339 (md_reader::read_file_fragment): New function.
34340 * read-md.h (md_reader::read_file_fragment): New decl.
34341 (md_reader::m_first_line): New field.
34342 (md_reader::m_last_line): New field.
34343 * read-rtl-function.c (function_reader::create_function): Only
34344 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34345 curr_properties. Set DECL_INITIAL to a dummy block.
34346 (read_rtl_function_body_from_file_range): New function.
34347 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34349 * run-rtl-passes.c: New file.
34350 * run-rtl-passes.h: New file.
34352 2017-01-24 Jeff Law <law@redhat.com>
34354 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34357 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34359 PR tree-optimization/79159
34360 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34361 (record_nonwrapping_iv): Improve boundary using above function if no
34362 value range information.
34364 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34365 Martin Jambor <mjambor@suse.cz>
34367 * brig-builtins.def: New file.
34368 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34369 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34370 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34371 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34372 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34373 * builtin-types.def (BT_INT8): New.
34374 (BT_INT16): Likewise.
34375 (BT_UINT8): Likewise.
34376 (BT_UINT16): Likewise.
34377 (BT_FN_ULONG): Likewise.
34378 (BT_FN_UINT_INT): Likewise.
34379 (BT_FN_UINT_ULONG): Likewise.
34380 (BT_FN_UINT_LONG): Likewise.
34381 (BT_FN_UINT_PTR): Likewise.
34382 (BT_FN_ULONG_PTR): Likewise.
34383 (BT_FN_INT8_FLOAT): Likewise.
34384 (BT_FN_INT16_FLOAT): Likewise.
34385 (BT_FN_UINT32_FLOAT): Likewise.
34386 (BT_FN_UINT16_FLOAT): Likewise.
34387 (BT_FN_UINT8_FLOAT): Likewise.
34388 (BT_FN_UINT64_FLOAT): Likewise.
34389 (BT_FN_UINT16_UINT32): Likewise.
34390 (BT_FN_UINT32_UINT16): Likewise.
34391 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34392 (BT_FN_INT_PTR_INT): Likewise.
34393 (BT_FN_UINT_PTR_UINT): Likewise.
34394 (BT_FN_LONG_PTR_LONG): Likewise.
34395 (BT_FN_ULONG_PTR_ULONG): Likewise.
34396 (BT_FN_VOID_UINT64_UINT64): Likewise.
34397 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34398 (BT_FN_INT8_INT8_INT8): Likewise.
34399 (BT_FN_INT16_INT16_INT16): Likewise.
34400 (BT_FN_INT_INT_INT): Likewise.
34401 (BT_FN_UINT_FLOAT_UINT): Likewise.
34402 (BT_FN_FLOAT_UINT_UINT): Likewise.
34403 (BT_FN_ULONG_UINT_UINT): Likewise.
34404 (BT_FN_ULONG_UINT_PTR): Likewise.
34405 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34406 (BT_FN_UINT_UINT_UINT): Likewise.
34407 (BT_FN_VOID_UINT_PTR): Likewise.
34408 (BT_FN_UINT_UINT_PTR: Likewise.
34409 (BT_FN_UINT32_UINT64_PTR): Likewise.
34410 (BT_FN_INT_INT_UINT_UINT): Likewise.
34411 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34412 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34413 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34414 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34415 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34416 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34417 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34418 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34419 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34420 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34421 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34422 * doc/frontends.texi: List BRIG FE.
34423 * doc/install.texi (Testing): Add BRIG tesring requirements.
34424 * doc/invoke.texi (Overall Options): Mention BRIG.
34425 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34427 2017-01-24 Richard Biener <rguenther@suse.de>
34429 PR translation/79208
34430 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34432 2017-01-24 Martin Jambor <mjambor@suse.cz>
34435 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34436 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34437 and known_contexts.
34439 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34441 PR middle-end/79123
34442 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34443 casts from signed to unsigned really don't have a range.
34445 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34447 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34448 GMP_RNDx for compatiblity.
34450 2017-01-24 Martin Liska <mliska@suse.cz>
34453 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34454 that would prevent us to call alloca with -1 as argument.
34456 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34458 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34459 Avoid -Wformat-security warning.
34461 2017-01-23 Andrew Pinski <apinski@cavium.com>
34463 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34466 2017-01-23 Martin Sebor <msebor@redhat.com>
34468 PR middle-end/78703
34469 * gimple-ssa-sprintf.c (warn_level): New global.
34470 (format_integer): Use it here and throughout the rest of the file.
34471 Use the same switch to compute sign as base.
34472 (maybe_warn): New function.
34473 (format_directive): Factor out warnings into maybe_warn.
34474 Add debugging output. Use warn_level.
34475 (add_bytes): Use warn_level.
34476 (pass_sprintf_length::compute_format_length): Add debugging output.
34477 (try_substitute_return_value): Same.
34478 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34480 PR middle-end/78703
34481 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34482 (struct fmtresult, format_integer, format_floating): Adjust.
34483 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34484 (get_string_length, format_string,format_directive): Same.
34485 (pass_sprintf_length::compute_format_length): Same.
34486 (try_substitute_return_value): Simplify slightly.
34488 PR middle-end/78703
34489 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34490 (fmtresult::operator+=): Outlined.
34491 (struct fmtresult): Add ctors.
34492 (struct conversion_spec): Rename...
34493 (struct directive): ...to this. Add and remove data members.
34494 (directive::set_width, directive::set_precision): New functions.
34495 (format_percent): Use fmtresult ctor.
34496 (get_width_and_precision): Remove.
34497 (format_integer): Make naming changes. Avoid computing width and
34499 (format_floating): Same. Adjust indentation.
34500 (format_character, format_none): New functions.
34501 (format_string): Moved character handling to format_character.
34502 (format_directive): Remove arguments, change return type.
34503 (parse_directive): New function.
34504 (pass_sprintf_length::compute_format_length): Move directive
34505 parsing to parse_directive.
34507 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34509 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34510 (assign_assembler_name_if_needed): ... this.
34511 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34512 (assign_assembler_name_if_needed): ... this.
34513 (free_lang_data_in_cgraph): Adjust callers.
34514 * cgraphunit.c (cgraph_node::analyze): Likewise.
34515 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34518 2017-01-23 Richard Biener <rguenther@suse.de>
34520 PR tree-optimization/79088
34521 PR tree-optimization/79188
34522 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34523 resetting loop bounds after last path deletion. Reset loop
34524 bounds of the target loop, make code match the comments.
34525 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34526 Make sure loops need no fixups.
34528 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34530 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34531 exponent support with double type for first argument.
34532 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34533 type returned by __builtin_vec_extract_sig,
34534 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34535 functions from "vector int" to "vector unsigned int" or from
34536 "vector long long int" to "vector unsigned long long int".
34537 Changed type returned by __builtin_vec_extract_exp,
34538 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34539 functions from "vector int" to "vector unsigned int" or from
34540 "vector long long int" to "vector unsigned long long int".
34541 Changed return type of __builtin_vec_test_data_class,
34542 __builtin_vec_test_data_class_sp, and
34543 __builtin_vec_test_data_class_dp from "vector int" to
34544 "vector bool int" or from "vector long long int" to "vector bool
34545 long long int" and changed second argument type from "unsigned
34546 int" to "int". Added new overloaded function forms "vector float
34547 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34548 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34549 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34550 double, vector unsigned long long int)" and "vector double
34551 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34552 long int)". Changed return type of
34553 __builtin_scalar_test_data_class and
34554 __builtin_scalar_test_data_class_sp and
34555 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34556 int" and changed second argument from "unsigned int" to "int".
34557 Changed type returned by __builtin_scalar_test_neg,
34558 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34559 from "int" to "bool int". Added new overloaded function form
34560 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34561 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34562 exponent double-precision with floating point first argument.
34563 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34564 documentation of scalar_test_data_class, scalar_test_neg,
34565 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34566 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34567 vec_test_data_class built-in functions to reflect refinements in
34568 their type signatures.
34570 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34572 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34574 (aarch64_elf_asm_destructor): Likewise.
34576 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34578 PR rtl-optimization/78634
34579 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34580 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34581 * ifcvt.c (noce_try_cmove): Add missing cost check.
34583 PR rtl-optimization/71724
34584 * combine.c (if_then_else_cond): Look for situations where it is
34585 beneficial to undo the work of one of the recursive calls.
34587 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34589 PR tree-optimization/70754
34590 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34591 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34592 combined stmt before it if not NULL.
34593 (combine_chains): Process refs reversely and compute dominance point
34596 2017-01-23 Martin Liska <mliska@suse.cz>
34598 PR tree-optimization/79196
34599 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34600 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34602 (strlen_optimize_stmt): Call the renamed function.
34604 2017-01-23 Michael Matz <matz@suse.de>
34606 PR tree-optimization/78384
34607 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34609 2017-01-23 Richard Biener <rguenther@suse.de>
34611 PR tree-optimization/79186
34612 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34613 both incoming edges before moving an assert.
34615 2017-01-23 Martin Jambor <mjambor@suse.cz>
34617 * ipa-prop.c (load_from_param_1): Removed.
34618 (load_from_unmodified_param): Bits from load_from_param_1 put back
34620 (load_from_param): Removed.
