1 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8 conditionals inside the function instead of around it. Call
9 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12 2017-09-26 Richard Biener <rguenther@suse.de>
14 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
16 (scop_detection::build_scop_breadth): ... this. Removed.
17 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
18 (scop_detection::harmful_stmt_in_bb): Likewise.
19 (scop_detection::graphite_can_represent_stmt): Likewise.
20 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
21 (scop_detection::can_represent_loop): Remove recursion, fold in ...
22 (scop_detection::can_represent_loop_1): ... this. Removed.
23 (scop_detection::harmful_loop_in_region): Simplify after inlining
24 the above and remove more quadraticness.
25 (build_scops): Adjust.
26 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
29 2017-09-26 Jakub Jelinek <jakub@redhat.com>
32 * config/i386/i386.c (ix86_print_operand_address_as): Only test
33 REGNO (base) == SP_REG if base is a REG.
36 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
37 if it is different SSA_NAME.
38 (optimize_range_tests_cmp_bitwise): New function.
39 (optimize_range_tests): Call it.
41 2017-09-26 Richard Biener <rguenther@suse.de>
43 PR tree-optimization/82321
44 * graphite.c (canonicalize_loop_closed_ssa): Properly check
45 for the def being inside the loop.
47 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
49 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
51 * config/s390/s390-builtins.def: Fix constraint on op4.
53 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
55 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
56 independent expanders.
57 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
58 ("vec_ordered", "vec_unordered"): New expanders.
60 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
62 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
65 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
67 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
69 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
71 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
73 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
74 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
75 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
77 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
79 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
81 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
83 ("*vec_slb<mode>"): New insn pattern.
84 ("vec_shr_<mode>"): New expander.
85 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
86 and force the shift count operand to V16QImode.
87 ("vec_srb<mode>"): Set shift count mode to V16QI.
89 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
91 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
92 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
93 ("vec_widen_smult_hi_<mode>"): New expander definitions.
95 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
98 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
100 2017-09-26 Richard Biener <rguenther@suse.de>
102 PR tree-optimization/82320
103 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
106 2017-09-25 Jeff Law <law@redhat.com>
108 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
109 prototype for new argument.
110 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
111 mostly extracted from rs6000_emit_allocate_stack.
112 (rs6000_emit_probe_stack_range_stack_clash): New function.
113 (rs6000_emit_allocate_stack): Call
114 rs6000_emit_probe_stack_range_stack_clash as needed.
115 (rs6000_emit_probe_stack_range): Add additional argument
116 to call to gen_probe_stack_range{si,di}.
117 (output_probe_stack_range): New.
118 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
119 (output_probe_stack_range_stack_clash): New.
120 (rs6000_emit_prologue): Emit notes into dump file as requested.
121 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
122 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
123 Add additional operand and pass it to output_probe_stack_range.
125 2017-09-25 Bin Cheng <bin.cheng@arm.com>
127 PR tree-optimization/82163
128 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
129 (checking_verify_loop_closed_ssa): New parameter.
130 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
131 (check_loop_closed_ssa_stmt): Delete.
132 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
133 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
134 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
137 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
139 * brig-builtins.def: Treat HSAIL barrier builtins as
140 setjmp/longjump style functions.
142 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
144 * target.def (constant_alignment): New hook.
145 * defaults.h (CONSTANT_ALIGNMENT): Delete.
146 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
147 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
148 * doc/tm.texi: Regenerate.
149 * targhooks.h (default_constant_alignment): Declare.
150 (constant_alignment_word_strings): Likewise.
151 * targhooks.c (default_constant_alignment): New function.
152 (constant_alignment_word_strings): Likewise.
153 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
154 instead of CONSTANT_ALIGNMENT.
155 * varasm.c (align_variable, get_variable_align, build_constant_desc)
156 (force_const_mem): Likewise.
157 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
158 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
159 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
160 (TARGET_CONSTANT_ALIGNMENT): Redefine.
161 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
163 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
164 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
165 constant_alignment_word_strings.
166 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
167 (CONSTANT_ALIGNMENT): Likewise.
168 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
169 (arm_constant_alignment): New function.
170 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
171 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
172 constant_alignment_word_strings.
173 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
174 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
175 constant_alignment_word_strings.
176 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
177 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
178 (cris_constant_alignment): New function.
179 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
180 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
181 (epiphany_constant_alignment): New function.
182 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
183 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
184 constant_alignment_word_strings.
185 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
186 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
187 constant_alignment_word_strings.
188 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
189 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
190 constant_alignment_word_strings.
191 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
192 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
193 * config/i386/i386.c (ix86_constant_alignment): Make static.
194 Use the same interface as the target hook.
195 (TARGET_CONSTANT_ALIGNMENT): Redefine.
196 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
197 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
198 constant_alignment_word_strings.
199 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
200 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
201 (TARGET_CONSTANT_ALIGNMENT): Redefine.
202 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
203 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
204 constant_alignment_word_strings.
205 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
206 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
207 constant_alignment_word_strings.
208 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
209 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
210 constant_alignment_word_strings.
211 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
212 * config/microblaze/microblaze.c (microblaze_constant_alignment):
214 (TARGET_CONSTANT_ALIGNMENT): Redefine.
215 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
216 * config/mips/mips.c (mips_constant_alignment): New function.
217 (TARGET_CONSTANT_ALIGNMENT): Redefine.
218 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
219 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
220 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
221 (mmix_constant_alignment): Make static. Use the same interface
223 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
224 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
225 constant_alignment_word_strings.
226 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
227 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
228 constant_alignment_word_strings.
229 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
230 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
231 constant_alignment_word_strings.
232 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
233 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
234 (rs6000_constant_alignment): New function.
235 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
236 * config/riscv/riscv.c (riscv_constant_alignment): New function.
237 (TARGET_CONSTANT_ALIGNMENT): Redefine.
238 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
239 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
240 (rs6000_constant_alignment): New function.
241 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
242 * config/s390/s390.c (s390_constant_alignment): New function.
243 (TARGET_CONSTANT_ALIGNMENT): Redefine.
244 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
245 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
246 constant_alignment_word_strings.
247 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
248 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
249 (sparc_constant_alignment): New function.
250 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
251 * config/spu/spu.c (spu_constant_alignment): New function.
252 (TARGET_CONSTANT_ALIGNMENT): Redefine.
253 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
254 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
255 constant_alignment_word_strings.
256 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
257 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
258 constant_alignment_word_strings.
259 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
260 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
261 constant_alignment_word_strings.
262 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
263 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
264 (visium_constant_alignment): New function.
265 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
266 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
267 (xtensa_constant_alignment): New function.
268 * system.h (CONSTANT_ALIGNMENT): Poison.
270 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
272 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
273 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
274 (rs6000_builtin_valid_without_lhs): New helper function.
275 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
276 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
278 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
280 * target.h (vec_perm_indices): Use unsigned short rather than
282 (auto_vec_perm_indices): Likewise.
283 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
284 Use unsigned int rather than unsigned char.
285 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
287 2017-09-25 Richard Biener <rguenther@suse.de>
289 * cfgloop.h (sort_sibling_loops): Declare.
290 * cfgloop.c (sort_sibling_loops_cmp): New helper.
291 (sort_sibling_loops): New function sorting the sibling loop list
293 * graphite.c (graphite_transform_loops): Sort sibling loops.
295 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
297 * target.def (vec_perm_const_ok): Change sel parameter to
299 * optabs-query.c (can_vec_perm_p): Update accordingly.
300 * doc/tm.texi: Regenerate.
301 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
302 auto_vec_perm_indices and remove separate nelt field.
303 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
304 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
305 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
306 (aarch64_expand_vec_perm_const): Update accordingly.
307 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
309 * config/arm/arm.c (expand_vec_perm_d): Change perm to
310 auto_vec_perm_indices and remove separate nelt field.
311 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
312 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
313 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
315 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
317 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
318 sel to vec_perm_indices.
319 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
320 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
321 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
323 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
326 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
329 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
330 on the FUNCTION_DECL function context if it has a DIE that is a
333 2017-09-25 Richard Biener <rguenther@suse.de>
335 PR tree-optimization/82285
336 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
339 2017-09-25 Tom de Vries <tom@codesourcery.com>
343 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
346 2017-09-25 Richard Biener <rguenther@suse.de>
348 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
349 ISL errors other than isl_error_quota happen. Dump if the
350 schedule is the same.
351 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
352 errors instead of aborting inside ISL.
354 2017-09-25 Iain Sandoe <iain@codesourcery.com>
357 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
358 of libgcc_eh for m64.
359 * config/i386/darwin64.h: Likewise.
361 2017-09-25 Richard Biener <rguenther@suse.de>
364 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
365 attribute for incomplete types nor twice for complete ones.
367 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
370 * config/i386/i386.c (ix86_print_operand_address_as): Encode
371 %esp as %rsp to avoid 0x67 prefix if there is no index or base
374 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
377 * config/i386/i386.opt (mprefer-avx256): Use
378 ix86_target_flags variable.
379 * config/i386/i386.c (ix86_target_string): Move
380 -mprefer-avx256 to flag2_opts.
382 2017-09-22 Jakub Jelinek <jakub@redhat.com>
385 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
386 and x != -1 | y != -1 into (x & y) != -1.
388 2017-09-22 Steve Ellcey <sellcey@cavium.com>
390 * config.gcc: Add new case statement to set
391 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
392 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
393 s390x-*-linux* case statements. Added aarch64 to the list of
394 supported architectures.
396 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
398 PR tree-optimization/82289
399 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
400 STMT_VINFO_RELEVANT_P.
402 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
403 Alan Hayward <alan.hayward@arm.com>
404 David Sherwood <david.sherwood@arm.com>
406 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
407 for VR_RANGE only; don't allow VR_ANTI_RANGE.
408 (extract_range_from_binary_expr_1): Don't call
409 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
411 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
412 Alan Hayward <alan.hayward@arm.com>
413 David Sherwood <david.sherwood@arm.com>
415 * target.def (preferred_vector_alignment): New hook.
416 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
418 * doc/tm.texi: Regenerate.
419 * targhooks.h (default_preferred_vector_alignment): Declare.
420 * targhooks.c (default_preferred_vector_alignment): New function.
421 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
423 (DR_TARGET_ALIGNMENT): New macro.
424 (aligned_access_p): Update commentary.
425 (vect_known_alignment_in_bytes): New function.
426 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
428 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
429 Calculate the misalignment based on the target alignment rather than
431 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
432 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
433 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
434 the target alignment, rather than masking the element misalignment
435 with the number of elements in a vector. Also use the target
436 alignment when calculating the maximum number of peels.
437 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
438 instead of TYPE_ALIGN_UNIT.
439 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
440 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
441 (vect_create_addr_base_for_vector_ref): Update call accordingly.
442 (vect_create_data_ref_ptr): Likewise.
443 (vect_setup_realignment): Realign by ANDing with
444 -DR_TARGET_MISALIGNMENT.
445 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
446 the number of peels based on DR_TARGET_ALIGNMENT.
447 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
448 with the guaranteed alignment boundary when deciding whether
450 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
451 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
452 (ensure_base_align): Remove stmt_info parameter. Get the
453 target base alignment from DR_TARGET_ALIGNMENT.
454 (vectorizable_store): Update call accordingly. Interpret
455 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
457 (vectorizable_load): Likewise.
459 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
460 Alan Hayward <alan.hayward@arm.com>
461 David Sherwood <david.sherwood@arm.com>
463 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
464 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
465 (vect_enhance_data_refs_alignment): Likewise.
467 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
469 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
470 error. Only quit immediately if parsing is complete.
471 (BEGIN): Initialize fatal_err and parse_done.
472 (begin fpu, end fpu): Check number of arguments.
473 (begin arch, end arch): Likewise.
474 (begin cpu, end cpu): Likewise.
475 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
476 (optalias): Likewise.
478 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
480 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
481 * config/arm/arm-isa.h: Delete. Move definitions to ...
482 * arm-cpus.in: ... here. Use new feature and fgroup values.
483 * config/arm/arm.c (arm_option_override): Use lower case for feature
485 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
486 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
487 * config/arm/parsecpu.awk (END): Add new command 'isa'.
489 (print_isa_bits_for): New function.
490 (gen_isa): New function.
491 (gen_comm_data): Use print_isa_bits_for.
492 (define feature): New keyword.
493 (define fgroup): New keyword.
494 * config/arm/t-arm (TM_H): Remove.
495 (GTM_H): Add arm-isa.h.
496 (arm-isa.h): Add rule to generate file.
497 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
498 case for feature bit names.
500 2017-09-22 Richard Biener <rguenther@suse.de>
502 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
504 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
505 print of no dependency loops ...
506 * graphite.c (graphite_transform_loops): ... here.
507 (canonicalize_loop_closed_ssa_form): Work from inner to outer
509 (same_close_phi_node, remove_duplicate_close_phi,
510 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
511 (canonicalize_loop_closed_ssa): ... here and simplify.
512 * graphite-optimize-isl.c: Include tree-vectorizer.h.
513 (optimize_isl): Use dump_printf_loc to tell when we stopped
514 optimizing because of an ISL timeout.
516 2017-09-22 Richard Biener <rguenther@suse.de>
518 PR tree-optimization/82291
519 * tree-if-conv.c (predicate_mem_writes): Make sure to
520 remove writes in blocks predicated with false.
522 2017-09-22 Richard Biener <rguenther@suse.de>
524 * sese.c: Include cfganal.h.
525 (if_region_set_false_region): Remove.
526 (create_if_region_on_edge): Likewise.
527 (move_sese_in_condition): Re-implement without destroying
530 2017-09-22 Richard Biener <rguenther@suse.de>
532 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
533 Verify both BBs contain loop PHI nodes before dispatching to
535 (graphite_regenerate_ast_isl): Do not recompute dominators,
536 do not verify three times. Restructure for clarity.
537 * graphite-scop-detection.c (same_close_phi_node,
538 remove_duplicate_close_phi, make_close_phi_nodes_unique,
539 defined_in_loop_p, canonicalize_loop_closed_ssa,
540 canonicalize_loop_closed_ssa_form): Simplify, remove excess
541 checking and SSA rewrite, move to ...
542 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
543 (graphite_initialize): Do not pass in ctx, do not reset the
544 SCEV cache, compute only dominators.
545 (graphite_transform_loops): Allocate ISL ctx after
546 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
547 Maintain post-dominators only around build_scops.
548 * sese.c (if_region_set_false_region): Make static. Free
549 and recompute dominators.
550 (move_sese_in_condition): Assert we don't get called with
551 post-dominators computed.
552 * sese.h (if_region_set_false_region): Remove.
554 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
556 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
557 mode attribute for TARGET_AVX512VL.
559 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
561 * config/i386/i386.opt (mprefer-avx256): New option.
562 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
564 (ix86_preferred_simd_mode): Return 256-bit AVX modes
565 for TARGET_PREFER_AVX256.
566 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
568 2017-09-21 Jeff Law <law@redhat.com>
570 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
571 Fix dump output if the only stack space is for pushed registers.
573 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
575 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
578 2017-09-21 Martin Sebor <msebor@redhat.com>
581 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
582 code (in C++) or code that triggers warnings.
584 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
586 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
588 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
590 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
591 * sched-rgn.c: Ditto.
592 * sel-sched-ir.c: Ditto.
594 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
596 * toplev.h (set_random_seed): Adjust return type.
597 * toplev.c (init_local_tick): Move eager initialization of random_seed
598 to get_random_seed. Adjust comment.
599 (init_random_seed): Inline to get_random_seed, delete.
600 (get_random_seed): Initialize random_seed lazily.
601 (set_random_seed): Do not return previous value.
602 (print_switch_value): Do not call get_random_seed.
604 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
606 * cgraph.c (delete_function_version): New, broken out from...
607 (cgraph_node::delete_function_version): ...here. Rename to
608 cgraph_node::delete_function_version_by_decl. Update all uses.
609 (cgraph_node::remove): Call delete_function_version.
611 2017-09-21 Jakub Jelinek <jakub@redhat.com>
614 * tree-inline.c (expand_call_inline): Emit clobber stmts for
615 VAR_DECLs to which addressable non-volatile parameters are mapped
616 and for id->retvar after the return value assignment. Clear
617 id->retval and id->retbnd after inlining.
619 2017-09-21 Richard Biener <rguenther@suse.de>
621 PR tree-optimization/82276
622 PR tree-optimization/82244
623 * tree-vrp.c (build_assert_expr_for): Set
624 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
626 (remove_range_assertions): Revert earlier change.
628 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
631 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
633 2017-09-21 Richard Biener <rguenther@suse.de>
635 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
636 Restore valid IL after code generation errors.
637 * graphite.c (graphite_transform_loops): Diagnose code
638 generation issues as MSG_MISSED_OPTIMIZATION and continue
639 with processing SCOPs.
641 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
642 Alan Hayward <alan.hayward@arm.com>
643 David Sherwood <david.sherwood@arm.com>
645 * calls.c (compute_argument_addresses): Use simplify_gen_binary
646 rather than choosing between plus_constant and gen_rtx_<CODE>.
647 * expr.c (emit_push_insn): Likewise.
648 (expand_expr_real_2): Likewise.
650 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * loop-unroll.c (split_iv): Call copy_rtx on the step.
656 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
661 calling tree_to_uhwi.
663 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
664 Alan Hayward <alan.hayward@arm.com>
665 David Sherwood <david.sherwood@arm.com>
667 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
668 INTEGER_CST rather than a negative test for ADDR_EXPR.
670 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
671 Alan Hayward <alan.hayward@arm.com>
672 David Sherwood <david.sherwood@arm.com>
674 * tree-vrp.c (extract_range_from_binary_expr_1): Check
675 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
677 2017-09-21 Richard Biener <rguenther@suse.de>
679 PR tree-optimization/71351
680 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
681 graphite_create_new_loop_guard): Remove, fold remaining parts
683 (translate_isl_ast_node_for): ... here and simplify.
685 2017-09-21 Jakub Jelinek <jakub@redhat.com>
688 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
689 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
690 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
691 alternative always use QI mode, for -Os imov (=R,R) alternative
692 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
695 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
696 Jeff Law <law@redhat.com>
698 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
699 (allocate_stack_space): New function, partially extracted from
701 (s390_emit_prologue): Track offset to most recent stack probe.
702 Code to allocate space moved into allocate_stack_space.
703 Dump actions when no stack is allocated.
704 (s390_prologue_plus_offset): New function.
705 (s390_emit_stack_probe): Likewise.
707 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
709 * common.opt (Wa, Wl, Wp, g, gz=): Add
711 (gno-column-info): Remove.
712 (gcolumn-info): Drop RejectNegative.
714 (gno-record-gcc-switches): Remove.
715 (grecord-gcc-switches): Drop RejectNegative.
716 (gno-split-dwarf): Remove.
717 (gsplit-dwarf): Drop RejectNegative.
718 (gno-strict-dwarf): Remove.
719 (gstrict-dwarf): Drop RejectNegative.
720 * config/darwin.opt (gfull, gused): Add RejectNegative.
721 * dwarf2out.c (gen_producer_string): Drop
722 gno-record-gcc-switches handler.
723 * optc-gen.awk: Add g to prefixes with negative forms.
724 * opts-common.c (remapping_prefix_p): New.
725 (find_opt): Check it.
726 (generate_canonical_option): Test g prefix.
727 (option_map): Add -gno- mapping.
728 (add_misspelling_candidates): Check remapping_prefix_p.
730 2017-09-20 Jeff Law <law@redhat.com>
732 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
733 thinko in stack clash protection support.
735 * explow.c (compute_stack_clash_protection_loop_data): Use
736 CONST_INT_P instead of explicit test. Verify object is a
737 CONST_INT_P before looking at INTVAL.
738 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
739 instead of explicit test.
741 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
744 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
745 address instead of to r1 and r11.
747 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
749 * config.gcc: Support "knm".
750 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
751 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
753 * config/i386/i386.c (m_KNM): Define.
754 (processor_target_table): Add "knm".
756 (ix86_option_override_internal): Add "knm".
757 (ix86_issue_rate): Add PROCESSOR_KNM.
758 (ix86_adjust_cost): Ditto.
759 (ia32_multipass_dfa_lookahead): Ditto.
760 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
761 (fold_builtin_cpu): Add M_INTEL_KNM.
762 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
763 (processor_type): Add PROCESSOR_KNM.
764 * config/i386/x86-tune.def: Add m_KNM.
765 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
767 2017-09-20 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/80213
770 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
771 are allowed in empty BBs as well.
772 (canonicalize_loop_closed_ssa): Also look for other complex
774 (scop_detection::get_sese): Include the loop-closed PHI block
776 (scop_detection::merge_sese): Remove code adding extra blocks.
777 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
778 (build_scops): Assert the final returned scop is invalid.
780 2017-09-20 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/82264
783 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
785 (vn_phi_lookup): Likewise.
786 (vn_phi_insert): Likewise.
788 2017-09-20 Jakub Jelinek <jakub@redhat.com>
790 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
791 that fits into uhwi or shwi, add DW_AT_const_value regardless
792 of early_dwarf without going through RTL, using add_AT_unsigned
795 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
796 (DEBUG_LTO_ABBREV_SECTION): Likewise.
797 (DEBUG_LTO_MACINFO_SECTION): Likewise.
798 (DEBUG_MACRO_SECTION): Likewise.
799 (DEBUG_LTO_MACRO_SECTION): Likewise.
800 (DEBUG_STR_DWO_SECTION): Likewise.
801 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
802 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
803 (DEBUG_LTO_DWO_LINE_SECTION): Define.
804 (DEBUG_LTO_LINE_STR_SECTION): Define.
805 (init_sections_and_labels): Initialize debug_line_str_section
806 variable. Initialize debug_loc_section for -gdwarf-5 to
807 DEBUG_LOCLISTS_SECTION. Formatting fixes.
809 2017-09-20 Richard Biener <rguenther@suse.de>
811 * graphite-sese-to-poly.c (extract_affine): Properly handle
812 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
814 2017-09-20 Richard Biener <rguenther@suse.de>
816 PR tree-optimization/81373
817 * graphite-scop-detection.c (build_cross_bb_scalars_def):
818 Force SESE live-out defs to be handled even if they are
821 2017-09-19 Jeff Law <law@redhat.com>
823 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
824 nothing for stack adjustments with REG_STACK_CHECK.
825 * sched-deps.c (parse_add_or_inc): Reject insns with
826 REG_STACK_CHECK from dependency breaking.
827 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
828 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
829 * reg-notes.def (STACK_CHECK): New note.
831 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
832 (ix86_expand_prologue): Dump stack clash info as needed.
833 Call ix86_adjust_stack_and_probe_stack_clash as needed.
835 * function.c (dump_stack_clash_frame_info): New function.
836 * function.h (dump_stack_clash_frame_info): Prototype.
837 (enum stack_clash_probes): New enum.
839 * config/alpha/alpha.c (alpha_expand_prologue): Also check
840 flag_stack_clash_protection.
841 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
842 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
843 (arm_frame_pointer_required): Likewise.
844 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
845 (ia64_expand_prologue): Likewise.
846 * config/mips/mips.c (mips_expand_prologue): Likewise.
847 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
848 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
849 (sparc_flat_expand_prologue): Likewise.
850 * config/spu/spu.c (spu_expand_prologue): Likewise.
852 * explow.c: Include "params.h".
853 (anti_adjust_stack_and_probe_stack_clash): New function.
854 (get_stack_check_protect): Likewise.
855 (compute_stack_clash_protection_loop_data): Likewise.
856 (emit_stack_clash_protection_loop_start): Likewise.
857 (emit_stack_clash_protection_loop_end): Likewise.
858 (allocate_dynamic_stack_space): Use get_stack_check_protect.
859 Use anti_adjust_stack_and_probe_stack_clash.
860 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
861 (emit_stack_clash_protection_loop_start): Likewise.
862 (emit_stack_clash_protection_loop_end): Likewise.
863 * rtl.h (get_stack_check_protect): Prototype.
864 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
865 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
866 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
868 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
870 * doc/tm.texi: Rebuilt.
871 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
872 get_stack_check_protect.
873 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
874 * config/arm/arm.c (arm_expand_prologue): Likewise.
875 (arm_frame_pointer_required): Likewise.
876 * config/i386/i386.c (ix86_expand_prologue): Likewise.
877 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
878 * config/mips/mips.c (mips_expand_prologue): Likewise.
879 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
880 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
881 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
882 (sparc_flat_expand_prologue): Likewise.
884 * common.opt (-fstack-clash-protection): New option.
885 * flag-types.h (enum stack_check_type): Note difference between
886 -fstack-check= and -fstack-clash-protection.
887 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
888 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
889 * toplev.c (process_options): Issue warnings/errors for cases
890 not handled with -fstack-clash-protection.
891 * doc/invoke.texi (-fstack-clash-protection): Document new option.
892 (-fstack-check): Note additional problem with -fstack-check=generic.
893 Note that -fstack-check is primarily for Ada and refer users
894 to -fstack-clash-protection for stack-clash-protection.
895 Document new params for stack clash protection.
897 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
899 * config/i386/i386.c (ix86_split_long_move): Do not handle
900 address used for LEA in a special way.
902 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
904 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
906 2017-09-19 Martin Sebor <msebor@redhat.com>
909 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
910 of incompatible types.
912 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
915 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
916 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
917 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
919 2017-09-19 Richard Biener <rguenther@suse.de>
921 PR tree-optimization/82244
922 * tree-vrp.c (remove_range_assertions): Do not propagate
923 a constant to abnormals but replace the assert with a copy.
925 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
927 PR rtl-optimization/57878
928 PR rtl-optimization/68988
929 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
930 avoidance test involving non_reload_pseudos. Move frequency test
931 below the general fragmentation avoidance test.
933 2017-09-19 Richard Biener <rguenther@suse.de>
935 PR tree-optimization/69728
936 * graphite-sese-to-poly.c (schedule_error): New global.
937 (add_loop_schedule): Handle empty domain by failing the
939 (build_original_schedule): Handle schedule_error.
941 2017-09-19 Richard Biener <rguenther@suse.de>
943 * graphite-scop-detection.c (scop_detection::can_represent_loop):
944 Do not iterate to sibling loops but only to siblings of inner
947 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
950 * config/m68k/m68k.md (moveq feeding equality comparison): Check
951 that the registers are different.
953 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
955 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
956 to processor_model and "amdfam17h" to arch_names_table.
957 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
959 2017-09-18 Jakub Jelinek <jakub@redhat.com>
962 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
964 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
965 Alan Hayward <alan.hayward@arm.com>
966 David Sherwood <david.sherwood@arm.com>
968 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
970 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
972 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
973 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
974 vect_schedule_slp_instance.
975 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
976 Update call to vect_slp_analyze_node_operations. Simplify return
978 (vect_slp_analyze_bb_1): Update call accordingly.
979 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
980 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
981 (vect_schedule_slp): Update call accordingly.
983 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
984 Alan Hayward <alan.hayward@arm.com>
985 David Sherwood <david.sherwood@arm.com>
987 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
988 with types that aren't in fact scalar.
990 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
992 * tree-vect-slp.c (vect_record_max_nunits): New function,
994 (vect_build_slp_tree_1): ...here.
995 (vect_build_slp_tree_2): Call it for phis too.
997 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
999 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1000 to vect_get_vec_def_for_operand when getting the mask operand.
1002 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1003 Alan Hayward <alan.hayward@arm.com>
1004 David Sherwood <david.sherwood@arm.com>
1006 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1010 Alan Hayward <alan.hayward@arm.com>
1011 David Sherwood <david.sherwood@arm.com>
1013 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1014 calculation for vector booleans.
1016 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1017 Alan Hayward <alan.hayward@arm.com>
1018 David Sherwood <david.sherwood@arm.com>
1020 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1022 (vect_transform_stmt): ...here.
1023 (vect_analyze_stmt): Use it instead of calling
1024 vectorizable_live_operation directly.
1026 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1028 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1029 non-SIMT targets in acc vector loops.
1031 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1033 * configure.ac: Add arc and check if assembler supports gdwarf2.
1034 * configure: Regenerate.
1036 2017-09-18 Richard Biener <rguenther@suse.de>
1038 PR tree-optimization/82220
1039 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1040 epilogue niters from the min_profitable_iters compute.
1042 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1045 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1046 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1047 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1049 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1052 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1053 switching to a new text section.
1055 2017-09-18 Richard Biener <rguenther@suse.de>
1057 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1059 (build_alias_set): Reject aliases with no access function.
1061 2017-09-18 Richard Biener <rguenther@suse.de>
1063 PR tree-optimization/79622
1064 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1066 (build_cross_bb_scalars_use): Likewise.
1068 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1070 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1072 2017-09-18 Alan Modra <amodra@gmail.com>
1075 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1076 stack_pointer_rtx for count 0. Update comments. Break up
1077 large rtl expression.
1079 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1081 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1082 Increase to 20 bytes.
1083 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1084 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1085 or avx version of the stub.
1087 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1090 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1091 compute the minimum stack alignment. Also update preferred stack
1092 boundary for leaf functions.
1094 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1096 PR tree-optimization/82228
1097 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1100 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1102 * common/config/nds32/nds32-common.c
1103 (nds32_option_optimization_table): Refine formatting.
1104 (nds32_option_optimization_table): Use -fsched-pressure and
1105 -fomit-frame-pointer for specific optimization level.
1107 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1109 * config/nds32/nds32.c: Refine formatting and comments.
1110 * config/nds32/nds32.h: Likewise.
1111 * config/nds32/nds32.md: Likewise.
1112 * config/nds32/nds32-cost.c: Likewise.
1113 * config/nds32/nds32-isr.c: Likewise.
1114 * config/nds32/nds32-md-auxiliary.c: Likewise.
1115 * config/nds32/nds32-multiple.md: Likewise.
1116 * config/nds32/nds32-predicates.c: Likewise.
1118 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1119 Jakub Jelinek <jakub@redhat.com>
1121 Add support for -std=c++2a.
1122 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1124 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1126 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1129 * doc/extend.texi (Common Function Attributes): Add
1130 references to ARM, AArch64, and S/390 specific attributes.
1131 (Function Specific Option Pragmas): Add AArch64 and S/390
1132 to list of back ends that support the target pragma.
1134 2017-09-15 Nathan Sidwell <nathan@acm.org>
1136 * doc/standards.texi: Fix C++17 description. Update URLs for
1139 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1141 * common.opt (Wcast-align=strict): New warning option.
1142 * doc/invoke.texi: Document -Wcast-align=strict.
1144 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1146 * cgraph.h (cgraph_thunk_info): Add comments.
1147 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1148 assert for VIRTUAL_* arguments stricter.
1150 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1152 PR tree-optimization/71026
1153 * match.pd: Move RDIV patterns from fold-const.c
1154 * fold-const.c (distribute_real_division): Removed.
1155 (fold_binary_loc): Remove calls to distribute_real_divison.
1157 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1159 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1160 c++1z and gnu++1z as deprecated. Change other references to
1161 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1162 Change -Wc++1z-compat to -Wc++17-compat.
1163 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1164 * dwarf2out.c (highest_c_language): Handle C++17.
1165 (gen_compile_unit_die): Likewise.
1167 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1169 PR rtl-optimization/82192
1170 * combine.c (make_extraction): Don't look through non-paradoxical
1171 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1174 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1175 Alan Hayward <alan.hayward@arm.com>
1176 David Sherwood <david.sherwood@arm.com>
1178 * target.def (function_arg_offset): New hook.
1179 * targhooks.h (default_function_arg_offset): Declare.
1180 * targhooks.c (default_function_arg_offset): New function.
1181 * function.c (locate_and_pad_parm): Use
1182 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1183 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1184 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1185 * doc/tm.texi: Regenerate.
1186 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1187 * config/spu/spu.c (spu_function_arg_offset): New function.
1188 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1189 * system.h (FUNCTION_ARG_OFFSET): Poison.
1191 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1192 Alan Hayard <alan.hayward@arm.com>
1193 David Sherwood <david.sherwood@arm.com>
1195 * target.def (truly_noop_truncation): New hook.
1196 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1197 than TRULY_NOOP_TRUNCATION.
1198 * hooks.h (hook_bool_uint_uint_true): Declare.
1199 * hooks.c (hook_bool_uint_uint_true): New function.
1200 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1201 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1202 * doc/tm.texi: Regenerate.
1203 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1204 rather than TRULY_NOOP_TRUNCATION in comments.
1205 (simplify_comparison): Likewise.
1206 (record_truncated_value): Likewise.
1207 * expmed.c (extract_bit_field_1): Likewise.
1208 (extract_split_bit_field): Likewise.
1209 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1210 instead of TRULY_NOOP_TRUNCATION.
1211 * function.c (assign_parm_setup_block): Likewise.
1212 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1213 * rtlhooks.c: Include target.h.
1214 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1215 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1216 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1217 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1218 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1219 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1220 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1221 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1222 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1223 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1224 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1225 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1226 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1227 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1228 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1229 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1230 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1231 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1232 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1233 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1234 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1235 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1236 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1237 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1238 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1239 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1240 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1241 rather than TRULY_NOOP_TRUNCATION in comments.
1242 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1243 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1244 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1245 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1246 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1247 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1248 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1249 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1250 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1251 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1252 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1253 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1254 rather than TRULY_NOOP_TRUNCATION in comments.
1255 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1256 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1257 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1258 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1259 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1260 TRULY_NOOP_TRUNCATION condition.
1261 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1262 (TRULY_NOOP_TRUNCATION): Delete.
1263 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1264 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1265 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1266 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1267 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1268 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1269 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1270 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1271 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1272 rather than TRULY_NOOP_TRUNCATION in comments.
1273 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1274 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1275 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1276 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1277 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1278 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1280 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1283 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1284 (*cmp_ior): Likewise.
1285 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1286 (*ior_scc_scc_cmp): Likewise.
1287 (*and_scc_scc): Likewise.
1288 (*and_scc_scc_cmp): Likewise.
1290 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1291 Alan Hayard <alan.hayward@arm.com>
1292 David Sherwood <david.sherwood@arm.com>
1294 * target.def (can_change_mode_class): New hook.
1295 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1296 (hard_regno_nregs): Likewise.
1297 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1298 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1299 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1300 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1301 (LOAD_EXTEND_OP): Update accordingly.
1302 * doc/tm.texi: Regenerate.
1303 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1304 CANNOT_CHANGE_MODE_CLASS.
1305 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1306 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1307 * combine.c (simplify_set): Update accordingly.
1308 * emit-rtl.c (validate_subreg): Likewise.
1309 * recog.c (general_operand): Likewise.
1310 * regcprop.c (mode_change_ok): Likewise.
1311 * reload1.c (choose_reload_regs): Likewise.
1312 (inherit_piecemeal_p): Likewise.
1313 * rtlanal.c (simplify_subreg_regno): Likewise.
1314 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1315 instead of CANNOT_CHANGE_MODE_CLASS.
1316 (reload_cse_simplify_operands): Likewise.
1317 * reload.c (push_reload): Use targetm.can_change_mode_class
1318 instead of CANNOT_CHANGE_MODE_CLASS.
1319 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1320 REG_CANNOT_CHANGE_MODE_P.
1321 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1322 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1323 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1324 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1325 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1326 (arm_can_change_mode_class): New function.
1327 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1328 than CANNOT_CHANGE_MODE_CLASS in comments.
1329 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1330 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1331 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1332 (ix86_can_change_mode_class): ...this new function, inverting the
1333 sense of the return value.
1334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1335 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1336 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1337 (ia64_can_change_mode_class): New function.
1338 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1339 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1340 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1341 (m32c_can_change_mode_class): ...this new function, inverting the
1342 sense of the return value.
1343 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1344 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1345 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1346 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1347 (mips_can_change_mode_class): ...this new function, inverting the
1348 sense of the return value.
1349 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1350 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1351 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1352 (msp430_can_change_mode_class): New function.
1353 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1354 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1355 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1356 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1357 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1358 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1359 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1360 (pa_cannot_change_mode_class): Replace with...
1361 (pa_can_change_mode_class): ...this new function, inverting the
1362 sense of the return value.
1363 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1364 than CANNOT_CHANGE_MODE_CLASS in comments.
1365 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1366 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1367 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1368 (pdp11_cannot_change_mode_class): Replace with...
1369 (pdp11_can_change_mode_class): ...this new function, inverting the
1370 sense of the return value.
1371 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1372 * config/powerpcspe/powerpcspe-protos.h
1373 (rs6000_cannot_change_mode_class_ptr): Delete.
1374 * config/powerpcspe/powerpcspe.c
1375 (rs6000_cannot_change_mode_class_ptr): Delete.
1376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1377 (rs6000_option_override_internal): Assign to
1378 targetm.can_change_mode_class instead of
1379 rs6000_cannot_change_mode_class_ptr.
1380 (rs6000_cannot_change_mode_class): Replace with...
1381 (rs6000_can_change_mode_class): ...this new function, inverting the
1382 sense of the return value.
1383 (rs6000_debug_cannot_change_mode_class): Replace with...
1384 (rs6000_debug_can_change_mode_class): ...this new function.
1385 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1386 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1387 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1388 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1389 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1391 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1392 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1393 (rs6000_option_override_internal): Assign to
1394 targetm.can_change_mode_class instead of
1395 rs6000_cannot_change_mode_class_ptr.
1396 (rs6000_cannot_change_mode_class): Replace with...
1397 (rs6000_can_change_mode_class): ...this new function, inverting the
1398 sense of the return value.
1399 (rs6000_debug_cannot_change_mode_class): Replace with...
1400 (rs6000_debug_can_change_mode_class): ...this new function.
1401 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1402 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1403 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1404 (s390_can_change_mode_class): ...this new function, inverting the
1405 sense of the return value.
1406 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1407 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1408 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1409 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1410 (sh_cannot_change_mode_class): Replace with...
1411 (sh_can_change_mode_class): ...this new function, inverting the
1412 sense of the return value.
1413 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1414 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1415 (sparc_can_change_mode_class): New function.
1416 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1417 * config/spu/spu.c (spu_can_change_mode_class): New function.
1418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1419 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1420 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1421 (visium_can_change_mode_class): New function.
1422 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1424 2017-09-15 Richard Biener <rguenther@suse.de>
1426 PR tree-optimization/82217
1427 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1428 but not undefined case.
1430 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1433 * postreload.c (reload_cse_simplify_operands): Skip
1434 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1436 2017-09-15 Richard Biener <rguenther@suse.de>
1438 PR tree-optimization/68823
1439 * graphite-scop-detection.c (build_alias_set): If we have a
1440 possible dependence check whether we can handle them by just
1441 looking at the DRs DR_ACCESS_FNs.
1442 (build_scops): If build_alias_set fails, fail the SCOP.
1444 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1446 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1447 to support float128 built-in functions that require the ISA 3.0
1449 (BU_FLOAT128_3_HW): Likewise.
1450 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1452 (FMAF128): Likewise.
1454 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1455 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1456 floating point instructions.
1457 (rs6000_invalid_builtin): Likewise.
1458 (rs6000_builtin_mask_names): Likewise.
1459 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1460 (RS6000_BTM_FLOAT128_HW): Likewise.
1461 (RS6000_BTM_COMMON): Likewise.
1462 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1464 * doc/extend.texi (RS/6000 built-in functions): Document the
1465 IEEE 128-bit floating point square root and fused multiply-add
1468 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1470 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1471 reg (r2) isn't in the set of registers defined in the prologue.
1473 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1478 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1479 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1480 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1481 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1483 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1484 max_vectorization_factor.
1485 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1487 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1488 Alan Hayward <alan.hayward@arm.com>
1489 David Sherwood <david.sherwood@arm.com>
1491 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1492 (vect_worthwhile_without_simd_p): Declare.
1493 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1494 (vectorizable_reduction): Use it.
1495 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1496 (vectorizable_operation): Likewise.
1498 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1499 Alan Hayward <alan.hayward@arm.com>
1500 David Sherwood <david.sherwood@arm.com>
1502 * tree-vectorizer.h (vect_get_num_copies): New function.
1503 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1504 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1505 (vectorizable_induction): Likewise.
1506 (vectorizable_live_operation): Likewise.
1507 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1508 (vectorizable_bswap): Likewise.
1509 (vectorizable_call): Likewise.
1510 (vectorizable_conversion): Likewise.
1511 (vectorizable_assignment): Likewise.
1512 (vectorizable_shift): Likewise.
1513 (vectorizable_operation): Likewise.
1514 (vectorizable_store): Likewise.
1515 (vectorizable_load): Likewise.
1516 (vectorizable_condition): Likewise.
1517 (vectorizable_comparison): Likewise.
1518 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1521 Alan Hayward <alan.hayward@arm.com>
1522 David Sherwood <david.sherwood@arm.com>
1524 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1525 of vect_init_vector.
1527 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1528 Alan Hayward <alan.hayward@arm.com>
1529 David Sherwood <david.sherwood@arm.com>
1531 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1532 an inline wrapper that provides a location.
1533 (gimple_build_vector): Likewise.
1534 * gimple-fold.c (gimple_build_vector_from_val): New function.
1535 (gimple_build_vector): Likewise.
1536 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1537 functions to build the initial value. Always return a gimple value.
1538 (get_initial_defs_for_reduction): Likewise. Only compute
1540 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1541 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1542 (vectorizable_induction): Use gimple_build_vector rather than
1545 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * target.h (vec_perm_indices): New typedef.
1550 (auto_vec_perm_indices): Likewise.
1551 * optabs-query.h: Include target.h
1552 (can_vec_perm_p): Take a vec_perm_indices *.
1553 * optabs-query.c (can_vec_perm_p): Likewise.
1554 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1555 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1556 * tree-vect-generic.c (lower_vec_perm): Likewise.
1557 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1558 (vect_grouped_load_supported): Likewise.
1559 (vect_shift_permute_load_chain): Likewise.
1560 (vect_permute_store_chain): Use auto_vec_perm_indices.
1561 (vect_permute_load_chain): Likewise.
1562 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
1563 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
1564 Update uses of can_vec_perm_p.
1565 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
1566 mode with a number of elements. Take a vec_perm_indices *.
1567 (vect_create_epilog_for_reduction): Update accordingly.
1568 Use auto_vec_perm_indices.
1569 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
1570 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
1571 (vect_transform_slp_perm_load): Likewise.
1572 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
1573 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
1574 (vect_gen_perm_mask_checked): Likewise.
1575 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
1576 (vect_gen_perm_mask_checked): Likewise.
1577 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
1578 (vectorizable_store): Likewise.
1579 (vectorizable_load): Likewise.
1580 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
1581 (vectorizable_bswap): Likewise.
1583 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1584 Alan Hayward <alan.hayward@arm.com>
1585 David Sherwood <david.sherwood@arm.com>
1587 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1588 * tree.c (build_vector): Likewise.
1589 (build_vector_from_ctor): Update accordingly.
1590 (build_vector_from_val): Likewise.
1591 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1592 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1593 * tree-vect-generic.c (add_rshift): Likewise.
1594 (expand_vector_divmod): Likewise.
1595 (optimize_vector_constructor): Likewise.
1596 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1597 (vect_transform_slp_perm_load): Likewise.
1598 (vect_schedule_slp_instance): Likewise.
1599 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1600 (vectorizable_call): Likewise.
1601 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1602 * expmed.c (make_tree): Likewise.
1603 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1604 a vector passed to build_vector.
1605 (fold_convert_const): Likewise.
1606 (exact_inverse): Likewise.
1607 (fold_ternary_loc): Likewise.
1608 (fold_relational_const): Likewise.
1609 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1610 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1611 (const_unop): Likewise. Store the reduction accumulator in a
1612 variable rather than an array.
1613 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1614 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1615 the new vector, rather than constructing it after the input arrays.
1616 (native_interpret_vector): Use auto_vec<tree> when building
1617 a vector passed to build_vector. Add elements in order.
1618 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1619 auto_vec<tree> when building a vector passed to build_vector.
1620 (vect_create_epilog_for_reduction): Likewise.
1621 (vectorizable_induction): Likewise.
1622 (get_initial_def_for_reduction): Likewise. Fix indentation of
1624 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1626 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1627 passed to build_vector.
1629 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1630 Alan Hayward <alan.hayward@arm.com>
1631 David Sherwood <david.sherwood@arm.com>
1633 * tree-core.h (tree_base::u): Add an "nelts" field.
1634 (tree_vector): Use VECTOR_CST_NELTS as the length.
1635 * tree.c (tree_size): Likewise.
1636 (make_vector): Initialize VECTOR_CST_NELTS.
1637 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1638 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1639 TYPE_VECTOR_SUBPARTS.
1640 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1641 as the number of units, setting it from VECTOR_CST_NELTS.
1642 (const_vector_from_tree): Likewise.
1643 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1644 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1645 (fold_negate_expr_1): Likewise.
1646 (fold_convert_const): Likewise.
1647 (const_binop): Likewise. Differentiate the number of output and
1649 (const_unop): Likewise.
1650 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1651 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1652 in cases that did the opposite.
1654 2017-09-14 Richard Biener <rguenther@suse.de>
1656 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1659 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1661 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1663 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1666 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1667 if and where to split a bb, except for splitting before debug insn
1668 sequences followed by non-label real insn. Delete debug insns
1669 in between basic blocks.
1671 * combine.c (make_compound_operation_int): Formatting fixes.
1673 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1674 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1675 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1676 * config/sol2.h (LINK_EH_SPEC): Likewise.
1677 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1678 * config/s390/linux.h (LINK_SPEC): Likewise.
1679 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1680 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1681 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1682 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1683 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1684 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1685 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1686 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1688 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1690 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1692 (ENDFILE_LINUX_SPEC): Likewise.
1693 (LINK_EH_SPEC): Likewise.
1694 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1695 (LINK_OS_LINUX_SPEC32): Likewise.
1696 (LINK_OS_LINUX_SPEC64): Likewise.
1697 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1698 (LINK_OS_LINUX_SPEC): Likewise.
1700 2017-09-13 Martin Liska <mliska@suse.cz>
1703 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1704 CASE_HIGH is NULL_TREE.
1706 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1707 Alan Hayward <alan.hayward@arm.com>
1708 David Sherwood <david.sherwood@arm.com>
1710 * target.def (secondary_memory_needed): New hook.
1711 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1712 instead of SECONDARY_MEMORY_NEEDED.
1713 (secondary_memory_needed_mode): Likewise.
1714 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1715 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1716 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1717 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1718 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1719 * doc/tm.texi: Regenerate.
1720 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1721 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1722 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1723 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1724 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1725 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1726 mode argument first and change the reg_class arguments to reg_class_t.
1727 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1728 Make static. Update the call to inline_secondary_memory_needed.
1729 (ix86_register_move_cost): Update the call to
1730 inline_secondary_memory_needed.
1731 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1732 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1734 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1735 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1737 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1738 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1739 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1740 and match hook interface. Add comment from mips.h.
1741 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1742 * config/mmix/mmix.md (truncdfsf2): Refer to
1743 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1745 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1746 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1747 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1748 (pa_secondary_memory_needed): New function.
1749 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1750 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1751 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1752 (pdp11_secondary_memory_needed): Make static and match hook interface.
1753 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1754 * config/powerpcspe/powerpcspe-protos.h
1755 (rs6000_secondary_memory_needed_ptr): Delete.
1756 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1758 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1759 (rs6000_option_override_internal): Assign to
1760 targetm.secondary_memory_needed rather than
1761 rs6000_secondary_memory_needed_ptr.
1762 (rs6000_secondary_memory_needed): Match hook interface.
1763 (rs6000_debug_secondary_memory_needed): Likewise.
1764 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1765 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1766 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1767 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1768 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1769 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1771 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1772 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1773 (rs6000_option_override_internal): Assign to
1774 targetm.secondary_memory_needed rather than
1775 rs6000_secondary_memory_needed_ptr.
1776 (rs6000_secondary_memory_needed): Match hook interface.
1777 (rs6000_debug_secondary_memory_needed): Likewise.
1778 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1779 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1780 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1781 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1782 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1783 (sparc_secondary_memory_needed): New function.
1784 * lra-constraints.c (check_and_process_move): Refer to
1785 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1787 (curr_insn_transform): Likewise.
1788 (process_alt_operands): Use targetm.secondary_memory_needed
1789 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1790 (check_secondary_memory_needed_p): Likewise.
1791 (choose_split_class): Likewise.
1792 * reload.c: Unconditionally include code that was previously
1793 conditional on SECONDARY_MEMORY_NEEDED.
1794 (push_secondary_reload): Use targetm.secondary_memory_needed
1795 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1796 (push_reload): Likewise.
1797 * reload1.c: Unconditionally include code that was previously
1798 conditional on SECONDARY_MEMORY_NEEDED.
1799 (choose_reload_regs): Use targetm.secondary_memory_needed
1800 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1801 (gen_reload): Likewise.
1802 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1804 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1805 Alan Hayward <alan.hayward@arm.com>
1806 David Sherwood <david.sherwood@arm.com>
1808 * target.def (secondary_memory_needed_mode): New hook:
1809 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1810 * targhooks.h (default_secondary_memory_needed_mode): New function.
1811 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1812 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1813 * doc/tm.texi: Regenerate.
1814 * lra-constraints.c (check_and_process_move): Use
1815 targetm.secondary_memory_needed_mode instead of
1816 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1817 (curr_insn_transform): Likewise.
1818 * reload.c (get_secondary_mem): Likewise.
1819 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1820 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1822 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1823 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1824 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1825 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1826 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1828 * config/powerpcspe/powerpcspe-protos.h
1829 (rs6000_secondary_memory_needed_mode): Delete.
1830 * config/powerpcspe/powerpcspe.c
1831 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1832 (rs6000_secondary_memory_needed_mode): Make static.
1833 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1834 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1836 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1838 (rs6000_secondary_memory_needed_mode): Make static.
1839 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1840 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1841 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1842 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1843 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1845 (sparc_secondary_memory_needed_mode): New function.
1846 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1848 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1850 * config/aarch64/constraints.md (Umq): New constraint.
1851 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1853 (mov<mode>): Update condition.
1855 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1857 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1858 when bitposition is the same.
1860 2017-09-13 Richard Biener <rguenther@suse.de>
1862 * dwarf2out.c (output_die_symbol): Remove.
1863 (output_die): Do not output a DIEs symbol.
1865 2017-09-13 Richard Biener <rguenther@suse.de>
1868 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1869 default-def to avoid breaking iterator update with the weird
1870 interaction with cgraph_update_edges_for_call_stmt_node.
1872 2017-09-13 Richard Biener <rguenther@suse.de>
1874 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1875 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1876 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1877 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1879 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1881 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1882 Disable pc relative literal load irrespective of
1883 TARGET_FIX_ERR_A53_84341 for default.
1885 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1887 * config/sparc/sparc.c (output_return): Output the source location of
1888 the insn in the delay slot, if any.
1889 (output_sibcall): Likewise.
1891 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1894 * common.opt (-static-pie): New alias.
1895 (shared): Negate static-pie.
1896 (-no-pie): Update help text.
1898 (static-pie): New option.
1899 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1900 -static-pie support.
1901 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1902 (LINK_EH_SPEC): Likewise.
1903 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1904 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1905 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1906 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1907 (init_gcc_specs): Likewise.
1908 (init_spec): Likewise.
1909 (display_help): Update help message for -pie.
1910 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1913 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1915 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1916 (movdi_aarch64): Likewise.
1917 (movti_aarch64): Likewise.
1919 2017-09-12 Simon Wright <simon@pushface.org>
1922 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1923 calculation of the minor version, always output as 0.
1925 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1928 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1929 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1930 on it early, rather than manual conversion late. For
1931 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1932 instead of performing manual conversion.
1934 2017-09-12 Carl Love <cel@us.ibm.com>
1936 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1937 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1939 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1940 VMULOSW): Add definitions.
1941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1942 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1943 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1945 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1947 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1949 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1951 (movti_aarch64): Likewise.
1952 (movdf_aarch64): Likewise.
1953 (movtf_aarch64): Likewise.
1954 (load_pairdi): Likewise.
1955 (store_pairdi): Likewise.
1956 (load_pairdf): Likewise.
1957 (store_pairdf): Likewise.
1958 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1959 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1960 (ldr_got_small_<mode>): Likewise.
1961 (ldr_got_small_28k_<mode>): Likewise.
1962 (ldr_got_tiny): Likewise.
1963 * config/aarch64/iterators.md (ldst_sz): New.
1964 (ldpstp_sz): Likewise.
1965 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1967 (thunderx_load): Split load_8 to load_16.
1968 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1970 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1971 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1972 (xgene1_store_pair): Split store_8 to store_16.
1973 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1974 (falkor_st_0_st_sd): Split store_8 to store_16.
1976 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1978 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1979 and store1/2/3/4 to store_4/8/12/16.
1980 * config/aarch64/aarch64.md: Update for rename.
1981 * config/arm/arm.md: Likewise.: Likewise.
1982 * config/arm/arm.c: Likewise.
1983 * config/arm/thumb1.md: Likewise.
1984 * config/arm/thumb2.md: Likewise.
1985 * config/arm/vfp.md: Likewise.
1986 * config/arm/arm-generic.md: Likewise.
1987 * config/arm/arm1020e.md: Likewise.
1988 * config/arm/arm1026ejs.md: Likewise.
1989 * config/arm/arm1136jfs.md: Likewise.
1990 * config/arm/arm926ejs.md: Likewise.
1991 * config/arm/cortex-a15.md: Likewise.
1992 * config/arm/cortex-a17.md: Likewise.
1993 * config/arm/cortex-a5.md: Likewise.
1994 * config/arm/cortex-a53.md: Likewise.
1995 * config/arm/cortex-a57.md: Likewise.
1996 * config/arm/cortex-a7.md: Likewise.
1997 * config/arm/cortex-a8.md: Likewise.
1998 * config/arm/cortex-a9.md: Likewise.
1999 * config/arm/cortex-m4.md: Likewise.
2000 * config/arm/cortex-m7.md: Likewise.
2001 * config/arm/cortex-r4.md: Likewise.
2002 * config/arm/exynos-m1.md: Likewise.
2003 * config/arm/fa526.md: Likewise.
2004 * config/arm/fa606te.md: Likewise.
2005 * config/arm/fa626te.md: Likewise.
2006 * config/arm/fa726te.md: Likewise.
2007 * config/arm/fmp626.md: Likewise.
2008 * config/arm/iwmmxt.md: Likewise.
2009 * config/arm/ldmstm.md: Likewise.
2010 * config/arm/marvell-pj4.md: Likewise.
2011 * config/arm/xgene1.md: Likewise.
2012 * config/aarch64/thunderx.md: Likewise.
2013 * config/aarch64/thunderx2t99.md: Likewise.
2014 * config/aarch64/falkor.md: Likewise.
2016 2017-09-12 Martin Liska <mliska@suse.cz>
2018 * attribs.c (private_lookup_attribute): New function.
2019 * attribs.h (private_lookup_attribute): Declared here.
2020 (lookup_attribute): Called from this place.
2022 2017-09-12 Richard Biener <rguenther@suse.de>
2024 PR tree-optimization/82157
2025 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2026 stmts with side-effects.
2028 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * target.def (hard_regno_nregs): New hook.
2033 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2034 * targhooks.h (default_hard_regno_nregs): Declare.
2035 * targhooks.c (default_hard_regno_nregs): New function.
2036 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2037 (TARGET_HARD_REGNO_NREGS): ...this hook.
2038 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2039 (CLASS_MAX_NREGS): Likewise.
2040 * doc/tm.texi: Regenerate.
2041 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2042 instead of HARD_REGNO_NREGS.
2043 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2044 HARD_REGNO_NREGS in the comment.
2045 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2046 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2047 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2048 Return an unsigned int.
2049 (TARGET_HARD_REGNO_NREGS): Redefine.
2050 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2051 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2052 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2053 (arc_hard_regno_nregs): New function.
2054 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2055 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2056 (arm_hard_regno_nregs): New function.
2057 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2058 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2059 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2060 (TARGET_HARD_REGNO_NREGS): Redefine.
2061 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2062 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2063 (HARD_REGNO_NREGS): Delete.
2064 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2065 (cr16_hard_regno_nregs): New function.
2066 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2067 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2068 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2069 (cris_hard_regno_nregs): New function.
2070 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2071 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2072 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2073 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2074 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2075 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2076 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2077 (frv_hard_regno_nregs): Make static. Take and return an
2079 (frv_class_max_nregs): Remove outdated copy of documentation.
2080 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2081 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2082 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2083 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2084 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2085 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2086 (TARGET_HARD_REGNO_NREGS): Redefine.
2087 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2088 (CLASS_MAX_NREGS): Update comment.
2089 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2090 (ia64_hard_regno_nregs): New function.
2091 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2092 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2093 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2094 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2095 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2097 (m32c_hard_regno_nregs): Likewise. Make static.
2098 (TARGET_HARD_REGNO_NREGS): Redefine.
2099 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2100 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2101 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2102 (m68k_hard_regno_nregs): New function.
2103 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2104 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2105 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2106 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2107 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2108 Take and return an unsigned int.
2109 (TARGET_HARD_REGNO_NREGS): Redefine.
2110 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2111 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2112 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2113 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2114 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2115 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2116 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2117 (msp430_hard_regno_nregs): Make static. Take and return an
2119 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2120 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2121 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2122 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2123 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2124 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2125 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2126 (TARGET_HARD_REGNO_NREGS): Redefine.
2127 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2128 (PA_HARD_REGNO_NREGS): ...this.
2129 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2130 (PA_HARD_REGNO_NREGS): ...this.
2131 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2132 (pa_hard_regno_nregs): New function.
2133 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2134 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2135 (pdp11_hard_regno_nregs): New function.
2136 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2137 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2138 (rs6000_hard_regno_nregs_hook): New function.
2139 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2140 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2141 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2142 Take and return an unsigned int. Move earlier in file.
2143 (TARGET_HARD_REGNO_NREGS): Redefine.
2144 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2145 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2146 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2147 (rl78_hard_regno_nregs): Make static. Take and return an
2149 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2150 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2151 (rs6000_hard_regno_nregs_hook): New function.
2152 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2153 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2154 (TARGET_HARD_REGNO_NREGS): Redefine.
2155 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2156 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2157 instead of HARD_REGNO_NREGS.
2158 (s390_hard_regno_nregs): New function.
2159 (s390_hard_regno_mode_ok): Add comment from s390.h.
2160 (TARGET_HARD_REGNO_NREGS): Redefine.
2161 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2162 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2163 (sh_hard_regno_nregs): New function.
2164 (sh_pass_in_reg_p): Use it.
2165 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2166 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2167 (sparc_hard_regno_nregs): New function.
2168 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2169 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2170 (spu_function_arg_advance): Use it, supplying a valid register number.
2171 (TARGET_HARD_REGNO_NREGS): Redefine.
2172 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2173 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2174 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2175 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2176 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2177 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2178 (CLASS_MAX_NREGS): Remove copy of old documentation.
2179 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2180 (visium_hard_regno_nregs): New function.
2181 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2182 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2183 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2184 xtensa_hard_regno_nregs): New function.
2185 * system.h (HARD_REGNO_NREGS): Poison.
2187 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2189 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2190 hard_regno_nregs instead of HARD_REGNO_NREGS.
2191 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2192 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2193 (c6x_expand_epilogue): Likewise.
2194 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2195 (frv_read_iacc_argument): Likewise.
2196 * config/sh/sh.c: Include regs.h.
2197 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2198 (regs_used): Likewise.
2199 (output_stack_adjust): Likewise.
2200 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2201 * expmed.c: Include regs.h.
2202 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2203 * ree.c: Include regs.h.
2204 (combine_reaching_defs): Use hard_regno_nregs instead of
2206 (add_removable_extension): Likewise.
2208 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2210 * regs.h (hard_regno_nregs): Turn into a function.
2211 (end_hard_regno): Update accordingly.
2212 * caller-save.c (setup_save_areas): Likewise.
2213 (save_call_clobbered_regs): Likewise.
2214 (replace_reg_with_saved_mem): Likewise.
2215 (insert_restore): Likewise.
2216 (insert_save): Likewise.
2217 * combine.c (can_change_dest_mode): Likewise.
2218 (move_deaths): Likewise.
2219 (distribute_notes): Likewise.
2220 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2221 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2222 (rs6000_split_multireg_move): Likewise.
2223 (rs6000_register_move_cost): Likewise.
2224 (rs6000_memory_move_cost): Likewise.
2225 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2226 (rs6000_split_multireg_move): Likewise.
2227 (rs6000_register_move_cost): Likewise.
2228 (rs6000_memory_move_cost): Likewise.
2229 * cselib.c (cselib_reset_table): Likewise.
2230 (cselib_lookup_1): Likewise.
2231 * emit-rtl.c (set_mode_and_regno): Likewise.
2232 * function.c (aggregate_value_p): Likewise.
2233 * ira-color.c (setup_profitable_hard_regs): Likewise.
2234 (check_hard_reg_p): Likewise.
2235 (calculate_saved_nregs): Likewise.
2236 (assign_hard_reg): Likewise.
2237 (improve_allocation): Likewise.
2238 (calculate_spill_cost): Likewise.
2239 * ira-emit.c (modify_move_list): Likewise.
2240 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2241 (ira_hard_reg_in_set_p): Likewise.
2242 * ira.c (setup_reg_mode_hard_regset): Likewise.
2243 (clarify_prohibited_class_mode_regs): Likewise.
2244 (check_allocation): Likewise.
2245 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2246 (lra_setup_reg_renumber): Likewise.
2247 (setup_try_hard_regno_pseudos): Likewise.
2248 (spill_for): Likewise.
2249 (assign_hard_regno): Likewise.
2250 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2251 * lra-constraints.c (in_class_p): Likewise.
2252 (lra_constraint_offset): Likewise.
2253 (simplify_operand_subreg): Likewise.
2254 (lra_constraints): Likewise.
2255 (split_reg): Likewise.
2256 (split_if_necessary): Likewise.
2257 (invariant_p): Likewise.
2258 (inherit_in_ebb): Likewise.
2259 * lra-lives.c (process_bb_lives): Likewise.
2260 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2261 (get_hard_regs): Likewise.
2262 (do_remat): Likewise.
2263 * lra-spills.c (assign_spill_hard_regs): Likewise.
2264 * mode-switching.c (create_pre_exit): Likewise.
2265 * postreload.c (reload_combine_recognize_pattern): Likewise.
2266 * recog.c (peep2_find_free_register): Likewise.
2267 * regcprop.c (kill_value_regno): Likewise.
2268 (set_value_regno): Likewise.
2269 (copy_value): Likewise.
2270 (maybe_mode_change): Likewise.
2271 (find_oldest_value_reg): Likewise.
2272 (copyprop_hardreg_forward_1): Likewise.
2273 * regrename.c (check_new_reg_p): Likewise.
2274 (regrename_do_replace): Likewise.
2275 * reload.c (push_reload): Likewise.
2276 (combine_reloads): Likewise.
2277 (find_dummy_reload): Likewise.
2278 (operands_match_p): Likewise.
2279 (find_reloads): Likewise.
2280 (find_equiv_reg): Likewise.
2281 (reload_adjust_reg_for_mode): Likewise.
2282 * reload1.c (count_pseudo): Likewise.
2283 (count_spilled_pseudo): Likewise.
2284 (find_reg): Likewise.
2285 (clear_reload_reg_in_use): Likewise.
2286 (free_for_value_p): Likewise.
2287 (allocate_reload_reg): Likewise.
2288 (choose_reload_regs): Likewise.
2289 (reload_adjust_reg_for_temp): Likewise.
2290 (emit_reload_insns): Likewise.
2291 (delete_output_reload): Likewise.
2292 * rtlanal.c (subreg_get_info): Likewise.
2293 * sched-deps.c (sched_analyze_reg): Likewise.
2294 * sel-sched.c (init_regs_for_mode): Likewise.
2295 (mark_unavailable_hard_regs): Likewise.
2296 (choose_best_reg_1): Likewise.
2297 (verify_target_availability): Likewise.
2298 * valtrack.c (dead_debug_insert_temp): Likewise.
2299 * var-tracking.c (track_loc_p): Likewise.
2300 (emit_note_insn_var_location): Likewise.
2301 * varasm.c (make_decl_rtl): Likewise.
2302 * reginfo.c (choose_hard_reg_mode): Likewise.
2303 (init_reg_modes_target): Refer directly to
2304 this_target_regs->x_hard_regno_nregs.
2306 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2308 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2309 instead of hard_regno_nregs.
2311 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2313 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2314 end_hard_regno instead of hard_regno_nregs.
2315 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2316 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2317 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2318 * ira-color.c (improve_allocation): Likewise.
2319 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2320 * lra-lives.c (mark_regno_live): Likewise.
2321 (mark_regno_dead): Likewise.
2322 * lra-remat.c (operand_to_remat): Likewise.
2323 * lra.c (collect_non_operand_hard_regs): Likewise.
2324 * postreload.c (reload_combine_note_store): Likewise.
2325 (move2add_valid_value_p): Likewise.
2326 * reload.c (regno_clobbered_p): Likewise.
2328 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2330 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2332 * config/v850/v850.c (v850_reorg): Likewise.
2333 * reload.c (refers_to_regno_for_reload_p): Likewise.
2334 (find_equiv_reg): Likewise.
2335 * reload1.c (reload_reg_reaches_end_p): Likewise.
2337 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2339 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2341 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2342 * config/arm/arm.c (output_move_neon): Likewise.
2343 (arm_attr_length_move_neon): Likewise.
2344 (neon_split_vcombine): Likewise.
2345 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2346 (c6x_mark_reg_written): Likewise.
2347 (c6x_dwarf_register_span): Likewise.
2348 * config/i386/i386.c (ix86_save_reg): Likewise.
2349 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2350 (rws_access_reg): Likewise.
2351 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2352 * mode-switching.c (create_pre_exit): Likewise.
2353 * ree.c (combine_reaching_defs): Likewise.
2354 (add_removable_extension): Likewise.
2355 * regcprop.c (find_oldest_value_reg): Likewise.
2356 (copyprop_hardreg_forward_1): Likewise.
2357 * reload.c (reload_inner_reg_of_subreg): Likewise.
2358 (push_reload): Likewise.
2359 (combine_reloads): Likewise.
2360 (find_dummy_reload): Likewise.
2361 (reload_adjust_reg_for_mode): Likewise.
2362 * reload1.c (find_reload_regs): Likewise.
2363 (forget_old_reloads_1): Likewise.
2364 (reload_reg_free_for_value_p): Likewise.
2365 (reload_adjust_reg_for_temp): Likewise.
2366 (emit_reload_insns): Likewise.
2367 (delete_output_reload): Likewise.
2368 * sel-sched.c (choose_best_reg_1): Likewise.
2369 (choose_best_pseudo_reg): Likewise.
2371 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2372 Alan Hayward <alan.hayward@arm.com>
2373 David Sherwood <david.sherwood@arm.com>
2375 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2376 * target.def (slow_unaligned_access): New hook.
2377 * targhooks.h (default_slow_unaligned_access): Declare.
2378 * targhooks.c (default_slow_unaligned_access): New function.
2379 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2380 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2381 * doc/tm.texi: Regenerate.
2382 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2383 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2384 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2386 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2387 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2389 (rs6000_slow_unaligned_access): New function.
2390 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2391 (expand_block_compare): Likewise.
2392 (expand_strn_compare): Likewise.
2393 (rs6000_rtx_costs): Likewise.
2394 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2395 (riscv_slow_unaligned_access): Likewise.
2396 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2397 (riscv_slow_unaligned_access_p): ...this and make static.
2398 (riscv_option_override): Update accordingly.
2399 (riscv_slow_unaligned_access): New function.
2400 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2401 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2402 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2403 (rs6000_slow_unaligned_access): New function.
2404 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2405 (rs6000_rtx_costs): Likewise.
2406 * config/rs6000/rs6000-string.c (expand_block_compare)
2407 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2408 of SLOW_UNALIGNED_ACCESS.
2409 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2410 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2411 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2412 of SLOW_UNALIGNED_ACCESS.
2413 * expmed.c (simple_mem_bitfield_p): Likewise.
2414 * expr.c (alignment_for_piecewise_move): Likewise.
2415 (emit_group_load_1): Likewise.
2416 (emit_group_store): Likewise.
2417 (copy_blkmode_from_reg): Likewise.
2418 (emit_push_insn): Likewise.
2419 (expand_assignment): Likewise.
2420 (store_field): Likewise.
2421 (expand_expr_real_1): Likewise.
2422 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2423 * lra-constraints.c (simplify_operand_subreg): Likewise.
2424 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2425 * gimple-ssa-store-merging.c: Likewise in block comment at start
2427 * tree-ssa-strlen.c: Include target.h.
2428 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2429 of SLOW_UNALIGNED_ACCESS.
2430 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2434 PR rtl-optimization/82185
2435 * expmed.c (emit_store_flag_int): Only test tem if it has been
2438 2017-09-12 Richard Biener <rguenther@suse.de>
2441 * match.pd ((FTYPE) N CMP CST): Fix typo.
2443 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2445 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2447 (mips_near_type_p): Add 'short_call' attribute as a synonym
2449 * doc/extend.texi (short_call): Document new function attribute.
2451 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2454 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2455 assertion check that in the condition.
2456 (get_atomic_generic_size): Likewise. Before testing if parameter
2457 has pointer type, if it has array type, call for C++
2458 default_conversion to perform array-to-pointer conversion.
2460 2017-09-12 Richard Biener <rguenther@suse.de>
2462 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2463 for operations we cannot scalarize.
2465 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2467 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2468 vectors heap vectors. Clean up comments.
2469 Make visited_bbs a reference.
2470 (profitable_jump_thread_path): Make GC
2471 vectors heap vectors. Clean up comments.
2473 (convert_and_register_jump_thread_path): Make GC vectors heap
2475 (check_subpath_and_update_thread_path): Same. Clean up comments.
2476 Make visited_bbs a reference.
2477 (handle_phi): Abstract common code to to
2478 register_jump_thread_path_if_profitable.
2479 Rename VAR_BB to DEF_BB.
2481 Make GC vectors heap vectors.
2482 Make visited_bbs a reference.
2483 (handle_assignment): Same.
2484 (register_jump_thread_path_if_profitable): New.
2485 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2487 Make GC vectors heap vectors. Clean up comments.
2488 Make visited_bbs a reference.
2489 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2490 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2493 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2496 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2497 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2499 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2501 Revert r251800 and r251799.
2503 2017-09-11 Martin Jambor <mjambor@suse.cz>
2506 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2507 arguments in advance.
2508 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2509 use it to find predecessor edges.
2510 (naive_outof_ssa): Collect vector of predecessors.
2512 2017-09-08 Jason Merrill <jason@redhat.com>
2514 PR c++/70029 - ICE with ref-qualifier and -flto
2515 * langhooks.h (struct lang_hooks_for_types): Add
2516 copy_lang_qualifiers.
2517 * attribs.c (build_type_attribute_qual_variant): Use it.
2518 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2520 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2521 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2523 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2526 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2527 (*mulsi3_sp64): New instruction.
2528 (mulsi3): New expander.
2530 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2532 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2534 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2536 * sancov.c: Include memmodel.h.
2538 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2541 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2544 2017-09-07 Carl Love <cel@us.ibm.com>
2546 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2547 the sldi instruction.
2549 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2551 * sancov.c: Include tm_p.h.
2553 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2556 * output.h (switch_to_other_text_partition): New declaration.
2557 * varasm.c (switch_to_other_text_partition): New function.
2558 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
2559 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
2560 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
2561 to the other text partition before emitting LCL label and switch back
2562 after emitting the word after it.
2564 2017-09-07 Richard Biener <rguenther@suse.de>
2566 * passes.def (pass_split_crit_edges): Remove instance before PRE.
2567 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
2568 critical edges here, after loop init.
2569 (pass_data_pre): Remove PROP_no_crit_edges flags.
2570 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
2571 for valueization of call args to avoid leaking VN_TOP.
2572 (visit_use): Assert we do not visit default defs.
2573 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
2574 Use error_mark_node to more easily detect leaking VN_TOP.
2575 All default-defs are varying, not VN_TOP. Mark them visited.
2576 (run_scc_vn): Make code match comment.
2578 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2580 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
2581 OPTION_MASK_FLOAT128_KEYWORD.
2582 (POWERPC_MASKS): Likewise.
2583 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2584 support for the -mfloat128-type option, and make -mfloat128
2585 default on PowerPC Linux systems. Define or undefine
2586 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2587 Define __float128 to be __ieee128 if IEEE 128-bit support is
2588 enabled, or undefine it.
2589 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2590 Delete defining __FLOAT128_TYPE__.
2591 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2592 -mfloat128-type option and make -mfloat128 default on PowerPC
2594 (TARGET_FLOAT128_TYPE): Likewise.
2595 (-mfloat128-type): Likewise.
2596 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2597 Delete the -mfloat128-type option and make -mfloat128 default on
2598 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2599 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2600 128-bit floating point is enabled. Change tests from using
2601 -mfloat128-type to -mfloat128.
2602 (rs6000_mangle_type): Use the correct mangling for the __float128
2603 type even if normal long double is restricted to 64-bits.
2604 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2606 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2607 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2608 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2610 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2611 documentation for -mfloat128.
2613 2017-09-06 Olivier Hainque <hainque@adacore.com>
2615 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2617 2017-09-06 Wish Wu <wishwu007@gmail.com>
2618 Jakub Jelinek <jakub@redhat.com>
2620 * asan.c (initialize_sanitizer_builtins): Add
2621 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2622 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2623 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2624 BT_FN_VOID_UINT64_PTR variables.
2625 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2626 (BT_FN_VOID_UINT16_UINT16): Likewise.
2627 (BT_FN_VOID_UINT32_UINT32): Likewise.
2628 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2629 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2630 (BT_FN_VOID_UINT64_PTR): Likewise.
2631 * common.opt (flag_sanitize_coverage): New variable.
2632 (fsanitize-coverage=trace-pc): Remove.
2633 (fsanitize-coverage=): Add.
2634 * flag-types.h (enum sanitize_coverage_code): New enum.
2635 * fold-const.c (fold_range_test): Disable non-short-circuit
2636 optimization if flag_sanitize_coverage.
2637 (fold_truth_andor): Likewise.
2638 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2639 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2640 (coverage_sanitizer_opts): New array.
2641 (get_closest_sanitizer_option): Add OPTS argument, handle also
2642 OPT_fsanitize_coverage_.
2643 (parse_sanitizer_options): Adjusted to also handle
2644 OPT_fsanitize_coverage_.
2645 (common_handle_option): Add OPT_fsanitize_coverage_.
2646 * sancov.c (instrument_comparison, instrument_switch): New function.
2647 (sancov_pass): Add trace-cmp support.
2648 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2649 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2650 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2651 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2652 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2653 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2654 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2655 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2656 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2657 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2659 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2661 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2662 error. Only quit immediately if parsing is complete.
2663 (BEGIN): Initialize fatal_err and parse_done.
2664 (begin fpu, end fpu): Check number of arguments.
2665 (begin arch, end arch): Likewise.
2666 (begin cpu, end cpu): Likewise.
2667 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2668 (optalias): Likewise.
2670 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2672 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2673 * config/arm/arm-isa.h: Delete. Move definitions to ...
2674 * arm-cpus.in: ... here. Use new feature and fgroup values.
2675 * config/arm/arm.c (arm_option_override): Use lower case for feature
2677 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2678 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2679 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2681 (print_isa_bits_for): New function.
2682 (gen_isa): New function.
2683 (gen_comm_data): Use print_isa_bits_for.
2684 (define feature): New keyword.
2685 (define fgroup): New keyword.
2686 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2687 (arm-isa.h): Add rule to generate file.
2688 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2689 case for feature bit names.
2691 2017-09-06 Richard Biener <rguenther@suse.de>
2693 * tree-ssa-pre.c (NECESSARY): Remove.
2694 (create_expression_by_pieces): Do not touch pass-local flags.
2695 (insert_into_preds_of_block): Likewise.
2696 (do_pre_regular_insertion): Likewise.
2697 (eliminate_insert): Likewise.
2698 (eliminate_dom_walker::before_dom_children): Likewise.
2699 (fini_eliminate): Do not look at inserted_exprs.
2700 (mark_operand_necessary): Remove.
2701 (remove_dead_inserted_code): Replace with simple work-list
2702 algorithm based on inserted_exprs and SSA uses.
2703 (pass_pre::execute): Re-order fini_eliminate and
2704 remove_dead_inserted_code.
2706 2017-09-06 Olivier Hainque <hainque@adacore.com>
2708 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2709 for VxWorks 7. Adjust surrounding comments.
2711 2017-09-06 Richard Biener <rguenther@suse.de>
2713 * gimple-ssa-strength-reduction.c
2714 (find_candidates_dom_walker::before_dom_children): Also allow
2717 2017-09-06 Richard Biener <rguenther@suse.de>
2719 PR tree-optimization/82108
2720 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2721 for gap in the non-permutation SLP case.
2723 2017-09-06 Martin Jambor <mjambor@suse.cz>
2725 PR tree-optimization/82078
2726 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2727 add_access_to_work_queue...
2728 (build_accesses_from_assign): ...here.
2729 (propagate_all_subaccesses): Make sure racc is the group
2730 representative, if there is one.
2732 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2735 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2738 2017-09-06 Richard Biener <rguenther@suse.de>
2740 * gimple-ssa-strength-reduction.c
2741 (find_candidates_dom_walker::before_doom_children): Use a
2742 type and not a mode check.
2744 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2747 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2749 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2751 2017-09-05 Jeff Law <law@redhat.com>
2753 PR tree-optimization/64910
2754 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2755 cases where we have 3 or more operands.
2757 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2760 * omp-low.c (lower_omp_for): Recompute tree invariant if
2761 gimple_omp_for_initial/final is ADDR_EXPR.
2764 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2765 into gimple val before gimplification fo the COND_EXPR.
2767 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2769 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2770 REGION_COPY argument.
2771 (thread_through_all_blocks): Remove unused argument to
2772 duplicate_thread_path.
2774 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2775 Alan Hayward <alan.hayward@arm.com>
2776 David Sherwood <david.sherwood@arm.com>
2778 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2779 Take a scalar_mode rather than a machine_mode.
2780 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2781 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2782 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2783 (aarch64_gen_adjusted_ldpstp): Likewise.
2784 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2786 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2787 Alan Hayward <alan.hayward@arm.com>
2788 David Sherwood <david.sherwood@arm.com>
2790 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2791 Take a scalar_int_mode instead of a machine_mode.
2792 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2793 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2794 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2795 (aarch64_simd_attr_length_rglist): Delete.
2796 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2797 a scalar_int_mode instead of a machine_mode.
2798 (aarch64_add_offset): Likewise.
2799 (aarch64_internal_mov_immediate): Likewise
2800 (aarch64_add_constant_internal): Likewise.