34621 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34622 with stmt. Reverted back to use of load_from_unmodified_param.
34624 2017-01-23 Martin Jambor <mjambor@suse.cz>
34627 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34628 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34629 field a pointer to garbage collected vector, mark lattices and
34630 ipcp_orig_node with GTY((skip)).
34631 (ipa_get_param_count): Adjust to descriptors being a pointer.
34632 (ipa_get_param): Likewise.
34633 (ipa_get_type): Likewise.
34634 (ipa_get_param_move_cost): Likewise.
34635 (ipa_set_param_used): Likewise.
34636 (ipa_get_controlled_uses): Likewise.
34637 (ipa_set_controlled_uses): Likewise.
34638 (ipa_is_param_used): Likewise.
34639 (ipa_node_params_t): Move into garbage collector. New methods insert
34641 (ipa_node_params_sum): Annotate wth GTY(()).
34642 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34644 (ipa_load_from_parm_agg): Adjust declaration.
34645 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34646 * ipa-profile.c (ipa_profile): Likewise.
34647 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34648 (ipa_populate_param_decls): Make descriptors parameter garbage
34650 (ipa_dump_param): Adjust to descriptors being a pointer.
34651 (ipa_alloc_node_params): Likewise.
34652 (ipa_initialize_node_params): Likewise.
34653 (load_from_param_1): Make descriptors parameter garbage collected.
34654 (load_from_unmodified_param): Likewise.
34655 (load_from_param): Likewise.
34656 (ipa_load_from_parm_agg): Likewise.
34657 (ipa_node_params::~ipa_node_params): Removed.
34658 (ipa_free_all_node_params): Remove call to delete operator.
34659 (ipa_node_params_t::insert): New.
34660 (ipa_node_params_t::remove): Likewise.
34661 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34662 copy known_csts and known_contexts vectors.
34663 (ipa_read_node_info): Adjust to descriptors being a pointer.
34664 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34666 (ipcp_transform_function): Make descriptors variable garbage
34669 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34671 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34672 * config/i386/avx512dqintrin.h: Ditto.
34673 * config/i386/avx512fintrin.h: Ditto.
34674 * config/i386/i386.c: Handle new builtins.
34675 * config/i386/i386-builtin.def: Add new builtins.
34676 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34677 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34679 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34680 Martin Liska <mliska@suse.cz>
34682 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34683 * asan.c (asan_expand_poison_ifn): Support stores and use
34684 appropriate ASAN report function.
34685 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34686 * internal-fn.def (ASAN_POISON_USE): Declare.
34687 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34688 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34689 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34690 ASAN_POISON calls w/o LHS.
34691 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34692 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34693 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34694 * gimplify.c (asan_poison_variables): Add attribute
34695 use_after_scope_memory to variables that really needs to live
34697 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34698 having the attribute.
34700 2017-01-23 Martin Liska <mliska@suse.cz>
34702 * asan.c (create_asan_shadow_var): New function.
34703 (asan_expand_poison_ifn): Likewise.
34704 * asan.h (asan_expand_poison_ifn): New declaration.
34705 * internal-fn.c (expand_ASAN_POISON): Likewise.
34706 * internal-fn.def (ASAN_POISON): New builtin.
34707 * sanopt.c (pass_sanopt::execute): Expand
34708 asan_expand_poison_ifn.
34709 * tree-inline.c (copy_decl_for_dup_finish): Make function
34711 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34712 * tree-ssa.c (is_asan_mark_p): New function.
34713 (execute_update_addresses_taken): Rewrite local variables
34714 (identified just by use-after-scope as addressable) into SSA.
34716 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34718 * doc/install.texi (Specific): opensource.apple.com uses https
34719 now. Remove trailing slash.
34721 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34723 * README.Portability: Remove note on an Irix compatibility issue.
34725 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34727 * gcov.c (INCLUDE_ALGORITHM): Define.
34728 (INCLUDE_VECTOR): Define.
34729 No longer include <vector> and <algorithm> directly.
34731 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34733 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34735 * doc/invoke.texi (Code Gen Options): Ditto.
34737 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34740 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34742 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34744 rtl-optimization/79125
34745 * cprop.c (local_cprop_pass): Handle cases where we make an
34746 unconditional trap.
34748 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34752 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34753 read from, for big endian.
34755 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34757 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34758 register pauth builtins for LP64 only.
34760 2017-01-20 Marek Polacek <polacek@redhat.com>
34763 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34766 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34768 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34770 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34771 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34772 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34774 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34777 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34778 in big-endian mode when they are not a single duplicated value.
34780 2017-01-20 Richard Biener <rguenther@suse.de>
34782 * BASE-VER: Bump to 7.0.1.
34784 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34786 * omp-low.c (omplow_simd_context): New struct. Use it...
34787 (lower_rec_simd_input_clauses): ...here and...
34788 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34789 references to idx, lane, max_vf, is_simt.
34791 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34793 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34796 2017-01-20 Martin Jambor <mjambor@suse.cz>
34798 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34799 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34801 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34802 (GTFILES): Rename hsa.c to hsa-common.c.
34803 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34804 * hsa-dump.c: Likewise.
34805 * hsa-gen.c: Likewise.
34806 * hsa-regalloc.c: Likewise.
34807 * ipa-hsa.c: Likewise.
34808 * omp-expand.c: Likewise.
34809 * omp-low.c: Likewise.
34810 * toplev.c: Likewise.
34812 2017-01-20 Marek Polacek <polacek@redhat.com>
34815 * doc/invoke.texi: Document -Wduplicated-branches.
34816 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34817 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34818 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34819 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34820 return 0 only when not OEP_LEXICOGRAPHIC.
34821 (fold_build_cleanup_point_expr): Use the expression
34822 location when building CLEANUP_POINT_EXPR.
34823 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34824 * tree.c (add_expr): Handle error_mark_node.
34826 2017-01-20 Martin Liska <mliska@suse.cz>
34829 * tree-profile.c (init_ic_make_global_vars): Do not call
34831 (gimple_init_gcov_profiler): Likewise.
34833 2017-01-20 Martin Liska <mliska@suse.cz>
34836 * cgraph.h (maybe_create_reference): Remove argument and
34838 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34840 * ipa-cp.c (create_specialized_node): Likewise.
34841 * symtab.c (symtab_node::maybe_create_reference): Handle
34842 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34844 2017-01-20 Martin Liska <mliska@suse.cz>
34846 * read-rtl-function.c (function_reader::create_function): Use
34847 build_decl instread of build_decl_stat.
34849 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34851 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34852 * config/i386/avx512dqintrin.h: Ditto.
34853 * config/i386/avx512fintrin.h: Ditto.
34854 * config/i386/i386-builtin-types.def: Add new types.
34855 * config/i386/i386.c: Handle new types.
34856 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34857 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34858 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34859 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34860 (__builtin_ia32_kshiftridi): New.
34861 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34863 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34867 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34868 define to rs6000_init_stack_protect_guard.
34869 (rs6000_init_stack_protect_guard): New function.
34871 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34872 Yunqiang Su <yunqiang.su@imgtec.com>
34874 * config.gcc (supported_defaults): Add madd4.
34875 (with_madd4): Add validation.
34876 (all_defaults): Add madd4.
34877 * config/mips/mips.opt (mmadd4): New option.
34878 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34880 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34882 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34883 (ISA_HAS_FUSED_MADD4): Likewise.
34884 * doc/invoke.texi (-mmadd4): Document the new option.
34885 * doc/install.texi (--with-madd4): Document the new option.
34887 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34889 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34890 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34891 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34892 (aarch64_init_pauth_hint_builtins): New.
34893 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34894 (aarch64_expand_builtin): Expand new builtins.
34896 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34898 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34899 * combine-stack-adj.c (no_unhandled_cfa): Handle
34900 REG_CFA_TOGGLE_RA_MANGLE.
34901 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34902 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34903 info for return address signing.
34904 (aarch64_expand_epilogue): Likewise.
34906 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34908 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34909 * config/aarch64/aarch64-protos.h
34910 (aarch64_return_address_signing_enabled): New declaration.
34911 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34913 (aarch64_expand_prologue): Sign return address before it's pushed onto
34915 (aarch64_expand_epilogue): Authenticate return address fetched from
34917 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34918 (aarch64_attributes): New function attributes for "sign-return-address".
34919 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34920 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34921 ("*do_return"): Generate combined instructions according to key index.
34922 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34923 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34925 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34926 * config/aarch64/aarch64.opt (msign-return-address=): New.
34927 * doc/extend.texi (AArch64 Function Attributes): Documents
34928 "sign-return-address=".
34929 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34931 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34933 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34934 overall option summary.
34936 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34938 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34939 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34940 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34941 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34943 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34945 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34946 -mpower9-minmax by default for -mcpu=power9.
34947 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34948 128-bit floating point.
34950 2017-01-20 Alan Modra <amodra@gmail.com>
34952 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34953 optimizing for size.
34955 2017-01-20 Alan Modra <amodra@gmail.com>
34958 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34959 for strcmp and strncmp from corresponding builtin decl.
34961 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34963 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34964 instead of i386/rtems-64.h.
34965 * config/i386/rtems-64.h: Remove.
34967 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34971 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34973 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34975 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34977 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34978 Change int to HOST_WIDE_INT.