2801 (aarch64_add_constant): Likewise.
2802 (aarch64_movw_imm): Likewise.
2803 (aarch64_rtx_arith_op_extract_p): Likewise.
2804 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2805 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2806 Remove assert that the mode isn't a vector.
2807 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2808 (aarch64_expand_mov_immediate): Update calls after above changes.
2809 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2810 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2811 (aarch64_move_imm): Likewise.
2812 (aarch64_can_const_movi_rtx_p): Likewise.
2813 (aarch64_strip_extend): Likewise.
2814 (aarch64_extr_rtx_p): Likewise.
2815 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2816 a CONST_INT when the mode parameter is VOIDmode.
2817 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2819 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2821 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2822 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2823 (bitwise_type_for_mode): Update accordingly.
2825 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2827 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2828 * stor-layout.c (mode_for_size_tree): Likewise.
2829 (mode_for_array): Update accordingly.
2830 (layout_decl): Likewise.
2831 (compute_record_mode): Likewise. Only set the mode once.
2833 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2835 * target.def (get_mask_mode): Change return type to opt_mode.
2837 * doc/tm.texi: Regenerate.
2838 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2839 * targhooks.c (default_get_mask_mode): Likewise.
2840 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2841 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2842 targetm.get_mask_mode.
2843 * tree.c (build_truth_vector_type): Likewise.
2845 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2847 * machmode.h (mode_for_vector): Return an opt_mode.
2848 * stor-layout.c (mode_for_vector): Likewise.
2849 (mode_for_int_vector): Update accordingly.
2850 (layout_type): Likewise.
2851 * config/i386/i386.c (emit_memmov): Likewise.
2852 (ix86_expand_set_or_movmem): Likewise.
2853 (ix86_expand_vector_init): Likewise.
2854 (ix86_get_mask_mode): Likewise.
2855 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2857 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2858 * expmed.c (extract_bit_field_1): Likewise.
2859 * expr.c (expand_expr_real_2): Likewise.
2860 * optabs-query.c (can_vec_perm_p): Likewise.
2861 (can_vec_mask_load_store_p): Likewise.
2862 * optabs.c (expand_vec_perm): Likewise.
2863 * targhooks.c (default_get_mask_mode): Likewise.
2864 * tree-vect-stmts.c (vectorizable_store): Likewise.
2865 (vectorizable_load): Likewise.
2866 (get_vectype_for_scalar_type_and_size): Likewise.
2868 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2870 * machmode.h (mode_for_int_vector): New function.
2871 * stor-layout.c (mode_for_int_vector): Likewise.
2872 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2873 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2874 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2875 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2876 (s390_expand_vcond): Likewise.
2878 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2880 * machmode.h (opt_machine_mode): New type.
2881 (opt_mode<T>): Allow construction from anything that can be
2883 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2884 (mode_for_size): Return an opt_machine_mode.
2885 * stor-layout.c (mode_for_size): Likewise.
2886 (mode_for_size_tree): Update call accordingly.
2887 (bitwise_mode_for_mode): Likewise.
2888 (make_fract_type): Likewise.
2889 (make_accum_type): Likewise.
2890 * caller-save.c (replace_reg_with_saved_mem): Update call
2892 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2893 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2894 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2895 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2896 * expmed.c (extract_bit_field_1): Likewise.
2897 * reload.c (get_secondary_mem): Likewise.
2898 * varasm.c (assemble_integer): Likewise.
2899 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2902 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2904 * machmode.h (decimal_float_mode_for_size): New function.
2905 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2906 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2907 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2908 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2909 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2910 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2912 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2914 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2915 (get_builtin_sync_mode): Likewise.
2916 (expand_ifn_atomic_compare_exchange): Likewise.
2917 (expand_builtin_atomic_clear): Likewise.
2918 (expand_builtin_atomic_test_and_set): Likewise.
2919 (fold_builtin_atomic_always_lock_free): Likewise.
2920 * calls.c (compute_argument_addresses): Likewise.
2921 (emit_library_call_value_1): Likewise.
2922 (store_one_arg): Likewise.
2923 * combine.c (combine_instructions): Likewise.
2924 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2925 * config/arm/arm.c (arm_function_value): Likewise.
2926 (aapcs_allocate_return_reg): Likewise.
2927 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2928 * config/i386/i386.c (construct_container): Likewise.
2929 (ix86_gimplify_va_arg): Likewise.
2930 (ix86_expand_sse_cmp): Likewise.
2931 (emit_memmov): Likewise.
2932 (emit_memset): Likewise.
2933 (expand_small_movmem_or_setmem): Likewise.
2934 (ix86_expand_pextr): Likewise.
2935 (ix86_expand_pinsr): Likewise.
2936 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2937 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2939 * config/mips/mips.c (mips_function_value_1) Likewise.
2940 (mips_block_move_straight): Likewise.
2941 (mips_expand_ins_as_unaligned_store): Likewise.
2942 * config/powerpcspe/powerpcspe.c
2943 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2944 (rs6000_darwin64_record_arg_flush): Likewise.
2945 * config/rs6000/rs6000.c
2946 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2947 (rs6000_darwin64_record_arg_flush): Likewise.
2948 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2949 (sparc_function_value_1): Likewise.
2950 * config/spu/spu.c (adjust_operand): Likewise.
2951 (spu_emit_branch_or_set): Likewise.
2952 (arith_immediate_p): Likewise.
2953 * emit-rtl.c (gen_lowpart_common): Likewise.
2954 * expr.c (expand_expr_real_1): Likewise.
2955 * function.c (assign_parm_setup_block): Likewise.
2956 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2957 * reload1.c (alter_reg): Likewise.
2958 * stor-layout.c (mode_for_vector): Likewise.
2959 (layout_type): Likewise.
2961 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2963 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2964 (spu_convert_move): Likewise.
2965 * lower-subreg.c (resolve_simple_move): Likewise.
2967 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2970 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2971 define_insn to a define_expand.
2972 (altivec_vsum2sws_direct): New define_insn.
2973 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2975 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2977 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2980 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2985 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2986 realign stack if stack alignment needed is less than incoming
2989 2017-09-05 Marek Polacek <polacek@redhat.com>
2992 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2995 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2997 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2999 2017-09-05 Richard Biener <rguenther@suse.de>
3001 PR tree-optimization/82084
3002 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3004 2017-09-05 Richard Biener <rguenther@suse.de>
3006 PR tree-optimization/82102
3007 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3009 2017-09-05 Martin Liska <mliska@suse.cz>
3011 PR tree-optimization/82032
3012 * tree-cfg.c (generate_range_test): New function.
3013 * tree-cfg.h (generate_range_test): Declared here.
3014 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3015 (cleanup_control_expr_graph): Use it.
3016 * tree-switch-conversion.c (try_switch_expansion): Remove
3018 (emit_case_nodes): Use generate_range_test.
3020 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3023 * config/i386/i386.md (*<btsc><mode>_mask): Add
3024 TARGET_USE_BT to insn constraint.
3025 (*btr<mode>_mask): Ditto.
3027 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3029 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3030 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3032 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3035 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3036 TARGET_NEON and TARGET_IWMMXT.
3037 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3038 TARGET_NEON and TARGET_IWMMXT.
3039 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3041 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3043 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3044 (ix86_rewrite_tls_address): Ditto.
3045 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3046 (ix86_rewrite_tls_address_1): Ditto.
3047 (ix86_rewrite_tls_address): Ditto.
3048 * config/i386/predicates.md (tls_address_pattern): New predicate.
3049 * config/i386/i386.md (TLS address splitter): New splitter.
3051 2017-09-04 Richard Biener <rguenther@suse.de>
3053 PR tree-optimization/82084
3054 * fold-const.h (can_native_encode_string_p): Declare.
3055 * fold-const.c (can_native_encode_string_p): Factor out from ...
3056 (native_encode_string): ... here.
3057 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3058 vectorizing stores from constants we later cannot handle.
3060 2017-09-04 Marek Polacek <polacek@redhat.com>
3063 * doc/invoke.texi: Update -Wtautological-compare documentation.
3065 2017-09-04 Jeff Law <law@redhat.com>
3067 PR tree-optimization/64910
3068 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3069 swap the first and last operand if the last is a constant.
3071 2017-09-04 Marek Polacek <polacek@redhat.com>
3074 * convert.c (do_narrow): When sanitizing signed integer overflows,
3075 bail out for signed types.
3076 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3078 2017-09-04 Richard Biener <rguenther@suse.de>
3080 PR tree-optimization/82060
3081 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3082 Move devirtualization after stmt folding and before EH/AB/noreturn
3083 cleanup to get the stmt refs canonicalized. Use a bool instead
3084 of gimple_modified_p since that doesn't work for NOPs. Schedule
3085 NOPs generated by folding for removal.
3087 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3088 Alan Hayward <alan.hayward@arm.com>
3089 David Sherwood <david.sherwood@arm.com>
3091 * coretypes.h (pad_direction): New enum.
3092 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3093 (FUNCTION_ARG_PADDING): Likewise.
3094 * target.def (function_arg_padding): New hook.
3095 * targhooks.h (default_function_arg_padding): Declare.
3096 * targhooks.c (default_function_arg_padding): New function.
3097 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3098 (TARGET_FUNCTION_ARG_PADDING): ...this.
3099 * doc/tm.texi: Regenerate.
3100 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3101 instead of direction.
3102 (compute_argument_addresses): Likewise.
3103 (load_register_parameters): Likewise.
3104 (emit_library_call_value_1): Likewise.
3105 (store_one_arg): Use targetm.calls.function_arg_padding instead
3106 of FUNCTION_ARG_PADDING.
3107 (must_pass_in_stack_var_size_or_pad): Likewise.
3108 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3109 (emit_group_store): Likewise.
3110 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3111 instead of FUNCTION_ARG_PADDING.
3112 (emit_push_insn): Likewise, and propagate enum change throughout
3114 * function.h (direction): Delete.
3115 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3117 * function.c (assign_parm_find_stack_rtl): Likewise.
3118 (assign_parm_setup_block_p): Likewise.
3119 (assign_parm_setup_block): Likewise.
3120 (gimplify_parameters): Likewise.
3121 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3122 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3124 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3125 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3126 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3127 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3128 (aarch64_function_arg_padding): ...this new function.
3129 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3130 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3131 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3132 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3133 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3134 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3135 (arm_pad_arg_upward): Replace with...
3136 (arm_function_arg_padding): ...this new function.
3137 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3139 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3140 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3141 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3142 (ia64_hpux_function_arg_padding): Replace with...
3143 (ia64_function_arg_padding): ...this new function. Use pad_direction
3144 instead of direction. Check for TARGET_HPUX.
3145 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3146 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3147 (iq2000_function_arg_padding): New function.
3148 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3149 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3150 (mips_function_arg_padding): ...this new function.
3151 (mips_pad_reg_upward): Update accordingly.
3152 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3153 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3154 targetm.calls.function_arg_padding.
3155 (FUNCTION_ARG_PADDING): Delete.
3156 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3157 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3158 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3159 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3160 (nios2_block_reg_padding): Return pad_direction instead of direction.
3161 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3162 instead of direction.
3163 (nios2_function_arg_padding): Likewise. Make static.
3164 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3165 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3166 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3167 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3168 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3169 (pa_function_arg_padding): Make static. Return pad_direction instead
3171 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3172 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3173 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3174 instead of direction. Use targetm.calls.function_arg_padding.
3175 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3176 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3177 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3178 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3179 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3181 (function_arg_padding): Rename to...
3182 (rs6000_function_arg_padding): ...this. Make static. Return
3183 pad_direction instead of direction.
3184 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3185 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3186 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3187 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3188 instead of direction. Use targetm.calls.function_arg_padding.
3189 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3190 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3191 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3192 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3193 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3194 (function_arg_padding): Rename to...
3195 (rs6000_function_arg_padding): ...this. Make static. Return
3196 pad_direction instead of direction.
3197 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3198 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3199 * config/s390/s390.c (s390_function_arg_padding): New function.
3200 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3201 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3202 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3203 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3204 (function_arg_padding): Rename to...
3205 (sparc_function_arg_padding): ...this. Make static. Return
3206 pad_direction instead of direction.
3207 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3208 * config/spu/spu.c (spu_function_arg_padding): New function.
3209 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3210 * system.h (FUNCTION_ARG_PADDING): Poison.
3212 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3213 Alan Hayward <alan.hayward@arm.com>
3214 David Sherwood <david.sherwood@arm.com>
3216 * target.def (modes_tieable_p): New hook.
3217 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3218 (TARGET_MODES_TIEABLE_P): ...this.
3219 * doc/tm.texi.in: Regenerate.
3220 * hooks.h (hook_bool_mode_mode_true): Declare.
3221 * hooks.c (hook_bool_mode_mode_true): New function.
3222 * combine.c (subst): Use targetm.modes_tieable_p instead of
3224 * dse.c (find_shift_sequence): Likewise.
3225 * expmed.c (extract_low_bits): Likewise.
3226 * lower-subreg.c: Include target.h.
3227 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3229 * rtlanal.c (rtx_cost): Likewise.
3230 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3231 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3232 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3233 (TARGET_MODES_TIEABLE_P): Redefine.
3234 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3235 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3236 (TARGET_MODES_TIEABLE_P): Redefine.
3237 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3238 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3239 (arc_modes_tieable_p): New function.
3240 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3241 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3242 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3243 (arm_modes_tieable_p): Make static.
3244 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3245 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3246 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3247 (TARGET_MODES_TIEABLE_P): Redefine.
3248 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3249 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3250 (TARGET_MODES_TIEABLE_P): Redefine.
3251 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3252 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3253 (cr16_modes_tieable_p): New function.
3254 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3255 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3256 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3257 (TRULY_NOOP_TRUNCATION): Update comment.
3258 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3259 (TRULY_NOOP_TRUNCATION): Update comment.
3260 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3261 (frv_modes_tieable_p): New function.
3262 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3263 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3264 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3265 (TARGET_MODES_TIEABLE_P): Redefine.
3266 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3267 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3268 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3269 (TARGET_MODES_TIEABLE_P): Redefine.
3270 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3271 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3272 (ia64_modes_tieable_p): New function.
3273 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3274 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3275 (iq2000_modes_tieable_p): New function.
3276 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3277 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3278 (lm32_modes_tieable_p): New function.
3279 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3280 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3281 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3282 (TARGET_MODES_TIEABLE_P): Redefine.
3283 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3284 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3285 (m32r_modes_tieable_p): New function.
3286 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3287 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3288 (m68k_modes_tieable_p): New function.
3289 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3290 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3291 (mcore_modes_tieable_p): New function.
3292 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3293 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3295 (TARGET_MODES_TIEABLE_P): Redefine.
3296 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3297 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3298 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3299 (TARGET_MODES_TIEABLE_P): Redefine.
3300 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3301 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3302 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3303 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3304 (mn10300_modes_tieable_p): ...this and make static.
3305 (TARGET_MODES_TIEABLE_P): Redefine.
3306 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3307 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3308 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3309 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3310 (msp430_modes_tieable_p): Make static.
3311 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3312 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3313 (TARGET_MODES_TIEABLE_P): Redefine.
3314 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3315 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3316 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3317 (TARGET_MODES_TIEABLE_P): Redefine.
3318 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3319 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3320 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3321 (TARGET_MODES_TIEABLE_P): Redefine.
3322 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3323 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3324 (pdp11_modes_tieable_p): New function.
3325 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3326 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3327 (rs6000_modes_tieable_p): New function.
3328 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3329 * config/powerpcspe/powerpcspe.md: Update comment.
3330 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3331 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3332 (TARGET_MODES_TIEABLE_P): Redefine.
3333 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3334 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3335 (rl78_modes_tieable_p): New function.
3336 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3337 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3338 (rs6000_modes_tieable_p): New function.
3339 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3340 * config/rs6000/rs6000.md: Update comment.
3341 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3342 * config/rx/rx.c (rx_modes_tieable_p): New function.
3343 (TARGET_MODES_TIEABLE_P): Redefine.
3344 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3345 * config/s390/s390.c (s390_modes_tieable_p): New function.
3346 (TARGET_MODES_TIEABLE_P): Redefine.
3347 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3348 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3349 (sh_modes_tieable_p): New function.
3350 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3351 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3352 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3353 (sparc_modes_tieable_p): Make static.
3354 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3355 * config/spu/spu.c (spu_modes_tieable_p): New function.
3356 (TARGET_MODES_TIEABLE_P): Redefine.
3357 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3358 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3359 (TARGET_MODES_TIEABLE_P): Redefine.
3360 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3361 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3362 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3363 * config/v850/v850.c (v850_modes_tieable_p): New function.
3364 (TARGET_MODES_TIEABLE_P): Redefine.
3365 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3366 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3367 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3368 (visium_modes_tieable_p): New function.
3369 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3370 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3371 (xtensa_modes_tieable_p): New function.
3372 * system.h (MODES_TIEABLE_P): Poison.
3374 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3375 Alan Hayward <alan.hayward@arm.com>
3376 David Sherwood <david.sherwood@arm.com>
3378 * target.def (hard_regno_mode_ok): New hook.
3379 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3380 (TARGET_HARD_REGNO_MODE_OK): ...this.
3381 * doc/tm.texi.in: Regenerate.
3382 * hooks.h (hook_bool_uint_mode_true): Declare.
3383 * hooks.c (hook_bool_uint_mode_true): New function.
3384 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3386 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3387 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3388 instead of HARD_REGNO_MODE_OK.
3389 * caller-save.c: Include target.h.
3390 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3392 * combine.c (can_combine_p): Likewise.
3393 (combinable_i3pat): Likewise.
3394 (can_change_dest_mode): Likewise.
3395 * expr.c (init_expr_target): Likewise.
3396 (convert_move): Likewise.
3397 (convert_modes): Likewise.
3398 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3399 (setup_prohibited_mode_move_regs): Likewise.
3400 * ira.h (target_ira): Likewise.
3401 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3402 * lra-constraints.c (process_alt_operands): Likewise.
3403 (split_reg): Likewise.
3404 * recog.c (peep2_find_free_register): Likewise.
3405 * ree.c (combine_reaching_defs): Likewise.
3406 * regcprop.c (maybe_mode_change): Likewise.
3407 * reginfo.c (init_reg_sets_1): Likewise.
3408 (choose_hard_reg_mode): Likewise.
3409 (simplifiable_subregs): Likewise.
3410 * regrename.c (check_new_reg_p): Likewise.
3411 * reload.c (find_valid_class): Likewise.
3412 (find_valid_class_1): Likewise.
3413 (reload_inner_reg_of_subreg): Likewise.
3414 (push_reload): Likewise.
3415 (combine_reloads): Likewise.
3416 (find_dummy_reload): Likewise.
3417 (find_reloads): Likewise.
3418 * reload1.c (find_reg): Likewise.
3419 (set_reload_reg): Likewise.
3420 (allocate_reload_reg): Likewise.
3421 (choose_reload_regs): Likewise.
3422 (reload_adjust_reg_for_temp): Likewise.
3423 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3424 (simplify_subreg_regno): Likewise.
3425 * sel-sched.c (init_regs_for_mode): Likewise.
3426 * varasm.c (make_decl_rtl): Likewise.
3427 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3428 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3430 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3431 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3433 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3434 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3436 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3437 (arc_mode_class): Delete.
3438 (HARD_REGNO_MODE_OK): Delete.
3439 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3440 (arc_hard_regno_mode_ok): Rename old array to...
3441 (arc_hard_regno_mode_ok_modes): ...this.
3442 (arc_conditional_register_usage): Update accordingly.
3443 (arc_mode_class): Make static.
3444 (arc_hard_regno_mode_ok): New function.
3445 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3446 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3447 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3448 (arm_hard_regno_mode_ok): Make static.
3449 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3451 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3452 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3453 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3455 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3456 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3457 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3458 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3459 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3460 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3461 * config/bfin/predicates.md (valid_reg_operand): Use
3462 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3463 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3464 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3465 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3466 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3467 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3468 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3469 (cr16_hard_regno_mode_ok): Make static and return a bool.
3470 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3471 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3472 (cris_hard_regno_mode_ok): New function.
3473 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3474 (epiphany_mode_class): Delete.
3475 (HARD_REGNO_MODE_OK): Delete.
3476 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3477 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3478 (hard_regno_mode_ok): Rename to...
3479 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3480 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3481 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3483 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3484 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3485 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3486 (frv_hard_regno_mode_ok): Make static and return a bool.
3487 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3489 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3490 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3491 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3492 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3495 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3496 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3497 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3500 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3501 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3502 (ia64_hard_regno_mode_ok): New function.
3503 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3504 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3505 (iq2000_hard_regno_mode_ok): New function.
3506 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3507 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3508 (lm32_hard_regno_mode_ok): New function.
3509 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3510 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3511 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3512 instead of HARD_REGNO_MODE_OK.
3513 (m32c_hard_regno_ok): Rename to...
3514 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3515 (m32c_cannot_change_mode_class): Update accordingly.
3516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3517 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3518 (m32r_mode_class): Delete.
3519 (HARD_REGNO_MODE_OK): Delete.
3520 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3521 (m32r_hard_regno_mode_ok): Rename to...
3522 (m32r_hard_regno_modes): ...this.
3523 (m32r_mode_class): Make static.
3524 (m32r_hard_regno_mode_ok): New function.
3525 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3526 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3527 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3528 (m68k_hard_regno_mode_ok): Make static.
3529 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3530 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3531 (mcore_hard_regno_mode_ok): New function.
3532 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3533 (HARD_REGNO_MODE_OK): Delete.
3534 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3536 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3537 (microblaze_hard_regno_mode_ok): New function.
3538 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3539 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3540 (mips_hard_regno_mode_ok): Delete.
3541 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3542 (mips_hard_regno_mode_ok_p): ...this and make static.
3543 (mips_hard_regno_mode_ok_p): Rename to...
3544 (mips_hard_regno_mode_ok_uncached): ...this.
3545 (mips_hard_regno_mode_ok): New function.
3546 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3547 of HARD_REGNO_MODE_OK.
3548 (mips_option_override): Update after above name changes.
3549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3550 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3551 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3552 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3553 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3555 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3556 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3557 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3558 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3559 (msp430_hard_regno_mode_ok): Make static and return a bool.
3560 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
3561 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
3562 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
3564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3565 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
3566 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
3567 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
3568 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
3569 (PA_HARD_REGNO_MODE_OK): ...this
3570 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
3571 (PA_HARD_REGNO_MODE_OK): ...this.
3572 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3573 (pa_hard_regno_mode_ok): New function.
3574 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
3575 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3576 (pdp11_hard_regno_mode_ok): New function.
3577 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
3578 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
3580 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
3582 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3583 (rs6000_hard_regno_mode_ok): Rename to...
3584 (rs6000_hard_regno_mode_ok_uncached): ...this.
3585 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3586 (rs6000_hard_regno_mode_ok): New function.
3587 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3588 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3589 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3590 (riscv_hard_regno_mode_ok): ...this and make static.
3591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3592 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3593 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3594 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3595 (rl78_hard_regno_mode_ok): Make static and return bool.
3596 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3597 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3599 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3600 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3601 (rs6000_hard_regno_mode_ok): Rename to...
3602 (rs6000_hard_regno_mode_ok_uncached): ...this.
3603 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3604 (rs6000_hard_regno_mode_ok): New function.
3605 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3606 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3608 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3609 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3610 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3612 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3613 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3614 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3615 (sh_hard_regno_mode_ok): Make static.
3616 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3617 instead of HARD_REGNO_MODE_OK.
3618 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3619 (sparc_mode_class): Delete.
3620 (HARD_REGNO_MODE_OK): Delete.
3621 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3622 (hard_regno_mode_classes): Make static.
3623 (sparc_mode_class): Likewise.
3624 (sparc_hard_regno_mode_ok): New function.
3625 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3626 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3627 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3630 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3631 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3632 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3633 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3635 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3636 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3637 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3638 (visium_hard_regno_mode_ok): New function.
3639 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3640 instead of HARD_REGNO_MODE_OK.
3641 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3642 (HARD_REGNO_MODE_OK): Delete.
3643 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3644 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3645 (xtensa_option_override): Update accordingly.
3646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3647 (xtensa_hard_regno_mode_ok): New function.
3648 * system.h (HARD_REGNO_MODE_OK): Poison.
3650 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3651 Alan Hayward <alan.hayward@arm.com>
3652 David Sherwood <david.sherwood@arm.com>
3654 * target.def (hard_regno_call_part_clobbered): New hook.
3655 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3656 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3657 * doc/tm.texi: Regenerate.
3658 * hooks.h (hook_bool_uint_mode_false): Declare.
3659 * hooks.c (hook_bool_uint_mode_false): New function.
3660 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3661 * cselib.c (cselib_process_insn): Use
3662 targetm.hard_regno_call_part_clobbered instead of
3663 HARD_REGNO_CALL_PART_CLOBBERED.
3664 * ira-conflicts.c (ira_build_conflicts): Likewise.
3665 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3666 * lra-constraints.c (need_for_call_save_p): Likewise.
3667 * lra-lives.c: Include target.h.
3668 (check_pseudos_live_through_calls): Use
3669 targetm.hard_regno_call_part_clobbered instead of
3670 HARD_REGNO_CALL_PART_CLOBBERED.
3671 * regcprop.c: Include target.h.
3672 (copyprop_hardreg_forward_1): Use
3673 targetm.hard_regno_call_part_clobbered instead of
3674 HARD_REGNO_CALL_PART_CLOBBERED.
3675 * reginfo.c (choose_hard_reg_mode): Likewise.
3676 * regrename.c (check_new_reg_p): Likewise.
3677 * reload.c (find_equiv_reg): Likewise.
3678 * reload1.c (emit_reload_insns): Likewise.
3679 * sched-deps.c (deps_analyze_insn): Likewise.
3680 * sel-sched.c (init_regs_for_mode): Likewise.
3681 (mark_unavailable_hard_regs): Likewise.
3682 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3683 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3684 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3686 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3687 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3688 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3690 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3692 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3693 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3694 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3696 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3697 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3698 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3700 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3701 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3703 * config/powerpcspe/powerpcspe.c
3704 (rs6000_hard_regno_call_part_clobbered): New function.
3705 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3706 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3707 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3709 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3710 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3711 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3713 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3714 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3715 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3717 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3718 Alan Hayward <alan.hayward@arm.com>
3719 David Sherwood <david.sherwood@arm.com>
3721 * rtl.h (subreg_memory_offset): Declare.
3722 * emit-rtl.c (subreg_memory_offset): New function.
3723 * expmed.c (store_bit_field_1): Use it.
3724 * expr.c (undefined_operand_subword_p): Likewise.
3725 * simplify-rtx.c (simplify_subreg): Likewise.
3727 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3729 PR rtl-optimization/57448
3732 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3733 using atomic_load pattern.
3734 (expand_atomic_store): Likewise.
3736 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3739 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3740 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3741 handling. Use replace_call_with_value with NULL instead of
3742 gsi_replace, unlink_stmt_vdef and release_defs.
3744 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3747 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3749 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3752 * rtl.h (emit_library_call_value_1): Declare.
3753 (emit_library_call): Replace declaration with a series of overloads.
3754 Remove the parameter count argument.
3755 (emit_library_call_value): Likewise.
3756 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3757 with an "rtx_mode_t *".
3758 (emit_library_call_value): Delete.
3759 (emit_library_call): Likewise.
3760 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3761 (asan_emit_allocas_unpoison): Likewise.
3762 * builtins.c (expand_builtin_powi): Likewise.
3763 (expand_asan_emit_allocas_unpoison): Likewise.
3764 * cfgexpand.c (expand_main_function): Likewise.
3765 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3766 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3767 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3768 * config/arm/arm.c (arm_trampoline_init): Likewise.
3769 (arm_call_tls_get_addr): Likewise.
3770 (arm_expand_divmod_libfunc): Likewise.
3771 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3772 (smulsi3_highpart): Likewise.
3773 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3774 (c6x_expand_compare): Likewise.
3775 (c6x_expand_movmem): Likewise.
3776 * config/frv/frv.c (frv_trampoline_init): Likewise.
3777 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3778 (ix86_expand_divmod_libfunc): Likewise.
3779 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3780 (ia64_expand_compare): Likewise.
3781 (ia64_profile_hook): Likewise.
3782 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3783 (nonlocal_goto): Likewise.
3784 (restore_stack_nonlocal): Likewise.
3785 * config/m32r/m32r.c (block_move_call): Likewise.
3786 (m32r_trampoline_init): Likewise.
3787 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3788 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3789 (m68k_call_m68k_read_tp): Likewise.
3790 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3791 (microblaze_expand_divide): Likewise.
3792 * config/mips/mips.h (mips_args): Likewise.
3793 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3794 (MIPS_ICACHE_SYNC): Likewise.
3795 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3796 (nios2_trampoline_init): Likewise.
3797 * config/pa/pa.c (hppa_tls_call): Likewise.
3798 (pa_trampoline_init): Likewise.
3799 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3800 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3801 (expand_strn_compare): Likewise.
3802 (rs6000_generate_compare): Likewise.
3803 (rs6000_expand_float128_convert): Likewise.
3804 (output_profile_hook): Likewise.
3805 (rs6000_trampoline_init): Likewise.
3806 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3807 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3808 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3809 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3810 (rs6000_generate_compare): Likewise.
3811 (rs6000_expand_float128_convert): Likewise.
3812 (output_profile_hook): Likewise.
3813 (rs6000_trampoline_init): Likewise.
3814 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3815 * config/sh/sh.c (sh_trampoline_init): Likewise.
3816 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3817 (sparc_emit_float_lib_cmp): Likewise.
3818 (sparc32_initialize_trampoline): Likewise.
3819 (sparc64_initialize_trampoline): Likewise.
3820 (sparc_profile_hook): Likewise.
3821 * config/spu/spu.c (ea_load_store): Likewise.
3822 * config/spu/spu.md (floatunssidf2): Likewise.
3823 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3824 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3825 * config/visium/visium.c (expand_block_move_4): Likewise.
3826 (expand_block_move_2): Likewise.
3827 (expand_block_move_1): Likewise.
3828 (expand_block_set_4): Likewise.
3829 (expand_block_set_2): Likewise.
3830 (expand_block_set_1): Likewise.
3831 (visium_trampoline_init): Likewise.
3832 (visium_profile_hook): Likewise.
3833 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3834 (xtensa_setup_frame_addresses): Likewise.
3835 (xtensa_trampoline_init): Likewise.
3836 * except.c (sjlj_emit_function_enter): Likewise.
3837 (sjlj_emit_function_exit): Likewise.
3838 * explow.c (allocate_dynamic_stack_space): Likewise.
3839 (probe_stack_range): Likewise.
3840 * expr.c (convert_mode_scalar): Likewise.
3841 * optabs.c (expand_binop): Likewise.
3842 (expand_twoval_binop_libfunc): Likewise.
3843 (expand_unop): Likewise.
3844 (prepare_cmp_insn): Likewise.
3845 (prepare_float_lib_cmp): Likewise.
3846 (expand_float): Likewise.
3847 (expand_fix): Likewise.
3848 (expand_fixed_convert): Likewise.
3849 (maybe_emit_sync_lock_test_and_set): Likewise.
3850 (expand_atomic_compare_and_swap): Likewise.
3851 (expand_mem_thread_fence): Likewise.
3852 (expand_atomic_fetch_op): Likewise.
3854 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3856 * doc/generic.texi (OpenACC): Adjust URL.
3857 * doc/invoke.texi (C Dialect Options): Ditto.
3859 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3861 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3862 predicate for operand 1. Add (m,<S>) constraint.
3863 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3864 Prevent memory operand 1 with register operand 2.
3866 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3868 PR rtl-optimization/82024
3869 * combine.c (try_combine): If the combination result is a PARALLEL,
3870 and we only need to retain the SET in there that would be placed
3871 at I2, check that we can place that at I3 instead, before doing so.
3873 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3876 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3878 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3879 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3882 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3883 Jeff Law <law@redhat.com>
3885 * varasm.c (bss_initializer_p): Do not put constants into .bss
3886 (categorize_decl_for_section): Handle bss_initializer_p returning
3887 false when DECL_INITIAL is NULL.
3889 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3892 * config/s390/s390.c (s390_can_inline_p): New function.
3894 2017-09-01 Jeff Law <law@redhat.com>
3896 PR tree-optimization/82052
3897 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3898 Always initialize the returned slot after a hash table miss
3899 when INSERT is true.
3901 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3903 * config/s390/s390.md (mem_signal_fence): Remove.
3904 * doc/md.texi (mem_signal_fence): Remove.
3905 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3907 * target-insns.def (mem_signal_fence): Remove.
3909 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3912 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3915 * asan.c (create_odr_indicator): Strip name encoding from assembler
3916 name before appending it after __odr_asan_.
3918 2017-09-01 Martin Liska <mliska@suse.cz>
3920 PR tree-optimization/82059
3921 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3922 frequency only when an edge is redirected.
3924 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3926 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3927 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3928 (arc_conditional_register_usage): Remove ARC600 lp_count
3930 (arc_file_start): Emit Tag_ARC_CPU_variation.
3931 (arc_can_use_doloop_p): New conditions to use ZOLs.
3932 (hwloop_fail): New function.
3933 (hwloop_optimize): Likewise.
3934 (hwloop_pattern_reg): Likewise.
3935 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3936 (arc_reorg_loops): New function, calls reorg_loops.
3937 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3938 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3940 (arc_loop_hazard): Remove function, functionality moved into
3942 (arc_hazard): Remove arc_loop_hazard call.
3943 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3944 into hwloop_optimize.
3945 (arc_label_align): Remove ZOL handling.
3946 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3947 * config/arc/arc.md (doloop_begin): Remove pattern.
3948 (doloop_begin_i): Likewise.
3949 (doloop_end_i): Likewise.
3950 (doloop_fallback): Likewise.
3951 (doloop_fallback_m): Likewise.
3952 (doloop_end): Reimplement expand.
3953 (arc_lp): New pattern for LP instruction.
3954 (loop_end): New pattern.
3955 (loop_fail): Likewise.
3956 (decrement_and_branch_until_zero): Likewise.
3957 * config/arc/arc.opt (mlpc-width): New option.
3958 * doc/invoke.texi (mlpc-width): Document option.
3960 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3962 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3963 (arc_ccfsm_advance): Fix checking for delay slots.
3964 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3966 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3968 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3970 (movhi_insn): Update store instruction constraint which are saving
3971 6-bit short immediates.
3972 (movsi_insn): Consider also short scaled load operations.
3973 (zero_extendhisi2_i): Use Usd constraint instead of T.
3974 (extendhisi2_i): Add q constraint.
3975 (arc_clzsi2): Add type and length attributes.
3976 (arc_ctzsi2): Likewise.
3977 * config/arc/constraints.md (Usc): Update constraint, the
3978 assembler can parse two relocations for a single instruction.
3980 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3982 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3983 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3985 2017-08-31 Olivier Hainque <hainque@adacore.com>
3987 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3988 match as powerpc-wrs-vxworks*.
3990 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3992 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3993 register constraint for by-element operand.
3994 (aarch64_mls_elt_merge<mode>): Likewise.
3996 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3998 * config/arc/arc.c (arc_can_follow_jump): Check for short
4001 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4003 * config.gcc: Use g.opt for arc.
4004 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4005 functionality moved to ...
4006 (legitimate_scaled_address_p): New function, ...here.
4007 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4008 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4009 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4011 (arc_override_options): Handle G option.
4012 (arc_output_pic_addr_const): Correct function definition.
4013 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4014 (arc_decl_anon_ns_mem_p): Delete.
4015 (arc_in_small_data_p): Overhaul this function to take into
4016 consideration the value given via G option.
4017 (arc_rewrite_small_data_1): Renamed and corrected old
4018 arc_rewrite_small_data function.
4019 (arc_rewrite_small_data): New function.
4020 (small_data_pattern): Don't use pic_offset_table_rtx.
4021 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4022 * config/arc/simdext.md (movmisalignv2hi): Use
4023 prepare_move_operands function.
4025 (movmisalign*): Likewise.
4026 * doc/invoke.texi (ARC options): Document -G option.
4028 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4030 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4032 * config/arc/arc.c (arc_print_operand): Output scalled address for
4033 sdata whenever is possible.
4034 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4035 load/stores are available.
4036 (compact_sda_memory_operand): Check for the alignment required by
4037 code density instructions.
4038 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4040 * config/arc/constraints.md (Usd): Update constraint.
4041 (Us0): New constraint.
4042 (Usc): Update constraint.
4044 2017-08-31 Richard Biener <rguenther@suse.de>
4047 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4050 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4052 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4053 Resize type_signature.
4055 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4056 Alan Hayward <alan.hayward@arm.com>
4057 David Sherwood <david.sherwood@arm.com>
4059 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4060 subregs whose inner modes can be stored in GPRs.
4061 (aarch64_classify_index): Likewise.
4063 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4064 Alan Hayward <alan.hayward@arm.com>
4065 David Sherwood <david.sherwood@arm.com>
4067 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4068 (V_INT_EQUIV): ...this.
4069 (v_cmp_result): Rename to...
4070 (v_int_equiv): ...this.
4071 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4072 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4073 (copysign<mode>3): Likewise.
4074 (aarch64_simd_bsl<mode>_internal): Likewise.
4075 (aarch64_simd_bsl<mode>): Likewise.
4076 (vec_cmp<mode><mode>): Likewise.
4077 (vcond<mode><mode>): Likewise.
4078 (vcond<v_cmp_mixed><mode>): Likewise.
4079 (vcondu<mode><v_cmp_mixed>): Likewise.
4080 (aarch64_cm<optab><mode>): Likewise.
4081 (aarch64_cmtst<mode>): Likewise.
4082 (aarch64_fac<optab><mode>): Likewise.
4083 (vec_perm_const<mode>): Likewise.
4084 (vcond_mask_<mode><v_cmp_result>): Rename to...
4085 (vcond_mask_<mode><v_int_equiv>): ...this.
4086 (vec_cmp<mode><v_cmp_result>): Rename to...
4087 (vec_cmp<mode><v_int_equiv>): ...this.
4089 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4090 Alan Hayward <alan.hayward@arm.com>
4091 David Sherwood <david.sherwood@arm.com>
4093 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4095 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4096 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4097 (UNSPEC_LD4_DREG): New unspecs.
4098 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4099 (aarch64_ld2<mode>_dreg_be): Replace with...
4100 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4102 (aarch64_ld3<mode>_dreg_le)
4103 (aarch64_ld3<mode>_dreg_be): Replace with...
4104 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4106 (aarch64_ld4<mode>_dreg_le)
4107 (aarch64_ld4<mode>_dreg_be): Replace with...
4108 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4111 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4113 PR tree-optimization/81987
4114 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4115 insert an initializer in a location not dominated by the stride
4118 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4120 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4121 on the entire header of the finally block in the fallthru case.
4123 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4125 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4127 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4129 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4130 rs6000_emit_move_from_cr and call renamed function.
4131 (rs6000_emit_prologue): Call renamed functions.
4132 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4133 movesi_from_cr, remove volatile CRs.
4135 2017-08-30 Jon Beniston <jon@beniston.com>
4136 Richard Biener <rguenther@suse.de>
4138 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4139 of VECTOR_MODE_P check.
4140 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4141 element vector types.
4143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4145 * df.h (df_read_modify_subreg_p): Remove in favor of...
4146 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4147 const_rtx instead of an rtx.
4148 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4149 * df-problems.c (df_word_lr_mark_ref): Likewise.
4150 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4151 (mark_pseudo_reg_dead): Likewise.