34979 * config/aarch64/aarch64-protos.h
34980 (aarch64_simd_gen_const_vector_dup): Likewise.
34981 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34983 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34985 * langhooks-def.h (lhd_type_for_size): New decl.
34986 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34987 * langhooks.c (lhd_type_for_size): New function, taken from
34990 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34992 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34993 define_bypass for CR latency.
34994 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34995 (power9-alu2): Add define_bypass for CR latency.
34997 (power9-mul): Update insn latency.
34998 (power9-mul-compare): Update insn latency, bypass latency and remove
35001 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35003 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35005 * config/aarch64/aarch64.md
35006 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35007 aarch64_nopcrelative_literal_loads.
35008 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35010 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35012 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35013 TARGET_LOONGSON_3A.
35014 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35016 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35019 * config.gcc (supported_defaults): Add lxc1-sxc1.
35020 (with_lxc1_sxc1): Add validation.
35021 (all_defaults): Add lxc1-sxc1.
35022 * config/mips/mips.opt (mlxc1-sxc1): New option.
35023 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35025 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35026 __mips_no_lxc1_sxc1.
35027 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35028 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35029 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35031 2017-01-19 Richard Biener <rguenther@suse.de>
35033 PR tree-optimization/72488
35034 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35035 sure to restore SSA info.
35036 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35038 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35040 PR rtl-optimization/79121
35041 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35042 of the inner type when shifting an extended value.
35044 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35047 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35048 interposable aliases.
35050 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35053 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35054 Use the evmergelohi instruction.
35055 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35056 (mov_sitf_e500_subreg8_2_be): Likewise.
35057 (mov_sitf_e500_subreg12_2_le): Likewise.
35058 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35059 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35060 (mov_sitf_e500_subreg8_2_le): Likewise.
35061 (mov_sitf_e500_subreg12_2_be): Likewise.
35063 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35065 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35066 attribute from vecsimple to vecperm.
35067 (altivec_vbpermq2): Likewise.
35069 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35072 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35074 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35075 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35076 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35077 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35078 case where N arg is SIZE_MAX.
35079 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35080 (cmpstrsi): Add pattern.
35082 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35084 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35085 __builtin_vec_revb builtins.
35086 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35087 built-in functions to support generation of the ISA 3.0 XXBR<x>
35088 vector byte reverse instructions.
35089 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35090 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35091 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35092 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35093 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35094 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35095 (P9V_BUILTIN_VEC_REVB): Likewise.
35096 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35097 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35098 (p9_xxbrq_v16qi): Likewise.
35099 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35100 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35101 (p9_xxbrh_v8hi): Likewise.
35102 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35103 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35104 vec_revb built-in functions.
35106 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35108 PR rtl-optimization/78952
35109 * config/i386/i386.md (any_extract): New code iterator.
35110 (*insvqi_2): Use any_extract for source operand.
35111 (*insvqi_3): Use any_shiftrt for source operand.
35113 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35115 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35117 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35119 2017-01-18 Matthias Klose <doko@ubuntu.com>
35121 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35123 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35125 * config/rs6000/altivec.h (vec_bperm): Change #define.
35126 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35127 (altivec_vbpermq2): New define_insn.
35128 (altivec_vbpermd): Likewise.
35129 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35130 function interface.
35131 (VBPERMD): Likewise.
35132 (VBPERM): New polymorphic function interface.
35133 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35134 Add entries for P9V_BUILTIN_VEC_VBPERM.
35135 * doc/extend.texi: Add interfaces for vec_bperm.
35137 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35139 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35140 first letter of error messages.
35141 (s390_resolve_overloaded_builtin): Likewise.
35142 * config/s390/s390.c (s390_expand_builtin): Likewise.
35143 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35144 (s390_valid_target_attribute_inner_p): Likewise.
35145 * config/s390/s390.md ("tabort"): Likewise.
35147 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35149 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35150 (ISA_AVOID_DIV_HILO): New macro.
35151 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35152 (ISA_HAS_DDIV): Likewise.
35154 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35156 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35158 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35160 * doc/invoke.texi (fabi-version): Spelling fix.
35162 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35165 * doc/invoke.texi (fabi-version): Mention mangling fix for
35168 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35171 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35173 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35176 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35177 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35179 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35181 (stack_protect_test): Ditto.
35182 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35183 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35185 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35186 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35187 -mstack-protector-guard-offset=.
35188 (RS/6000 and PowerPC Options): Ditto.
35190 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35192 * config/i386/i386.h (MASK_CLASS_P): New define.
35193 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35194 there are no registers from different register sets also when
35195 mask registers are used. Update function comment.
35196 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35197 to (*k/*r) and (*k/*km) alternatives.
35199 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35201 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35202 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35203 (EH_RETURN_HANDLER_RTX): New define.
35204 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35205 Force frame pointer in EH return functions.
35206 (aarch64_expand_epilogue): Add barrier for eh_return.
35207 (aarch64_final_eh_return_addr): Remove.
35208 (aarch64_eh_return_handler_rtx): New function.
35209 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35211 (aarch64_eh_return_handler_rtx): New prototype.
35213 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35215 * config/rs6000/altivec.h (vec_rlmi): New #define.
35216 (vec_vrlnm): Likewise.
35217 (vec_rlnm): Likewise.
35218 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35219 (UNSPEC_VRLNM): Likewise.
35220 (VIlong): New mode iterator.
35221 (altivec_vrl<VI_char>mi): New define_insn.
35222 (altivec_vrl<VI_char>nm): Likewise.
35223 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35225 (VRLDNM): Likewise.
35226 (RLNM): New polymorphic function entry.
35227 (VRLWMI): New monomorphic function entry.
35228 (VRLDMI): Likewise.
35229 (RLMI): New polymorphic function entry.
35230 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35231 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35232 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35235 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35238 * dwarf2out.c (field_byte_offset): Restore the
35239 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35240 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35241 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35244 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35247 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35249 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35252 * dwarf2out.c (add_data_member_location_attribute): For constant
35253 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35254 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35255 DW_AT_byte_size attributes.
35257 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35259 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35260 after forcing to constant memory when the code model is medium.
35262 2017-01-17 Julia Koval <julia.koval@intel.com>
35265 * config/i386/avx512fintrin.h
35266 (_mm512_i32gather_ps): Change __addr type to void const*.
35267 (_mm512_mask_i32gather_ps): Ditto.
35268 (_mm512_i32gather_pd): Ditto.
35269 (_mm512_mask_i32gather_pd): Ditto.
35270 (_mm512_i64gather_ps): Ditto.
35271 (_mm512_mask_i64gather_ps): Ditto.
35272 (_mm512_i64gather_pd): Ditto.
35273 (_mm512_mask_i64gather_pd): Ditto.
35274 (_mm512_i32gather_epi32): Ditto.
35275 (_mm512_mask_i32gather_epi32): Ditto.
35276 (_mm512_i32gather_epi64): Ditto.
35277 (_mm512_mask_i32gather_epi64): Ditto.
35278 (_mm512_i64gather_epi32): Ditto.
35279 (_mm512_mask_i64gather_epi32): Ditto.
35280 (_mm512_i64gather_epi64): Ditto.
35281 (_mm512_mask_i64gather_epi64): Ditto.
35282 (_mm512_i32scatter_ps): Change __addr type to void*.
35283 (_mm512_mask_i32scatter_ps): Ditto.
35284 (_mm512_i32scatter_pd): Ditto.
35285 (_mm512_mask_i32scatter_pd): Ditto.
35286 (_mm512_i64scatter_ps): Ditto.
35287 (_mm512_mask_i64scatter_ps): Ditto.
35288 (_mm512_i64scatter_pd): Ditto.
35289 (_mm512_mask_i64scatter_pd): Ditto.
35290 (_mm512_i32scatter_epi32): Ditto.
35291 (_mm512_mask_i32scatter_epi32): Ditto.
35292 (_mm512_i32scatter_epi64): Ditto.
35293 (_mm512_mask_i32scatter_epi64): Ditto.
35294 (_mm512_i64scatter_epi32): Ditto.
35295 (_mm512_mask_i64scatter_epi32): Ditto.
35296 (_mm512_i64scatter_epi64): Ditto.
35297 (_mm512_mask_i64scatter_epi64): Ditto.
35298 * config/i386/avx512pfintrin.h
35299 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35300 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35301 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35302 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35303 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35304 (_mm512_prefetch_i32scatter_ps): Ditto.
35305 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35306 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35307 (_mm512_prefetch_i64scatter_pd): Ditto.
35308 (_mm512_prefetch_i64scatter_ps): Ditto.
35309 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35310 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35311 * config/i386/avx512vlintrin.h
35312 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35313 (_mm_mmask_i32gather_ps): Ditto.
35314 (_mm256_mmask_i32gather_pd): Ditto.
35315 (_mm_mmask_i32gather_pd): Ditto.
35316 (_mm256_mmask_i64gather_ps): Ditto.
35317 (_mm_mmask_i64gather_ps): Ditto.
35318 (_mm256_mmask_i64gather_pd): Ditto.
35319 (_mm_mmask_i64gather_pd): Ditto.
35320 (_mm256_mmask_i32gather_epi32): Ditto.
35321 (_mm_mmask_i32gather_epi32): Ditto.
35322 (_mm256_mmask_i32gather_epi64): Ditto.
35323 (_mm_mmask_i32gather_epi64): Ditto.
35324 (_mm256_mmask_i64gather_epi32): Ditto.
35325 (_mm_mmask_i64gather_epi32): Ditto.
35326 (_mm256_mmask_i64gather_epi64): Ditto.
35327 (_mm_mmask_i64gather_epi64): Ditto.
35328 (_mm256_i32scatter_ps): Change __addr type to void*.
35329 (_mm256_mask_i32scatter_ps): Ditto.
35330 (_mm_i32scatter_ps): Ditto.
35331 (_mm_mask_i32scatter_ps): Ditto.
35332 (_mm256_i32scatter_pd): Ditto.
35333 (_mm256_mask_i32scatter_pd): Ditto.
35334 (_mm_i32scatter_pd): Ditto.
35335 (_mm_mask_i32scatter_pd): Ditto.
35336 (_mm256_i64scatter_ps): Ditto.
35337 (_mm256_mask_i64scatter_ps): Ditto.
35338 (_mm_i64scatter_ps): Ditto.
35339 (_mm_mask_i64scatter_ps): Ditto.
35340 (_mm256_i64scatter_pd): Ditto.
35341 (_mm256_mask_i64scatter_pd): Ditto.
35342 (_mm_i64scatter_pd): Ditto.
35343 (_mm_mask_i64scatter_pd): Ditto.
35344 (_mm256_i32scatter_epi32): Ditto.
35345 (_mm256_mask_i32scatter_epi32): Ditto.
35346 (_mm_i32scatter_epi32): Ditto.
35347 (_mm_mask_i32scatter_epi32): Ditto.
35348 (_mm256_i32scatter_epi64): Ditto.
35349 (_mm256_mask_i32scatter_epi64): Ditto.
35350 (_mm_i32scatter_epi64): Ditto.
35351 (_mm_mask_i32scatter_epi64): Ditto.
35352 (_mm256_i64scatter_epi32): Ditto.
35353 (_mm256_mask_i64scatter_epi32): Ditto.
35354 (_mm_i64scatter_epi32): Ditto.
35355 (_mm_mask_i64scatter_epi32): Ditto.
35356 (_mm256_i64scatter_epi64): Ditto.
35357 (_mm256_mask_i64scatter_epi64): Ditto.
35358 (_mm_i64scatter_epi64): Ditto.
35359 (_mm_mask_i64scatter_epi64): Ditto.
35360 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35361 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35362 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35363 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35364 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35365 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35366 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35367 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35368 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35369 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35370 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35371 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35372 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35373 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35374 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35375 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35376 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35377 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35378 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35379 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35380 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35381 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35382 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35383 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35384 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35385 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35386 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35387 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35388 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35389 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35390 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35391 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35392 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35393 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35394 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35395 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35396 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35397 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35398 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35399 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35400 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35401 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35402 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35403 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35404 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35405 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35406 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35407 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35408 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35409 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35410 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35411 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35412 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35413 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35414 definitions accordingly.
35416 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35417 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35420 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35423 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35426 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35427 hard regs for inner regno.
35429 2017-01-17 Martin Liska <mliska@suse.cz>
35432 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35433 assumption and add comment.
35435 2017-01-17 Nathan Sidwell <nathan@acm.org>
35437 * ipa-visibility.c (localize_node): New function, broken out of ...
35438 (function_and_variable_visibility): ... here. Call it.
35440 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35442 PR middle-end/77445
35443 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35444 correctly set frequency of oudgoing edge.
35445 (duplicate_thread_path): Fix profile updating.
35447 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35450 * configure.ac: Add GCC_BASE_VER.
35451 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35452 version from BASE-VER file.
35453 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35454 (gcc.o): Depend on $(BASEVER).
35455 * common.opt (dumpfullversion): New option.
35456 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35457 * doc/invoke.texi: Document -dumpfullversion.
35458 * doc/install.texi: Document --with-gcc-major-version-only.
35459 * configure: Regenerated.
35461 2017-01-17 Richard Biener <rguenther@suse.de>
35463 PR tree-optimization/71433
35464 * tree-vrp.c (register_new_assert_for): Merge same asserts
35465 on all incoming edges.
35466 (process_assert_insertions_for): Handle insertions at the
35469 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35471 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35472 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35474 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35477 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35480 2017-01-17 Alan Modra <amodra@gmail.com>
35483 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35484 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35485 symbolic stack limit when pic.
35487 2017-01-16 Martin Sebor <msebor@redhat.com>
35489 PR tree-optimization/78608
35490 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35492 2017-01-16 Jeff Law <law@redhat.com>
35495 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35496 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35497 for several include directories that may be relative to sysroot.
35498 * config/i386/x-mingw32 (gplus_includedir): Define.
35499 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35500 (native_system_includedir): Likewise.
35501 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35502 override if TARGET_SYSTEM_ROOT is defined.
35503 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35505 PR tree-optimization/79090
35506 PR tree-optimization/33562
35507 PR tree-optimization/61912
35508 PR tree-optimization/77485
35509 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35510 and computed trims into the dump file.
35512 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35514 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35516 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35519 * gimplify.c (gimplify_init_constructor): If want_value and
35520 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35524 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35525 sequence. Formatting fixes.
35526 (doloop_optimize): Formatting fixes.
35529 * gcc.c (debug_level_greater_than_spec_func): New function.
35530 (static_spec_functions): Add debug-level-gt spec function.
35531 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35533 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35534 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35535 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35536 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35537 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35538 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35540 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35542 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35543 QImode fixups to general and mask registers only.
35545 2017-01-16 Carl Love <cel@us.ibm.com>
35547 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35548 for built-in functions
35549 vector signed char vec_nabs (vector signed char)
35550 vector signed short vec_nabs (vector signed short)
35551 vector signed int vec_nabs (vector signed int)
35552 vector signed long long vec_nabs (vector signed long long)
35553 vector float vec_nabs (vector float)
35554 vector double vec_nabs (vector double)
35555 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35557 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35558 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35559 * doc/extend.texi: Update the documentation file for the new built-in
35562 2017-01-16 Martin Sebor <msebor@redhat.com>
35564 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35567 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35569 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35570 UNSPEC_VSX__XXSPLTD to require special splat handling.
35572 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35575 * system.h: Poison strndup.
35577 2017-01-16 Alan Modra <amodra@gmail.com>
35580 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35583 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35585 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35587 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35589 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35590 call recog here. Assert that INSN_CODE (insn) is non-negative.
35592 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35595 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35597 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35598 in the currently scheduled RTL fragment.
35600 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35602 PR rtl-optimization/78751
35603 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35606 2017-01-14 Jeff Law <law@redhat.com>
35608 PR tree-optimization/79090
35609 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35610 variable length stores.
35611 (compute_trims): Delete dead assignment to *trim_tail.
35612 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35615 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35617 PR rtl-optimization/78626
35618 PR rtl-optimization/78727
35619 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35620 of a block, and split such blocks after everything else is finished.
35622 2017-01-14 Alan Modra <amodra@gmail.com>
35625 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35626 target legitimate_combined_insn.
35627 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35628 (rs6000_legitimate_combined_insn): New function.
35629 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35631 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35632 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35633 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35635 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35637 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35639 2017-01-13 Jeff Law <law@redhat.com>
35641 PR tree-optimization/33562
35642 PR tree-optimization/61912
35643 PR tree-optimization/77485
35644 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35646 (delete_dead_assignment): Likewise.
35647 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35648 statement to delete_dead_call and delete_dead_assignment.
35650 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35653 * substring-locations.c (format_warning_va): Strengthen case 1 so
35654 that both endpoints of the substring must be within the format
35655 range for just the substring to be printed.
35657 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35659 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35660 * config/i386/i386.c (ix86_target_string): Add missing options
35661 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35662 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35663 flags_other and ix86_target_other to flags2_other. Display unknown
35665 (ix86_valid_target_attribute_inner_p): Add missing options and
35666 reorder options by implied ISAs, as in ix86_target_string.
35668 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35670 * hash-table.h (hash_table::too_empty_p): New function.
35671 (hash_table::expand): Use it.
35672 (hash_table::traverse): Likewise.
35673 (hash_table::empty_slot): Use sizeof (value_type) instead of
35674 sizeof (PTR) to convert bytes to elements. Shrink the table
35675 if the current size is excessive for the current number of
35678 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35680 * ira-costs.c (record_reg_classes): Break from the inner loop
35681 early once alt_fail is known to be true. Update outer loop
35682 handling accordingly.
35684 2017-01-13 Jeff Law <law@redhat.com>
35686 * tree-ssa-dse.c (decrement_count): New function.
35687 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35688 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35689 when we know the partially dead statement is a mem* function.
35691 PR tree-optimization/61912
35692 PR tree-optimization/77485
35693 * tree-ssa-dse.c: Include expr.h.
35694 (maybe_trim_constructor_store): New function.
35695 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35697 PR tree-optimization/33562
35698 PR tree-optimization/61912
35699 PR tree-optimization/77485
35700 * doc/invoke.texi: Document new dse-max-object-size param.
35701 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35702 * tree-ssa-dse.c: Include params.h.