4152 (mark_ref_dead): Likewise.
4153 * reginfo.c (init_subregs_of_mode): Likewise.
4154 * sched-deps.c (sched_analyze_1): Likewise.
4155 * df-scan.c (df_def_record_1): Likewise.
4156 (df_uses_record): Likewise.
4157 (df_read_modify_subreg_p): Remove in favor of...
4158 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4159 const_rtx instead of an rtx.
4161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4162 Alan Hayward <alan.hayward@arm.com>
4163 David Sherwood <david.sherwood@arm.com>
4165 * rtl.h (partial_subreg_p): New function.
4166 * caller-save.c (save_call_clobbered_regs): Use it.
4167 * calls.c (expand_call): Likewise.
4168 * combine.c (combinable_i3pat): Likewise.
4169 (simplify_set): Likewise.
4170 (make_extraction): Likewise.
4171 (make_compound_operation_int): Likewise.
4172 (gen_lowpart_or_truncate): Likewise.
4173 (force_to_mode): Likewise.
4174 (make_field_assignment): Likewise.
4175 (reg_truncated_to_mode): Likewise.
4176 (record_truncated_value): Likewise.
4177 (move_deaths): Likewise.
4178 * cse.c (record_jump_cond): Likewise.
4179 (cse_insn): Likewise.
4180 * cselib.c (cselib_lookup_1): Likewise.
4181 * expmed.c (extract_bit_field_using_extv): Likewise.
4182 * function.c (assign_parm_setup_reg): Likewise.
4183 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4184 * ira-build.c (create_insn_allocnos): Likewise.
4185 * lra-coalesce.c (merge_pseudos): Likewise.
4186 * lra-constraints.c (match_reload): Likewise.
4187 (simplify_operand_subreg): Likewise.
4188 (curr_insn_transform): Likewise.
4189 * lra-lives.c (process_bb_lives): Likewise.
4190 * lra.c (new_insn_reg): Likewise.
4191 (lra_substitute_pseudo): Likewise.
4192 * regcprop.c (mode_change_ok): Likewise.
4193 (maybe_mode_change): Likewise.
4194 (copyprop_hardreg_forward_1): Likewise.
4195 * reload.c (push_reload): Likewise.
4196 (find_reloads): Likewise.
4197 (find_reloads_subreg_address): Likewise.
4198 * reload1.c (alter_reg): Likewise.
4199 (eliminate_regs_1): Likewise.
4200 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4202 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4204 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4205 back to if statements, including unpack.
4207 2017-08-30 Martin Liska <mliska@suse.cz>
4210 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4212 (func_checker::compare_asm_inputs_outputs): ... this function.
4213 (func_checker::compare_gimple_asm): Use the function to compare
4214 also ASM constrains.
4215 * ipa-icf-gimple.h: Rename the function.
4217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4218 Alan Hayward <alan.hayward@arm.com>
4219 David Sherwood <david.sherwood@arm.com>
4221 * coretypes.h (complex_mode): New type.
4222 * gdbhooks.py (build_pretty_printer): Handle it.
4223 * machmode.h (complex_mode): New class.
4224 (complex_mode::includes_p): New function.
4225 (is_complex_int_mode): Likewise.
4226 (is_complex_float_mode): Likewise.
4227 * genmodes.c (get_mode_class): Handle complex mode classes.
4228 * function.c (expand_function_end): Use is_complex_int_mode.
4230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4231 Alan Hayward <alan.hayward@arm.com>
4232 David Sherwood <david.sherwood@arm.com>
4234 * coretypes.h (scalar_mode_pod): New typedef.
4235 * gdbhooks.py (build_pretty_printer): Handle it.
4236 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4237 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4238 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4239 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4243 Alan Hayward <alan.hayward@arm.com>
4244 David Sherwood <david.sherwood@arm.com>
4246 * machmode.h (mode_for_vector): Take a scalar_mode instead
4248 * stor-layout.c (mode_for_vector): Likewise.
4249 * explow.c (promote_mode): Use as_a <scalar_mode>.
4250 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * target.def (preferred_simd_mode): Take a scalar_mode
4257 instead of a machine_mode.
4258 * targhooks.h (default_preferred_simd_mode): Likewise.
4259 * targhooks.c (default_preferred_simd_mode): Likewise.
4260 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4261 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4262 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4263 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4264 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4265 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4266 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4267 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4269 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4270 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4271 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4272 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4273 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4274 * doc/tm.texi: Regenerate.
4275 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4279 Alan Hayward <alan.hayward@arm.com>
4280 David Sherwood <david.sherwood@arm.com>
4282 * target.def (scalar_mode_supported_p): Take a scalar_mode
4283 instead of a machine_mode.
4284 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4285 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4286 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4287 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4288 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4289 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4290 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4291 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4292 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4293 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4294 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4295 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4296 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4297 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4299 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4300 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4301 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4302 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4303 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4305 * doc/tm.texi: Regenerate.
4307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4308 Alan Hayward <alan.hayward@arm.com>
4309 David Sherwood <david.sherwood@arm.com>
4311 * coretypes.h (opt_scalar_mode): New typedef.
4312 * gdbhooks.py (build_pretty_printers): Handle it.
4313 * machmode.h (mode_iterator::get_2xwider): Add overload for
4315 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4317 * expr.c (convert_mode_scalar): Likewise.
4318 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4319 * optabs.c (expand_float): Likewise.
4320 (expand_fix): Likewise.
4321 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4324 Alan Hayward <alan.hayward@arm.com>
4325 David Sherwood <david.sherwood@arm.com>
4327 * optabs.c (expand_float): Explicitly check for scalars before
4328 using a branching expansion.
4329 (expand_fix): Likewise.
4331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4332 Alan Hayward <alan.hayward@arm.com>
4333 David Sherwood <david.sherwood@arm.com>
4335 * expr.c (convert_mode): Split scalar handling out into...
4336 (convert_mode_scalar): ...this new function. Treat the modes
4339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4340 Alan Hayward <alan.hayward@arm.com>
4341 David Sherwood <david.sherwood@arm.com>
4343 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4345 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4348 Alan Hayward <alan.hayward@arm.com>
4349 David Sherwood <david.sherwood@arm.com>
4351 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4352 rather than a machine_mode.
4353 (fixed_from_string): Likewise.
4354 (fixed_convert): Likewise.
4355 (fixed_convert_from_int): Likewise.
4356 (fixed_convert_from_real): Likewise.
4357 (real_convert_from_fixed): Likewise.
4358 * fixed-value.c (fixed_from_double_int): Likewise.
4359 (fixed_from_string): Likewise.
4360 (fixed_convert): Likewise.
4361 (fixed_convert_from_int): Likewise.
4362 (fixed_convert_from_real): Likewise.
4363 (real_convert_from_fixed): Likewise.
4364 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4367 Alan Hayward <alan.hayward@arm.com>
4368 David Sherwood <david.sherwood@arm.com>
4370 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4371 of separate mode class checks. Do not allow vector modes here.
4372 (immed_wide_int_const): Use as_a <scalar_mode>.
4373 * explow.c (trunc_int_for_mode): Likewise.
4374 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4375 (wi::shwi): Likewise.
4376 (wi::min_value): Likewise.
4377 (wi::max_value): Likewise.
4378 * dwarf2out.c (loc_descriptor): Likewise.
4379 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4383 Alan Hayward <alan.hayward@arm.com>
4384 David Sherwood <david.sherwood@arm.com>
4386 * tree.h (SCALAR_TYPE_MODE): New macro.
4387 * expr.c (expand_expr_addr_expr_1): Use it.
4388 (expand_expr_real_2): Likewise.
4389 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4390 (fold_convert_const_fixed_from_int): Likewise.
4391 (fold_convert_const_fixed_from_real): Likewise.
4392 (native_encode_fixed): Likewise
4393 (native_encode_complex): Likewise
4394 (native_encode_vector): Likewise.
4395 (native_interpret_fixed): Likewise.
4396 (native_interpret_real): Likewise.
4397 (native_interpret_complex): Likewise.
4398 (native_interpret_vector): Likewise.
4399 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4400 (simd_clone_adjust_argument_types): Likewise.
4401 (simd_clone_init_simd_arrays): Likewise.
4402 (simd_clone_adjust): Likewise.
4403 * stor-layout.c (layout_type): Likewise.
4404 * tree.c (build_minus_one_cst): Likewise.
4405 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4406 * tree-inline.c (estimate_move_cost): Likewise.
4407 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4408 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4409 (vectorizable_reduction): Likewise.
4410 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4411 (vect_recog_mixed_size_cond_pattern): Likewise.
4412 (check_bool_pattern): Likewise.
4413 (adjust_bool_pattern): Likewise.
4414 (search_type_for_mask_1): Likewise.
4415 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4416 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4417 (vectorizable_load): Likewise.
4418 (vectorizable_store): Likewise.
4419 * ubsan.c (ubsan_encode_value): Likewise.
4420 * varasm.c (output_constant): Likewise.
4422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4423 Alan Hayward <alan.hayward@arm.com>
4424 David Sherwood <david.sherwood@arm.com>
4426 * coretypes.h (scalar_mode): New class.
4427 * machmode.h (scalar_mode): Likewise.
4428 (scalar_mode::includes_p): New function.
4429 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4430 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4431 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4432 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4433 * expmed.c (store_bit_field_1): Likewise.
4434 (extract_bit_field_1): Likewise.
4435 * expr.c (write_complex_part): Likewise.
4436 (read_complex_part): Likewise.
4437 (emit_move_complex_push): Likewise.
4438 (expand_expr_real_2): Likewise.
4439 * function.c (assign_parm_setup_reg): Likewise.
4440 (assign_parms_unsplit_complex): Likewise.
4441 * optabs.c (expand_binop): Likewise.
4442 * rtlanal.c (subreg_get_info): Likewise.
4443 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4444 * varasm.c (output_constant_pool_2): Likewise.
4446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4447 Alan Hayward <alan.hayward@arm.com>
4448 David Sherwood <david.sherwood@arm.com>
4450 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4452 (expmed_mult_highpart_optab): Likewise.
4453 (expmed_mult_highpart): Likewise.
4455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4456 Alan Hayward <alan.hayward@arm.com>
4457 David Sherwood <david.sherwood@arm.com>
4459 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4460 instead of a machine_mode.
4461 (builtin_memset_read_str): Likewise.
4462 * builtins.c (c_readstr): Likewise.
4463 (builtin_memcpy_read_str): Likewise.
4464 (builtin_strncpy_read_str): Likewise.
4465 (builtin_memset_read_str): Likewise.
4466 (builtin_memset_gen_str): Likewise.
4467 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4468 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4469 instead of a machine_mode.
4470 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4472 (make_extraction): Likewise.
4473 (try_widen_shift_mode): Take and return scalar_int_modes instead
4475 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4476 a scalar_int_mode instead of a machine_mode.
4477 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4478 (avr_addr_space_pointer_mode): Likewise.
4479 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4480 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4481 (msp430_unwind_word_mode): Likewise.
4482 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4483 (spu_addr_space_pointer_mode): Likewise.
4484 (spu_addr_space_address_mode): Likewise.
4485 (spu_libgcc_cmp_return_mode): Likewise.
4486 (spu_libgcc_shift_count_mode): Likewise.
4487 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4488 (rl78_addr_space_pointer_mode): Likewise.
4489 (fl78_unwind_word_mode): Likewise.
4490 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4492 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4493 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4494 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4495 (mips_valid_pointer_mode): Likewise.
4496 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4497 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4498 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4500 (ft32_addr_space_address_mode): Likewise.
4501 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4502 scalar_int_mode instead of a machine_mode.
4503 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4505 (m32c_addr_space_address_mode): Likewise.
4506 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4507 (rs6000_eh_return_filter_mode): Likewise.
4508 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4509 (rs6000_eh_return_filter_mode): Likewise.
4510 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4511 (s390_libgcc_shift_count_mode): Likewise.
4512 (s390_unwind_word_mode): Likewise.
4513 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4515 * target.def (mode_rep_extended): Likewise.
4516 (valid_pointer_mode): Likewise.
4517 (addr_space.valid_pointer_mode): Likewise.
4518 (eh_return_filter_mode): Return a scalar_int_mode rather than
4520 (libgcc_cmp_return_mode): Likewise.
4521 (libgcc_shift_count_mode): Likewise.
4522 (unwind_word_mode): Likewise.
4523 (addr_space.pointer_mode): Likewise.
4524 (addr_space.address_mode): Likewise.
4525 * doc/tm.texi: Regenerate.
4526 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4528 (do_jump): Use scalar_int_mode for local variables.
4529 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4530 rather than a machine_mode.
4531 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4532 (scompare_loc_descriptor_wide): Likewise.
4533 (scompare_loc_descriptor_narrow): Likewise.
4534 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4536 * except.c (sjlj_emit_dispatch_table): Likewise.
4537 (expand_builtin_eh_copy_values): Likewise.
4538 * explow.c (convert_memory_address_addr_space_1): Likewise.
4539 Take a scalar_int_mode rather than a machine_mode.
4540 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4541 than a machine_mode.
4542 (memory_address_addr_space): Use scalar_int_mode for local variables.
4543 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4544 rather than a machine_mode.
4545 * expmed.c (mask_rtx): Likewise.
4546 (init_expmed_one_conv): Likewise.
4547 (expand_mult_highpart_adjust): Likewise.
4548 (extract_high_half): Likewise.
4549 (expmed_mult_highpart_optab): Likewise.
4550 (expmed_mult_highpart): Likewise.
4551 (expand_smod_pow2): Likewise.
4552 (expand_sdiv_pow2): Likewise.
4553 (emit_store_flag_int): Likewise.
4554 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4556 (extract_low_bits): Likewise.
4557 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
4559 * expr.c (pieces_addr::adjust): Likewise.
4560 (can_store_by_pieces): Likewise.
4561 (store_by_pieces): Likewise.
4562 (clear_by_pieces_1): Likewise.
4563 (expand_expr_addr_expr_1): Likewise.
4564 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
4565 (expand_expr_real_1): Likewise.
4566 (try_casesi): Likewise.
4567 * final.c (shorten_branches): Likewise.
4568 * fold-const.c (fold_convert_const_int_from_fixed): Change the
4569 type of "mode" to machine_mode.
4570 * internal-fn.c (expand_arith_overflow_result_store): Take a
4571 scalar_int_mode rather than a machine_mode.
4572 (expand_mul_overflow): Use scalar_int_mode for local variables.
4573 * loop-doloop.c (doloop_modify): Likewise.
4574 (doloop_optimize): Likewise.
4575 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
4576 than a machine_mode.
4577 (expand_doubleword_shift_condmove): Likewise.
4578 (expand_doubleword_shift): Likewise.
4579 (expand_doubleword_clz): Likewise.
4580 (expand_doubleword_popcount): Likewise.
4581 (expand_doubleword_parity): Likewise.
4582 (expand_absneg_bit): Use scalar_int_mode for local variables.
4583 (prepare_float_lib_cmp): Likewise.
4584 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4585 rather than a machine_mode.
4586 (convert_memory_address_addr_space): Likewise.
4587 (get_mode_bounds): Likewise.
4588 (get_address_mode): Return a scalar_int_mode rather than a
4590 * rtlanal.c (get_address_mode): Likewise.
4591 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4592 than a machine_mode.
4593 * targhooks.c (default_mode_rep_extended): Likewise.
4594 (default_valid_pointer_mode): Likewise.
4595 (default_addr_space_valid_pointer_mode): Likewise.
4596 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4597 than a machine_mode.
4598 (default_libgcc_cmp_return_mode): Likewise.
4599 (default_libgcc_shift_count_mode): Likewise.
4600 (default_unwind_word_mode): Likewise.
4601 (default_addr_space_pointer_mode): Likewise.
4602 (default_addr_space_address_mode): Likewise.
4603 * targhooks.h (default_eh_return_filter_mode): Likewise.
4604 (default_libgcc_cmp_return_mode): Likewise.
4605 (default_libgcc_shift_count_mode): Likewise.
4606 (default_unwind_word_mode): Likewise.
4607 (default_addr_space_pointer_mode): Likewise.
4608 (default_addr_space_address_mode): Likewise.
4609 (default_mode_rep_extended): Take a scalar_int_mode rather than
4611 (default_valid_pointer_mode): Likewise.
4612 (default_addr_space_valid_pointer_mode): Likewise.
4613 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4615 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4616 rather than a machine_mode.
4617 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4618 for local variables.
4619 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4620 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4621 than a machine_mode.
4623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4624 Alan Hayward <alan.hayward@arm.com>
4625 David Sherwood <david.sherwood@arm.com>
4627 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4628 the mode argument to scalar_int_mode.
4629 (do_jump_by_parts_zero_rtx): Likewise.
4630 (do_jump_by_parts_equality_rtx): Likewise.
4631 (do_jump_by_parts_greater): Take a mode argument.
4632 (do_jump_by_parts_equality): Likewise.
4633 (do_jump_1): Update calls accordingly.
4635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4636 Alan Hayward <alan.hayward@arm.com>
4637 David Sherwood <david.sherwood@arm.com>
4639 * is-a.h (safe_dyn_cast): New function.
4640 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4641 (jump_table_for_label): Likewise.
4642 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4643 instead of an rtx_insn *.
4644 (shorten_branches): Use dyn_cast instead of LABEL_P and
4645 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4646 rtx_jump_table_data::get_data_mode.
4647 (final_scan_insn): Likewise.
4649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4650 Alan Hayward <alan.hayward@arm.com>
4651 David Sherwood <david.sherwood@arm.com>
4653 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4654 trying to combine a full-register integer set with a subreg
4657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4658 Alan Hayward <alan.hayward@arm.com>
4659 David Sherwood <david.sherwood@arm.com>
4661 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4662 that is always either address_mode or pointer_mode.
4664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4665 Alan Hayward <alan.hayward@arm.com>
4666 David Sherwood <david.sherwood@arm.com>
4668 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4669 when the two are known to be equal.
4671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4672 Alan Hayward <alan.hayward@arm.com>
4673 David Sherwood <david.sherwood@arm.com>
4675 * simplify-rtx.c (simplify_const_unary_operation): Use
4676 is_a <scalar_int_mode> instead of checking for a nonzero
4677 precision. Forcibly convert op_mode to a scalar_int_mode
4678 in that case. More clearly differentiate the operand and
4679 result modes and use the former when deciding what the value
4680 of a count-bits operation should be. Use is_int_mode instead
4681 of checking for a MODE_INT. Remove redundant check for whether
4682 this mode has a zero precision.
4684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4685 Alan Hayward <alan.hayward@arm.com>
4686 David Sherwood <david.sherwood@arm.com>
4688 * optabs.c (widen_leading): Change the type of the mode argument
4689 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4690 (widen_bswap): Likewise.
4691 (expand_parity): Likewise.
4692 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4693 (expand_ffs): Likewise.
4694 (epand_unop): Check for scalar integer modes before calling the
4697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4698 Alan Hayward <alan.hayward@arm.com>
4699 David Sherwood <david.sherwood@arm.com>
4701 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4703 (expand_expr_real_1): Update call accordingly.
4705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4706 Alan Hayward <alan.hayward@arm.com>
4707 David Sherwood <david.sherwood@arm.com>
4709 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4710 value_mode arguments. Use scalar_int_mode internally.
4711 (store_bit_field_1): Rename the new integer mode from imode
4712 to op0_mode and use it instead of GET_MODE (op0). Update calls
4713 to store_split_bit_field, store_bit_field_using_insv and
4714 store_fixed_bit_field.
4715 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4716 Use scalar_int_mode internally. Use a bit count rather than a mode
4717 when calculating the largest bit size for get_best_mode.
4718 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4719 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4720 Remove assertion that OP0 has a scalar integer mode.
4721 (store_split_bit_field): Add op0_mode and value_mode arguments.
4722 Update calls to extract_fixed_bit_field.
4723 (extract_bit_field_using_extv): Add an op0_mode argument.
4724 Use scalar_int_mode internally.
4725 (extract_bit_field_1): Rename the new integer mode from imode to
4726 op0_mode and use it instead of GET_MODE (op0). Update calls to
4727 extract_split_bit_field, extract_bit_field_using_extv and
4728 extract_fixed_bit_field.
4729 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4730 to extract_split_bit_field and extract_fixed_bit_field_1.
4731 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4732 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4734 (extract_split_bit_field): Add an op0_mode argument. Update call
4735 to extract_fixed_bit_field.
4737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4738 Alan Hayward <alan.hayward@arm.com>
4739 David Sherwood <david.sherwood@arm.com>
4741 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4742 * explow.c (hard_function_value): Likewise.
4743 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4744 convert_to_mode call outside the loop.
4745 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4746 for the mode iterator. Require the mode specified by max_pieces
4748 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4750 (copy_blkmode_to_reg): Likewise.
4751 (set_storage_via_setmem): Likewise.
4752 * optabs.c (prepare_cmp_insn): Likewise.
4753 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4754 * stor-layout.c (finish_bitfield_representative): Likewise.
4756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4757 Alan Hayward <alan.hayward@arm.com>
4758 David Sherwood <david.sherwood@arm.com>
4760 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4761 * expr.c (convert_move): Use them.
4762 (convert_modes): Likewise.
4763 (store_expr_with_bounds): Likewise.
4765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4766 Alan Hayward <alan.hayward@arm.com>
4767 David Sherwood <david.sherwood@arm.com>
4769 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4770 parameter for the mode of "x". Remove the "known_x", "known_mode"
4771 and "known_ret" arguments. Change the type of the mode argument
4773 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4774 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4775 (reg_num_sign_bit_copies_for_combine): Likewise.
4776 * rtlanal.c (nonzero_bits1): Likewise.
4777 (num_sign_bit_copies1): Likewise.
4778 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4779 (reg_num_sign_bit_copies_general): Likewise.
4780 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4781 (reg_nonzero_bits_general): Likewise.
4783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4784 Alan Hayward <alan.hayward@arm.com>
4785 David Sherwood <david.sherwood@arm.com>
4787 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4788 than in subroutines. Return 1 for non-integer modes.
4789 (cached_num_sign_bit_copies): Change the type of the mode parameter
4791 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4792 classes. Handle CONST_INT_P first and then check whether X also
4793 has a scalar integer mode. Check the same thing for inner registers
4794 of a SUBREG and for values that are being extended or truncated.
4796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4797 Alan Hayward <alan.hayward@arm.com>
4798 David Sherwood <david.sherwood@arm.com>
4800 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4801 in subroutines. Return the mode mask for non-integer modes.
4802 (cached_nonzero_bits): Change the type of the mode parameter
4804 (nonzero_bits1): Likewise. Remove early exit for other mode
4805 classes. Handle CONST_INT_P first and then check whether X
4806 also has a scalar integer mode.
4808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4809 Alan Hayward <alan.hayward@arm.com>
4810 David Sherwood <david.sherwood@arm.com>
4812 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4813 Return a scalar_int_mode and assert that the size is greater than
4815 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4816 (op_by_pieces_d::op_by_pieces_d): Likewise.
4817 (op_by_pieces_d::run): Likewise.
4818 (can_store_by_pieces): Likewise.
4820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4821 Alan Hayward <alan.hayward@arm.com>
4822 David Sherwood <david.sherwood@arm.com>
4824 * combine.c (extract_left_shift): Add a mode argument and update
4826 (make_compound_operation_int): Change the type of the mode parameter
4827 to scalar_int_mode and update the call to extract_left_shift.
4829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4830 Alan Hayward <alan.hayward@arm.com>
4831 David Sherwood <david.sherwood@arm.com>
4833 * combine.c (simplify_and_const_int): Change the type of the mode
4834 parameter to scalar_int_mode.
4835 (simplify_and_const_int_1): Likewise. Update recursive call.
4837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4838 Alan Hayward <alan.hayward@arm.com>
4839 David Sherwood <david.sherwood@arm.com>
4841 * combine.c (simplify_compare_const): Check that the mode is a
4842 scalar_int_mode (rather than VOIDmode) before testing its
4844 (simplify_comparison): Move COMPARISON_P handling out of the
4845 loop and restrict the latter part of the loop to scalar_int_modes.
4846 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4847 and when considering SUBREG_REGs. Use is_int_mode instead of
4848 checking GET_MODE_CLASS against MODE_INT.
4850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4851 Alan Hayward <alan.hayward@arm.com>
4852 David Sherwood <david.sherwood@arm.com>
4854 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4855 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4856 shift_unit_mode and for modes involved in scalar shifts.
4858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4859 Alan Hayward <alan.hayward@arm.com>
4860 David Sherwood <david.sherwood@arm.com>
4862 * combine.c (force_int_to_mode): New function, split out from...
4863 (force_to_mode): ...here. Keep xmode up-to-date and use it
4864 instead of GET_MODE (x).
4866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4867 Alan Hayward <alan.hayward@arm.com>
4868 David Sherwood <david.sherwood@arm.com>
4870 * optabs-query.h (extraction_insn::struct_mode): Change type to
4871 opt_scalar_int_mode and update comment.
4872 (extraction_insn::field_mode): Change type to scalar_int_mode.
4873 (extraction_insn::pos_mode): Likewise.
4874 * combine.c (make_extraction): Update accordingly.
4875 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4876 (get_optab_extraction_insn): Likewise.
4877 * recog.c (simplify_while_replacing): Likewise.
4878 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4879 parameter to opt_scalar_int_mode.
4881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4882 Alan Hayward <alan.hayward@arm.com>
4883 David Sherwood <david.sherwood@arm.com>
4885 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4886 to a scalar_int_mode instead of a machine_mode.
4887 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4888 (get_best_mode): Return a boolean and use a pointer argument to store
4889 the selected mode. Replace the limit mode parameter with a bit limit.
4890 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4891 for the values returned by bit_field_mode_iterator::next_mode.
4892 (store_bit_field): Update call to get_best_mode.
4893 (store_fixed_bit_field): Likewise.
4894 (extract_fixed_bit_field): Likewise.
4895 * expr.c (optimize_bitfield_assignment_op): Likewise.
4896 * fold-const.c (optimize_bit_field_compare): Likewise.
4897 (fold_truth_andor_1): Likewise.
4898 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4899 Update for new type of m_mode.
4900 (get_best_mode): As above.
4902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4903 Alan Hayward <alan.hayward@arm.com>
4904 David Sherwood <david.sherwood@arm.com>
4906 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4907 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4908 (store_bit_field): Check is_a <scalar_int_mode> before calling
4909 strict_volatile_bitfield_p.
4910 (extract_bit_field): Likewise.
4912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4913 Alan Hayward <alan.hayward@arm.com>
4914 David Sherwood <david.sherwood@arm.com>
4916 * target.def (cstore_mode): Return a scalar_int_mode.
4917 * doc/tm.texi: Regenerate.
4918 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4919 * targhooks.h (default_cstore_mode): Likewise.
4920 * targhooks.c (default_cstore_mode): Likewise, using a forced
4922 * expmed.c (emit_cstore): Expect the target of the cstore to be
4925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4926 Alan Hayward <alan.hayward@arm.com>
4927 David Sherwood <david.sherwood@arm.com>
4929 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4931 (niter_desc): Likewise mode.
4932 (iv_analyze): Add a mode parameter.
4934 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4935 and change its type to scalar_int_mode.
4936 * loop-iv.c: Update commentary at head of file.
4937 (iv_constant): Pass the mode paraeter before the rtx it describes
4938 and change its type to scalar_int_mode. Remove VOIDmode handling.
4939 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4940 (iv_extend): Likewise.
4941 (shorten_into_mode): Likewise.
4942 (iv_add): Use scalar_int_mode.
4943 (iv_mult): Likewise.
4944 (iv_shift): Likewise.
4945 (canonicalize_iv_subregs): Likewise.
4946 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4947 it describes and change its mode to scalar_int_mode. Also change
4948 the type of the returned inner_mode to scalar_int_mode.
4949 (get_biv_step): Likewise, turning outer_mode from a pointer
4950 into a direct parameter. Update call to get_biv_step_1.
4951 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4952 iv_constant and get_biv_step.
4953 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4954 and change its type to scalar_int_mode. Don't initialise iv->mode
4955 to VOIDmode and remove later checks for its still being VOIDmode.
4956 Update calls to iv_analyze_op and iv_analyze_expr. Check
4957 is_a <scalar_int_mode> when changing the mode under consideration.
4958 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4959 Update call to iv_analyze_expr.
4960 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4961 inner register is not also a scalar_int_mode. Update call to
4963 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4964 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4965 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4966 separate mode class checks. Update calls to iv_analyze. Remove
4967 fix-up of VOIDmodes after iv_analyze_biv.
4968 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4969 don't have a scalar_int_mode. Update call to biv_p.
4971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4972 Alan Hayward <alan.hayward@arm.com>
4973 David Sherwood <david.sherwood@arm.com>
4975 * cfgexpand.c (convert_debug_memory_address): Use
4976 as_a <scalar_int_mode>.
4977 * combine.c (expand_compound_operation): Likewise.
4978 (make_extraction): Likewise.
4979 (change_zero_ext): Likewise.
4980 (simplify_comparison): Likewise.
4981 * cse.c (cse_insn): Likewise.
4982 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4983 (mem_loc_descriptor): Likewise.
4984 (loc_descriptor): Likewise.
4985 * expmed.c (init_expmed_one_mode): Likewise.
4986 (synth_mult): Likewise.
4987 (emit_store_flag_1): Likewise.
4988 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4989 of a comparison with size.
4990 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4991 (reduce_to_bit_field_precision): Likewise.
4992 * function.c (expand_function_end): Likewise.
4993 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4994 * loop-doloop.c (doloop_modify): Likewise.
4995 * optabs.c (expand_binop): Likewise.
4996 (expand_unop): Likewise.
4997 (expand_copysign_absneg): Likewise.
4998 (prepare_cmp_insn): Likewise.
4999 (maybe_legitimize_operand): Likewise.
5000 * recog.c (const_scalar_int_operand): Likewise.
5001 * rtlanal.c (get_address_mode): Likewise.
5002 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5003 (simplify_cond_clz_ctz): Likewise.
5004 * tree-nested.c (get_nl_goto_field): Likewise.
5005 * tree.c (build_vector_type_for_mode): Likewise.
5006 * var-tracking.c (use_narrower_mode): Likewise.
5008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5009 Alan Hayward <alan.hayward@arm.com>
5010 David Sherwood <david.sherwood@arm.com>
5012 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5013 * builtins.c (expand_builtin_signbit): Use it.
5014 * cfgexpand.c (expand_debug_expr): Likewise.
5015 * dojump.c (do_jump): Likewise.
5016 (do_compare_and_jump): Likewise.
5017 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5018 * expmed.c (make_tree): Likewise.
5019 * expr.c (expand_expr_real_2): Likewise.
5020 (expand_expr_real_1): Likewise.
5021 (try_casesi): Likewise.
5022 * fold-const-call.c (fold_const_call_ss): Likewise.
5023 * fold-const.c (unextend): Likewise.
5024 (extract_muldiv_1): Likewise.
5025 (fold_single_bit_test): Likewise.
5026 (native_encode_int): Likewise.
5027 (native_encode_string): Likewise.
5028 (native_interpret_int): Likewise.
5029 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5030 * internal-fn.c (expand_addsub_overflow): Likewise.
5031 (expand_neg_overflow): Likewise.
5032 (expand_mul_overflow): Likewise.
5033 (expand_arith_overflow): Likewise.
5034 * match.pd: Likewise.
5035 * stor-layout.c (layout_type): Likewise.
5036 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5037 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5038 * tree-ssanames.c (get_range_info): Likewise.
5039 * tree-switch-conversion.c (array_value_type) Likewise.
5040 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5041 (vect_recog_divmod_pattern): Likewise.
5042 (vect_recog_mixed_size_cond_pattern): Likewise.
5043 * tree-vrp.c (extract_range_basic): Likewise.
5044 (simplify_float_conversion_using_ranges): Likewise.
5045 * tree.c (int_fits_type_p): Likewise.
5046 * ubsan.c (instrument_bool_enum_load): Likewise.
5047 * varasm.c (mergeable_string_section): Likewise.
5048 (narrowing_initializer_constant_valid_p): Likewise.
5049 (output_constant): Likewise.
5051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5052 Alan Hayward <alan.hayward@arm.com>
5053 David Sherwood <david.sherwood@arm.com>
5055 * machmode.h (NARROWEST_INT_MODE): New macro.
5056 * expr.c (alignment_for_piecewise_move): Use it instead of
5057 GET_CLASS_NARROWEST_MODE (MODE_INT).
5058 (push_block): Likewise.
5059 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5061 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5064 Alan Hayward <alan.hayward@arm.com>
5065 David Sherwood <david.sherwood@arm.com>
5067 * postreload.c (move2add_valid_value_p): Change the type of the
5068 mode parameter to scalar_int_mode.
5069 (move2add_use_add2_insn): Add a mode parameter and use it instead
5071 (move2add_use_add3_insn): Likewise.
5072 (reload_cse_move2add): Update accordingly.
5074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5075 Alan Hayward <alan.hayward@arm.com>
5076 David Sherwood <david.sherwood@arm.com>
5078 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5080 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5081 * optabs.c (expand_unop): Likewise.
5083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5084 Alan Hayward <alan.hayward@arm.com>
5085 David Sherwood <david.sherwood@arm.com>
5087 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5088 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5089 (popcount_loc_descriptor): Likewise.
5090 (bswap_loc_descriptor): Likewise.
5091 (rotate_loc_descriptor): Likewise.
5092 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5093 calling the functions above.
5095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5096 Alan Hayward <alan.hayward@arm.com>
5097 David Sherwood <david.sherwood@arm.com>
5099 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5101 (try_combine): Likewise.
5102 (simplify_if_then_else): Likewise.
5103 * cse.c (cse_insn): Likewise.
5104 * dwarf2out.c (mem_loc_descriptor): Likewise.
5105 * emit-rtl.c (gen_lowpart_common): Likewise.
5106 * simplify-rtx.c (simplify_truncation): Likewise.
5107 (simplify_binary_operation_1): Likewise.
5108 (simplify_const_relational_operation): Likewise.
5109 (simplify_ternary_operation): Likewise.
5110 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5113 Alan Hayward <alan.hayward@arm.com>
5114 David Sherwood <david.sherwood@arm.com>
5116 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5117 * reload.c (push_reload): Likewise.
5118 (find_reloads): Likewise.
5120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5121 Alan Hayward <alan.hayward@arm.com>
5122 David Sherwood <david.sherwood@arm.com>
5124 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5125 (make_compound_operation_int): Likewise.
5126 (change_zero_ext): Likewise.
5127 * expr.c (convert_move): Likewise.
5128 (convert_modes): Likewise.
5129 * fwprop.c (forward_propagate_subreg): Likewise.
5130 * loop-iv.c (get_biv_step_1): Likewise.
5131 * optabs.c (widen_operand): Likewise.
5132 * postreload.c (move2add_valid_value_p): Likewise.
5133 * recog.c (simplify_while_replacing): Likewise.
5134 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5135 (simplify_binary_operation_1): Likewise. Remove redundant
5136 mode equality check.
5138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5139 Alan Hayward <alan.hayward@arm.com>
5140 David Sherwood <david.sherwood@arm.com>
5142 * combine.c (combine_simplify_rtx): Add checks for
5143 is_a <scalar_int_mode>.
5144 (simplify_if_then_else): Likewise.
5145 (make_field_assignment): Likewise.
5146 (simplify_comparison): Likewise.
5147 * ifcvt.c (noce_try_bitop): Likewise.
5148 * loop-invariant.c (canonicalize_address_mult): Likewise.
5149 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5152 Alan Hayward <alan.hayward@arm.com>
5153 David Sherwood <david.sherwood@arm.com>
5155 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5156 is_a <scalar_int_mode> instead of != BLKmode.
5158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5159 Alan Hayward <alan.hayward@arm.com>
5160 David Sherwood <david.sherwood@arm.com>
5162 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5163 instead of != VOIDmode.
5164 * combine.c (if_then_else_cond): Likewise.
5165 (change_zero_ext): Likewise.
5166 * dwarf2out.c (mem_loc_descriptor): Likewise.
5167 (loc_descriptor): Likewise.
5168 * rtlanal.c (canonicalize_condition): Likewise.
5169 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5172 Alan Hayward <alan.hayward@arm.com>
5173 David Sherwood <david.sherwood@arm.com>
5175 * simplify-rtx.c (simplify_binary_operation_1): Use
5176 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5179 Alan Hayward <alan.hayward@arm.com>
5180 David Sherwood <david.sherwood@arm.com>
5182 * wide-int.h (int_traits<unsigned char>) New class.
5183 (int_traits<unsigned short>) Likewise.
5184 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5185 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5187 * combine.c (set_nonzero_bits_and_sign_copies): Use
5188 is_a <scalar_int_mode>.
5189 (find_split_point): Likewise.
5190 (combine_simplify_rtx): Likewise.
5191 (simplify_logical): Likewise.
5192 (expand_compound_operation): Likewise.
5193 (expand_field_assignment): Likewise.
5194 (make_compound_operation): Likewise.
5195 (extended_count): Likewise.
5196 (change_zero_ext): Likewise.
5197 (simplify_comparison): Likewise.
5198 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5199 (ucompare_loc_descriptor): Likewise.
5200 (minmax_loc_descriptor): Likewise.
5201 (mem_loc_descriptor): Likewise.
5202 (loc_descriptor): Likewise.
5203 * expmed.c (init_expmed_one_mode): Likewise.
5204 * lra-constraints.c (lra_constraint_offset): Likewise.
5205 * optabs.c (prepare_libcall_arg): Likewise.
5206 * postreload.c (move2add_note_store): Likewise.
5207 * reload.c (operands_match_p): Likewise.
5208 * rtl.h (load_extend_op): Likewise.
5209 * rtlhooks.c (gen_lowpart_general): Likewise.
5210 * simplify-rtx.c (simplify_truncation): Likewise.
5211 (simplify_unary_operation_1): Likewise.
5212 (simplify_binary_operation_1): Likewise.
5213 (simplify_const_binary_operation): Likewise.
5214 (simplify_const_relational_operation): Likewise.
5215 (simplify_subreg): Likewise.
5216 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5217 * var-tracking.c (adjust_mems): Likewise.
5218 (prepare_call_arguments): Likewise.
5220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5221 Alan Hayward <alan.hayward@arm.com>
5222 David Sherwood <david.sherwood@arm.com>
5224 * machmode.h (is_int_mode): New fuction.
5225 * combine.c (find_split_point): Use it.
5226 (combine_simplify_rtx): Likewise.
5227 (simplify_if_then_else): Likewise.
5228 (simplify_set): Likewise.
5229 (simplify_shift_const_1): Likewise.
5230 (simplify_comparison): Likewise.
5231 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5232 * cse.c (notreg_cost): Likewise.
5233 (cse_insn): Likewise.
5234 * cselib.c (cselib_lookup_1): Likewise.
5235 * dojump.c (do_jump_1): Likewise.
5236 (do_compare_rtx_and_jump): Likewise.
5237 * dse.c (get_call_args): Likewise.
5238 * dwarf2out.c (rtl_for_decl_init): Likewise.
5239 (native_encode_initializer): Likewise.
5240 * expmed.c (emit_store_flag_1): Likewise.
5241 (emit_store_flag): Likewise.
5242 * expr.c (convert_modes): Likewise.
5243 (store_field): Likewise.
5244 (expand_expr_real_1): Likewise.
5245 * fold-const.c (fold_read_from_constant_string): Likewise.
5246 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5247 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5248 * optabs.c (expand_binop): Likewise.