35703 (dse_store_status): New enum.
35704 (initialize_ao_ref_for_dse): New, partially extracted from
35706 (valid_ao_ref_for_dse, normalize_ref): New.
35707 (setup_live_bytes_from_ref, compute_trims): Likewise.
35708 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35709 (maybe_trim_partially_dead_store): Likewise.
35710 (maybe_trim_complex_store): Likewise.
35711 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35712 Track what bytes live from the original store. Return tri-state
35713 for dead, partially dead or live.
35714 (dse_dom_walker): Add constructor, destructor and new private members.
35715 (delete_dead_call, delete_dead_assignment): New extracted from
35717 (dse_optimize_stmt): Make a member of dse_dom_walker.
35718 Use initialize_ao_ref_for_dse.
35720 PR tree-optimization/33562
35721 PR tree-optimization/61912
35722 PR tree-optimization/77485
35723 * sbitmap.h (bitmap_count_bits): Prototype.
35724 (bitmap_clear_range, bitmap_set_range): Likewise.
35725 * sbitmap.c (bitmap_clear_range): New function.
35726 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35728 2017-01-13 Martin Liska <mliska@suse.cz>
35731 * function.c (set_cfun): Add new argument force.
35732 * function.h (set_cfun): Likewise.
35733 * ipa-inline-transform.c (inline_call): Use the function when
35734 strict alising from is dropped for function we inline to.
35736 2017-01-13 Richard Biener <rguenther@suse.de>
35738 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35739 for dumping GIMPLE INTEGER_CSTs.
35741 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35743 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35744 to 201112L since C++17.
35746 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35749 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35750 if -fsanitize=kernel-address is present.
35752 2017-01-13 Richard Biener <rguenther@suse.de>
35754 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35755 as _Literal ( type ) number in case usual suffixes do not
35756 preserve all information.
35758 2017-01-13 Richard Biener <rguenther@suse.de>
35760 PR tree-optimization/77283
35761 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35762 and ssa-iterators.h.
35763 (is_feasible_trace): Implement a cost model based on joiner
35766 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35769 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35770 char or short to __float128/_Float128 directly.
35772 2017-01-12 Martin Sebor <msebor@redhat.com>
35774 to -Wformat-overflow.
35775 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35776 (min_bytes_remaining): Same.
35777 (get_string_length): Same.
35778 (format_string): Same.
35779 (format_directive): Same.
35781 (pass_sprintf_length::handle_gimple_call): Same.
35783 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35785 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35786 info.nowrite calls with no lhs that can't throw. Return bool
35787 whether gsi_remove has been called or not.
35788 (pass_sprintf_length::handle_gimple_call): Return bool whether
35789 try_substitute_return_value called gsi_remove. Formatting fix.
35790 (pass_sprintf_length::execute): Don't use gsi_remove if
35791 handle_gimple_call returned true.
35794 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35795 be removed due to side-effects, don't remove following barrier nor
35796 turn the successor edge into fallthru edge.
35798 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35801 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35802 element-reversing loads and stores as not swappable.
35804 2017-01-12 Nathan Sidwell <nathan@acm.org>
35805 Nicolai Stange <nicstange@gmail.com>
35807 * combine.c (try_combine): Don't ignore result of overlap checking
35808 loop. Combine overlap & asm check into single loop.
35810 2017-01-12 Richard Biener <rguenther@suse.de>
35812 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35813 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35815 2017-01-12 Richard Biener <rguenther@suse.de>
35817 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35818 and TS_TARGET_OPTION directly derive from TS_BASE.
35819 * tree-core.h (tree_optimization_option): Derive from tree_base.
35820 (tree_target_option): Likewise.
35822 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35824 * config/i386/i386.c (memory_address_length): Increase len
35825 only when rip_relative_addr_p returns false.
35827 2017-01-11 Julia Koval <julia.koval@intel.com>
35829 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35830 (OPTION_MASK_ISA_SGX_SET): New.
35831 (ix86_handle_option): Handle OPT_msgx.
35832 * config.gcc: Added sgxintrin.h.
35833 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35834 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35835 * config/i386/i386.c (ix86_target_string): Add -msgx.
35837 (ix86_option_override_internal): Handle new options.
35838 (ix86_valid_target_attribute_inner_p): Add sgx.
35839 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35840 * config/i386/i386.opt: Add msgx.
35841 * config/i386/sgxintrin.h: New file.
35842 * config/i386/x86intrin.h: Add sgxintrin.h.
35844 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35847 * fold-const.c (maybe_nonzero_address): Return 1 for function
35849 (tree_single_nonzero_warnv_p): Don't handle function local objects
35853 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35856 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35859 * opts.c: Include "spellcheck.h"
35860 (struct string_fragment): New struct.
35861 (struct edit_distance_traits<const string_fragment &>): New
35863 (get_closest_sanitizer_option): New function.
35864 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35866 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35868 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35870 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35871 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35872 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35873 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35874 for initial die_offset if dwarf_split_debug_info.
35875 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35876 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35878 (output_skeleton_debug_sections): Formatting fix. Use
35879 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35880 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35882 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35884 * config/arm/cortex-a53.md: Add bypasses for
35885 cortex_a53_r2f_cvt.
35886 (cortex_a53_r2f): Only use for transfers.
35887 (cortex_a53_f2r): Likewise.
35888 (cortex_a53_r2f_cvt): Add reservation for conversions.
35889 (cortex_a53_f2r_cvt): Likewise.
35891 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35893 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35894 to all inlined functions, change static to extern.
35896 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35899 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35901 (arm_assemble_integer): Likewise.
35903 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35905 * config.gcc: Use new awk script to check CPU, FPU and architecture
35906 parameters for --with-... options.
35907 * config/arm/parsecpu.awk: New file
35908 * config/arm/arm-cpus.in: New file.
35909 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35911 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35913 * config/arm/t-arm: Update dependency rules.
35914 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35915 of processing .def files.
35916 * config/arm/genopt.sh: Deleted.
35917 * config/arm/gentune.sh: Deleted.
35918 * config/arm/arm-cores.def: Deleted.
35919 * config/arm/arm-arches.def: Deleted.
35920 * config/arm/arm-fpus.def: Deleted.
35921 * config/arm/arm-tune.md: Regenerated.
35922 * config/arm/arm-tables.opt: Regenerated.
35923 * config/arm/arm-cpu.h: New generated file.
35924 * config/arm/arm-cpu-data.h: New generated file.
35925 * config/arm/arm-cpu-cdata.h: New generated file.
35927 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35930 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35932 (input_varpool_node): Unpack dynamically_initialized bit.
35934 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35936 PR rtl-optimization/79032
35937 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35938 the alignment of the adjusted memory reference against that of MODE,
35939 instead of the alignment of the original memory reference.
35941 2017-01-11 Martin Jambor <mjambor@suse.cz>
35943 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35945 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35946 decorated functions.
35948 2017-01-11 Richard Biener <rguenther@suse.de>
35950 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35951 set range/nonnull info for PHI results. Do not set it on
35952 stmts marked for removal.
35954 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35956 * expr.c (store_field): In the bitfield case, fetch the return value
35957 from the registers before applying a single big-endian adjustment.
35958 Always do a final load for a BLKmode value not larger than a word.
35960 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35963 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35964 that we correctly handle column numbers greater than
35965 LINE_MAP_MAX_COLUMN_NUMBER.
35967 2017-01-10 Martin Sebor <msebor@redhat.com>
35969 PR middle-end/78245
35970 * gimple-ssa-sprintf.c (get_destination_size): Call
35971 {init,fini}object_sizes.
35972 * tree-object-size.c (addr_object_size): Adjust.
35973 (pass_through_call): Adjust.
35974 (pass_object_sizes::execute): Adjust.
35975 * tree-object-size.h (fini_object_sizes): Declare.
35977 2017-01-10 Martin Sebor <msebor@redhat.com>
35979 PR tree-optimization/78775
35980 * builtins.c (get_size_range): Move...
35981 * calls.c: ...to here.
35982 (alloc_max_size): Accept zero argument.
35983 (operand_signed_p): Remove.
35984 (maybe_warn_alloc_args_overflow): Call get_size_range.
35985 * calls.h (get_size_range): Declare.
35987 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35989 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35990 from TI's devices.csv file as of September 2016.
35991 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35993 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35995 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35996 * doc/invoke.texi: Likewise.
35997 * doc/md.texi: Likewise.
35998 * doc/objc.texi: Likewise.
36000 2017-01-10 Joshua Conner <joshconner@google.com>
36002 * config/arm/fuchsia-elf.h: New file.
36003 * config/fuchsia.h: New file.
36004 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36005 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36007 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36009 2016-01-10 Richard Biener <rguenther@suse.de>
36011 PR tree-optimization/79034
36012 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36013 Propagate out degenerate PHIs in the joiner.
36015 2017-01-10 Martin Liska <mliska@suse.cz>
36017 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36018 (sort_congruence_classes_by_decl_uid): Likewise.
36019 (sort_congruence_class_groups_by_decl_uid): Likewise.
36020 (sem_item_optimizer::merge_classes): Sort class, groups in these
36021 classes and members in the groups by DECL_UID of declarations.