5249 (expand_unop): Likewise.
5250 (expand_abs_nojump): Likewise.
5251 (expand_one_cmpl_abs_nojump): Likewise.
5252 * simplify-rtx.c (mode_signbit_p): Likewise.
5253 (val_signbit_p): Likewise.
5254 (val_signbit_known_set_p): Likewise.
5255 (val_signbit_known_clear_p): Likewise.
5256 (simplify_relational_operation_1): Likewise.
5257 * tree.c (vector_type_mode): Likewise.
5259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5260 Alan Hayward <alan.hayward@arm.com>
5261 David Sherwood <david.sherwood@arm.com>
5263 * machmode.h (smallest_mode_for_size): Fix formatting.
5264 (smallest_int_mode_for_size): New function.
5265 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5266 instead of smallest_mode_for_size.
5267 * combine.c (make_extraction): Likewise.
5268 * config/arc/arc.c (arc_expand_movmem): Likewise.
5269 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5270 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5271 * config/s390/s390.c (s390_expand_insv): Likewise.
5272 * config/sparc/sparc.c (assign_int_registers): Likewise.
5273 * config/spu/spu.c (spu_function_value): Likewise.
5274 (spu_function_arg): Likewise.
5275 * coverage.c (get_gcov_type): Likewise.
5276 (get_gcov_unsigned_t): Likewise.
5277 * dse.c (find_shift_sequence): Likewise.
5278 * expmed.c (store_bit_field_1): Likewise.
5279 * expr.c (convert_move): Likewise.
5280 (store_field): Likewise.
5281 * internal-fn.c (expand_arith_overflow): Likewise.
5282 * optabs-query.c (get_best_extraction_insn): Likewise.
5283 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5284 * stor-layout.c (layout_type): Likewise.
5285 (initialize_sizetypes): Likewise.
5286 * targhooks.c (default_get_mask_mode): Likewise.
5287 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5290 Alan Hayward <alan.hayward@arm.com>
5291 David Sherwood <david.sherwood@arm.com>
5293 * machmode.h (opt_mode::else_blk): New function.
5294 (int_mode_for_mode): Declare.
5295 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5296 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5298 * cfgexpand.c (expand_debug_expr): Likewise.
5299 * combine.c (gen_lowpart_or_truncate): Likewise.
5300 (gen_lowpart_for_combine): Likewise.
5301 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5302 * config/avr/avr.c (avr_to_int_mode): Likewise.
5303 (avr_out_plus_1): Likewise.
5304 (avr_out_plus): Likewise.
5305 (avr_out_round): Likewise.
5306 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5307 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5308 (s390_expand_vcond): Likewise.
5309 * config/spu/spu.c (spu_split_immediate): Likewise.
5310 (spu_expand_mov): Likewise.
5311 * dse.c (get_stored_val): Likewise.
5312 * expmed.c (store_bit_field_1): Likewise.
5313 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5315 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5316 (extract_low_bits): Likewise.
5317 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5318 handling rather than repeating the check.
5319 (emit_group_store): Likewise.
5320 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5321 * optabs.c (expand_absneg_bit): Likewise.
5322 (expand_copysign_absneg): Likewise.
5323 (expand_copysign_bit): Likewise.
5324 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5325 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5326 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5327 * var-tracking.c (prepare_call_arguments): Likewise.
5328 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5329 int_mode_for_mode instead of mode_for_size.
5330 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5333 Alan Hayward <alan.hayward@arm.com>
5334 David Sherwood <david.sherwood@arm.com>
5336 * machmode.h (int_mode_for_size): New function.
5337 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5338 instead of mode_for_size.
5339 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5341 * combine.c (expand_field_assignment): Use int_mode_for_size
5342 instead of mode_for_size.
5343 (make_extraction): Likewise.
5344 (simplify_shift_const_1): Likewise.
5345 (simplify_comparison): Likewise.
5346 * dojump.c (do_jump): Likewise.
5347 * dwarf2out.c (mem_loc_descriptor): Likewise.
5348 * emit-rtl.c (init_derived_machine_modes): Likewise.
5349 * expmed.c (flip_storage_order): Likewise.
5350 (convert_extracted_bit_field): Likewise.
5351 * expr.c (copy_blkmode_from_reg): Likewise.
5352 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5353 * internal-fn.c (expand_mul_overflow): Likewise.
5354 * lower-subreg.c (simple_move): Likewise.
5355 * optabs-libfuncs.c (init_optabs): Likewise.
5356 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5357 * tree.c (vector_type_mode): Likewise.
5358 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5359 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5360 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5361 * tree-vect-stmts.c (vectorizable_load): Likewise.
5362 (vectorizable_store): Likewise.
5364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5365 Alan Hayward <alan.hayward@arm.com>
5366 David Sherwood <david.sherwood@arm.com>
5368 * coretypes.h (pod_mode): New type.
5369 (scalar_int_mode_pod): New typedef.
5370 * machmode.h (pod_mode): New class.
5371 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5372 * genmodes.c (emit_mode_int_n): Update accordingly.
5373 * lower-subreg.h (target_lower_subreg): Change type to
5374 scalar_int_mode_pod.
5375 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5376 scalar_int_mode_pod.
5378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5379 Alan Hayward <alan.hayward@arm.com>
5380 David Sherwood <david.sherwood@arm.com>
5382 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5383 machine_mode to scalar_int_mode.
5384 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5385 (rs6000_option_override_internal): Remove cast to int.
5386 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5387 machine_mode to scalar_int_mode.
5388 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5389 (rs6000_option_override_internal): Remove cast to int.
5390 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5391 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5393 * config/s390/s390.c (s390_expand_builtin): Likewise.
5394 * coretypes.h (scalar_int_mode): New type.
5395 (opt_scalar_int_mode): New typedef.
5396 * machmode.h (scalar_int_mode): New class.
5397 (scalar_int_mode::includes_p): New function.
5398 (byte_mode): Change type to scalar_int_mode.
5399 (word_mode): Likewise.
5400 (ptr_mode): Likewise.
5401 * emit-rtl.c (byte_mode): Likewise.
5402 (word_mode): Likewise.
5403 (ptr_mode): Likewise.
5404 (init_derived_machine_modes): Update accordingly.
5405 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5406 and MODE_PARTIAL_INT.
5407 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5408 opt_scalar_int_mode.
5410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5411 Alan Hayward <alan.hayward@arm.com>
5412 David Sherwood <david.sherwood@arm.com>
5414 * target.def (libgcc_floating_mode_supported_p): Take a
5416 * doc/tm.texi: Regenerate.
5417 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5419 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5420 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5424 Alan Hayward <alan.hayward@arm.com>
5425 David Sherwood <david.sherwood@arm.com>
5427 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5428 * doc/tm.texi: Regenerate.
5429 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5430 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5431 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5432 * targhooks.h (default_floatn_mode): Likewise.
5433 * targhooks.c (default_floatn_mode): Likewise.
5434 * tree.c (build_common_tree_nodes): Update accordingly.
5436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5437 Alan Hayward <alan.hayward@arm.com>
5438 David Sherwood <david.sherwood@arm.com>
5440 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5441 (mode_iterator::iterate_p): Likewise.
5442 (mode_iterator::get_wider): Likewise.
5443 * expr.c (init_expr_target): Use opt_scalar_float_mode.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * coretypes.h (opt_scalar_float_mode): New typedef.
5450 * machmode.h (float_mode_for_size): New function.
5451 * emit-rtl.c (double_mode): Delete.
5452 (init_emit_once): Use float_mode_for_size.
5453 * stor-layout.c (layout_type): Likewise.
5454 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5457 Alan Hayward <alan.hayward@arm.com>
5458 David Sherwood <david.sherwood@arm.com>
5460 * output.h (assemble_real): Take a scalar_float_mode.
5461 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5462 * config/arm/arm.md (consttable_4): Likewise.
5463 (consttable_8): Likewise.
5464 (consttable_16): Likewise.
5465 * config/mips/mips.md (consttable_float): Likewise.
5466 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5467 * varasm.c (assemble_real): Take a scalar_float_mode.
5468 (output_constant_pool_2): Update accordingly.
5469 (output_constant): Likewise.
5471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5472 Alan Hayward <alan.hayward@arm.com>
5473 David Sherwood <david.sherwood@arm.com>
5475 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5476 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5477 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5478 (native_encode_real): Likewise.
5479 (native_interpret_real): Likewise.
5480 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5481 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5484 Alan Hayward <alan.hayward@arm.com>
5485 David Sherwood <david.sherwood@arm.com>
5487 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5488 <scalar_float_mode>. Simplify.
5489 (gen_extend_conv_libfunc): Likewise.
5491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5492 Alan Hayward <alan.hayward@arm.com>
5493 David Sherwood <david.sherwood@arm.com>
5495 * coretypes.h (scalar_float_mode): New type.
5496 * machmode.h (mode_traits::from_int): Use machine_mode if
5497 USE_ENUM_MODES is defined.
5498 (is_a): New function.
5500 (dyn_cast): Likewise.
5501 (scalar_float_mode): New class.
5502 (scalar_float_mode::includes_p): New function.
5503 (is_float_mode): Likewise.
5504 * gdbhooks.py (MachineModePrinter): New class.
5505 (build_pretty_printer): Use it for scalar_float_mode.
5506 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5507 (format_helper::format_helper): Turn into a template.
5508 * genmodes.c (get_mode_class): New function.
5509 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5510 or machine_mode if none.
5511 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5512 as_a <scalar_float_mode>.
5513 * dwarf2out.c (mem_loc_descriptor): Likewise.
5514 (insert_float): Likewise.
5515 (add_const_value_attribute): Likewise.
5516 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5517 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5518 (expand_unop): Update accordingly.
5519 (expand_abs_nojump): Likewise.
5520 (expand_copysign_absneg): Take a scalar_float_mode.
5521 (expand_copysign_bit): Likewise.
5522 (expand_copysign): Update accordingly.
5524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5525 Alan Hayward <alan.hayward@arm.com>
5526 David Sherwood <david.sherwood@arm.com>
5528 * coretypes.h (opt_mode): New class.
5529 * machmode.h (opt_mode): Likewise.
5530 (opt_mode::else_void): New function.
5531 (opt_mode::require): Likewise.
5532 (opt_mode::exists): Likewise.
5533 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5534 (GET_MODE_2XWIDER_MODE): Likewise.
5535 (mode_iterator::get_wider): Update accordingly.
5536 (mode_iterator::get_2xwider): Likewise.
5537 (mode_iterator::get_known_wider): Likewise, turning into a template.
5538 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5539 forcing a wider mode to exist.
5540 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5541 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5542 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5543 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5544 * lower-subreg.c (init_lower_subreg): Likewise.
5545 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5546 on the final iteration.
5547 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5548 a wider mode exists before asking for a move pattern.
5549 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5550 forcing a wider mode to exist.
5551 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5552 returning false if no such mode exists.
5553 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5554 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5555 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5556 Avoid checking for a MODE_INT if we already know the mode is not a
5558 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
5559 forcing a wider mode to exist.
5560 (expmed_mult_highpart_optab): Likewise.
5561 (expmed_mult_highpart): Likewise.
5562 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
5564 * lto-streamer-in.c (lto_input_mode_table): Likewise.
5565 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
5566 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5567 * internal-fn.c (expand_mul_overflow): Update use of
5568 GET_MODE_2XWIDER_MODE.
5569 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5570 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
5571 GET_MODE_WIDER_MODE.
5572 (convert_plusminus_to_widen): Likewise.
5573 * tree-switch-conversion.c (array_value_type): Likewise.
5574 * var-tracking.c (emit_note_insn_var_location): Likewise.
5575 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5576 Return false inside rather than outside the loop if no wider mode
5578 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
5579 and GET_MODE_2XWIDER_MODE
5580 (can_compare_p): Use else_void.
5581 * gdbhooks.py (OptMachineModePrinter): New class.
5582 (build_pretty_printer): Use it for opt_mode.
5584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5585 Alan Hayward <alan.hayward@arm.com>
5586 David Sherwood <david.sherwood@arm.com>
5588 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5589 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5592 Alan Hayward <alan.hayward@arm.com>
5593 David Sherwood <david.sherwood@arm.com>
5595 * machmode.h (mode_traits): New structure.
5596 (get_narrowest_mode): New function.
5597 (mode_iterator::start): Likewise.
5598 (mode_iterator::iterate_p): Likewise.
5599 (mode_iterator::get_wider): Likewise.
5600 (mode_iterator::get_known_wider): Likewise.
5601 (mode_iterator::get_2xwider): Likewise.
5602 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5603 (FOR_EACH_MODE): Likewise.
5604 (FOR_EACH_MODE_FROM): Likewise.
5605 (FOR_EACH_MODE_UNTIL): Likewise.
5606 (FOR_EACH_WIDER_MODE): Likewise.
5607 (FOR_EACH_2XWIDER_MODE): Likewise.
5608 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5609 * combine.c (simplify_comparison): Likewise
5610 * config/i386/i386.c (type_natural_mode): Likewise.
5611 * cse.c (cse_insn): Likewise.
5612 * dse.c (find_shift_sequence): Likewise.
5613 * emit-rtl.c (init_derived_machine_modes): Likewise.
5614 (init_emit_once): Likewise.
5615 * explow.c (hard_function_value): Likewise.
5616 * expmed.c (extract_fixed_bit_field_1): Likewise.
5617 (extract_bit_field_1): Likewise.
5618 (expand_divmod): Likewise.
5619 (emit_store_flag_1): Likewise.
5620 * expr.c (init_expr_target): Likewise.
5621 (convert_move): Likewise.
5622 (alignment_for_piecewise_move): Likewise.
5623 (widest_int_mode_for_size): Likewise.
5624 (emit_block_move_via_movmem): Likewise.
5625 (copy_blkmode_to_reg): Likewise.
5626 (set_storage_via_setmem): Likewise.
5627 (compress_float_constant): Likewise.
5628 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5629 * optabs-query.c (get_best_extraction_insn): Likewise.
5630 * optabs.c (expand_binop): Likewise.
5631 (expand_twoval_unop): Likewise.
5632 (expand_twoval_binop): Likewise.
5633 (widen_leading): Likewise.
5634 (widen_bswap): Likewise.
5635 (expand_parity): Likewise.
5636 (expand_unop): Likewise.
5637 (prepare_cmp_insn): Likewise.
5638 (prepare_float_lib_cmp): Likewise.
5639 (expand_float): Likewise.
5640 (expand_fix): Likewise.
5641 (expand_sfix_optab): Likewise.
5642 * postreload.c (move2add_use_add2_insn): Likewise.
5643 * reg-stack.c (reg_to_stack): Likewise.
5644 * reginfo.c (choose_hard_reg_mode): Likewise.
5645 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5646 * stor-layout.c (mode_for_size): Likewise.
5647 (smallest_mode_for_size): Likewise.
5648 (mode_for_vector): Likewise.
5649 (finish_bitfield_representative): Likewise.
5650 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5651 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5652 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5653 * var-tracking.c (prepare_call_arguments): Likewise.
5655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5656 Alan Hayward <alan.hayward@arm.com>
5657 David Sherwood <david.sherwood@arm.com>
5659 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5660 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5661 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5662 * machmode.h (mode_size): Move earlier in file.
5663 (mode_precision): Likewise.
5664 (mode_inner): Likewise.
5665 (mode_nunits): Likewise.
5666 (mode_unit_size): Likewise.
5667 (unit_unit_precision): Likewise.
5668 (mode_wider): Likewise.
5669 (mode_2xwider): Likewise.
5670 (machine_mode): New class.
5671 (mode_to_bytes): New function.
5672 (mode_to_bits): Likewise.
5673 (mode_to_precision): Likewise.
5674 (mode_to_inner): Likewise.
5675 (mode_to_unit_size): Likewise.
5676 (mode_to_unit_precision): Likewise.
5677 (mode_to_nunits): Likewise.
5678 (GET_MODE_SIZE): Use mode_to_bytes.
5679 (GET_MODE_BITSIZE): Use mode_to_bits.
5680 (GET_MODE_PRECISION): Use mode_to_precision.
5681 (GET_MODE_INNER): Use mode_to_inner.
5682 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5683 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5684 (GET_MODE_NUNITS): Use mode_to_nunits.
5685 * system.h (ALWAYS_INLINE): New macro.
5686 * config/powerpcspe/powerpcspe-c.c
5687 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5688 int for arg1_mode and arg2_mode.
5690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5691 Alan Hayward <alan.hayward@arm.com>
5692 David Sherwood <david.sherwood@arm.com>
5694 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5695 Prefix mode names with E_ in case statements.
5696 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5697 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5698 (aarch64_split_simd_move): Likewise.
5699 (aarch64_gen_storewb_pair): Likewise.
5700 (aarch64_gen_loadwb_pair): Likewise.
5701 (aarch64_gen_store_pair): Likewise.
5702 (aarch64_gen_load_pair): Likewise.
5703 (aarch64_get_condition_code_1): Likewise.
5704 (aarch64_constant_pool_reload_icode): Likewise.
5705 (get_rsqrte_type): Likewise.
5706 (get_rsqrts_type): Likewise.
5707 (get_recpe_type): Likewise.
5708 (get_recps_type): Likewise.
5709 (aarch64_gimplify_va_arg_expr): Likewise.
5710 (aarch64_simd_container_mode): Likewise.
5711 (aarch64_emit_load_exclusive): Likewise.
5712 (aarch64_emit_store_exclusive): Likewise.
5713 (aarch64_expand_compare_and_swap): Likewise.
5714 (aarch64_gen_atomic_cas): Likewise.
5715 (aarch64_emit_bic): Likewise.
5716 (aarch64_emit_atomic_swap): Likewise.
5717 (aarch64_emit_atomic_load_op): Likewise.
5718 (aarch64_evpc_trn): Likewise.
5719 (aarch64_evpc_uzp): Likewise.
5720 (aarch64_evpc_zip): Likewise.
5721 (aarch64_evpc_ext): Likewise.
5722 (aarch64_evpc_rev): Likewise.
5723 (aarch64_evpc_dup): Likewise.
5724 (aarch64_gen_ccmp_first): Likewise.
5725 (aarch64_gen_ccmp_next): Likewise.
5726 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5727 (alpha_emit_xfloating_libcall): Likewise.
5728 (emit_insxl): Likewise.
5729 (alpha_arg_type): Likewise.
5730 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5731 (arc_preferred_simd_mode): Likewise.
5732 (arc_secondary_reload): Likewise.
5733 (get_arc_condition_code): Likewise.
5734 (arc_print_operand): Likewise.
5735 (arc_legitimate_constant_p): Likewise.
5736 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5737 * config/arc/arc.md (casesi_load): Likewise.
5738 (casesi_compact_jump): Likewise.
5739 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5740 (cc_use_register): Likewise.
5741 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5742 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5743 (arm_init_iwmmxt_builtins): Likewise.
5744 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5745 (neon_expand_vector_init): Likewise.
5746 (arm_attr_length_move_neon): Likewise.
5747 (maybe_get_arm_condition_code): Likewise.
5748 (arm_emit_vector_const): Likewise.
5749 (arm_preferred_simd_mode): Likewise.
5750 (arm_output_iwmmxt_tinsr): Likewise.
5751 (thumb1_output_casesi): Likewise.
5752 (thumb2_output_casesi): Likewise.
5753 (arm_emit_load_exclusive): Likewise.
5754 (arm_emit_store_exclusive): Likewise.
5755 (arm_expand_compare_and_swap): Likewise.
5756 (arm_evpc_neon_vuzp): Likewise.
5757 (arm_evpc_neon_vzip): Likewise.
5758 (arm_evpc_neon_vrev): Likewise.
5759 (arm_evpc_neon_vtrn): Likewise.
5760 (arm_evpc_neon_vext): Likewise.
5761 (arm_validize_comparison): Likewise.
5762 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5763 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5764 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5765 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5766 (c6x_preferred_simd_mode): Likewise.
5767 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5768 (epiphany_rtx_costs): Likewise.
5769 * config/epiphany/predicates.md (proper_comparison_operator):
5771 * config/frv/frv.c (condexec_memory_operand): Likewise.
5772 (frv_emit_move): Likewise.
5773 (output_move_single): Likewise.
5774 (output_condmove_single): Likewise.
5775 (frv_hard_regno_mode_ok): Likewise.
5776 (frv_matching_accg_mode): Likewise.
5777 * config/h8300/h8300.c (split_adds_subs): Likewise.
5778 (h8300_rtx_costs): Likewise.
5779 (h8300_print_operand): Likewise.
5780 (compute_mov_length): Likewise.
5781 (output_logical_op): Likewise.
5782 (compute_logical_op_length): Likewise.
5783 (compute_logical_op_cc): Likewise.
5784 (h8300_shift_needs_scratch_p): Likewise.
5785 (output_a_shift): Likewise.
5786 (compute_a_shift_length): Likewise.
5787 (compute_a_shift_cc): Likewise.
5788 (expand_a_rotate): Likewise.
5789 (output_a_rotate): Likewise.
5790 * config/i386/i386.c (classify_argument): Likewise.
5791 (function_arg_advance_32): Likewise.
5792 (function_arg_32): Likewise.
5793 (function_arg_64): Likewise.
5794 (function_value_64): Likewise.
5795 (ix86_gimplify_va_arg): Likewise.
5796 (ix86_legitimate_constant_p): Likewise.
5797 (put_condition_code): Likewise.
5798 (split_double_mode): Likewise.
5799 (ix86_avx256_split_vector_move_misalign): Likewise.
5800 (ix86_expand_vector_logical_operator): Likewise.
5801 (ix86_split_idivmod): Likewise.
5802 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5803 (ix86_build_const_vector): Likewise.
5804 (ix86_build_signbit_mask): Likewise.
5805 (ix86_match_ccmode): Likewise.
5806 (ix86_cc_modes_compatible): Likewise.
5807 (ix86_expand_branch): Likewise.
5808 (ix86_expand_sse_cmp): Likewise.
5809 (ix86_expand_sse_movcc): Likewise.
5810 (ix86_expand_int_sse_cmp): Likewise.
5811 (ix86_expand_vec_perm_vpermi2): Likewise.
5812 (ix86_expand_vec_perm): Likewise.
5813 (ix86_expand_sse_unpack): Likewise.
5814 (ix86_expand_int_addcc): Likewise.
5815 (ix86_split_to_parts): Likewise.
5816 (ix86_vectorize_builtin_gather): Likewise.
5817 (ix86_vectorize_builtin_scatter): Likewise.
5818 (avx_vpermilp_parallel): Likewise.
5819 (inline_memory_move_cost): Likewise.
5820 (ix86_tieable_integer_mode_p): Likewise.
5821 (x86_maybe_negate_const_int): Likewise.
5822 (ix86_expand_vector_init_duplicate): Likewise.
5823 (ix86_expand_vector_init_one_nonzero): Likewise.
5824 (ix86_expand_vector_init_one_var): Likewise.
5825 (ix86_expand_vector_init_concat): Likewise.
5826 (ix86_expand_vector_init_interleave): Likewise.
5827 (ix86_expand_vector_init_general): Likewise.
5828 (ix86_expand_vector_set): Likewise.
5829 (ix86_expand_vector_extract): Likewise.
5830 (emit_reduc_half): Likewise.
5831 (ix86_emit_i387_round): Likewise.
5832 (ix86_mangle_type): Likewise.
5833 (ix86_expand_round_sse4): Likewise.
5834 (expand_vec_perm_blend): Likewise.
5835 (canonicalize_vector_int_perm): Likewise.
5836 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5837 (expand_vec_perm_1): Likewise.
5838 (expand_vec_perm_interleave3): Likewise.
5839 (expand_vec_perm_even_odd_pack): Likewise.
5840 (expand_vec_perm_even_odd_1): Likewise.
5841 (expand_vec_perm_broadcast_1): Likewise.
5842 (ix86_vectorize_vec_perm_const_ok): Likewise.
5843 (ix86_expand_vecop_qihi): Likewise.
5844 (ix86_expand_mul_widen_hilo): Likewise.
5845 (ix86_expand_sse2_abs): Likewise.
5846 (ix86_expand_pextr): Likewise.
5847 (ix86_expand_pinsr): Likewise.
5848 (ix86_preferred_simd_mode): Likewise.
5849 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5850 * config/i386/sse.md (*andnot<mode>3): Likewise.
5851 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5852 (*<code><mode>3): Likewise.
5853 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5854 (ia64_expand_atomic_op): Likewise.
5855 (ia64_arg_type): Likewise.
5856 (ia64_mode_to_int): Likewise.
5857 (ia64_scalar_mode_supported_p): Likewise.
5858 (ia64_vector_mode_supported_p): Likewise.
5859 (expand_vec_perm_broadcast): Likewise.
5860 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5861 (iq2000_function_arg_advance): Likewise.
5862 (iq2000_function_arg): Likewise.
5863 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5864 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5865 (m68k_libcall_value): Likewise.
5866 (m68k_function_value): Likewise.
5867 (sched_attr_op_type): Likewise.
5868 * config/mcore/mcore.c (mcore_output_move): Likewise.
5869 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5871 (microblaze_function_arg): Likewise.
5872 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5873 (mips_print_operand): Likewise.
5874 (mips_mode_ok_for_mov_fmt_p): Likewise.
5875 (mips_vector_mode_supported_p): Likewise.
5876 (mips_preferred_simd_mode): Likewise.
5877 (mips_expand_vpc_loongson_even_odd): Likewise.
5878 (mips_expand_vec_unpack): Likewise.
5879 (mips_expand_vi_broadcast): Likewise.
5880 (mips_expand_vector_init): Likewise.
5881 (mips_expand_vec_reduc): Likewise.
5882 (mips_expand_msa_cmp): Likewise.
5883 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5884 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5885 (cc_flags_for_mode): Likewise.
5886 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5887 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5888 (nds32_output_casesi_pc_relative): Likewise.
5889 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5890 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5891 (nvptx_gen_unpack): Likewise.
5892 (nvptx_gen_pack): Likewise.
5893 (nvptx_gen_shuffle): Likewise.
5894 (nvptx_gen_wcast): Likewise.
5895 (nvptx_preferred_simd_mode): Likewise.
5896 * config/pa/pa.c (pa_secondary_reload): Likewise.
5897 * config/pa/predicates.md (base14_operand): Likewise.
5898 * config/powerpcspe/powerpcspe-c.c
5899 (altivec_resolve_overloaded_builtin): Likewise.
5900 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5902 (rs6000_preferred_simd_mode): Likewise.
5903 (output_vec_const_move): Likewise.
5904 (rs6000_expand_vector_extract): Likewise.
5905 (rs6000_split_vec_extract_var): Likewise.
5906 (reg_offset_addressing_ok_p): Likewise.
5907 (rs6000_legitimate_offset_address_p): Likewise.
5908 (rs6000_legitimize_address): Likewise.
5909 (rs6000_emit_set_const): Likewise.
5910 (rs6000_const_vec): Likewise.
5911 (rs6000_emit_move): Likewise.
5912 (spe_build_register_parallel): Likewise.
5913 (rs6000_darwin64_record_arg_recurse): Likewise.
5914 (swap_selector_for_mode): Likewise.
5915 (spe_init_builtins): Likewise.
5916 (paired_init_builtins): Likewise.
5917 (altivec_init_builtins): Likewise.
5918 (do_load_for_compare): Likewise.
5919 (rs6000_generate_compare): Likewise.
5920 (rs6000_expand_float128_convert): Likewise.
5921 (emit_load_locked): Likewise.
5922 (emit_store_conditional): Likewise.
5923 (rs6000_output_function_epilogue): Likewise.
5924 (rs6000_handle_altivec_attribute): Likewise.
5925 (rs6000_function_value): Likewise.
5926 (emit_fusion_gpr_load): Likewise.
5927 (emit_fusion_p9_load): Likewise.
5928 (emit_fusion_p9_store): Likewise.
5929 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5930 (fusion_gpr_mem_load): Likewise.
5931 (fusion_addis_mem_combo_load): Likewise.
5932 (fusion_addis_mem_combo_store): Likewise.
5933 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5934 (fusion_gpr_mem_load): Likewise.
5935 (fusion_addis_mem_combo_load): Likewise.
5936 (fusion_addis_mem_combo_store): Likewise.
5937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5939 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5940 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5941 (rs6000_preferred_simd_mode): Likewise.
5942 (output_vec_const_move): Likewise.
5943 (rs6000_expand_vector_extract): Likewise.
5944 (rs6000_split_vec_extract_var): Likewise.
5945 (reg_offset_addressing_ok_p): Likewise.
5946 (rs6000_legitimate_offset_address_p): Likewise.
5947 (rs6000_legitimize_address): Likewise.
5948 (rs6000_emit_set_const): Likewise.
5949 (rs6000_const_vec): Likewise.
5950 (rs6000_emit_move): Likewise.
5951 (rs6000_darwin64_record_arg_recurse): Likewise.
5952 (swap_selector_for_mode): Likewise.
5953 (paired_init_builtins): Likewise.
5954 (altivec_init_builtins): Likewise.
5955 (rs6000_expand_float128_convert): Likewise.
5956 (emit_load_locked): Likewise.
5957 (emit_store_conditional): Likewise.
5958 (rs6000_output_function_epilogue): Likewise.
5959 (rs6000_handle_altivec_attribute): Likewise.
5960 (rs6000_function_value): Likewise.
5961 (emit_fusion_gpr_load): Likewise.
5962 (emit_fusion_p9_load): Likewise.
5963 (emit_fusion_p9_store): Likewise.
5964 * config/rx/rx.c (rx_gen_move_template): Likewise.
5965 (flags_from_mode): Likewise.
5966 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5967 (s390_slb_comparison): Likewise.
5968 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5969 (s390_vector_mode_supported_p): Likewise.
5970 (s390_cc_modes_compatible): Likewise.
5971 (s390_match_ccmode_set): Likewise.
5972 (s390_canonicalize_comparison): Likewise.
5973 (s390_emit_compare_and_swap): Likewise.
5974 (s390_branch_condition_mask): Likewise.
5975 (s390_rtx_costs): Likewise.
5976 (s390_secondary_reload): Likewise.
5977 (__SECONDARY_RELOAD_CASE): Likewise.
5978 (s390_expand_cs): Likewise.
5979 (s390_preferred_simd_mode): Likewise.
5980 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5981 * config/sh/sh.c (sh_print_operand): Likewise.
5982 (dump_table): Likewise.
5983 (sh_secondary_reload): Likewise.
5984 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5985 * config/sh/sh.md (casesi_worker_1): Likewise.
5986 (casesi_worker_2): Likewise.
5987 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5988 (fcc_comparison_operator): Likewise.
5989 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5990 (emit_soft_tfmode_cvt): Likewise.
5991 (sparc_preferred_simd_mode): Likewise.
5992 (output_cbranch): Likewise.
5993 (sparc_print_operand): Likewise.
5994 (sparc_expand_vec_perm_bmask): Likewise.
5995 (vector_init_bshuffle): Likewise.
5996 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5997 (spu_vector_mode_supported_p): Likewise.
5998 (spu_expand_insv): Likewise.
5999 (spu_emit_branch_or_set): Likewise.
6000 (spu_handle_vector_attribute): Likewise.
6001 (spu_builtin_splats): Likewise.
6002 (spu_builtin_extract): Likewise.
6003 (spu_builtin_promote): Likewise.
6004 (spu_expand_sign_extend): Likewise.
6005 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6006 (tilegx_simd_int): Likewise.
6007 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6008 (tilepro_simd_int): Likewise.
6009 * config/v850/v850.c (const_double_split): Likewise.
6010 (v850_print_operand): Likewise.
6011 (ep_memory_offset): Likewise.
6012 * config/vax/vax.c (vax_rtx_costs): Likewise.
6013 (vax_output_int_move): Likewise.
6014 (vax_output_int_add): Likewise.
6015 (vax_output_int_subtract): Likewise.
6016 * config/visium/predicates.md (visium_branch_operator): Likewise.
6017 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6018 (visium_print_operand_address): Likewise.
6019 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6020 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6021 (xtensa_expand_conditional_branch): Likewise.
6022 (xtensa_copy_incoming_a7): Likewise.
6023 (xtensa_output_literal): Likewise.
6024 * dfp.c (decimal_real_maxval): Likewise.
6025 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6028 Alan Hayward <alan.hayward@arm.com>
6029 David Sherwood <david.sherwood@arm.com>
6031 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6032 (mode_nunits_inline): Likewise.
6033 (mode_inner_inline): Likewise.
6034 (mode_unit_size_inline): Likewise.
6035 (mode_unit_precision_inline): Likewise.
6036 (emit_insn_modes_h): Likewise. Also emit a #define of the
6038 (emit_mode_wider): Add an E_ prefix to mode names.
6039 (emit_mode_complex): Likewise.
6040 (emit_mode_inner): Likewise.
6041 (emit_mode_adjustments): Likewise.
6042 (emit_mode_int_n): Likewise.
6043 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6044 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6045 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6046 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6047 (CRC32_BUILTIN, ENTRY): Likewise.
6048 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6049 (aarch64_pop_regs): Likewise.
6050 (aarch64_process_components): Likewise.
6051 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6052 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6053 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6054 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6055 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6056 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6058 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6059 (pcmpistr): Likewise.
6060 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6061 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6062 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6064 * config/rl78/rl78.c (mduc_regs): Likewise.
6065 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6066 (htm_expand_builtin): Likewise.
6067 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6068 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6069 * config/xtensa/xtensa.c (print_operand): Likewise.
6070 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6071 (NUM_MODE_VECTOR_INT): Likewise.
6072 * genoutput.c (null_operand): Likewise.
6073 (output_operand_data): Likewise.
6074 * genrecog.c (print_parameter_value): Likewise.
6075 * lra.c (debug_operand_data): Likewise.
6077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6079 * dwarf2out.c (scompare_loc_descriptor_wide)
6080 (scompare_loc_descriptor_narrow): New functions, split out from...
6081 (scompare_loc_descriptor): ...here.
6082 * expmed.c (emit_store_flag_int): New function, split out from...
6083 (emit_store_flag): ...here.
6085 2017-08-30 Richard Biener <rguenther@suse.de>
6087 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6088 (dwarf2out_early_finish): Move setting of AT_pubnames from
6089 early debug output to early finish.
6091 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6093 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6094 and -mdata-region to the assembler.
6096 2017-08-30 Richard Biener <rguenther@suse.de>
6098 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6100 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6101 (dwarf2out_early_global_decl): Only generate a DIE for the
6102 abstract origin if it doesn't already exist or is a declaration DIE.
6103 (resolve_addr): Do not add the linkage name twice when
6104 generating a stub DIE for the DW_TAG_GNU_call_site target.
6106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6108 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6109 Use machine_mode rather than int for arg1_mode.
6111 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6114 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6115 that the second argument of the built-in functions to unpack
6116 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6117 switch statement instead a lot of if statements.
6118 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6119 Allow 64-bit values to be in Altivec registers as well as
6120 traditional floating point registers.
6121 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6123 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6125 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6126 MAX_REGS_PER_ADDRESS == 1.
6128 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6130 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6131 * config/i386/i386.c (ix86_option_override_internal): Simplify
6132 setting of opts->x_flag_entry.
6134 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6135 Jakub Jelinek <jakub@redhat.com>
6136 Richard Biener <rguenther@suse.de>
6138 PR tree-optimization/81503
6139 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6140 folded constant fits in the target type; reorder tests for clarity.
6142 2017-08-29 Martin Liska <mliska@suse.cz>
6144 * passes.def: Include pass_lower_switch.
6145 * stmt.c (dump_case_nodes): Remove and move to
6146 tree-switch-conversion.
6147 (case_values_threshold): Likewise.
6148 (expand_switch_as_decision_tree_p): Likewise.
6149 (emit_case_decision_tree): Likewise.
6150 (expand_case): Likewise.
6151 (balance_case_nodes): Likewise.
6152 (node_has_low_bound): Likewise.
6153 (node_has_high_bound): Likewise.
6154 (node_is_bounded): Likewise.
6155 (emit_case_nodes): Likewise.
6156 (struct simple_case_node): New struct.
6157 (add_case_node): Remove.
6158 (emit_case_dispatch_table): Use vector instead of case_list.
6159 (reset_out_edges_aux): Remove.
6160 (compute_cases_per_edge): Likewise.
6161 (expand_case): Build list of simple_case_node.
6162 (expand_sjlj_dispatch_table): Use it.
6163 * tree-switch-conversion.c (struct case_node): Moved from
6164 stmt.c and adjusted.
6165 (emit_case_nodes): Likewise.
6166 (node_has_low_bound): Likewise.
6167 (node_has_high_bound): Likewise.
6168 (node_is_bounded): Likewise.
6169 (case_values_threshold): Likewise.
6170 (reset_out_edges_aux): Likewise.
6171 (compute_cases_per_edge): Likewise.
6172 (add_case_node): Likewise.
6173 (dump_case_nodes): Likewise.
6174 (balance_case_nodes): Likewise.
6175 (expand_switch_as_decision_tree_p): Likewise.
6176 (emit_jump): Likewise.
6177 (emit_case_decision_tree): Likewise.
6178 (try_switch_expansion): Likewise.
6179 (do_jump_if_equal): Likewise.
6180 (emit_cmp_and_jump_insns): Likewise.
6181 (fix_phi_operands_for_edge): New function.
6182 (record_phi_operand_mapping): Likewise.
6183 (class pass_lower_switch): New pass.
6184 (pass_lower_switch::execute): New function.
6185 (make_pass_lower_switch): Likewise.
6186 (conditional_probability):
6187 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6188 * tree-pass.h: Add make_pass_lower_switch.
6190 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6193 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6196 2017-08-29 Richard Biener <rguenther@suse.de>
6198 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6199 we do not add a DW_AT_inline attribute twice.
6200 (gen_subprogram_die): Remove code setting DW_AT_inline on
6201 DECL_ABSTRACT_P nodes.
6203 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6205 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6206 calls to internal functions.
6207 (gimplify_modify_expr): Likewise.
6208 * tree-call-cdce.c (use_internal_fn): Likewise.
6209 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6210 (convert_to_divmod): Set the nothrow flag.
6211 * tree-if-conv.c (predicate_mem_writes): Likewise.
6212 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6213 (vectorizable_call): Likewise.
6214 (vectorizable_store): Likewise.
6215 (vectorizable_load): Likewise.
6216 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6217 (vect_recog_mask_conversion_pattern): Likewise.
6219 2017-08-29 Martin Liska <mliska@suse.cz>
6222 * gcc.c (driver_handle_option): Add new argument.
6223 * opts-common.c (handle_option): Pass
6224 target_option_override_hook.
6225 * opts-global.c (lang_handle_option): Add new option.
6226 (set_default_handlers): Add new argument.
6227 (decode_options): Likewise.
6228 * opts.c (target_handle_option): Likewise.
6229 (common_handle_option): Call target_option_override_hook.
6230 * opts.h (struct cl_option_handler_func): Add hook for
6231 target option override.
6232 (struct cl_option_handlers): Likewise.
6233 (set_default_handlers): Add new argument.
6234 (decode_options): Likewise.
6235 (common_handle_option): Likewise.
6236 (target_handle_option): Likewise.
6237 * toplev.c (toplev::main): Pass targetm.target_option.override
6240 2017-08-29 Richard Biener <rguenther@suse.de>
6241 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6243 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6244 life to the active subtree.
6246 2017-08-28 Jeff Law <law@redhat.com>
6248 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6249 derive_equivalences.
6250 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6252 (edge_info::derive_equivalences): New private member function
6254 * tree-ssa-dom.c (class edge_info): Changed from a struct
6255 to a class. Add ctor/dtor, methods and data members.