36022 This would make merge operations stable.
36024 2017-01-10 Martin Liska <mliska@suse.cz>
36026 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36027 usage of m_classes_vec.
36028 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36029 (sem_item_optimizer::get_group_by_hash): Likewise.
36030 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36031 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36032 (sem_item_optimizer::verify_classes): Likewise.
36033 (sem_item_optimizer::process_cong_reduction): Likewise.
36034 (sem_item_optimizer::dump_cong_classes): Likewise.
36035 (sem_item_optimizer::merge_classes): Likewise.
36036 * ipa-icf.h (congruence_class_hash): Rename from
36037 congruence_class_group_hash. Remove declaration of m_classes_vec.
36039 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36041 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36042 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36043 * config.gcc: Add avx512vpopcntdqintrin.h.
36044 * config/i386/avx512vpopcntdqintrin.h: New.
36045 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36046 * config/i386/i386-builtin-types.def: Add new types.
36047 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36048 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36049 __builtin_ia32_vpopcountq_v8di_mask): New.
36050 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36051 __AVX512VPOPCNTDQ__.
36052 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36053 (PTA_AVX512VPOPCNTDQ): Define.
36054 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36055 TARGET_AVX512VPOPCNTDQ_P): Define.
36056 * config/i386/i386.opt: Add mavx512vpopcntdq.
36057 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36058 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36060 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36062 PR middle-end/77484
36063 * predict.def (PRED_CALL): Set to 67.
36065 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36067 * expr.c (store_field): In the bitfield case, if the value comes from
36068 a function call and is of an aggregate type returned in registers, do
36069 not modify the field mode; extract the value in all cases if the mode
36070 is BLKmode and the size is not larger than a word.
36072 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36075 * config/i386/cpuid.h: Fix undefined behavior.
36077 2017-01-04 Jeff Law <law@redhat.com>
36079 PR tree-optimization/79007
36080 PR tree-optimization/67955
36081 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36082 conservative for pt.null when flag_non_call_exceptions is on.
36084 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36086 PR translation/79019
36087 PR translation/79020
36088 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36089 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36090 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36092 * config/avr/avr.opt (maccumulate-args): Likewise.
36093 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36094 * common.opt (freport-bug): Likewise.
36095 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36096 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36097 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36098 translatable string.
36099 * config/i386/i386.c (function_value_32): Likewise.
36100 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36101 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36103 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36104 * common/config/msp430/msp430-common.c (msp430_handle_option):
36106 * symtab.c (symtab_node::verify_base): Likewise.
36107 * opts.c (set_debug_level): Likewise.
36108 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36109 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36110 missing whitespace to translatable strings.
36111 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36112 * config/sh/superh.h: Likewise.
36113 * config/i386/xopintrin.h: Likewise.
36114 * config/i386/znver1.md: Likewise.
36115 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36116 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36117 * double-int.h (struct double_int): Likewise.
36118 * double-int.c (div_and_round_double): Likewise.
36119 * wide-int.cc: Likewise.
36120 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36121 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36122 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36123 (crossjumps_occurred): ... this.
36124 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36127 PR tree-optimization/78899
36128 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36129 returning bool return struct loop *, NULL for failure and the new
36131 (versionable_outer_loop_p): Don't version outer loop if it has
36132 dont_vectorized bit set.
36133 (tree_if_conversion): When versioning outer loop, ensure
36134 tree_if_conversion is performed also on the inner loop of the
36135 non-vectorizable outer loop copy.
36136 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36137 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36138 prevent vectorization of it.
36139 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36140 the outer loop vectorization of the non-scalar version is attempted
36141 before vectorization of the inner loop in scalar version. If
36142 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36143 vectorization of its inner loop.
36144 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36145 has 2 inner loops, rename also on edges from bb whose single pred
36146 is outer_loop->header. Fix typo in function comment.
36148 2017-01-09 Martin Sebor <msebor@redhat.com>
36151 * asan.c (asan_emit_stack_protection): Increase local buffer size
36152 to avoid snprintf truncation warning.
36154 2017-01-09 Andrew Pinski <apinski@cavium.com>
36156 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36157 to reference thunderx2t99 for the tuning structure
36158 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36160 (thunderx2t99_extra_costs): This.
36161 * config/aarch64/aarch64-tune.md: Regenerate.
36162 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36163 (vulcan_addrcost_table): This.
36164 (vulcan_regmove_cost): Rename to ...
36165 (thunderx2t99_regmove_cost): This.
36166 (vulcan_vector_cost): Rename to ...
36167 (thunderx2t99_vector_cost): this.
36168 (vulcan_branch_cost): Rename to ...
36169 (thunderx2t99_branch_cost): This.
36170 (vulcan_tunings): Rename to ...
36171 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36172 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36174 2017-01-09 Martin Jambor <mjambor@suse.cz>
36178 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36179 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36180 (propagate_vr_accross_jump_function): Use the above function for all
36181 value range computations for pass-through jump functions and type
36182 converasion from explicit value range values.
36183 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36184 parameters from TYPE_ARG_TYPES.
36185 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36186 (ipa_write_node_info): Stream type of the actual argument.
36187 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36189 2017-01-09 Martin Liska <mliska@suse.cz>
36192 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36193 (lookup_compiler): Do not show error message with have_E.
36195 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36197 PR tree-optimization/78938
36198 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36199 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36200 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36201 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36204 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36206 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36209 2017-01-09 Richard Biener <rguenther@suse.de>
36211 PR tree-optimization/78997
36212 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36213 name condition properly.
36215 2017-01-09 Richard Biener <rguenther@suse.de>
36218 * dwarf2out.c (is_cxx): New overload with context.
36219 (is_naming_typedef_decl): Use it.
36221 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36223 * invoke.texi (Option Summary): Correct spacing in option lists
36224 and add line breaks to fix over-long lines.
36226 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36228 PR middle-end/17660
36230 * extend.texi (Common Variable Attributes): Add xref to GCC
36231 Internals manual to explain mode attribute keywords.
36233 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36236 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36237 and Preprocessor Options.
36238 (Options for Linking): Document -pthread here....
36239 (RS/6000 and PowerPC Options): ...not here.
36240 (Solaris 2 Options): ...or here.
36241 * doc/cppopts.texi: Document -pthread.
36243 2017-01-08 Martin Sebor <msebor@redhat.com>
36245 PR middle-end/77708
36246 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36247 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36248 New member functions.
36249 (format_directive): Used them.
36251 (pass_sprintf_length::handle_gimple_call): Same.
36252 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36253 to avoid truncation for any argument.
36254 (extract_affine_mul): Same.
36255 * tree.c (get_file_function_name): Same.
36257 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36259 PR middle-end/77484
36260 * predict.def (PRED_INDIR_CALL): Set to 86.
36262 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36264 PR preprocessor/54124
36265 * doc/cppopts.texi: Reformat -d subtable to list the full name
36266 of the options. Add cross-reference to the docs for the general
36267 compiler -d options.
36268 * doc/invoke.texi (Developer Options): Add cross-reference to the
36269 preprocessor-specific -d option documentation.
36271 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36273 PR preprocessor/13498
36274 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36275 redudant material, and reflect new command-line options.
36276 (System Headers): Likewise.
36278 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36280 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36281 -isystem, and -idirafter. Copy-edit.
36282 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36283 default for -ftrack-macro-expansion. Delete obsolete and
36284 badly-formatted implementation details about -fdebug-cpp output.
36285 * doc/cppwarnopts.texi: Copy-edit.
36287 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36290 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36291 that the transition from a max line width >= 1<<10 to narrower
36292 lines works correctly.
36294 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36296 * doc/options.texi (PerFunction): New.
36297 * opt-functions.awk (switch_flags): Map both Optimization and
36298 PerFunction to CL_OPTIMIZATION.
36299 * opth-gen.awk: Test for PerFunction flag along with
36301 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36302 it only when the latter is present. Skip those that don't in
36303 the hash function generator.
36304 * common.opt (fvar-tracking): Mark as PerFunction instead of
36306 (fvar-tracking-assignments): Likewise.
36307 (fvar-tracking-assignments-toggle): Likewise.
36308 (fvar-tracking-uninit): Likewise.
36310 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36312 PR translation/79018
36313 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36316 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36319 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36320 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36321 TARGET_LONG_JUMP_TABLE_OFFSETS.
36322 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36323 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36324 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36325 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36326 * config/m68k/m68k.md (tablejump expander): Likewise.
36327 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36328 TARGET_LONG_JUMP_TABLE_OFFSETS.
36329 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36330 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36332 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36333 David Holsgrove <david.holsgrove@xilinx.com>
36335 * common/config/microblaze/microblaze-common.c
36336 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36337 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36339 * config/microblaze/microblaze.c (microblaze_must_save_register)
36340 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36342 (microblaze_eh_return): New function.
36343 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36344 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36345 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36346 * config/microblaze/microblaze.md (eh_return): New pattern.
36348 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36350 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36351 GCC_DIAGNOSTIC_STRINGIFY): Define.
36353 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36355 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36357 * config/arm/arm.md (<mcrr>): New.
36359 * config/arm/arm.c (arm_arch5te): New.
36360 (arm_option_override): Set arm_arch5te.
36361 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36363 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36364 (arm_mcrr_qualifiers): ... this. New.