6256 (edge_info::edge_info): Renamed from allocate_edge_info.
6257 Initialize additional members.
6258 (edge_info::~edge_info): New.
6259 (free_dom_edge_info): Delete the edge info.
6260 (record_edge_info): Use new class & associated member functions.
6261 Tighten forms for testing for edge equivalences.
6262 (record_temporary_equivalences): Iterate over the simple
6263 equivalences rather than assuming there's only one per edge.
6264 (cprop_into_successor_phis): Iterate over the simple
6265 equivalences rather than assuming there's only one per edge.
6266 (optimize_stmt): Use operand_equal_p rather than pointer
6267 equality for mini-DSE code.
6269 2017-08-28 Nathan Sidwell <nathan@acm.org>
6271 * gcc.c (execute): Fold SIGPIPE handling into switch
6272 statement. Adjust internal error message.
6274 2017-08-28 Richard Biener <rguenther@suse.de>
6277 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6278 Do nothing for removed DIEs.
6280 2017-08-28 Richard Biener <rguenther@suse.de>
6282 PR tree-optimization/81977
6283 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6286 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6289 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6290 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6291 using targetm.gen_mem_thread_fence.
6293 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6296 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6297 predicate to register_operand. Reorder operands.
6298 (*btr<mode>): Ditto.
6299 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6300 (*btr<mode>_mask): Ditto.
6302 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6304 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6305 * config/rs6000/xmmintrin.h: New file.
6306 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6308 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6311 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6312 parameter and_insn and return it.
6313 (recombine_lvx_pattern): Insert a copy to ensure availability of
6314 the base register of the copied masking operation at the point of
6315 the instruction replacement.
6316 (recombine_stvx_pattern): Likewise.
6318 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6320 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6321 undocumented switches.
6322 (-mpower9-dform-vector): Likewise.
6323 (-mpower9-dform): Likewise.
6324 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6325 comments to delete references to -mpower9-dform* switches.
6326 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6327 Delete reference to -mpower9-dform* switches, test for
6328 -mpower9-vector instead.
6329 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6330 (OTHER_P9_VECTOR_MASKS): Likewise.
6331 (POWERPC_MASKS): Likewise.
6332 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6333 tests against -mpower9-dform* to -mpower9-vector. Delete code
6334 that checked for -mpower9-dform* consistancy with other options.
6335 Add test for -mpower9-misc to enable other power9 switches.
6336 (rs6000_init_hard_regno_mode_ok): Likewise.
6337 (rs6000_option_override_internal): Likewise.
6338 (rs6000_emit_prologue): Likewise.
6339 (rs6000_emit_epilogue): Likewise.
6340 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6341 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6342 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6344 (emit_fusion_p9_store): Likewise.
6345 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6346 resetting these macros if the assembler does not support ISA 3.0
6348 (TARGET_P9_DFORM_VECTOR): Likewise.
6349 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6350 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6352 2017-08-25 Alan Modra <amodra@gmail.com>
6354 PR rtl-optimization/81747
6355 * cse.c (cse_extended_basic_block): Don't attempt to record
6356 equivalences for degenerate conditional branches that branch
6357 to their fall-through.
6359 2017-08-24 Martin Sebor <msebor@redhat.com>
6362 * gimple-fold.c (size_must_be_zero_p): New function.
6363 (gimple_fold_builtin_memory_op): Call it.
6365 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6367 * config/rs6000/mm_malloc.h: New file.
6369 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6371 PR tree-optimization/81913
6372 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6373 analysis when either IVs in condition can wrap.
6375 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6377 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6378 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6380 2017-08-24 Richard Biener <rguenther@suse.de>
6383 * targhooks.c (default_target_can_inline_p): Properly
6384 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6385 is present and always compare.
6386 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6387 infer -mfpmath=sse from TARGET_SSE_P.
6388 (ix86_can_inline_p): Properly use target_option_default_node when
6389 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6391 2017-08-24 Richard Biener <rguenther@suse.de>
6394 * dwarf2out.c (output_die): Handle flag_generate_offload like
6396 (output_comp_unit): Likewise.
6397 (gen_array_type_die): Likewise.
6398 (dwarf2out_early_finish): Likewise.
6399 (note_variable_value_in_expr): Likewise.
6400 (dwarf2out_finish): Likewise. Adjust assert.
6401 * cgraphunit.c (symbol_table::compile): Move setting of
6402 flag_generate_offload earlier ...
6403 (symbol_table::finalize_compilation_unit): ... here, before
6404 early debug finalization.
6406 2017-08-24 Richard Biener <rguenther@suse.de>
6408 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6409 and ipa-fnsummary.h.
6410 (ix86_can_inline_p): When ix86_fpmath flags do not match
6411 check whether the callee uses FP math at all.
6413 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6416 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6417 instead of TYPE_PRECISION.
6419 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6420 Alan Hayward <alan.hayward@arm.com>
6421 David Sherwood <david.sherwood@arm.com>
6423 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6425 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6427 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6429 (ix86_frame::stack_realign_allocate): New field.
6430 (struct machine_frame_state): Modify comments.
6431 (machine_frame_state::sp_realigned_fp_end): New field.
6432 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6434 (sp_valid_at): Add assertion to assure no attempt to access invalid
6435 offset of a realigned stack.
6436 (fp_valid_at): Likewise.
6437 (choose_baseaddr): Modify comments.
6438 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6439 ix86_expand_prologue.
6440 (ix86_expand_prologue): Modify stack realignment and allocation.
6441 (ix86_expand_epilogue): Modify comments.
6442 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6443 avx2_runtime, avx512f, and avx512f_runtime.
6445 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6447 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6448 (mstackrealign): Do not init to -1.
6449 * config/i386/i386.c (ix86_option_override_internal):
6450 Check opts_set, not opts when setting default value of
6451 opts->x_ix86_force_align_arg_pointer.
6453 2017-08-23 Richard Biener <rguenther@suse.de>
6455 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6456 lang_hooks.decl_printable_name.
6457 * print-rtl-function.c (print_rtx_function): Likewise.
6458 * tree-pretty-print.c (dump_function_header): Likewise.
6460 2017-08-23 Richard Biener <rguenther@suse.de>
6463 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6464 -g0 at compile-time.
6466 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6469 * doc/sourcebuild.texi (Other hardware attributes):
6472 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6475 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6476 Fix single-use check.
6478 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6480 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6482 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6484 * doc/install.texi: Modify to add more details on running selected
6487 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6489 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6490 is combined with -mabi=ms.
6491 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6494 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6496 PR tree-optimization/81488
6497 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6498 and cached_basis fields.
6499 (MAX_SPREAD): New constant.
6500 (alloc_cand_and_find_basis): Initialize new fields.
6501 (clear_visited): New function.
6502 (create_phi_basis_1): Rename from create_phi_basis, set visited
6503 and cached_basis fields.
6504 (create_phi_basis): New wrapper function.
6505 (phi_add_costs_1): Rename from phi_add_costs, add spread
6506 parameter, set visited field, short-circuit when limits reached.
6507 (phi_add_costs): New wrapper function.
6508 (record_phi_increments_1): Rename from record_phi_increments, set
6510 (record_phi_increments): New wrapper function.
6511 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6512 (phi_incr_cost): New wrapper function.
6513 (all_phi_incrs_profitable_1): Rename from
6514 all_phi_incrs_profitable, set visited field.
6515 (all_phi_incrs_profitable): New wrapper function.
6517 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6522 that takes the outer and inner modes.
6523 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6524 comparison as the canonical test for a paradoxical subreg.
6525 * combine.c (simplify_set): Use paradoxical_subreg_p.
6526 (make_extraction): Likewise.
6527 (force_to_mode): Likewise.
6528 (rtx_equal_for_field_assignment_p): Likewise.
6529 (gen_lowpart_for_combine): Likewise.
6530 (simplify_comparison): Likewise.
6531 * cse.c (equiv_constant): Likewise.
6532 * expmed.c (store_bit_field_1): Likewise.
6533 * final.c (alter_subreg): Likewise.
6534 * fwprop.c (propagate_rtx): Likewise.
6535 (forward_propagate_subreg): Likewise.
6536 * ira-conflicts.c (ira_build_conflicts): Likewise.
6537 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6538 * lra-constraints.c (curr_insn_transform): Likewise.
6539 (split_reg): Likewise.
6540 * lra-eliminations.c (move_plus_up): Likewise.
6541 (lra_eliminate_regs_1): Likewise.
6542 * recog.c (general_operand): Likewise.
6543 * ree.c (combine_reaching_defs): Likewise.
6544 * reload.c (push_reload): Likewise.
6545 (find_reloads): Likewise.
6546 * reload1.c (elimination_effects): Likewise.
6547 (compute_reload_subreg_offset): Likewise.
6548 (choose_reload_regs): Likewise.
6549 * rtlanal.c (subreg_lsb_1): Likewise.
6550 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6551 (simplify_subreg): Likewise.
6552 * var-tracking.c (track_loc_p): Likewise.
6553 * emit-rtl.c (byte_lowpart_offset): Likewise.
6554 (paradoxical_subreg_p): Delete out-of-line definition.
6556 2017-08-22 Jeff Law <law@redhat.com>
6558 PR tree-optimization/81741
6559 PR tree-optimization/71947
6560 * tree-ssa-dom.c: Include tree-inline.h.
6561 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
6562 equivalences if one is more expensive to compute than the other.
6563 * tree-ssa-scopedtables.h (class const_or_copies): Make
6564 record_const_or_copy_raw method private.
6565 (class avail_exprs_stack): New method simplify_binary_operation.
6566 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
6567 avail_exprs_stack::simplify_binary_operation as needed.
6568 (avail_exprs_stack::simplify_binary_operation): New function.
6570 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
6572 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
6573 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
6574 (DOT_SYMBOLS): Likewise.
6575 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
6576 (RELOCATABLE_NEEDS_FIXUP): Likewise.
6577 (RS6000_ABI_NAME): Likewise.
6578 (TARGET_CMODEL): Likewise.
6579 (TOC_SECTION_ASM_OP): Likewise.
6580 (SET_CMODEL): New macro.
6581 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6583 2017-08-22 Richard Biener <rguenther@suse.de>
6585 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6586 to allow for free-lang-data replacements similar to verify_type_variant.
6588 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6589 Alan Hayward <alan.hayward@arm.com>
6590 David Sherwood <david.sherwood@arm.com>
6592 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6593 VOIDmode for the LABEL_REF.
6595 2017-08-22 Richard Biener <rguenther@suse.de>
6597 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6599 2017-08-22 Richard Biener <rguenther@suse.de>
6601 * common.opt (feliminate-dwarf2-dups): Ignore.
6602 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6603 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6604 same_die_p_wrap, compute_section_prefix,
6605 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6606 (comdat_symbol_id, comdat_symbol_number): Likewise.
6607 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6609 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6610 (output_die): Mark unreachable path unreachable.
6611 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6612 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6613 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6614 (dwarf2out_early_finish): Likewise.
6616 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6618 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6620 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6623 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6624 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6625 (avr_attribute_table) <io, io_low, address>: Initialize
6626 .decl_required with true.
6628 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6630 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6631 undocumented debugging options.
6632 (-mvsx-scalar-double): Likewise.
6633 (-mallow-df-permute): Likewise.
6634 (-mvectorize-builtins): Likewise.
6635 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6636 (rs6000_builtin_vectorized_function): Likewise.
6637 (rs6000_builtin_md_vectorized_function): Likewise.
6638 (rs6000_opt_vars): Likewise.
6640 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6643 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6644 (*btrq_imm): Rename from *btrq.
6645 (*btcq_imm): Rename from *btcq.
6646 (btsc): New code attribute.
6647 (*<btsc><mode>): New insn pattern.
6648 (*btr<mode>): Ditto.
6649 (*<btsc><mode>_mask): New insn_and_split pattern.
6650 (*btr<mode>_mask): Ditto.
6652 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6653 Alan Hayward <alan.hayward@arm.com>
6654 David Sherwood <david.sherwood@arm.com>
6656 * function.c (pad_below): Simplify padding calculation.
6658 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6659 Alan Hayward <alan.hayward@arm.com>
6660 David Sherwood <david.sherwood@arm.com>
6662 * target.def (function_prologue): Remove frame size argument.
6663 (function_epilogue): Likewise.
6664 * doc/tm.texi: Regenerate.
6665 * final.c (final_start_function): Update call to function_prologue.
6666 (final_end_function): Update call to function_epilogue.
6667 (default_function_pro_epilogue): Remove frame size argument.
6668 * output.h (default_function_pro_epilogue): Likewise.
6669 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6670 (arm_output_function_prologue): Likewise.
6671 * config/frv/frv.c (frv_function_prologue): Likewise.
6672 (frv_function_epilogue): Likewise.
6673 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6674 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6675 (ia64_output_function_epilogue): Likewise.
6676 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6677 (m32r_output_function_epilogue): Likewise.
6678 * config/microblaze/microblaze.c (microblaze_function_prologue)
6679 (microblaze_function_epilogue): Likewise.
6680 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6681 (mips_output_function_epilogue): Likewise.
6682 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6683 (mmix_target_asm_function_epilogue): Likewise.
6684 * config/msp430/msp430.c (msp430_start_function): Likewise.
6685 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6686 (nds32_asm_function_epilogue): Likewise.
6687 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6688 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6689 (pa_output_function_epilogue): Likewise.
6690 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6691 (rs6000_output_function_epilogue): Likewise.
6692 * config/rl78/rl78.c (rl78_start_function): Likewise.
6693 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6694 (rs6000_output_function_epilogue): Likewise.
6695 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6696 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6697 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6698 (sparc_asm_function_epilogue): Likewise.
6700 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6702 * tree.h (type_has_mode_precision_p): New function.
6703 * convert.c (convert_to_integer_1): Use it.
6704 * expr.c (expand_expr_real_2): Likewise.
6705 (expand_expr_real_1): Likewise.
6706 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6707 * match.pd: Likewise.
6708 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6709 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6710 * tree-tailcall.c (process_assignment): Likewise.
6711 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6712 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6713 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6714 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6715 (vectorizable_assignment): Likewise.
6716 (vectorizable_shift): Likewise.
6717 (vectorizable_operation): Likewise.
6718 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6720 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6722 * match.pd: Add pow (C, x) simplification.
6724 2017-08-21 Richard Biener <rguenther@suse.de>
6726 PR tree-optimization/81900
6727 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6728 for blocks with abnormal predecessors.
6729 (compute_antic): Do not set visited flag prematurely.
6731 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6734 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6736 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6738 * stor-layout.h (vector_type_mode): Move to...
6739 * tree.h (vector_type_mode): ...here.
6740 * stor-layout.c (vector_type_mode): Move to...
6741 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6743 2017-08-21 Richard Biener <rguenther@suse.de>
6745 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6746 register_external_die hooks.
6747 (debug_false_tree_charstarstar_uhwistar): Declare.
6748 (debug_nothing_tree_charstar_uhwi): Likewise.
6749 * debug.c (do_nothing_debug_hooks): Adjust.
6750 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6751 (debug_nothing_tree_charstar_uhwi): Likewise.
6752 * dbxout.c (dbx_debug_hooks): Adjust.
6753 (xcoff_debug_hooks): Likewise.
6754 * sdbout.c (sdb_debug_hooks): Likewise.
6755 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6756 * dwarf2out.c (macinfo_label_base): New global.
6757 (dwarf2out_register_external_die): New function for the
6758 register_external_die hook.
6759 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6760 (dwarf2_debug_hooks): Use them.
6761 (dwarf2_lineno_debug_hooks): Adjust.
6762 (struct die_struct): Add with_offset flag.
6763 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6764 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6765 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6766 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6767 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6768 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6769 defining section names for the early LTO debug variants.
6770 (reset_indirect_string): New helper.
6771 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6772 (print_dw_val): Add support for offsetted symbol references.
6773 (get_ultimate_context): Split out from is_cxx.
6774 (is_cxx): Use get_ultimate_context.
6775 (is_fortran): Add decl overload.
6776 (compute_comp_unit_symbol): Split out worker from
6777 compute_section_prefix.
6778 (compute_section_prefix): Call compute_comp_unit_symbol and
6779 set comdat_type_p here.
6780 (output_die): Skip DIE symbol output for the LTO added one.
6781 Handle DIE symbol references with offset.
6782 (output_comp_unit): Guard section name mangling properly.
6783 For LTO debug sections emit a symbol at the section beginning
6784 which we use to refer to its DIEs.
6785 (add_abstract_origin_attribute): For DIEs registered via
6786 dwarf2out_register_external_die directly refer to the early
6787 DIE rather than indirectly through the shadow one we created.
6788 Remove obsolete call to dwarf2out_abstract_function for
6789 non-function/block origins.
6790 (gen_array_type_die): When generating early LTO debug do
6791 not emit DW_AT_string_length.
6792 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6793 late when in LTO. As suggested place a gcc_unreachable for
6794 the DECL_ABSTRACT_P case.
6795 (gen_subprogram_die): Avoid another specification DIE
6796 for early built declarations/definitions for the late LTO case.
6797 (gen_variable_die): Add type references for late duplicated VLA dies
6799 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6800 we have the abstract instance already.
6801 (process_scope_var): Adjust decl DIE contexts in LTO which
6802 first puts them in limbo.
6803 (gen_decl_die): Do not generate type DIEs late apart from
6804 types for VLAs or for decls we do not yet have a DIE. Do not
6805 call dwarf2out_abstract_function late.
6806 (dwarf2out_early_global_decl): Make sure to create DIEs
6807 for abstract instances of a decl first.
6808 (dwarf2out_late_global_decl): Adjust comment.
6809 (output_macinfo_op): With multiple macro sections use
6810 macinfo_label_base to distinguish labels.
6811 (output_macinfo): Likewise. Update macinfo_label_base.
6812 Pass in the line info label.
6813 (note_variable_value_in_expr): When generating LTO resolve
6814 all variable values here by generating DIEs as needed.
6815 (init_sections_and_labels): Add early LTO debug flag parameter
6816 and generate different sections and names if set. Add generation
6817 counter for the labels so we can have multiple of them.
6818 (reset_dies): Helper to allow DIEs to be output multiple times.
6819 (dwarf2out_finish): When outputting DIEs to the fat part of an
6820 LTO object first reset DIEs.
6821 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6822 (modified_type_die): Check for decl_ultimate_origin being self
6824 (gen_type_die_with_usage): Likewise.
6825 (gen_typedef_die): Allow decl_ultimate_origin being self.
6826 (set_decl_abstract_flags): Remove.
6827 (set_block_abstract_flags): Likewise.
6828 (dwarf2out_abstract_function): Treat the early generated DIEs
6829 as the abstract copy and only add DW_AT_inline and
6830 DW_AT_artificial here and call set_decl_origin_self.
6831 If the DIE has an abstract origin don't do anything.
6832 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6833 if we have none yet (Go fails to build one, PR78628).
6834 (variably_modified_type_p): Prevent endless recursion for Ada
6835 cyclic pointer types.
6836 * lto-streamer-in.c: Include debug.h.
6837 (dref_queue): New global.
6838 (lto_read_tree_1): Stream in DIE references.
6839 (lto_input_tree): Register DIE references.
6840 (input_function): Stream DECL_DEBUG_ARGS.
6841 * lto-streamer-out.c: Include debug.h.
6842 (lto_write_tree_1): Output DIE references.
6843 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6844 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6845 (output_function): Stream DECL_DEBUG_ARGS.
6846 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6847 Stream DECL_ABSTRACT_ORIGIN.
6848 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6849 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6850 DECL_CONTEXT for file-scope decls.
6851 * lto-streamer.h (struct dref_entry): Declare.
6852 (dref_queue): Likewise.
6853 * cfgexpand.c (pass_expand::execute): Do not call the
6854 outlining_inline_function hook here.
6855 * lto-wrapper.c (debug_obj): New global.
6856 (tool_cleanup): Unlink it if required.
6857 (debug_objcopy): New function.
6858 (run_gcc): Handle early debug sections in the IL files by
6859 extracting them to separate files, partially linkin them and
6860 feeding the result back as result to the linker.
6861 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6862 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6863 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6864 sections into a separate segment.
6865 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6867 (darwin_asm_dwarf_section): Likewise.
6868 (darwin_asm_output_dwarf_offset): Likewise.
6869 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6871 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6872 Alan Hayward <alan.hayward@arm.com>
6873 David Sherwood <david.sherwood@arm.com>
6875 * read-md.h (md_reader::record_potential_iterator_use): Replace
6876 pointer argument with an rtx and an index.
6877 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6878 (apply_mode_iterator): Likewise.
6879 (apply_code_iterator): Likewise.
6880 (apply_int_iterator): Likewise.
6881 (apply_subst_iterator): Likewise.
6882 (record_iterator_use): Likewise.
6883 (record_attribute_use): Likewise.
6884 (md_reader::record_potential_iterator_use): Likewise. Update calls
6885 to record_iterator_use and apply_iterator.
6886 (iterator_use): Replace ptr with x and index.
6887 (attribute_use): Likewise.
6888 (apply_attribute_uses): Update calls to apply_iterator.
6889 (apply_iterators): Likewise. Update initialization of iterator_use.
6890 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6891 and record_potential_iterator_use.
6892 (rtx_reader::read_rtx_operand): Likewise.
6894 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6895 Alan Hayward <alan.hayward@arm.com>
6896 David Sherwood <david.sherwood@arm.com>
6898 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6901 2017-08-21 Richard Biener <rguenther@suse.de>
6904 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6905 at struct end conservatively when comparing common bases.
6907 2017-08-21 Richard Biener <rguenther@suse.de>
6909 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6910 (mem_ref_in_stmt): Remove.
6911 (determine_max_movement): Use ref index to get at the reference.
6912 (invariantness_dom_walker::before_dom_children): Deal with
6913 lim data already initialized.
6914 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6916 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6918 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6919 (TARGET_ISA_ROUND): Ditto.
6920 (TARGET_ROUND): Ditto.
6921 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6922 * config/i386/i386.md: Ditto.
6923 * config/i386/sse.md: Ditto.
6924 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6925 with OPTION_MASK_ISA_SSE4_1.
6927 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6930 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6931 __builtin_ia32_lzcnt_u16.
6933 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6936 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6937 (rs6000_set_current_function): Rewrite function to use it.
6939 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6942 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6943 and TYPE_WARN_IF_NOT_ALIGN.
6944 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6945 (handle_warn_if_not_align): New.
6946 (place_union_field): Call handle_warn_if_not_align.
6947 (place_field): Call handle_warn_if_not_align.
6948 Copy TYPE_WARN_IF_NOT_ALIGN.
6949 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6950 (layout_type): Likewise.
6951 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6953 (tree_decl_common): Add warn_if_not_align.
6954 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6955 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6956 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6957 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6958 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6959 * doc/extend.texi: Document warn_if_not_aligned attribute.
6960 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6962 2017-08-17 Martin Liska <mliska@suse.cz>
6965 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6966 (get_data_dependence): Use it as pointer type.
6967 (distribute_loop): Likewise.
6969 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6971 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6972 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6973 (p8_vmrgow_<mode>_direct): New define_insn.
6974 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6975 handle endianness for vmrgew and vmrgow permute patterns.
6977 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6979 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6980 * config/rs6000/rs6000-cpus.def: Remove comment.
6981 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6982 (POWERPC_MASKS): Likewise.
6983 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6984 use of TARGET_VSX_TIMODE.
6985 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6986 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6987 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6988 (rs6000_option_override_internal): Remove dead code.
6989 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6990 (rs6000_legitimize_reload_address): Likewise.
6991 (rs6000_legitimate_address_p): Likewise.
6992 (rs6000_opt_masks): Delete "vsx-timode".
6993 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6994 from function comment.
6995 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6996 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6997 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6999 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7000 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7001 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7004 (define_peephole2): Likewise.
7006 2017-08-17 Martin Sebor <msebor@redhat.com>
7009 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7010 past the end of an array.
7011 (test_pp_format): Add test cases.
7013 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7015 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7016 missing ECF_NOTHROW flags.
7018 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7021 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7022 operands residing in integer registers.
7023 (*vsx_le_perm_load_<mode>): Likewise.
7024 (*vsx_le_perm_store_<mode>): Likewise.
7025 (define_peephole2): Add peepholes to optimize the above.
7027 2017-08-17 Marek Polacek <polacek@redhat.com>
7030 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7031 to mimic what shorten_compare did. Change the return type to bool.
7032 (fold_cond_expr_with_comparison): Update call to
7033 operand_equal_for_comparison_p.
7034 (fold_ternary_loc): Likewise.
7036 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7038 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7040 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7042 2017-08-17 Richard Biener <rguenther@suse.de>
7044 * tree-ssa-structalias.c (solve_graph): When propagating
7045 to successors update the graphs succ edges and avoid duplicate work.
7047 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7050 * config/i386/i386.c (ix86_option_override_internal): Save target
7051 specific options after ix86_stack_protector_guard_reg was changed.
7053 2017-08-17 Richard Biener <rguenther@suse.de>
7055 PR tree-optimization/81827
7056 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7057 (new_var_info): Initialize it conservatively.
7058 (get_call_vi): Mark register vars.
7059 (new_scalar_tmp_constraint_exp): Likewise.
7060 (handle_rhs_call): Likewise.
7061 (handle_const_call): Likewise.
7062 (create_function_info_for): Likewise.
7063 (solve_constraints): Sort varinfos to separate register from
7064 non-register vars to pack points-to solution bitmaps during
7067 2017-08-17 Marek Polacek <polacek@redhat.com>
7069 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7071 2017-08-17 Richard Biener <rguenther@suse.de>
7073 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7074 to true when overflow is undefined and we saturated the result.
7076 2017-08-17 Alan Modra <amodra@gmail.com>
7079 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7080 Don't use store multiple if only one reg needs saving.
7081 (interesting_frame_related_regno): New function.
7082 (rs6000_frame_related): Don't emit frame info for regs that
7084 (rs6000_emit_epilogue): Likewise.
7086 2017-08-16 Nathan Sidwell <nathan@acm.org>
7088 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7089 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7090 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7091 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7092 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7093 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7094 (verify_type): Adjust for TYPE_BINFO move.
7095 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7096 process TYPE_BINFO directly.
7097 (hash_tree): Likewise.
7098 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7100 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7103 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7105 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7107 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7110 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7111 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7112 Add ix86_binary_operator_ok to insn constraint.
7113 (*iordi_1_bts): Ditto.
7114 (*xordi_1_btc): Ditto.
7115 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7116 Update corresponding peephole2 pattern.
7120 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7122 PR tree-optimization/81832
7123 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7124 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7126 2017-08-16 Marek Polacek <polacek@redhat.com>
7129 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7130 regarding size_zero_node.
7132 2017-08-16 Martin Liska <mliska@suse.cz>
7135 * config.gcc: Respect previously set extra_objs in case
7138 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7140 PR tree-optimization/81835
7141 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7142 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7143 not depend on the phi.
7145 2017-08-16 Alan Modra <amodra@gmail.com>
7147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7150 2017-08-16 Alan Modra <amodra@gmail.com>
7152 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7153 (save_reg_p): ..into this. Update all callers.
7154 (first_reg_to_save): Simplify.
7156 2017-08-16 Alan Modra <amodra@gmail.com>
7158 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7161 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7165 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7166 constant count if that count is less than 32.
7168 2017-08-15 Nathan Sidwell <nathan@acm.org>
7170 * gcc.c (execute): Emit friendlier message if inferior is killed
7171 by an external cause.
7173 2017-08-15 Richard Biener <rguenther@suse.de>
7175 PR tree-optimization/81790
7176 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7177 CONSTRUCTORs from simplifying and VN.
7179 2017-08-14 Martin Sebor <msebor@redhat.com>
7181 * builtin-attrs.def: Add comments.
7183 2017-08-14 Martin Sebor <msebor@redhat.com>
7186 * doc/extend.texi (attribute nonstring): Document new attribute.
7188 2017-08-14 Martin Sebor <msebor@redhat.com>
7191 * tree-diagnostic.c (default_tree_printer): Handle %G.
7192 * gimple-pretty-print.h (percent_G_format): Declare new function.
7193 * gimple-pretty-print.c (percent_G_format): Define.
7194 * tree-pretty-print.c (percent_K_format): Add argument.
7196 2017-08-14 Martin Sebor <msebor@redhat.com>
7198 PR translation/79998
7199 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7200 Remove a stray space.
7202 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7205 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7206 (*iordi_1_bts): Ditto.
7207 (*xordi_1_btc): Ditto.
7209 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7212 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7216 quoted strings, and make more translator-friendly.
7217 (darwin_rs6000_override_options): Likewise.
7218 (rs6000_option_override_internal): Likewise.
7219 (rs6000_return_in_memory): Fix overlong line.
7220 (init_cmulative_args): Use quoted strings, and make more
7221 translator-friendly.
7222 (rs6000_pass_by_reference): Fix overlong line.
7223 (def_builtin): Use quoted strings.
7224 (altivec_expand_predicate_builtin): Use quoted strings, and make
7225 more translator-friendly.
7226 (htm_expand_builtin): Use quoted strings.
7227 (cpu_expand_builtin): Use quoted strings, and make more
7228 translator-friendly.
7229 (altivec_expand_builtin): Likewise.
7230 (paired_expand_predicate_builtin): Likewise.
7231 (rs6000_invalid_builtin): Likewise.
7232 (builtin_function_type): Use quoted strings.
7233 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7234 more translator-friendly.
7235 (rs6000_trampoline_init): Likewise.
7236 (rs6000_handle_altivec_attribute): Likewise.
7237 (rs6000_inner_target_options): Use quoted strings.
7238 (rs6000_disable_incompatible_switches): Likewise.
7239 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7240 strings, and make more translator-friendly.
7241 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7243 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7245 PR tree-optimization/81799
7246 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7247 cond_expr to simple gimple operand.
7249 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7252 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7254 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7258 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7259 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7260 TARGET_GASISR_PROLOGUES.
7261 * config/avr/avr.c (avr_option_override): Same.
7262 (avr_pass_pre_proep::execute): Same.
7264 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7267 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7268 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7270 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7272 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7273 define_insn_and_split. Split to a memory load from 0 in
7274 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7275 using PTR mode iterator.
7276 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7277 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7278 (*add_tp_<mode>): Redefine as define_insn_and_split.
7279 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7280 address space. Merge with *add_tp_x32 using PTR mode iterator.
7281 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7282 Split to an add with a memory load from 0 in
7283 DEFAULT_TLS_SEG_REG address space.
7285 2017-08-12 Andrew Pinski <apinski@cavium.com>
7287 * config/aarch64/aarch64-option-extensions.def (rdma):
7288 Fix feature string to what Linux prints out in /proc/cpuinfo.
7290 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7293 * dwarf2out.c (modified_type_die): For C typedef types that have
7294 an ultimate origin, process the ultimate origin instead of the
7296 (gen_typedef_die): Assert that input DECLs have no ultimate
7298 (gen_type_die_with_usage): For typedef variants that have an
7299 ultimate origin, just call gen_decl_die on the original DECL.
7300 (process_scope_var): Avoid creating DIEs for local typedefs and
7301 concrete static variables.
7303 2017-08-12 Alan Modra <amodra@gmail.com>
7307 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7308 match gnu-user.h startfile.
7309 (ENDFILE_LINUX_SPEC): Similarly.
7311 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7314 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7316 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7318 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7319 * config/aarch64/aarch64.md (mov<mode>): Change.
7320 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7321 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7322 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7324 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7326 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7327 for storage order barriers.
7329 2017-08-11 Martin Liska <mliska@suse.cz>
7332 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7333 variables of void type.
7335 2017-08-11 Martin Liska <mliska@suse.cz>
7337 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7338 TARGET_SUPPORTS_ALIASES.
7339 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7340 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7341 (optimize_weakref): Likewise.
7342 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7343 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7344 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7346 2017-08-11 Martin Liska <mliska@suse.cz>
7349 * config/i386/i386.c (make_resolver_func): Do complete
7350 refactoring of the function.
7352 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7355 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7356 * config/i386/i386.c (ix86_stack_protect_guard): Use
7357 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7358 * doc/invoke.texi (x86 Options): Document
7359 -mstack-protector-guard-symbol= option.
7361 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7363 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7364 * config/i386/i386.c (ix86_split_stack_guard): New function.
7365 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7366 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7367 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7368 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7369 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7370 (split_stack_space_check): Call ix86_split_stack_guard.
7372 2017-08-10 Martin Sebor <msebor@redhat.com>
7374 * print-tree.c (print_node): Print location using the established
7376 Replace spaces with colons.
7377 (debug_raw, debug): Ditto.
7379 2017-08-10 Martin Sebor <msebor@redhat.com>
7382 * pretty-print.c (pp_format): Correct the handling of %s precision.
7384 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7387 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7389 (ix86_finalize_stack_frame_flags): This. Also clear
7390 frame_pointer_needed if -fno-omit-frame-pointer is used without
7392 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7393 with ix86_finalize_stack_frame_flags.
7394 (ix86_expand_epilogue): Likewise.
7395 (ix86_expand_split_stack_prologue): Likewise.
7396 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7398 2017-08-10 Martin Liska <mliska@suse.cz>
7401 * c-attribs.c (handle_target_attribute):
7402 Report warning for an empty string argument of target attribute.
7404 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7407 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7409 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7410 or DECL_NONLOCAL labels.
7411 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7412 or DECL_NONLOCAL labels here.
7414 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7417 to indicate when early gimple folding has been disabled.
7418 (rs6000_gimple_fold_builtin): Add debug content.
7419 (rs6000_invalid_builtin): Fix whitespace.
7420 (rs6000_expand_builtin): Fix whitespace.
7421 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7423 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7426 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7427 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7429 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7431 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7432 (qdf24xx): Likewise.
7433 * config/aarch64/aarch64.md: Include falkor.md.
7434 * config/aarch64/falkor.md: New.
7436 2017-08-09 Marek Polacek <polacek@redhat.com>
7439 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7440 * diagnostic.c (emit_diagnostic): Add a comment.
7441 (emit_diagnostic_valist): New function.
7443 2017-08-09 Marek Polacek <polacek@redhat.com>
7446 * input.c (make_location): New overload.
7447 * input.h (make_location): Declare.
7449 2017-08-08 Alan Modra <amodra@gmail.com>
7450 H.J. Lu <hongjiu.lu@intel.com>
7453 * gcc.c (NO_PIE_SPEC): Delete.
7454 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7456 (LINK_PIE_SPEC): ..to here.
7457 (LINK_COMMAND_SPEC): Support -no-pie.
7458 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7459 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7461 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7462 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7463 (ENDFILE_CRTEND_SPEC): Similarly.
7465 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7468 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7469 (mstack-protector-guard-offset=): Ditto.
7470 * config/i386/i386.c (ix86_option_override): Handle
7471 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7473 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7474 ix86_stack_protect_guard_offset variables.
7475 (TARGET_STACK_PROTECT_GUARD): Always define.
7476 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7477 and -mstack-protector-guard-offset= options.
7479 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7481 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7482 boundary case for the last candidate.
7484 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7486 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7487 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7489 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7492 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7493 * config/aarch64/aarch64-builtins.c
7494 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7495 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7496 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7498 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7499 Andrew Pinski <pinskia@gmail.com>
7502 * internal-fn.def (XORSIGN): New.
7503 * optabs.def (xorsign_optab): New.
7504 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7505 (convert_expand_mult_copysign): New.
7506 (pass_optimize_widening_mul::execute): Call
7507 convert_expand_mult_copysign.
7509 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7511 PR tree-optimization/81354
7512 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7513 Insert on edges rather than explicitly creating landing pads.
7514 (analyze_candidates_and_replace): Commit edge inserts.
7516 2017-08-08 Richard Biener <rguenther@suse.de>
7519 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7520 (expand_simple_operations): Also look through ADDR_EXPRs with
7521 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7523 2017-08-08 Richard Biener <rguenther@suse.de>
7525 PR tree-optimization/81723
7526 * tree-vect-slp.c (struct bst_traits): New hash traits.
7527 (bst_fail): New global.
7528 (vect_build_slp_tree_2): New worker, split out from ...
7529 (vect_build_slp_tree): ... this now wrapping it with using
7530 bst_fail set to cache SLP tree build fails. Properly handle
7532 (vect_analyze_slp_instance): Allocate and free bst_fail.
7534 2017-08-08 Martin Liska <mliska@suse.cz>
7537 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7538 LABEL_DECLs that can be from a different function.
7540 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7542 PR tree-optimization/81744
7543 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7544 loop's number of iterations.
7546 2017-08-08 Martin Liska <mliska@suse.cz>
7548 * asan.c: Include header files.
7549 * attribs.c (build_decl_attribute_variant): New function moved
7551 (build_type_attribute_qual_variant): Likewise.
7552 (cmp_attrib_identifiers): Likewise.
7553 (simple_cst_list_equal): Likewise.
7554 (omp_declare_simd_clauses_equal): Likewise.
7555 (attribute_value_equal): Likewise.
7556 (comp_type_attributes): Likewise.
7557 (build_type_attribute_variant): Likewise.
7558 (lookup_ident_attribute): Likewise.
7559 (remove_attribute): Likewise.
7560 (merge_attributes): Likewise.
7561 (merge_type_attributes): Likewise.
7562 (merge_decl_attributes): Likewise.
7563 (merge_dllimport_decl_attributes): Likewise.
7564 (handle_dll_attribute): Likewise.
7565 (attribute_list_equal): Likewise.
7566 (attribute_list_contained): Likewise.
7567 * attribs.h (lookup_attribute): New function moved from tree.[ch].
7568 (lookup_attribute_by_prefix): Likewise.
7569 * bb-reorder.c: Include header files.
7570 * builtins.c: Likewise.
7571 * calls.c: Likewise.
7572 * cfgexpand.c: Likewise.
7573 * cgraph.c: Likewise.
7574 * cgraphunit.c: Likewise.
7575 * convert.c: Likewise.
7576 * dwarf2out.c: Likewise.
7577 * final.c: Likewise.
7578 * fold-const.c: Likewise.
7579 * function.c: Likewise.
7580 * gimple-expr.c: Likewise.
7581 * gimple-fold.c: Likewise.
7582 * gimple-pretty-print.c: Likewise.
7583 * gimple.c: Likewise.
7584 * gimplify.c: Likewise.
7585 * hsa-common.c: Likewise.
7586 * hsa-gen.c: Likewise.
7587 * internal-fn.c: Likewise.
7588 * ipa-chkp.c: Likewise.
7589 * ipa-cp.c: Likewise.
7590 * ipa-devirt.c: Likewise.
7591 * ipa-fnsummary.c: Likewise.
7592 * ipa-inline.c: Likewise.
7593 * ipa-visibility.c: Likewise.
7595 * lto-cgraph.c: Likewise.
7596 * omp-expand.c: Likewise.
7597 * omp-general.c: Likewise.
7598 * omp-low.c: Likewise.
7599 * omp-offload.c: Likewise.
7600 * omp-simd-clone.c: Likewise.
7601 * opts-global.c: Likewise.
7602 * passes.c: Likewise.
7603 * predict.c: Likewise.
7604 * sancov.c: Likewise.
7605 * sanopt.c: Likewise.
7606 * symtab.c: Likewise.
7607 * toplev.c: Likewise.
7608 * trans-mem.c: Likewise.
7609 * tree-chkp.c: Likewise.
7610 * tree-eh.c: Likewise.
7611 * tree-into-ssa.c: Likewise.
7612 * tree-object-size.c: Likewise.
7613 * tree-parloops.c: Likewise.
7614 * tree-profile.c: Likewise.
7615 * tree-ssa-ccp.c: Likewise.
7616 * tree-ssa-live.c: Likewise.
7617 * tree-ssa-loop.c: Likewise.
7618 * tree-ssa-sccvn.c: Likewise.
7619 * tree-ssa-structalias.c: Likewise.
7620 * tree-ssa.c: Likewise.
7621 * tree-streamer-in.c: Likewise.
7622 * tree-vectorizer.c: Likewise.
7623 * tree-vrp.c: Likewise.
7625 * ubsan.c: Likewise.
7626 * varasm.c: Likewise.
7627 * varpool.c: Likewise.
7628 * tree.c: Remove functions moved to attribs.[ch].
7630 * config/aarch64/aarch64.c: Add attrs.h header file.
7631 * config/alpha/alpha.c: Likewise.
7632 * config/arc/arc.c: Likewise.
7633 * config/arm/arm.c: Likewise.
7634 * config/avr/avr.c: Likewise.
7635 * config/bfin/bfin.c: Likewise.
7636 * config/c6x/c6x.c: Likewise.
7637 * config/cr16/cr16.c: Likewise.
7638 * config/cris/cris.c: Likewise.
7639 * config/darwin.c: Likewise.
7640 * config/epiphany/epiphany.c: Likewise.
7641 * config/fr30/fr30.c: Likewise.
7642 * config/frv/frv.c: Likewise.
7643 * config/ft32/ft32.c: Likewise.
7644 * config/h8300/h8300.c: Likewise.
7645 * config/i386/winnt.c: Likewise.
7646 * config/ia64/ia64.c: Likewise.
7647 * config/iq2000/iq2000.c: Likewise.
7648 * config/lm32/lm32.c: Likewise.
7649 * config/m32c/m32c.c: Likewise.
7650 * config/m32r/m32r.c: Likewise.
7651 * config/m68k/m68k.c: Likewise.
7652 * config/mcore/mcore.c: Likewise.
7653 * config/microblaze/microblaze.c: Likewise.
7654 * config/mips/mips.c: Likewise.
7655 * config/mmix/mmix.c: Likewise.
7656 * config/mn10300/mn10300.c: Likewise.
7657 * config/moxie/moxie.c: Likewise.
7658 * config/msp430/msp430.c: Likewise.
7659 * config/nds32/nds32-isr.c: Likewise.
7660 * config/nds32/nds32.c: Likewise.
7661 * config/nios2/nios2.c: Likewise.
7662 * config/nvptx/nvptx.c: Likewise.
7663 * config/pa/pa.c: Likewise.
7664 * config/pdp11/pdp11.c: Likewise.
7665 * config/powerpcspe/powerpcspe.c: Likewise.
7666 * config/riscv/riscv.c: Likewise.
7667 * config/rl78/rl78.c: Likewise.
7668 * config/rx/rx.c: Likewise.
7669 * config/s390/s390.c: Likewise.
7670 * config/sh/sh.c: Likewise.
7671 * config/sol2.c: Likewise.
7672 * config/sparc/sparc.c: Likewise.
7673 * config/spu/spu.c: Likewise.
7674 * config/stormy16/stormy16.c: Likewise.
7675 * config/tilegx/tilegx.c: Likewise.
7676 * config/tilepro/tilepro.c: Likewise.
7677 * config/v850/v850.c: Likewise.
7678 * config/vax/vax.c: Likewise.
7679 * config/visium/visium.c: Likewise.
7680 * config/xtensa/xtensa.c: Likewise.
7682 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7685 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7686 constraints since the -mupper-regs-* switches have been
7688 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7689 into a vector from a double word element that was extracted from
7690 another vector, and eliminate extra XXPERMDI instructions.
7691 (vsx_concat_<mode>_2): Likewise.
7692 (vsx_concat_<mode>_3): Likewise.
7693 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7694 concat to allow optimizing inserts from previous extracts.
7696 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7698 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7699 memory reference to a SSP offset in TLS address space.
7700 (ix86_print_operand) <case '@'>: Remove.
7701 (ix86_print_operand_punct_valid_p): Remove '@' code.
7702 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7704 (stack_tls_protect_set_<mode>): Remove.
7705 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7706 (stack_tls_protect_test_<mode>): Remove.
7707 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7709 2017-08-07 Olivier Hainque <hainque@adacore.com>
7712 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7714 2017-08-07 Douglas Rupp <rupp@adacore.com>
7716 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7717 variable was referenced as multidir in command.
7719 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7722 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7725 2017-08-07 Martin Liska <mliska@suse.cz>
7727 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7728 * config/rl78/rl78.c: Add include of attribs.h.
7729 * config/sh/sh.c: Likewise.
7730 * config/v850/v850.c: Likewise.
7732 2017-08-07 Tom de Vries <tom@codesourcery.com>
7735 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7737 2017-08-07 Martin Liska <mliska@suse.cz>
7739 * config/mips/mips.c: Include attribs.h.
7741 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7744 * doc/invoke.texi: Document change in behvaior for -Ofast for
7747 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7749 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7751 (aarch64_pop_regs): Likewise.
7752 (aarch64_gen_load_pair): Likewise.
7753 (aarch64_save_callee_saves): Likewise.
7754 (aarch64_restore_callee_saves): Likewise.
7756 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7758 * config/i386/i386.c: Revert the last change.
7760 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7763 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7765 (ix86_finalize_stack_frame_flags): This. Also clear
7766 frame_pointer_needed if -fno-omit-frame-pointer is used without
7768 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7769 with ix86_finalize_stack_frame_flags.
7770 (ix86_expand_epilogue): Likewise.
7771 (ix86_expand_split_stack_prologue): Likewise.
7773 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7776 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7777 to P_AES for Westmere.
7779 2017-08-07 Jonathan Yong <10walls@gmail.com>
7781 * config/i386/mingw.opt (fset-stack-executable): Removed.
7782 * config/i386/cygming.opt (fset-stack-executable): Moved
7784 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7786 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7788 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7790 2017-08-07 Marek Polacek <polacek@redhat.com>
7793 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7795 2017-08-07 Martin Liska <mliska@suse.cz>
7797 * attribs.h (canonicalize_attr_name): New function.
7798 (cmp_attribs): Move from c-format.c and adjusted.
7799 (is_attribute_p): Moved from tree.h.
7800 * tree-inline.c: Add new includes.
7801 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7802 (private_is_attribute_p): Remove.
7803 (private_lookup_attribute): Likewise.
7804 (private_lookup_attribute_by_prefix): Simplify.
7805 (remove_attribute): Use is_attribute_p.
7806 * tree.h: Remove removed declarations.
7808 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7811 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7812 instead of computing it in the function. Formatting fix.
7813 (expand_case): Don't rely on default_edge being the first edge,
7814 clear it if removing it, pass default_edge to
7815 emit_case_dispatch_table.
7816 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7819 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7821 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7822 insn in the function, emit NOP after the insn.
7824 2017-08-06 Tom de Vries <tom@codesourcery.com>
7826 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7829 2017-08-06 Tom de Vries <tom@codesourcery.com>
7831 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7834 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7836 PR tree-optimization/57371
7837 * match.pd: New pattern.
7839 2017-08-04 Marek Polacek <polacek@redhat.com>
7842 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7843 perform the computation in offset_int.
7845 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7847 PR tree-optimization/81136
7848 * tree-vectorizer.h: Include tree-hash-traits.h.
7849 (vec_base_alignments): New typedef.
7850 (vec_info): Add a base_alignments field.
7851 (vect_record_base_alignments): Declare.
7852 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7854 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7855 (create_data_ref): Add an is_conditional_in_stmt argument.
7856 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7857 the is_conditional_in_stmt field.
7858 (data_ref_loc): Add an is_conditional_in_stmt field.
7859 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7860 (find_data_references_in_stmt): Update call to create_data_ref.
7861 (graphite_find_data_references_in_stmt): Likewise.
7862 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7863 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7864 (vect_record_base_alignment): New function.
7865 (vect_record_base_alignments): Likewise.
7866 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7867 for nested statements even if we fail to compute a misalignment.
7868 Use pooled base alignments for unconditional references.
7869 (vect_find_same_alignment_drs): Compare base addresses instead
7871 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7872 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7874 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7876 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7877 Add an explicit name for the enum. Use auto_vec for slp_instances
7879 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7881 (_bb_vec_info): Add a constructor and destructor.
7882 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7883 (destroy_loop_vec_info): Delete.
7884 (vect_destroy_datarefs): Likewise.
7885 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7886 (vec_info::vec_info): New function.
7887 (vec_info::~vec_info): Likewise.
7888 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7889 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7890 destroy_loop_vec_info.
7891 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7892 (_loop_vec_info::_loop_vec_info): ...this.
7893 (destroy_loop_vec_info): Replace with...
7894 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7895 the stmt_vec_infos. Leave handling of vec_info information to its
7896 destructor. Remove explicit vector releases.
7897 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7898 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7899 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7900 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7901 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7902 (destroy_bb_vec_info): Replace with...
7903 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7904 information to its destructor.
7905 (vect_slp_analyze_bb_1): Use new and delete instead of
7906 new_bb_vec_info and destroy_bb_vec_info.
7907 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7910 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7912 * tree-data-ref.h (subscript): Add access_fn field.
7913 (data_dependence_relation): Add could_be_independent_p.
7914 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7915 (same_access_functions): Move to tree-data-ref.c.
7916 * tree-data-ref.c (ref_contains_union_access_p): New function.
7917 (access_fn_component_p): Likewise.
7918 (access_fn_components_comparable_p): Likewise.
7919 (dr_analyze_indices): Add a reference to access_fn_component_p.
7920 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7922 (constant_access_functions): Likewise.
7923 (add_other_self_distances): Likewise.
7924 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7925 (initialize_data_dependence_relation): Use XCNEW and remove
7926 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7927 of access functions that have the same type. Allow the
7928 subsequence to end with different bases in some circumstances.
7929 Record the chosen access functions in SUB_ACCESS_FN.
7930 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7931 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7932 (subscript_dependence_tester_1): Likewise dra and drb.
7933 (build_classic_dist_vector): Update calls accordingly.
7934 (subscript_dependence_tester): Likewise.
7935 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7936 DDR_COULD_BE_INDEPENDENT_P.
7937 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7938 comp_alias_ddrs instead of may_alias_ddrs.
7939 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7941 (vect_analyze_data_ref_dependence): Use it if
7942 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7943 distance vectors if that fails.
7944 (dependence_distance_ge_vf): New function.
7945 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7946 LOOP_VINFO_MAY_ALIAS_DDRS.
7948 2017-08-04 Richard Biener <rguenther@suse.de>
7951 * fold-const.c (fold_binary_loc): Properly restrict
7952 minus_var0 && minus_var1 case when associating undefined overflow
7955 2017-08-04 Tom de Vries <tom@codesourcery.com>
7957 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7959 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7961 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7962 Don't start diagnostic messages with a capital letter.
7963 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7965 (rs6000_invalid_builtin): Likewise.
7966 (rs6000_trampoline_init): Likewise.
7968 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7971 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7972 after setting changeable df flags.
7974 2017-08-03 Richard Biener <rguenther@suse.de>
7976 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7977 up if the use is in USE - X.
7979 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7981 * toplev.c (dumpfile.h): New include.
7982 (internal_error_reentered): New static function. Use it...
7983 (internal_error_function): ...here to handle reentered internal_error.
7985 2017-08-03 Richard Biener <rguenther@suse.de>
7988 * fold-const.c (split_tree): Add minus_var and minus_con
7989 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7990 here but always use minus_*.
7991 (associate_trees): Assert we never associate with MINUS_EXPR
7992 and NULL first operand. Do not recurse for PLUS_EXPR operands
7993 when associating as MINUS_EXPR either.
7994 (fold_binary_loc): Track minus_var and minus_con.
7996 2017-08-03 Tom de Vries <tom@codesourcery.com>
7999 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8000 ACCEL_COMPILER, apply finish_options on
8001 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8003 2017-08-03 Tom de Vries <tom@codesourcery.com>
8006 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8007 function_entry_patch_area_size > 0.
8009 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8012 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8013 instead of 10??LU, perform unit multiplication in wide_int,
8014 don't change alloc_object_size_limit if the limit is larger
8017 PR tree-optimization/81655
8018 PR tree-optimization/81588
8019 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8020 the case when ranges[i].low and high are 1 for unsigned type with
8024 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8025 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8027 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8029 * tree-vrp.h: Add include guard.
8031 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8034 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8035 (ud2): New insn pattern.
8036 * config/i386/i386.c (ix86_expand_epilogue):
8037 For naked functions, generate ud2 instead of trap insn.
8039 2017-08-02 Marek Polacek <polacek@redhat.com>
8042 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8044 2017-08-02 Tom de Vries <tom@codesourcery.com>
8045 Cesar Philippidis <cesar@codesourcery.com>
8047 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8048 Add missing edge probabilities.
8050 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8052 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8055 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8058 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8059 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8060 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8061 prologue_seq sequences - if any.
8063 2017-08-02 Richard Biener <rguenther@suse.de>
8065 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8066 via vectors if supported, integer extracts via punning if supported
8067 or otherwise vector extracts.
8069 2017-08-02 Richard Biener <rguenther@suse.de>
8071 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8073 (bitmap_insert_into_set): ... this.
8075 2017-08-02 Richard Biener <rguenther@suse.de>
8077 PR tree-optimization/81633
8079 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8081 PR tree-optimization/71752
8082 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8084 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8086 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8087 (machine_function::call_ms2sysv_pad_out): Remove field.
8088 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8089 (ix86_compute_frame_layout): Likewise.
8091 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8094 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8095 attribute with interrupt attribute.
8097 2017-08-01 Andrew Pinski <apinski@cavium.com>
8099 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8100 BIT_INSERT_EXPR's operand 1
8101 to see if the types precision matches.
8103 2017-08-01 Martin Liska <mliska@suse.cz>
8106 * builtins.c (expand_builtin_memcpy_args): Remove.
8107 (expand_builtin_memcpy): Call newly added function
8108 expand_builtin_memory_copy_args.
8109 (expand_builtin_memcpy_with_bounds): Likewise.
8110 (expand_builtin_mempcpy): Remove last argument.
8111 (expand_builtin_mempcpy_with_bounds): Likewise.
8112 (expand_builtin_memory_copy_args): New function created from
8113 expand_builtin_mempcpy_args with small modifications.
8114 (expand_builtin_mempcpy_args): Remove.
8115 (expand_builtin_stpcpy): Remove unused argument.
8116 (expand_builtin): Likewise.
8117 (expand_builtin_with_bounds): Likewise.
8119 2017-08-01 Martin Liska <mliska@suse.cz>
8122 Make mempcpy more optimal (PR middle-end/70140).
8124 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8127 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8128 __builtin_vec_cmpne verify both arguments are compatible vectors
8129 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8130 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8131 move computation of aligned to after checking the argument types.
8135 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8136 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8139 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8140 Doug Rupp <rupp@adacore.com>
8141 Olivier Hainque <hainque@adacore.com>
8143 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8144 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8146 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8147 for TARGET_OS_CPP_BUILTIN.
8148 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8149 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8151 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8152 passing required abi options to the assembler for EABI configurations.
8153 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8154 of .text.hot and .text.unlikely sections for kernel modules when
8155 using ARM style exceptions.
8156 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8157 options. Add EXTRA_CC1_SPEC.
8158 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8160 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8162 (ARM_TARGET2_DWARF_FORMAT): Define.
8163 * config/arm/t-vxworks: Adjust multilib control to removal of the
8164 Diab command line options.
8166 2017-08-01 Martin Liska <mliska@suse.cz>
8168 PR gcov-profile/81561
8169 * gcov.c (unblock): Make unblocking safe as we need to preserve
8170 index correspondence of blocks and block_lists.
8172 2017-08-01 Richard Biener <rguenther@suse.de>
8174 PR tree-optimization/81181
8175 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8176 (compute_antic): ... end of iteration here.
8178 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8180 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8181 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8182 (ftree-slp-vectorize): Likewise.
8183 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8184 can no longer be set independent of flag_tree_loop_vectorize.
8185 * omp-general.c (emp_max_vf): Likewise.
8186 * opts.c (enable_fdo_optimizations): Remove references to
8187 flag_tree_vectorize, these are now implicit.
8188 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8189 and leave it for the options machinery.
8191 2017-08-01 Martin Liska <mliska@suse.cz>
8194 * builtins.c (expand_builtin_memcpy_args): Remove.
8195 (expand_builtin_memcpy): Call newly added function
8196 expand_builtin_memory_copy_args.
8197 (expand_builtin_memcpy_with_bounds): Likewise.
8198 (expand_builtin_mempcpy): Remove last argument.
8199 (expand_builtin_mempcpy_with_bounds): Likewise.
8200 (expand_builtin_memory_copy_args): New function created from
8201 expand_builtin_mempcpy_args with small modifications.
8202 (expand_builtin_mempcpy_args): Remove.
8203 (expand_builtin_stpcpy): Remove unused argument.
8204 (expand_builtin): Likewise.
8205 (expand_builtin_with_bounds): Likewise.
8207 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8210 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8211 print "ds:" only for immediates in generic address space.
8213 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8216 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8217 (ix86_function_ok_for_sibcall): Return false for naked functions.
8219 2017-08-01 Richard Biener <rguenther@suse.de>
8221 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8222 (compute_antic): Seed worklist with exit block predecessors.
8223 * cfganal.c (dfs_find_deadend): For a cycle return the source
8224 of the edge closing it.
8226 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8228 * config/aarch64/aarch64.c
8229 (aarch64_can_const_movi_rtx_p): Move 0 check.
8231 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8233 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8234 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8236 * match.pd: Ditto in address comparison pattern.
8238 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8240 PR tree-optimization/81627
8241 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8242 closed ssa form for store-store chain.
8244 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8246 PR tree-optimization/81620
8247 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8248 for store-store chain.
8250 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8252 PR tree-optimization/81588
8253 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8254 ranges[i].in_p, invert comparison code ccode. For >/>=,
8255 swap rhs1 and rhs2 and comparison code unconditionally,
8256 for </<= don't do that. Don't swap rhs1/rhs2 again if
8257 ranges[i].in_p, instead invert comparison code ccode if
8258 opcode or oe->rank is BIT_IOR_EXPR.
8261 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8262 a direct optab to conversion optab.
8263 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8264 with GET_MODE_INNER as last argument instead of optab_handler.
8265 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8266 vector extraction if possible and optab is available.
8267 * expr.c (store_constructor): Use convert_optab_handler instead
8268 of optab_handler. Use vector initialization from smaller
8269 vectors if possible and optab is available.
8270 * tree-vect-stmts.c (vectorizable_load): Likewise.
8271 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8273 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8274 of vec_init from half-sized vectors with the same element mode.
8275 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8276 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8277 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8278 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8279 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8280 after mode in gen_vec_extract* calls.
8281 (vec_extract<mode>): Renamed to ...
8282 (vec_extract<mode><ssescalarmodelower>): ... this.
8283 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8284 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8285 element mode after mode in gen_vec_init* calls.
8286 (VEC_INIT_HALF_MODE): New mode iterator.
8287 (vec_init<mode>): Renamed to ...
8288 (vec_init<mode><ssescalarmodelower>): ... this.
8289 (vec_init<mode><ssehalfvecmodelower>): New expander.
8290 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8291 (vec_extractv2sfsf): ... this.
8292 (vec_initv2sf): Renamed to ...
8293 (vec_initv2sfsf): ... this.
8294 (vec_extractv2si): Renamed to ...
8295 (vec_extractv2sisi): ... this.
8296 (vec_initv2si): Renamed to ...
8297 (vec_initv2sisi): ... this.
8298 (vec_extractv4hi): Renamed to ...
8299 (vec_extractv4hihi): ... this.
8300 (vec_initv4hi): Renamed to ...
8301 (vec_initv4hihi): ... this.
8302 (vec_extractv8qi): Renamed to ...
8303 (vec_extractv8qiqi): ... this.
8304 (vec_initv8qi): Renamed to ...
8305 (vec_initv8qiqi): ... this.
8306 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8307 (vec_init<mode>): Renamed to ...
8308 (vec_init<mode><VEC_base_l>): ... this.
8309 (vec_extract<mode>): Renamed to ...
8310 (vec_extract<mode><VEC_base_l>): ... this.
8311 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8312 (vec_initv2sfsf): ... this.
8313 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8314 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8315 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8316 element mode after mode in gen_vec_init* calls.
8317 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8318 (vec_init<mode><Vel>): ... this.
8319 (vec_extract<mode>): Renamed to ...
8320 (vec_extract<mode><Vel>): ... this.
8321 * config/aarch64/iterators.md (Vel): New mode attribute.
8322 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8323 Add element mode after mode in gen_vec_extract* calls.
8324 * config/s390/vector.md (non_vec_l): New mode attribute.
8325 (vec_extract<mode>): Renamed to ...
8326 (vec_extract<mode><non_vec_l>): ... this.
8327 (vec_init<mode>): Renamed to ...
8328 (vec_init<mode><non_vec_l>): ... this.
8329 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8330 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8332 * config/arm/iterators.md (V_elem_l): New mode attribute.
8333 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8334 (vec_extract<mode><V_elem_l>): ... this.
8335 (vec_extractv2di): Renamed to ...
8336 (vec_extractv2didi): ... this.
8337 (vec_init<mode>): Renamed to ...
8338 (vec_init<mode><V_elem_l>): ... this.
8339 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8340 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8341 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8342 Add element mode after gen_vec_extract* calls.
8343 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8344 (vec_init<mode><unitmode>): ... this.
8345 (vec_extract<mode>): Renamed to ...
8346 (vec_extract<mode><unitmode>): ... this.
8347 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8348 (vec_init<mode><unitmode>): ... this.
8349 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8350 (vec_initv2sfsf): ... this.
8351 (vec_extractv2sf): Renamed to ...
8352 (vec_extractv2sfsf): ... this.
8353 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8354 Add element mode after gen_vec_extract* calls.
8355 * config/mips/mips.md (unitmode): New mode iterator.
8356 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8357 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8358 * config/spu/spu.md (inner_l): New mode attribute.
8359 (vec_init<mode>): Renamed to ...
8360 (vec_init<mode><inner_l>): ... this.
8361 (vec_extract<mode>): Renamed to ...
8362 (vec_extract<mode><inner_l>): ... this.
8363 * config/sparc/sparc.md (veltmode): New mode iterator.
8364 (vec_init<VMALL:mode>): Renamed to ...
8365 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8366 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8367 (vec_initv2sisi): ... this.
8368 (vec_initv2sf): Renamed to ...
8369 (vec_initv2sfsf): ... this.
8370 (vec_extractv2sf): Renamed to ...
8371 (vec_extractv2sfsf): ... this.
8372 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8373 (vec_init<mode>): Renamed to ...
8374 (vec_init<mode><VEC_base_l>): ... this.
8375 (vec_extract<mode>): Renamed to ...
8376 (vec_extract<mode><VEC_base_l>): ... this.
8377 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8378 (vec_initv2sfsf): ... this.
8379 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8380 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8381 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8382 gen_vec_init* calls.
8384 2017-08-01 Richard Biener <rguenther@suse.de>
8386 PR tree-optimization/81297
8387 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8388 TREE_OVERFLOW from INTEGER_CSTs.
8390 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8392 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8394 2017-07-31 Carl Love <cel@us.ibm.com>
8396 * config/rs6000/rs6000-c: Add support for built-in functions
8397 vector signed char vec_xl_be (signed long long, signed char *);
8398 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8399 vector signed int vec_xl_be (signed long long, signed int *);
8400 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8401 vector signed long long vec_xl_be (signed long long, signed long long *);
8402 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8403 vector signed short vec_xl_be (signed long long, signed short *);
8404 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8405 vector double vec_xl_be (signed long long, double *);
8406 vector float vec_xl_be (signed long long, float *);
8407 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8408 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8409 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8411 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8412 (altivec_expand_builtin): Add switch statement to call
8413 altivec_expand_xl_be for each builtin.
8414 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8415 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8416 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8417 __builtin_vsx_le_be_v16qi.
8418 * doc/extend.texi: Update the built-in documentation file for the
8419 new built-in functions.
8421 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8424 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8426 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8428 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8430 * config.gcc: Add z14.
8431 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8432 CPU model numbers for z13s and z14.
8433 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8435 * config/s390/s390-opts.h (enum processor_type): Rename
8436 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8437 * config/s390/s390.c (processor_table): Add field for CPU name to
8438 be passed to Binutils.
8439 (s390_asm_output_machine_for_arch): Use the new field in
8440 processor_table for Binutils.
8441 (s390_expand_builtin): Replace arch12 with z14.
8442 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8443 (s390_get_sched_attrmask): Likewise.
8444 (s390_get_unit_mask): Likewise.
8445 * config/s390/s390.opt: Add z14 to processor_type enum.
8447 2017-07-31 Martin Jambor <mjambor@suse.cz>
8450 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8451 regardless of optimization level.
8453 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8454 Martin Liska <mliska@suse.cz>
8456 * predict.def: Remove old comment and adjust probability.
8457 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8460 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8463 * config/i386/i386.c (ix86_function_naked): New function.
8464 (ix86_can_use_return_insn_p): Return false for naked functions.
8465 (ix86_expand_prologue): Skip prologue for naked functions.
8466 (ix86_expand_epilogue): Skip epilogue for naked functions
8467 and emit trap instruction.
8468 (ix86_warn_func_return): New function.
8469 (ix86_attribute_table): Add "naked" attribute specification.
8470 (TARGET_WARN_FUNC_RETURN): Define.
8471 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8473 2017-07-31 Martin Liska <mliska@suse.cz>
8475 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8476 (dump_gimple_bb_header): Always dump BB info.
8477 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8479 2017-07-31 Martin Liska <mliska@suse.cz>
8482 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8483 also with current_function_decl non-null equality.
8485 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8488 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8489 change type to the element type, instead add eltype variable and
8490 use it where we are interested in the element type.
8492 PR tree-optimization/81603
8493 * ipa-polymorphic-call.c
8494 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8495 offset arithmetic in offset_int, bail out if the resulting bit offset
8496 doesn't fit into shwi.
8498 2017-07-31 Martin Liska <mliska@suse.cz>
8500 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8501 (gimplify_save_expr): Fix comment.
8503 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8506 * config/i386/i386.c (ix86_function_arg): Update arguments for
8508 (ix86_compute_frame_layout): Set the initial stack offset to
8509 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8510 INCOMING_FRAME_SP_OFFSET.
8511 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8512 stack before exception handler returns.
8513 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8514 the 'ERROR_CODE' for exception handler.
8516 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8518 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8519 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8520 (ASM_OUTPUT_REG_POP): Ditto.
8521 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8522 instead of asm_fprintf to output pure string.
8524 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8526 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8527 to imported_module_or_decl hook.
8528 (debug_nothing_tree_tree_tree_bool): Remove.
8529 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8530 * debug.c (do_nothing_debug_hooks): Use
8531 debug_nothing_tree_tree_tree_bool_bool instead of
8532 debug_nothing_tree_tree_tree_bool.
8533 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8534 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8535 * sdbout.c (sdb_debug_hooks): Likewise.
8536 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8537 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8539 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8540 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8541 attribute, don't add anything.
8543 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8545 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8546 (fold_build2_stat_loc): Likewise.
8547 (fold_build3_stat_loc): Likewise.
8548 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8549 (fold_build1_loc): Remove macro.
8550 (fold_build2_loc): Likewise.
8551 (fold_build3_loc): Likewise.
8553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8555 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8556 (gimple_build_debug_bind_source_stat): Likewise.
8557 * gimple.h (gimple_build_debug_bind): Remove macro.
8558 (gimple_build_debug_bind_source): Likewise.
8560 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8562 * bitmap.c (bitmap_alloc): Adjust.
8563 (bitmap_gc_alloc): Likewise.
8564 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
8566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8568 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
8569 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
8570 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
8571 (bitmap_gc_alloc_stat): Likewise.
8572 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
8574 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8576 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
8577 * rtl.h (shallow_copy_rtx): Remove macro.
8579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8581 * emit-rtl.c (gen_raw_REG): Adjust.
8582 * gengenrtl.c (gendef): Likewise.
8583 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8584 * rtl.h (rtx_alloc): Remove macro.
8586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8588 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8589 (build_tree_list_stat): Likewise.
8590 * tree.h (build_tree_list): Remove macro.
8591 (build_tree_list_vec): Likewise.
8593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8595 * tree.c (make_vector_stat): Remove _stat from name.
8596 (build_vector_stat): Likewise.
8597 * tree.h (make_vector_stat): Remove macro.
8598 (build_vector_stat): Likewise.
8600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8602 * tree.h (build_var_debug_value): Remove prototype.
8604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8606 * tree.c (tree_cons_stat): Remove _stat from name.
8607 * tree.h (tree_cons): Remove macro.
8609 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8611 * tree.c (build_vl_exp_stat): Remove _stat from name.
8612 * tree.h (build_vl_exp): Remove macro.
8614 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8616 * tree.c (build_decl_stat): Remove _stat from name.
8617 * tree.h (build_decl): Remove macro.
8619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8621 * gimple.c (gimple_build_with_ops_stat): Adjust.
8622 (gimple_alloc_stat): Remove _stat from name.
8623 * gimple.h (gimple_alloc): Remove macro.
8625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8627 * tree.c (make_tree_vec_stat): Remove _stat from name.
8628 (grow_tree_vec_stat): Likewise.
8629 * tree.h (make_tree_vec_stat): Adjust prototype.
8630 (grow_tree_vec_stat): Likewise.
8631 (make_tree_vec): Remove macro.
8632 (grow_tree_vec): Likewise.
8634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8636 * fold-const.c (fold_build1_stat_loc): Adjust.
8637 (fold_build2_stat_loc): Likewise.
8638 (fold_build3_stat_loc): Likewise.
8639 * tree.c (build0_stat): Remove _stat from name.
8640 (build1_stat): Likewise.
8641 (build2_stat): Likewise.
8642 (build3_stat): Likewise.
8643 (build4_stat): Likewise.
8644 (build5_stat): Likewise.
8645 * tree.h (build1_loc): Remove macro, and rename _stat function
8647 (build2_loc): Likewise.
8648 (build3_loc): Likewise.
8649 (build4_loc): Likewise.
8650 (build5_loc): Likewise.
8652 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8654 * tree.c (make_int_cst_stat): Remove _stat from name.
8655 * tree.h (make_int_cst_stat): Adjust prototype.
8656 (make_int_cst): Remove macro.
8658 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8660 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8661 * tree.h (make_tree_binfo_stat): Adjust prototype.
8662 (make_tree_binfo): Remove.
8664 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8666 * tree.c (copy_node_stat): Rename to copy_node.
8667 (build_distinct_type_copy): Adjust.
8668 * tree.h (copy_node_stat): Adjust prototype.
8669 (copy_node): Remove macro.
8671 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8673 * tree.c (make_node_stat): rename to make_node.
8674 (build_tree_list_stat): Adjust.
8675 (build0_stat): Likewise.
8676 (build2_stat): Likewise.
8677 (build3_stat): Likewise.
8678 (build4_stat): Likewise.
8679 (build5_stat): Likewise.
8680 (build_decl_stat): Likewise.
8681 * tree.h (make_node_stat): Adjust prototype.
8682 (make_node): remove macro.
8684 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8686 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8687 (PPC_FEATURE2_SCV): Likewise.
8688 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8690 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8692 * config/aarch64/aarch64.c
8693 (aarch64_internal_mov_immediate): Add new special pattern.
8694 * config/aarch64/aarch64.md (*movdi_aarch64):
8695 Add reg/32bit const mov case.
8697 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8698 Richard Sandiford <richard.sandiford@linaro.org>
8700 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8701 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8702 Add integer and movi cases.
8703 (movi-split-hf-df-sf split, fp16): New.
8704 (enabled): Added TARGET_FP_F16INST.
8705 * config/aarch64/iterators.md (GPF_HF): New.
8706 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8708 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8710 * config/aarch64/aarch64.c
8711 (aarch64_simd_container_mode): Add prototype.
8712 (aarch64_expand_mov_immediate): Add HI support.
8713 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8714 (aarch64_can_const_movi_rtx_p): New.
8715 (aarch64_preferred_reload_class):
8716 Remove restrictions of using FP registers for certain SIMD operations.
8717 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8718 (aarch64_valid_floating_const): Add integer move validation.
8719 (aarch64_simd_imm_scalar_p): Remove.
8720 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8721 (aarch64_legitimate_constant_p): Expand list of supported cases.
8722 * config/aarch64/aarch64-protos.h
8723 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8724 (aarch64_reinterpret_float_as_int): New.
8725 (aarch64_simd_imm_scalar_p): Remove.
8726 * config/aarch64/constraints.md (Uvi): New.
8727 (Dd): Split into Ds and new Dd.
8728 * config/aarch64/aarch64.md (*movsi_aarch64):
8730 (*movdi_aarch64): Add SIMD mov case.
8732 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8734 * tree-predcom.c: (struct chain): Handle store-store chain in which
8735 stores for elimination only store loop invariant values.
8736 (execute_pred_commoning_chain): Ditto.
8737 (prepare_initializers_chain_store_elim): Ditto.
8738 (prepare_finalizers): Ditto.
8739 (is_inv_store_elimination_chain): New function.
8740 (initialize_root_vars_store_elim_1): New function.
8742 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8744 * tree-predcom.c: Revise general description of the pass.
8745 (enum chain_type): New enum type for store elimination.
8746 (struct chain): New field supporting store elimination.
8747 (struct component): Ditto.
8748 (dump_chain): Dump store-stores chain.
8749 (release_chain): Release resources.
8750 (split_data_refs_to_components): Compute and create component
8751 contains only stores for elimination.
8752 (get_chain_last_ref_at): New function.
8753 (make_invariant_chain): Initialization.
8754 (make_rooted_chain): Specify chain type in parameter and record it.
8755 (add_looparound_copies): Skip for store-stores chain.
8756 (determine_roots_comp): Compute type of chain and pass it to
8758 (initialize_root_vars_store_elim_2): New function.
8759 (finalize_eliminated_stores): New function.
8760 (remove_stmt): Handle store for elimination.
8761 (execute_pred_commoning_chain): Execute predictive commoning on
8763 (determine_unroll_factor): Skip unroll for store-stores chain.
8764 (prepare_initializers_chain_store_elim): New function.
8765 (prepare_initializers_chain): Hanlde store-store chain.
8766 (prepare_finalizers_chain, prepare_finalizers): New function.
8767 (tree_predictive_commoning_loop): Return integer value indicating
8768 if loop is unrolled or lcssa form is corrupted.
8769 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8771 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8773 * tree-predcom.c (initialize_root): Delete.
8774 (execute_pred_commoning_chain): Initialize root vars and replace
8775 reference of non-combined chain directly, rather than call above
8778 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8780 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8781 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8783 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8785 * tree-predcom.c (struct chain): New field init_seq.
8786 (release_chain): Release init_seq.
8787 (prepare_initializers_chain): Record intialization stmts in above
8789 (insert_init_seqs): New function.
8790 (tree_predictive_commoning_loop): Call insert_init_seqs.
8792 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8794 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8796 2017-07-28 Richard Biener <rguenther@suse.de>
8798 * match.pd: Remove superfluous :c.
8799 * genmatch.c (simplify::id): Add member.
8800 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8802 (current_id): New global.
8803 (dt_node::parent): Move from ...
8804 (dt_operand::parent): ... here. Add for_id member.
8805 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8806 (decision_tree::find_node): Relax order requirement when
8807 merging DT_TRUE nodes to ones inbetween the current simplify
8808 and the one we try to merge with. Add diagnostic whenever
8809 we need to enforce pattern order by not merging.
8810 (decision_tree::insert): Set current_id.
8811 (decision_tree::print_node): Dump parent node and for_id.
8812 (parser::last_id): Add member.
8813 (parser::push_simplify): Assign unique id.
8814 (parser::parser): Initialize last_id.
8816 2017-07-28 Martin Liska <mliska@suse.cz>
8819 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8820 gimple_build_debug_bind.
8822 2017-07-28 Richard Biener <rguenther@suse.de>
8824 PR tree-optimization/81502
8825 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8827 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8829 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8830 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8831 for BIT_FIELD_REF args.
8832 * fold-const.c (make_bit_field_ref): Likewise.
8833 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8835 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8838 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8839 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8840 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8841 Or it into SANITIZER_UNDEFINED.
8842 * ubsan.c: Include gimple-fold.h and varasm.h.
8843 (ubsan_expand_ptr_ifn): New function.
8844 (instrument_pointer_overflow): New function.
8845 (maybe_instrument_pointer_overflow): New function.
8846 (instrument_object_size): Formatting fix.
8847 (pass_ubsan::execute): Call instrument_pointer_overflow
8848 and maybe_instrument_pointer_overflow.
8849 * internal-fn.c (expand_UBSAN_PTR): New function.
8850 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8851 * sanitizer.def (__ubsan_handle_pointer_overflow,
8852 __ubsan_handle_pointer_overflow_abort): New builtins.
8853 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8854 * internal-fn.def (UBSAN_PTR): New internal function.
8855 * opts.c (sanitizer_opts): Add pointer-overflow.
8856 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8857 * fold-const.c (build_range_check): Compute pointer range check in
8858 integral type if pointer arithmetics would be needed. Formatting
8861 2017-07-28 Martin Liska <mliska@suse.cz>
8864 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8865 parameters that are of a variable-length.
8867 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8869 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8871 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8872 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8873 (CRT_CALL_STATIC_FUNCTION): Likewise.
8874 (ASM_DEFAULT_SPEC): New define.
8875 (ASM_SPEC32): Likewise.
8876 (ASM_SPEC64): Likewise.
8877 (ASM_SPEC_COMMON): Likewise.
8878 (ASM_SPEC): Likewise.
8879 (INVALID_64BIT): Likewise.
8880 (LINK_OS_DEFAULT_SPEC): Likewise.
8881 (LINK_OS_SPEC32): Likewise.
8882 (LINK_OS_SPEC64): Likewise.
8883 (POWERPC_LINUX): Likewise.
8884 (PTRDIFF_TYPE): Likewise.
8885 (RESTORE_FP_PREFIX): Likewise.
8886 (RESTORE_FP_SUFFIX): Likewise.
8887 (SAVE_FP_PREFIX): Likewise.
8888 (SAVE_FP_SUFFIX): Likewise.
8889 (SIZE_TYPE): Likewise.
8890 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8891 (TARGET_64BIT): Likewise.
8892 (TARGET_64BIT): Likewise.
8893 (TARGET_AIX): Likewise.
8894 (WCHAR_TYPE_SIZE): Likewise.
8895 (WCHAR_TYPE): Undefine.
8896 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8897 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8898 (CPP_OS_RTEMS_SPEC): Delete.
8899 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8900 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8902 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8904 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8906 PR tree-optimization/81578
8907 * tree-parloops.c (build_new_reduction): Bail out if
8908 reduction_code isn't one of the standard OpenMP reductions.
8909 Move the details printing after that decision.
8911 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8913 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8914 related to reload_in_progress.
8915 (splat_input_operand): Likewise.
8916 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8918 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8920 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8921 (TARGET_INSTANTIATE_DECLS): Likewise.
8922 (legitimate_indexed_address_p): Delete reload_in_progress code.
8923 (rs6000_debug_legitimate_address_p): Likewise.