36365 (MRRC_QUALIFIERS): Define to...
36366 (arm_mrrc_qualifiers): ... this. New.
36367 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36369 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36370 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36371 (MRRCI, mrrc, MRRC): New.
36372 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36373 VUNSPEC_MRRC2): New.
36375 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36377 * config/arm/arm.md (<mcr>): New.
36379 * config/arm/arm.c (arm_coproc_builtin_available): Add
36380 support for mcr, mrc, mcr2 and mrc2.
36381 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36382 (arm_mcr_qualifiers): ... this. New.
36383 (MRC_QUALIFIERS): Define to ...
36384 (arm_mrc_qualifiers): ... this. New.
36385 (MCR_QUALIFIERS): Define to ...
36386 (arm_mcr_qualifiers): ... this. New.
36387 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36389 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36390 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36391 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36392 VUNSPEC_MRC2): New.
36394 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36396 * config/arm/arm.md (*ldc): New.
36400 * config/arm/arm.c (arm_coproc_builtin_available): Add
36401 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36402 (arm_coproc_ldc_stc_legitimate_address): New.
36403 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36404 'qualifier_const_pointer'.
36405 (LDC_QUALIFIERS): Define to...
36406 (arm_ldc_qualifiers): ... this. New.
36407 (STC_QUALIFIERS): Define to...
36408 (arm_stc_qualifiers): ... this. New.
36409 * config/arm/arm-protos.h
36410 (arm_coproc_ldc_stc_legitimate_address): New.
36411 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36412 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36413 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36414 stc2, stcl, stc2l): New.
36415 * config/arm/constraints.md (Uz): New.
36416 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36417 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36418 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36419 VUNSPEC_STC2L): New.
36421 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36423 * config/arm/arm.md (<cdp>): New.
36424 * config/arm/arm.c (neon_const_bounds): Rename this ...
36425 (arm_const_bounds): ... this.
36426 (arm_coproc_builtin_available): New.
36427 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36428 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36429 (CDP_QUALIFIERS): Define to...
36430 (arm_cdp_qualifiers): ... this. New.
36432 (arm_expand_builtin_args): Add case for 6 arguments.
36433 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36434 (arm_const_bounds): ... this.
36435 (arm_coproc_builtin_available): New.
36436 * config/arm/arm_acle.h (__arm_cdp): New.
36438 * config/arm/arm_acle_builtins.def (cdp): New.
36440 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36441 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36442 'arm_const_bounds'.
36443 * config/arm/types.md (coproc): New.
36444 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36445 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36446 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36447 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36449 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36451 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36452 (UBINOP_QUALIFIERS): New.
36454 (acle_builtin_data): New. Change comment.
36455 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36456 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36457 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36458 arm_acle_builtins.def.
36459 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36460 (arm_init_acle_builtins): New.
36461 (CRC32_BUILTIN): Remove.
36462 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36463 crc32cb, crc32ch and crc32cw.
36464 (arm_init_crc32_builtins): Remove.
36465 (arm_init_builtins): Use arm_init_acle_builtins rather
36466 than arm_init_crc32_builtins.
36467 (arm_expand_acle_builtin): New.
36468 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36469 * config/arm/arm_acle_builtins.def: New.
36471 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36473 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36474 (arm_builtin_datum): ... this.
36475 (arm_init_neon_builtin): Rename to ...
36476 (arm_init_builtin): ... this. Add a new parameters PREFIX
36477 and USE_SIG_IN_NAME.
36478 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36479 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36480 'arm_builtin_datum'.
36481 (arm_init_vfp_builtins): Likewise.
36482 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36483 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36484 (arm_expand_neon_args): Rename to ...
36485 (arm_expand_builtin_args): ... this. Rename builtin_arg
36486 enum values and differentiate between ARG_BUILTIN_MEMORY
36487 and ARG_BUILTIN_NEON_MEMORY.
36488 (arm_expand_neon_builtin_1): Rename to ...
36489 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36490 values, arm_expand_builtin_args and add bool parameter NEON.
36491 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36492 (arm_expand_vfp_builtin): Likewise.
36493 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36495 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36497 PR middle-end/77484
36498 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36499 * predict.c (tree_estimate_probability_bb): Reverse direction of
36500 polymorphic call predictor.
36502 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36504 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36505 (determine_pass_name_match): ...this new function and...
36506 (should_skip_pass_p): ...this new function.
36508 2017-01-06 Nathan Sidwell <nathan@acm.org>
36510 * ipa-visibility.c (function_and_variable_visibility): Reformat
36511 comments and long lines. Remove extrneous if.
36512 * symtab.c (symtab_node::make_decl_local): Fix code format.
36513 (symtab_node::set_section_for_node): Fix comment typo.
36515 2017-01-06 Martin Liska <mliska@suse.cz>
36518 * lra-constraints.c: Rename invariant to lra_invariant.
36519 * predict.c (set_even_probabilities): Initialize e to NULL.
36521 2017-01-05 Martin Sebor <msebor@redhat.com>
36523 PR tree-optimization/78910
36524 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36525 (format_integer): Correct off-by-one error in the handling
36526 of precision with negative numbers in signed conversions..
36528 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36530 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36532 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36534 PR tree-optimization/71016
36535 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36536 factor_out_conditional_conversion. Formatting fix.
36537 (factor_out_conditional_conversion): Add cond_stmt argument.
36538 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36539 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36542 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36544 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36545 read-rtl-function.o, and selftest-rtl.o.
36546 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36547 (selftest::aarch64_test_loading_full_dump): New function.
36548 (selftest::aarch64_run_selftests): New function.
36549 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36550 selftest::aarch64_run_selftests.
36551 * config/i386/i386.c
36552 (selftest::ix86_test_loading_dump_fragment_1): New function.
36553 (selftest::ix86_test_loading_call_insn): New function.
36554 (selftest::ix86_test_loading_full_dump): New function.
36555 (selftest::ix86_test_loading_unspec): New function.
36556 (selftest::ix86_run_selftests): Call the new functions.
36557 * emit-rtl.c (maybe_set_max_label_num): New function.
36558 * emit-rtl.h (maybe_set_max_label_num): New decl.
36559 * function.c (instantiate_decls): Guard call to
36560 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36561 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36563 * gensupport.c (gen_reader::gen_reader): Pass "false"
36564 for new "compact" param of rtx_reader.
36565 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36566 rather than an empty string for NULL strings.
36567 * read-md.c: Potentially include config.h rather than bconfig.h.
36568 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36569 (have_error): New global, copied from errors.c.
36570 (md_reader::read_name): Rename to...
36571 (md_reader::read_name_1): ...this, adding "out_loc" param,
36572 and converting "missing name or number" to returning false, rather
36574 (md_reader::read_name): Reimplement in terms of read_name_1.
36575 (md_reader::read_name_or_nil): New function.
36576 (md_reader::read_string): Handle "(nil)" by returning NULL.
36577 (md_reader::md_reader): Add new param "compact".
36578 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36579 (md_reader::read_file): New method.
36580 * read-md.h (md_reader::md_reader): Add new param "compact".
36581 (md_reader::read_file): New method.
36582 (md_reader::is_compact): New accessor.
36583 (md_reader::read_name): Convert return type from void to file_location.
36584 (md_reader::read_name_or_nil): New decl.
36585 (md_reader::read_name_1): New decl.
36586 (md_reader::m_compact): New field.
36587 (noop_reader::noop_reader): Pass "false" for new "compact" param
36589 (rtx_reader::rtx_reader): Add new "compact" param.
36590 (rtx_reader::read_rtx_operand): Make virtual and convert return
36591 type from void to rtx.
36592 (rtx_reader::read_until): New decl.
36593 (rtx_reader::handle_any_trailing_information): New virtual function.
36594 (rtx_reader::postprocess): New virtual function.
36595 (rtx_reader::finalize_string): New virtual function.
36596 (rtx_reader::m_in_call_function_usage): New field.
36597 (rtx_reader::m_reuse_rtx_by_id): New field.
36598 * read-rtl-function.c: New file.
36599 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36600 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36601 (selftest::verify_three_block_rtl_cfg): New decl.
36602 * read-rtl-function.h: New file.
36603 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36604 For host, include function.h, memmodel.h, and emit-rtl.h.
36605 (one_time_initialization): New function.
36606 (struct compact_insn_name): New struct.
36607 (compact_insn_names): New array.
36608 (find_code): Handle insn codes in compact dumps.
36609 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36610 (bind_subst_iter_and_attr): Likewise.
36611 (add_condition_to_string): Likewise.
36612 (add_condition_to_rtx): Likewise.
36613 (apply_attribute_uses): Likewise.
36614 (add_current_iterators): Likewise.
36615 (apply_iterators): Likewise.
36616 (initialize_iterators): Guard usage of apply_subst_iterator with
36617 #ifdef GENERATOR_FILE.
36618 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36619 (md_reader::read_mapping): Likewise.
36620 (add_define_attr_for_define_subst): Likewise.
36621 (add_define_subst_attr): Likewise.
36622 (read_subst_mapping): Likewise.
36623 (check_code_iterator): Likewise.
36624 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36626 (one_time_initialization): New function.
36627 (rtx_reader::read_until): New method.
36628 (read_flags): New function.
36629 (parse_reg_note_name): New function.
36630 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36631 Handle reuse_rtx ids.
36632 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36633 Add parsing support for RTL dumps, mirroring the special-cases in
36634 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36635 values, and calling handle_any_trailing_information.
36636 (rtx_reader::read_rtx_operand): Convert return type from void
36637 to rtx, returning return_rtx. Handle case 'e'. Call
36638 finalize_string on XSTR and XTMPL fields.
36639 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36640 "(nil)" values were omitted. Call the postprocess vfunc on the
36642 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36643 class ctor. Initialize m_in_call_function_usage. Call
36644 one_time_initialization.
36645 * rtl-tests.c (selftest::test_uncond_jump): Call
36646 set_new_first_and_last_insn.
36647 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36648 * selftest-rtl.c: New file.
36649 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36650 (selftest::get_insn_by_uid): New decl.
36651 * selftest-run-tests.c (selftest::run_tests): Call
36652 read_rtl_function_c_tests.
36653 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36654 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36657 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36659 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36660 operands in a special way. Assert that pos+len <= mode precision.
36662 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36664 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36665 3 argument Alias with unlimited for the negative form.
36666 (fno-vect-cost-model): Removed.
36668 2017-01-05 Martin Liska <mliska@suse.cz>
36670 * hsa-gen.c (gen_hsa_divmod): New function.
36671 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36673 2017-01-05 Martin Liska <mliska@suse.cz>
36676 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36679 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36681 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36682 small constant length operands.
36684 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36686 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36687 between loop iterations.
36689 2017-01-05 Martin Liska <mliska@suse.cz>
36692 * gimplify.c (gimplify_decl_expr): Compare to
36693 asan_poisoned_variables instread of checking flags.
36694 (gimplify_target_expr): Likewise.
36695 (gimplify_expr): Likewise.
36696 (gimplify_function_tree): Conditionally initialize
36697 asan_poisoned_variables.
36699 2017-01-04 Jeff Law <law@redhat.com>
36701 PR tree-optimizatin/78812
36702 * rtl.h (contains_mem_rtx_p): Prototype.
36703 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36704 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36705 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36706 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36707 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36709 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36711 * input.c (assert_char_at_range): Default-initialize actual_range.
36713 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36715 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36716 to match the caller.
36718 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36720 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36721 insns after final jump in test to emit dummy move.
36723 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36725 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36726 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36728 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36730 * multiple_target.c (create_dispatcher_calls): Init e_next.
36731 * tree-ssa-loop-split.c (split_loop): Init border.
36732 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36735 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36740 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36741 (altivec_register_operand): Do not return true if the operand
36742 contains a SUBREG mixing SImode and SFmode.
36743 (vsx_register_operand): Likewise.
36744 (vsx_reg_sfsubreg_ok): New predicate.
36745 (vfloat_operand): Do not return true if the operand contains a
36746 SUBREG mixing SImode and SFmode.
36747 (vint_operand): Likewise.
36748 (vlogical_operand): Likewise.
36749 (gpc_reg_operand): Likewise.
36750 (int_reg_operand): Likewise.
36751 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36752 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36753 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36755 (rs6000_emit_move_si_sf_subreg): New helper function.
36756 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36757 fixup SUBREGs involving SImode and SFmode.
36758 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36759 numbers for the new peephole2 optimization.
36760 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36761 the GLIBC math library that do AND/IOR/XOR operations on single
36762 precision floating point.
36763 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36764 target macros to say whether we need to avoid SUBREGs mixing
36766 (TARGET_ALLOW_SF_SUBREG): Likewise.
36767 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36768 (UNSPEC_SI_FROM_SF): Likewise.
36769 (iorxor): Change spacing.
36770 (and_ior_xor): New iterator for AND, IOR, and XOR.
36771 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36772 (movdi_from_sf_zero_ext): Likewise.
36773 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36774 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36775 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36776 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36777 (fms<mode>4): Likewise.
36778 (fnma<mode>4): Likewise.
36779 (fnms<mode>4): Likewise.
36780 (nfma<mode>4): Likewise.
36781 (nfms<mode>4): Likewise.
36783 2017-01-04 Marek Polacek <polacek@redhat.com>
36786 * doc/invoke.texi: Document -Wpointer-compare.
36788 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36790 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36793 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36794 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36797 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36800 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36801 documentation of the powerpc_popcntb_ok attribute.
36802 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36803 code to issue warning messages if a requested CPU configuration is
36804 not supported by the binary (assembler and loader) toolchain.
36805 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36806 made to define a built-in function that has been disabled.
36807 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36808 made to define a built-in function that has been disabled.
36809 (altivec_init_builtins): Add comment explaining why definition
36810 of the DST built-in functions is not preceded by an assertion
36811 check. Add assertions to prevent ICE if attempts are made to
36812 define an altivec predicate or an abs* built-in function that has
36814 (htm_init_builtins): Add comment explaining why definition of the
36815 htm built-in functions is not preceded by an assertion check.
36817 2017-01-04 Jeff Law <law@redhat.com>
36819 PR tree-optimizatin/67955
36820 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36821 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36822 the points-to solution does not include pt_null. Use DECL_PT_UID
36825 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36827 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36828 Use gen_int_mode instead of gen_lopwart for const_int operands.
36830 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36832 PR tree-optimization/71563
36833 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36834 out of range value - has low bits known to be zero.
36836 2017-01-04 Alan Modra <amodra@gmail.com>
36838 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36839 * configure: Regenerate.
36840 * config.in: Regenerate.
36842 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36845 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36846 a substring of the message, but strcmp with the whole message. Ifdef
36847 ENABLE_NLS, translate the message first using dgettext.
36849 2017-01-03 Jeff Law <law@redhat.com>
36851 PR tree-optimizatin/78856
36852 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36853 (mark_threaded_blocks): Remove code to truncate thread paths that
36854 cross multiple loop headers. Instead invalidate the cached loop
36855 iteration information and handle case of a thread path walking
36856 into an irreducible region.
36858 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36861 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36862 assertions. Add support for doing the signbit if the IEEE 128-bit
36863 floating point value is in a GPR.
36864 * config/rs6000/rs6000.md (Fsignbit): Delete.
36865 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36866 Update the length attribute if the value is in a GPR.
36867 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36868 the sign or zero extension instruction, since the value is always 0/1.
36869 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36872 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36873 extracting SImode to a GPR register so that we can generate a
36874 store, limit the vector to be in a traditional Altivec register
36875 for the vextuwrx instruction.
36877 2017-01-03 Ian Lance Taylor <iant@google.com>
36879 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36881 2017-01-03 Martin Sebor <msebor@redhat.com>
36883 PR tree-optimization/78696
36884 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36885 precision. Use MPFR for %f for greater fidelity. Correct handling
36887 (pass_sprintf_length::compute_format_length): Set width and precision
36888 specified by asrerisk to void_node for vararg functions.
36889 (try_substitute_return_value): Adjust dump output.
36891 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36893 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36895 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36897 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36898 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36899 -mlra/-mno-lra was passed to the compiler.
36901 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36903 PR rtl-optimization/65618
36904 * emit-rtl.c (try_split): Move initialization of "before" and
36905 "after" to just before the call to emit_insn_after_setloc.
36907 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36909 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36911 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36913 * dwarf2out.c (gen_enumeration_type_die): When
36914 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36916 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36918 PR tree-optimization/78965
36919 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36920 Change first argument from const call_info & to call_info &. For %n
36921 set info.nowrite to false.
36923 PR middle-end/78901
36924 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36925 possibly throwing calls.
36927 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36928 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36929 and fns handling, rather than in a separate case SSA_NAME.
36931 2017-01-02 Jeff Law <law@redhat.com>
36933 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36934 fixes for first_period and second_period variables.
36936 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36939 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36940 (*insvqi_1): New insn pattern.
36941 (*insvqi_1_mem_rex64): Ditto.
36942 (*insvqi_2): Ditto.
36943 (*insvqi_3): Rename from *insvqi.
36945 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36947 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36949 * doc/cfg.texi (Edges): Remove reference to Java.
36950 (Maintaining the CFG): Ditto.
36952 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36954 PR middle-end/77674
36955 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36956 transparent aliases.
36958 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36960 PR middle-end/77484
36961 * predict.def (PRED_CALL): Update hitrate.
36962 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36963 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36964 into direct/indirect/polymorphic variants.
36966 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36968 Update copyright years.
36970 * gcc.c (process_command): Update copyright notice dates.
36971 * gcov-dump.c (print_version): Ditto.
36972 * gcov.c (print_version): Ditto.
36973 * gcov-tool.c (print_version): Ditto.
36974 * gengtype.c (create_file): Ditto.
36975 * doc/cpp.texi: Bump @copying's copyright year.
36976 * doc/cppinternals.texi: Ditto.
36977 * doc/gcc.texi: Ditto.
36978 * doc/gccint.texi: Ditto.
36979 * doc/gcov.texi: Ditto.
36980 * doc/install.texi: Ditto.
36981 * doc/invoke.texi: Ditto.
36983 Copyright (C) 2017 Free Software Foundation, Inc.
36985 Copying and distribution of this file, with or without modification,
36986 are permitted in any medium without royalty provided the copyright
36987 notice and this notice are preserved.