8924 (rs6000_eliminate_indexed_memrefs): Likewise.
8925 (rs6000_emit_le_vsx_store): Likewise.
8926 (rs6000_emit_move_si_sf_subreg): Likewise.
8927 (rs6000_emit_move): Likewise.
8928 (register_to_reg_type): Likewise.
8929 (rs6000_pre_atomic_barrier): Likewise.
8930 (rs6000_machopic_legitimize_pic_address): Likewise.
8931 (rs6000_allocate_stack_temp): Likewise.
8932 (rs6000_address_for_fpconvert): Likewise.
8933 (rs6000_address_for_altivec): Likewise.
8934 (rs6000_secondary_memory_needed_rtx): Delete function.
8935 (rs6000_check_sdmode): Likewise.
8936 (rs6000_alloc_sdmode_stack_slot): Likewise.
8937 (rs6000_instantiate_decls): Likewise.
8938 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8939 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8940 Delete reload_in_progress.
8941 (*vec_reload_and_plus_<mptrsize>): Likewise.
8942 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8943 (vsx_div_v2di): Likewise.
8944 (vsx_udiv_v2di): Likewise.
8946 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8948 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8949 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8950 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8951 (rs6000_debug_reg_global): Delete print of LRA status.
8952 (rs6000_option_override_internal): Delete dead LRA related code.
8953 (rs6000_lra_p): Delete function.
8954 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8956 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8958 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8959 * config/riscv/rtems.h: New file.
8961 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8962 Sudakshina Das <sudi.das@arm.com>
8964 * config/aarch64/aarch64.md
8965 (define_split for and<mode>3nr_compare): Move
8966 non aarch64_logical_operand to a register.
8967 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8968 register immediate operand to a register.
8969 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8971 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8974 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8976 2017-07-27 Richard Biener <rguenther@suse.de>
8978 PR tree-optimization/81573
8979 PR tree-optimization/81494
8980 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8981 multi defuse cycle case.
8983 2017-07-27 Richard Biener <rguenther@suse.de>
8985 PR tree-optimization/81571
8986 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8989 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8991 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8992 earlier and only if MASK_FPU is set. Adjust formatting.
8994 2017-07-27 Martin Liska <mliska@suse.cz>
8996 * opt-functions.awk: Add validation of value of Init.
8997 * optc-gen.awk: Pass new argument.
8999 2017-07-27 Martin Liska <mliska@suse.cz>
9001 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9002 Fix wrong condition.
9004 2017-07-27 Martin Liska <mliska@suse.cz>
9006 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9007 BBs and edges seen by autoFDO.
9009 2017-07-27 Richard Biener <rguenther@suse.de>
9011 PR tree-optimization/81502
9012 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9013 with incompatible but same sized type.
9014 (execute_update_addresses_taken): Likewise.
9016 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9018 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9019 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9021 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9024 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9025 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9026 Change s_operand to memory_operand.
9028 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9030 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9031 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9032 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9033 Emit instructions rather than returning an expression. Handle TFmode
9034 and KFmode by casting to TImode.
9035 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9036 (rs6000_emit_le_vsx_store): Likewise.
9037 * config/rs6000/vsx.md (VSX_TI): New iterator.
9038 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9039 (*vsx_le_undo_permute_<mode>): Likewise.
9040 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9041 emit the split sequence.
9042 (*vsx_le_perm_store_<mode>): Likewise.
9044 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9046 PR tree-optimization/81555
9047 PR tree-optimization/81556
9048 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9049 if true, force CHANGED for the recursive invocation.
9050 (reassociate_bb): Remember original length of ops array, pass
9051 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9053 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9054 attributes for noipa attribute. For naked attribute use
9055 lookup_attribute first before lookup_attribute_spec.
9056 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9058 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9059 for functions with noipa attribute.
9060 (cgraph_externally_visible_p): Return true for functions with noipa
9062 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9063 for functions with noipa attribute.
9064 * doc/extend.texi: Document noipa function attribute.
9065 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9066 also for functions with noipa attribute.
9067 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9069 2017-07-26 Andrew Pinski <apinski@cavium.com>
9071 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9072 vec_unalign_load_cost and vec_unalign_store_cost.
9074 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9076 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9077 -mvsx-small-integer option.
9078 (ISA_3_0_MASKS_IEEE): Likewise.
9079 (OTHER_VSX_VECTOR_MASKS): Likewise.
9080 (POWERPC_MASKS): Likewise.
9081 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9082 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9083 code, only testing for DImode being allowed in non-VSX floating
9085 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9086 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9088 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9089 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9090 TARGET_P8_VECTOR test.
9091 (rs6000_secondary_reload_simple_move): Likewise.
9092 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9093 since TARGET_P9_VECTOR was already tested.
9094 (rs6000_opt_masks): Remove -mvsx-small-integer.
9095 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9096 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9098 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9099 test for TARGET_VEXTRACTUB was used, and that uses
9101 (p9 extract splitter): Likewise.
9102 (vsx_extract_<mode>_di_p9): Likewise.
9103 (vsx_extract_<mode>_store_p9): Likewise.
9104 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9105 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9106 the elimination of TARGET_VSX_SMALL_INTEGER.
9107 (vsx_extract_<mode>_p8): Likewise.
9108 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9109 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9110 (vsx_set_<mode>_p9): Likewise.
9111 (vsx_set_v4sf_p9): Likewise.
9112 (vsx_set_v4sf_p9_zero): Likewise.
9113 (vsx_insert_extract_v4sf_p9): Likewise.
9114 (vsx_insert_extract_v4sf_p9_2): Likewise.
9115 * config/rs6000/rs6000.md (sign extend splitter): Change
9116 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9117 (floatsi<mode>2_lfiwax_mem): Likewise.
9118 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9119 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9120 since a test for TARGET_P9_VECTOR was used.
9121 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9122 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9123 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9124 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9125 TARGET_P8_VECTOR test.
9126 (fix_trunc<mode>si2_stfiwx): Likewise.
9127 (fix_trunc<mode>si2_internal): Likewise.
9128 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9129 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9131 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9132 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9133 TARGET_P8_VECTOR test.
9134 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9135 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9136 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9138 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9139 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9140 since a test for TARGET_P9_VECTOR was used.
9141 (splitter for loading small constants): Likewise.
9143 2017-07-26 Andrew Pinski <apinski@cavium.com>
9145 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9148 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9151 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9152 (fp_valid_at): Likewise.
9154 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9156 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9157 (qdf24xx_addrcost_table): Likewise.
9158 (cortexa57_tunings): Update to use generic_branch_cost.
9159 (cortexa72_tunings): Likewise.
9160 (cortexa73_tunings): Likewise.
9161 (qdf24xx_tunings): Likewise.
9163 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9165 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9166 (thunderx2t99_branch_cost): Likewise.
9167 (cortexa35_tunings): Update to use generic_branch_cost.
9168 (cortexa53_tunings): Likewise.
9169 (cortexa57_tunings): Likewise.
9170 (cortexa72_tunings): Likewise.
9171 (cortexa73_tunings): Likewise.
9172 (thunderx2t99_tunings): Likewise.
9174 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9176 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9177 (sparc_option_override): Honour MASK_FSMULD.
9178 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9179 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9180 * config/sparc/sparc.opt (mfsmuld): New option.
9181 * doc/invoke.texi (mfsmuld): Document option.
9183 2017-07-26 Marek Polacek <polacek@redhat.com>
9186 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9188 2017-07-26 Richard Biener <rguenther@suse.de>
9190 * gimple-match-head.c (do_valueize): Return OP if valueize
9192 (get_def): New helper to get at the def stmt of a SSA name
9194 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9195 do_valueize to get at the def stmt.
9196 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9198 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9201 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9203 2017-07-26 Martin Liska <mliska@suse.cz>
9206 * function.c (expand_function_start): Make expansion of
9207 nonlocal_goto_save_area after parm_birth_insn.
9209 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9211 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9212 from all CPU target flags enable members.
9214 2017-07-26 Richard Biener <rguenther@suse.de>
9216 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9217 (decision_tree::gen): Make 'type' const.
9218 (write_predicate): Likewise.
9220 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9222 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9223 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9224 (rs6000_option_override_internal): Likewise.
9225 (rs6000_expand_vector_set): Likewise.
9226 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9227 (TARGET_UPPER_REGS_SF): Likewise.
9228 (TARGET_UPPER_REGS_DI): Likewise.
9229 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9230 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9231 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9232 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9233 (Splitters for DI constants in Altivec registers): Likewise.
9234 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9235 (vsx_set_v4sf_p9): Likewise.
9236 (vsx_set_v4sf_p9_zero): Likewise.
9237 (vsx_insert_extract_v4sf_p9): Likewise.
9238 (vsx_insert_extract_v4sf_p9_2): Likewise.
9240 2017-07-25 Carl Love <cel@us.ibm.com>
9242 * doc/extend.texi: Update the built-in documentation file for the
9243 existing built-in functions
9244 vector signed char vec_cnttz (vector signed char);
9245 vector unsigned char vec_cnttz (vector unsigned char);
9246 vector signed short vec_cnttz (vector signed short);
9247 vector unsigned short vec_cnttz (vector unsigned short);
9248 vector signed int vec_cnttz (vector signed int);
9249 vector unsigned int vec_cnttz (vector unsigned int);
9250 vector signed long long vec_cnttz (vector signed long long);
9251 vector unsigned long long vec_cnttz (vector unsigned long long);
9253 2017-07-25 Andrew Pinski <apinski@cavium.com>
9255 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9256 accesses where the use is for the first operand of a BIT_INSERT.
9258 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9261 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9262 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9264 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9266 * config/i386/gstabs.h: Delete.
9267 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9269 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9271 * config/i386/i386.c (ix86_decompose_address): Do not check for
9272 register RTX when looking at index_reg or base_reg.
9273 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9275 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9277 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9278 to update EH info here.
9280 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9282 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9284 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9286 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9288 2017-07-25 Torsten Duwe <duwe@suse.de>
9290 * common.opt: Introduce -fpatchable-function-entry
9291 command line option, and its variables function_entry_patch_area_size
9292 and function_entry_patch_area_start.
9293 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9294 including a two-value parser.
9295 * target.def (print_patchable_function_entry): New target hook.
9296 * targhooks.h (default_print_patchable_function_entry): New function.
9297 * targhooks.c (default_print_patchable_function_entry): Likewise.
9298 * toplev.c (process_options): Switch off IPA-RA if
9299 patchable function entries are being generated.
9300 * varasm.c (assemble_start_function): Look at the
9301 patchable-function-entry command line switch and current
9302 function attributes and maybe generate NOP instructions by
9303 calling the print_patchable_function_entry hook.
9304 * doc/extend.texi: Document patchable_function_entry attribute.
9305 * doc/invoke.texi: Document -fpatchable_function_entry
9306 command line option.
9307 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9309 * doc/tm.texi: Re-generate.
9311 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9314 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9315 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9317 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9319 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9321 2017-07-25 Richard Biener <rguenther@suse.de>
9323 PR tree-optimization/81455
9324 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9325 not walk in cycles when looking for guards.
9327 2017-07-25 Richard Biener <rguenther@suse.de>
9329 PR tree-optimization/81529
9330 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9331 when optimizing backedge uses.
9333 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9335 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9337 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9338 to dl_section_ref. On AIX, append an expression to subtract
9339 the size of the section length to dl_section_ref.
9341 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9343 * configure.ac: If any of the config.* scripts fail, exit 1.
9344 * configure: Regenerate.
9346 2017-07-25 Richard Biener <rguenther@suse.de>
9349 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9350 of immediate uses, be more verbose on errors.
9352 2017-07-25 Richard Biener <rguenther@suse.de>
9354 PR tree-optimization/81510
9355 * tree-vect-loop.c (vect_is_simple_reduction): When the
9356 reduction stmt is not inside the loop bail out.
9358 2017-07-25 Richard Biener <rguenther@suse.de>
9360 PR tree-optimization/81303
9361 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9362 profitability check against LOOP_VINFO_NITERSM1.
9364 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9366 * domwalk.c (cmp_bb_postorder): Simplify.
9367 (sort_bbs_postorder): New function. Use it...
9368 (dom_walker::walk): ...here to optimize common cases.
9370 2017-07-25 Martin Liska <mliska@suse.cz>
9373 * ipa-visibility.c (function_and_variable_visibility): Make the
9374 redirection just on target that supports aliasing.
9375 Fix GNU coding style.
9377 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9380 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9382 * config/arm/rtems.h: Likewise.
9383 * config/bfin/rtems.h: Likewise.
9384 * config/i386/rtemself.h: Likewise.
9385 * config/lm32/rtems.h: Likewise.
9386 * config/m32c/rtems.h: Likewise.
9387 * config/m68k/rtemself.h: Likewise.
9388 * config/microblaze/rtems.h: Likewise.
9389 * config/mips/rtems.h: Likewise.
9390 * config/moxie/rtems.h: Likewise.
9391 * config/nios2/rtems.h: Likewise.
9392 * config/powerpcspe/rtems.h: Likewise.
9393 * config/rs6000/rtems.h: Likewise.
9394 * config/rtems.h: Likewise.
9395 * config/sh/rtems.h: Likewise.
9396 * config/sh/rtemself.h: Likewise.
9397 * config/sparc/rtemself.h: Likewise.
9399 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9402 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9403 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9404 * tree-ssa-structalias.c (alias_get_name): Same.
9406 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9409 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9410 instructions if no du chain is found.
9412 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9414 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9416 2017-07-25 Richard Biener <rguenther@suse.de>
9419 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9422 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9424 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9426 (ISA_2_7_MASKS_SERVER): Likewise.
9427 (ISA_3_0_MASKS_IEEE): Likewise.
9428 (OTHER_P8_VECTOR_MASKS): Likewise.
9429 (OTHER_VSX_VECTOR_MASKS): Likewise.
9430 (POWERPC_MASKS): Likewise.
9431 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9432 duplicate list of options.
9433 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9434 explicit -mupper-regs options.
9435 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9436 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9437 alias for -mupper-regs-df.
9438 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9439 (rs6000_init_hard_regno_mode_ok): Likewise.
9440 (rs6000_option_override_internal): Likewise.
9441 (rs6000_opt_masks): Likewise.
9442 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9443 options in terms of whether -mvsx or -mpower8-vector was used.
9444 (TARGET_UPPER_REGS_DI): Likewise.
9445 (TARGET_UPPER_REGS_SF): Likewise.
9446 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9447 -mupper-regs-* options.
9449 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9451 * passes.c (emergency_dump_function): Print some empty lines and a
9452 header before the RTL dump.
9454 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9456 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9458 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9461 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9462 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9464 2017-07-24 Carl Love <cel@us.ibm.com>
9466 * config/rs6000/rs6000-c.c: Add support for built-in functions
9467 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9468 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9469 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9470 vec_extract_fp_from_shortl): Add defines for the two builtins.
9471 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9472 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9474 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9475 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9476 * doc/extend.texi: Update the built-in documentation file for the
9477 new built-in function.
9479 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9482 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9484 * doc/generic.texi: Likewise.
9485 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9486 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9488 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9490 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9491 (aarch64_mls_elt_merge<mode>): Likewise.
9493 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9495 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9496 having __cxa_atexit.
9498 2017-07-23 Michael Collison <michael.collison@arm.com>
9500 * config/arm/arm.c (arm_option_override): Deprecate
9501 use of -mstructure-size-boundary.
9502 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9503 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9505 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9508 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9509 Reduce cost estimate for direct moves.
9511 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9514 * config/i386/i386.c (ix86_option_override_internal): Disable
9515 BMI, BMI2 and TBM instructions for -m16.
9517 2017-07-21 Carl Love <cel@us.ibm.com>
9519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9520 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9521 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9523 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9524 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9525 VMULOSW): New enum "unspec" values.
9526 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9527 altivec_vmulosw): New patterns.
9528 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9529 VMULOSW): Add definitions.
9531 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9533 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9534 (qdf24xx): Likewise.
9535 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9536 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9537 (AARCH64_FL_V8_1): Renumber.
9538 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9539 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9540 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9541 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9542 rdma to feature modifiers list.
9544 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9547 * ipa-visibility (function_and_variable_visibility): Convert
9548 recursive PLT call to direct call if appropriate.
9550 2017-07-21 Andrew Pinski <apinski@cavium.com>
9552 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9553 operand 1 to see if the types precision matches.
9554 * fold-const.c (operand_equal_p): Likewise.
9556 2017-07-21 Richard Biener <rguenther@suse.de>
9558 PR tree-optimization/81303
9559 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
9560 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
9561 (vect_peeling_hash_get_lowest_cost): Adjust.
9562 (vect_enhance_data_refs_alignment): Likewise. Use
9563 vect_get_peeling_costs_all_drs to compute the penalty for no
9564 peeling to match up costs.
9566 2017-07-21 Richard Biener <rguenther@suse.de>
9568 PR tree-optimization/81500
9569 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
9570 we didn't identify a reduction path.
9572 2017-07-21 Tom de Vries <tom@codesourcery.com>
9573 Cesar Philippidis <cesar@codesourcery.com>
9575 PR gcov-profile/81442
9576 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
9579 2017-07-21 Tom de Vries <tom@codesourcery.com>
9582 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9584 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9585 nvptx_override_options_after_change.
9587 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9589 * dwarf2out.c (output_file_names): Avoid double testing for
9592 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9594 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9596 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9598 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9600 (try_crossjump_to_edge): Do not punt on partitioned functions.
9602 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9604 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9605 Put all BBs reachable only via paths crossing cold region to cold
9607 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9609 2016-07-21 Richard Biener <rguenther@suse.de>
9611 PR tree-optimization/81303
9612 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9613 into account prologue and epilogue iterations when raising
9614 min_profitable_iters to sth at least covering one vector iteration.
9616 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9618 * config/arm/arm.c (arm_test_cpu_arch_dat):
9621 2017-07-20 Nathan Sidwell <nathan@acm.org>
9623 Remove TYPE_METHODS.
9624 * tree.h (TYPE_METHODS): Delete.
9625 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9626 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9627 (dbxout_type_methods): Scan TYPE_FIELDS.
9628 (dbxout_type): Don't check TYPE_METHODS here.
9629 * function.c (use_register_for_decl): Always ignore register for
9630 class types when not optimizing.
9631 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9632 * tree.c (free_lang_data_in_type): Stitch out member functions and
9633 templates from TYPE_FIELDS.
9634 (build_distinct_type_copy, verify_type_variant,
9635 verify_type): Member fns are on TYPE_FIELDS.
9636 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9637 * tree-pretty-print.c (dump_generic_node): Likewise.
9639 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9642 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9643 V2TImode and V4TImode.
9644 (ix86_expand_vector_extract): Likewise.
9645 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9646 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9647 (ssescalarmode): Handle V4TImode and V2TImode.
9648 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9649 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9650 (VEXTRACTI128_MODE): New mode iterator.
9651 (splitter for *vec_extractv?ti first element): New.
9652 (VEC_INIT_MODE): New mode iterator.
9653 (vec_init<mode>): Consolidate 3 expanders into one using
9654 VEC_INIT_MODE mode iterator.
9656 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9658 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9659 non_spilled_static_chain_regno_p.
9661 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9663 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9665 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9667 * bb-reorder.c (connect_traces): Allow copying of blocks within
9670 2017-07-20 Richard Biener <rguenther@suse.de>
9672 * gimple.h (gimple_phi_result): Add gphi * overload.
9673 (gimple_phi_result_ptr): Likewise.
9674 (gimple_phi_arg): Likewise. Adjust index assert to only
9675 allow actual argument accesses rather than all slots available
9677 (gimple_phi_arg_def): Add gphi * overload.
9678 * tree-phinodes.c (make_phi_node): Initialize only actual
9680 (resize_phi_node): Clear memory not covered by old node,
9681 do not initialize excess argument slots.
9682 (reserve_phi_args_for_new_edge): Initialize new argument slot
9685 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9687 PR tree-optimization/81388
9689 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9691 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9692 by removing computation of may_be_zero.
9694 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9695 Tom de Vries <tom@codesourcery.com>
9698 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9699 when gimple level profile disagrees with what RTL expander did.
9701 2017-07-20 Richard Biener <rguenther@suse.de>
9703 PR tree-optimization/61171
9704 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9705 (vect_analyze_stmt): Add slp instance parameter.
9706 (vectorizable_reduction): Likewise.
9707 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9708 (vect_is_simple_reduction): Deal with chains not detected
9709 as SLP reduction chain, specifically not properly associated
9710 chains containing a mix of plus/minus.
9711 (get_reduction_op): Remove.
9712 (get_initial_defs_for_reduction): Simplify, pass in whether
9713 this is a reduction chain, pass in the SLP node for the PHIs.
9714 (vect_create_epilog_for_reduction): Get the SLP instance as
9716 (vectorizable_reduction): Get the SLP instance as arg.
9717 During analysis remember the SLP node with the PHIs in the
9718 instance. Simplify getting at the vectorized reduction PHIs.
9719 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9720 through SLP instance.
9721 (vect_slp_analyze_operations): Likewise.
9722 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9723 (vect_transform_stmt): Likewise.
9725 2017-07-20 Tom de Vries <tom@codesourcery.com>
9727 PR tree-optimization/81489
9728 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9729 read of phi arg location to before loop that modifies phi.
9731 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9733 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9736 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9739 * except.c (execute): Fix ordering issue.
9741 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9743 PR rtl-optimization/81423
9744 * combine.c (make_compound_operation_int): Don't try to optimize
9745 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9747 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9749 PR rtl-optimization/81423
9750 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9751 with a constant that is -1 in the truncated to mode.
9753 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9755 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9756 (determine_unlikely_bbs): ... here.
9757 * predict.h (propagate_unlikely_bbs_forward): Declare.
9758 * cfgexpand.c (pass_expand::execute): Use it.
9759 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9761 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9762 propagate_unlikely_bbs_forward.
9764 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9767 * except.c (maybe_add_nop_after_section_switch): New function.
9770 2017-07-19 Tom de Vries <tom@codesourcery.com>
9772 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9774 2017-07-19 Tom de Vries <tom@codesourcery.com>
9776 * gimple.h (gimple_phi_arg): Make assert more strict.
9778 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9780 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9781 * config/rs6000/mmintrin.h: New file.
9782 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9784 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9786 PR tree-optimization/81346
9787 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9789 2017-07-19 Tom de Vries <tom@codesourcery.com>
9791 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9793 2017-07-19 Tom de Vries <tom@codesourcery.com>
9795 * config/nvptx/nvptx-modes.def: Add V2DImode.
9796 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9798 (nvptx_output_mov_insn): Handle lack of mov.b128.
9799 (nvptx_print_operand): Handle 'H' and 'L' codes.
9800 (nvptx_vector_mode_supported): Allow V2DImode.
9801 (nvptx_preferred_simd_mode): New function.
9802 (nvptx_data_alignment): New function.
9803 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9804 nvptx_preferred_simd_mode.
9805 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9807 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9809 2017-07-19 Tom de Vries <tom@codesourcery.com>
9811 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9812 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9813 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9814 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9815 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9816 (mov<VECIM>_insn): New define_insn.
9817 (define_expand "mov<VECIM>): New define_expand.
9819 2017-07-19 Tom de Vries <tom@codesourcery.com>
9821 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9823 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9825 PR tree-optimization/81346
9826 * fold-const.h (fold_div_compare, range_check_type): Declare.
9827 * fold-const.c (range_check_type): New function.
9828 (build_range_check): Use range_check_type.
9829 (fold_div_compare): No longer static, rewritten into
9830 a match.pd helper function.
9831 (fold_comparison): Don't call fold_div_compare here.
9832 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9835 2017-07-19 Nathan Sidwell <nathan@acm.org>
9837 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9838 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9839 * tree.c (find_decls_types_r, verify_type): Use
9840 TYPE_{MIN,MAX}_VALUE_RAW.
9841 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9842 (hash_tree): Likewise.
9843 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9845 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9848 2017-07-18 Tom de Vries <tom@codesourcery.com>
9851 * omp-expand.c (expand_omp_for_static_chunk): Handle
9852 equal-argument loop exit phi.
9854 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9857 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9858 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9859 operand 2 predicate.
9860 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9861 operand 2 predicate.
9862 (ror,rol -> rorx splitters): Use const_int_operand as
9863 operand 2 predicate.
9865 2017-06-18 Richard Biener <rguenther@suse.de>
9867 PR tree-optimization/81410
9868 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9869 the gap in the ! slp_perm SLP case after each group.
9871 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9874 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9877 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9880 * predict.c (set_even_probabilities): Cleanup; do not affect
9881 probabilities that are already known.
9882 (combine_predictions_for_bb): Call even when count is set.
9884 2017-07-18 Nathan Sidwell <nathan@acm.org>
9886 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9889 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9892 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9893 optimization for loop niter analysis.
9895 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9898 * config/avr/avr.c (avr_optimize_casesi): Don't use
9899 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9901 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9903 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9904 body_cost_vec from _vect_peel_extended_info.
9905 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9906 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9909 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9911 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9913 2017-07-18 Richard Biener <rguenther@suse.de>
9915 PR tree-optimization/80620
9916 PR tree-optimization/81403
9917 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9918 info when re-using a VN table entry.
9920 2017-07-18 Richard Biener <rguenther@suse.de>
9922 PR tree-optimization/81418
9923 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9924 vectype_in. Verify that with lane-reducing reduction operations
9925 we have a single def-use cycle.
9927 2017-07-17 Carl Love <cel@us.ibm.com>
9929 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9931 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9932 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9933 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9934 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9935 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9936 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9937 VMULOSW): New enum "unspec" values.
9938 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9939 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9940 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9941 altivec_vmulosw): New patterns.
9942 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9943 VMULOSW): Add definitions.
9944 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9946 * config/alpha/alpha.c: Include predict.h.
9948 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9950 * tree-vrp.c (compare_assert_loc): Fix comparison function
9951 to return predictable results.
9953 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9955 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9958 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9959 * doc/invoke.texi (mexpand-adddi): Update text.
9961 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9963 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9964 that also clobbers the CC register. The old expand code is moved
9966 (*arc_clzsi2): ... here.
9967 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9968 the CC register. The old expand code is moved to ...
9969 (arc_ctzsi2): ... here.
9971 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9973 * config/arc/arc.opt (mindexed-loads): Use initial value
9974 TARGET_INDEXED_LOADS_DEFAULT.
9975 (mauto-modify-reg): Use initial value
9976 TARGET_AUTO_MODIFY_REG_DEFAULT.
9977 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9978 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9979 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9980 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9982 2017-07-17 Martin Liska <mliska@suse.cz>
9985 * opts.c (finish_options): Do not allow -fgnu-tm
9986 w/ -fsanitize={kernel-,}address. Say sorry.
9988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9991 * tree-loop-distribution.c (classify_partition): Only assert on
9992 numer of iterations.
9993 (merge_dep_scc_partitions): Delete prameter. Update function call.
9994 (distribute_loop): Remove code handling loop with unknown niters.
9995 (pass_loop_distribution::execute): Skip loop with unknown niters.
9997 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10000 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10001 function sort_partitions_by_post_order.
10003 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10005 PR tree-optimization/81374
10006 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10007 the max index of basic blocks, rather than number of basic blocks.
10009 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10011 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10013 (arc_legitimate_pic_operand_p): Likewise.
10014 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10016 (arc_needs_pcl_p): Likewise.
10017 (arc_legitimate_pc_offset_p): Likewise.
10018 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10019 function is also used in constrains.md.
10020 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10021 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10022 PLUS. Only return true/false in known cases, otherwise assert.
10023 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10024 is already called in arc_legitimate_constant_p.
10025 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10027 (LEGITIMATE_PIC_OPERAND_P): Use
10028 arc_raw_symbolic_reference_mentioned_p function.
10029 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10034 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10035 Andrew Burgess <andrew.burgess@embecosm.com>
10037 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10038 (arc_return_address_register): New function.
10039 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10040 (arc_handle_fndecl_attribute): Add naked attribute.
10041 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10042 (TARGET_WARN_FUNC_RETURN): Likewise.
10043 (arc_allocate_stack_slots_for_args): New function.
10044 (arc_warn_func_return): Likewise.
10045 (machine_function): Change type fn_type.
10046 (arc_compute_function_type): Consider new naked function type,
10047 change function return type.
10048 (arc_must_save_register): Adapt to handle new
10049 arc_compute_function_type's return type.
10050 (arc_expand_prologue): Likewise.
10051 (arc_expand_epilogue): Likewise.
10052 (arc_return_address_regs): Delete.
10053 (arc_return_address_register): New function.
10054 (arc_epilogue_uses): Use above function.
10055 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10056 (arc_function_type): Change encoding, add naked type.
10057 (ARC_INTERRUPT_P): Change to handle the new encoding.
10058 (ARC_FAST_INTERRUPT_P): Likewise.
10059 (ARC_NORMAL_P): Define.
10060 (ARC_NAKED_P): Likewise.
10061 (arc_compute_function_type): Delete prototype.
10062 * config/arc/arc.md (in_ret_delay_slot): Use
10063 arc_return_address_register function.
10064 (simple_return): Likewise.
10065 (p_return_i): Likewise.
10067 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10069 PR tree-optimization/81428
10070 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10071 can't be built for those types.
10073 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10075 Remove stuff dead since r239246.
10077 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10078 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10079 (avr_inform_devices): Remove dead stuff.
10081 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10083 * config/arm/arm_neon.h: Fix softp typo.
10085 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10087 PR tree-optimization/81365
10088 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10089 aggregate moves onto bb predecessor edges, make sure there are no
10090 loads that could alias the lhs in between the start of bb and the
10093 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10096 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10097 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10098 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10100 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10102 PR tree-optimization/81396
10103 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10104 (init_symbolic_number): Initialize it to 1.
10105 (perform_symbolic_merge): Add n_ops from both operands into the new
10107 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10108 without base_addr as useless if they need more than one operation.
10109 (bswap_replace): Handle !bswap case for NULL base_addr.
10111 2017-07-17 Tom de Vries <tom@codesourcery.com>
10114 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10117 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10119 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10120 conditional builtin define __FIX_LEON3FT_B2BST.
10122 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10124 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10125 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10128 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10130 PR rtl-optimization/81424
10131 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10132 to remove potential trapping from operands if -fnon-call-exceptions.
10134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10136 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10137 profile_proability for scalling.
10138 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10140 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10142 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10144 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10146 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10147 fixpoint arithmetics.
10149 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10151 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10152 fixpoint arithmetics.
10154 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10156 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10157 fixpoint arithmetics.
10159 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10161 * profile-count.h (profile_probability::from_reg_br_prob_note,
10162 profile_probability::to_reg_br_prob_note): New functions.
10163 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10164 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10165 * predict.c (probability_reliable_p): Update.
10166 (edge_probability_reliable_p): Update.
10167 (br_prob_note_reliable_p): Update.
10168 (invert_br_probabilities): Update.
10169 (add_reg_br_prob_note): New function.
10170 (combine_predictions_for_insn): Update.
10171 * asan.c (asan_clear_shadow): Update.
10172 * cfgbuild.c (compute_outgoing_frequencies): Update.
10173 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10174 (update_br_prob_note): Update.
10175 (rtl_verify_edges): Update.
10176 (purge_dead_edges): Update.
10177 (fixup_reorder_chain): Update.
10178 * emit-rtl.c (try_split): Update.
10179 * ifcvt.c (cond_exec_process_insns): Update.
10180 (cond_exec_process_if_block): Update.
10181 (dead_or_predicable): Update.
10182 * internal-fn.c (expand_addsub_overflow): Update.
10183 (expand_neg_overflow): Update.
10184 (expand_mul_overflow): Update.
10185 * loop-doloop.c (doloop_modify): Update.
10186 * loop-unroll.c (compare_and_jump_seq): Update.
10187 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10188 * predict.h: Update.
10189 * reorg.c (mostly_true_jump): Update.
10191 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10192 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10193 * config/arc/arc.c: (emit_unlikely_jump): Update.
10194 * config/arm/arm.c: (emit_unlikely_jump): Update.
10195 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10196 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10197 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10198 (ix86_print_operand): Update.
10199 (ix86_split_fp_branch): Update.
10200 (predict_jump): Update.
10201 * config/ia64/ia64.c (ia64_print_operand): Update.
10202 * config/mmix/mmix.c (mmix_print_operand): Update.
10203 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10204 (rs6000_expand_split_stack_prologue): Update.
10205 * config/rs6000/rs6000.c: Update.
10206 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10207 (s390_expand_vec_movstr): Update.
10208 (s390_expand_cs_tdsi): Update.
10209 (s390_expand_split_stack_prologue): Update.
10210 * config/sh/sh.c (sh_print_operand): Update.
10211 (expand_cbranchsi4): Update.
10212 (expand_cbranchdi4): Update.
10213 * config/sparc/sparc.c (output_v9branch): Update.
10214 * config/spu/spu.c (get_branch_target): Update.
10215 (ea_load_store_inline): Update.
10216 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10217 * config/tilepro/tilepro.c: Update.
10219 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10221 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10222 (gimplify_save_expr): Likewise.
10224 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10226 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10228 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10230 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10232 * timevar.def (TV_IPA_FNSUMMARY): Define.
10234 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10236 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10237 to back store errata sensitive sequence from being generated.
10238 (sqrtdf2_fix): Likewise.
10240 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10242 * tree-ssa-threadupdate.c (compute_path_counts,
10243 update_joiner_offpath_counts): Use profile_probability.
10245 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10248 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10250 * config/arm/arm-c.c (arm_cpu_builtins): Define
10251 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10253 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10255 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10256 array entries to represent __ieee128 versions of the
10257 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10258 scalar_extract_sig, and scalar_insert_exp built-in functions.
10259 (altivec_resolve_overloaded_builtin): Add special case handling
10260 for the __builtin_scalar_insert_exp function, as represented by
10261 the P9V_BUILTIN_VEC_VSIEDP constant.
10262 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10263 exponent support for __ieee128 argument.
10264 (VSESQP): Add scalar extract signature support for __ieee128
10266 (VSTDCNQP): Add scalar test negative support for __ieee128
10268 (VSIEQP): Add scalar insert exponent support for __int128 argument
10269 with __ieee128 result.
10270 (VSIEQPF): Add scalar insert exponent support for __ieee128
10271 argument with __ieee128 result.
10272 (VSTDCQP): Add scalar test data class support for __ieee128
10274 (VSTDCNQP): Add overload support for scalar test negative with
10275 __ieee128 argument.
10276 (VSTDCQP): Add overload support for scalar test data class
10277 __ieee128 argument.
10278 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10279 UNSPEC_VSX_SXSIGDP.
10280 (UNSPEC_VSX_SIEXPQP): New constant.
10281 (xsxexpqp): New insn for VSX scalar extract exponent quad
10283 (xsxsigqp): New insn for VSX scalar extract significand quad
10285 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10286 precision with floating point argument.
10287 (xststdcqp): New expand for VSX scalar test data class quad
10289 (xststdcnegqp): New expand for VSX scalar test negative quad
10291 (xststdcqp): New insn to match expansions for VSX scalar test data
10292 class quad precision and VSX scalar test negative quad precision.
10293 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10294 special case operand checking to enforce that second operand of
10295 VSX scalar test data class with quad precision argument is a 7-bit
10297 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10298 prototypes and descriptions of __ieee128 versions of
10299 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10300 scalar_test_data_class, and scalar_test_neg built-in functions.
10302 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10304 PR tree-optimization/81162
10305 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10306 replace a negate with an add.
10308 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10310 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10312 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10314 * config/arm/arm-c.c (arm_cpu_builtins): Define
10315 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10317 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10319 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10320 (armv8-r): Set ARM Cortex-R52 as default CPU.
10321 * config/arm/arm-tables.opt: Regenerate.
10322 * config/arm/arm-tune.md: Regenerate.
10323 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10325 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10326 extension for -mcpu=cortex-r52.
10328 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10330 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10331 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10332 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10333 (fp-armv8): Define it as FP_ARMv8 only.
10334 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10335 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10337 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10338 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10339 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10340 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10341 than TARGET_FPU_ARMV8.
10342 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10343 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10344 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10346 * config/arm/neon.md (neon_vrint): Likewise.
10347 (neon_vcvt): Likewise.
10348 (neon_<fmaxmin_op><mode>): Likewise.
10349 (<fmaxmin><mode>3): Likewise.
10350 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10351 * config/arm/predicates.md (arm_cond_move_operator): Check against
10352 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10354 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10356 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10357 to top of function.
10359 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10361 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10362 loop in comment with memset.
10364 2017-07-14 Martin Liska <mliska@suse.cz>
10366 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10367 * dwarf2out.c (is_java): Remove the function.
10368 (output_pubname): Remove usage of the function.
10369 (lower_bound_default): Remove usage of DW_LANG_Java.
10370 (gen_compile_unit_die): Likewise.
10371 * gcc.c: Remove compiler defaults for .java and .zip files.
10372 * gimple-expr.c (remove_suffix): Change as there's no longer
10373 extension than 4-letter one.
10374 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10375 (gimplify_save_expr): Likewise.
10376 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10377 as it's possible even for other languages than Java.
10378 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10379 * lto-opts.c (lto_write_options): Remove reference to Java.
10380 * opts.c (strip_off_ending): Update file extension handling.
10381 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10382 * tree-eh.c (lower_resx): Likewise.
10383 * tree.c (free_lang_data_in_type): Remove dead code.
10384 (find_decls_types_r): Likewise.
10385 (build_common_builtin_nodes): Remove Java from a comment.
10386 (verify_type): Remove dead code.
10387 * varasm.c (assemble_external): Remove Java from a comment.
10389 2017-07-14 Martin Liska <mliska@suse.cz>
10391 * opts.c (finish_options): Add quotes.
10392 (common_handle_option): Likewise.
10394 2017-07-14 Martin Liska <mliska@suse.cz>
10396 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10397 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10398 Remove N_SO_PASCAL.
10399 * dwarf2out.c (lower_bound_default): Do not handle
10401 (gen_compile_unit_die): Likewise.
10402 * gcc.c: Remove default extension binding for GNU Pascal.
10403 * stmt.c: Remove Pascal language from a comment.
10404 * xcoffout.c: Likewise.
10406 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10409 * diagnostic-show-locus.c (fixit_cmp): New function.
10410 (layout::layout): Sort m_fixit_hints.
10411 (column_range::column_range): Assert that the values are valid.
10412 (struct char_span): New struct.
10413 (correction::overwrite): New method.
10414 (struct source_line): New struct.
10415 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10416 calls in terms of classes source_line and char_span, and
10417 correction::overwrite.
10418 (selftest::test_overlapped_fixit_printing_2): New function.
10419 (selftest::diagnostic_show_locus_c_tests): Call it.
10421 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10423 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10424 early if there is no lhs.
10426 2017-07-13 Martin Liska <mliska@suse.cz>
10428 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10429 (gen_reference_type_die): Likewise.
10430 * stor-layout.c: Remove Pascal-related comment.
10432 2017-07-13 Martin Liska <mliska@suse.cz>
10434 * opts.c (finish_options): Add quotes to error messages.
10435 (parse_sanitizer_options): Likewise.
10437 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10439 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10441 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10443 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10445 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10447 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10449 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10451 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10454 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10455 provides the hardware capability bits, define the macro
10456 __BUILTIN_CPU_SUPPORTS__.
10457 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10458 if GLIBC does not provide the hardware capability bits. Add a
10459 gcc_unreachable call if the built-in cpu function is neither
10460 __builtin_cpu_is nor __builtin_cpu_supports.
10461 (rs6000_get_function_versions_dispatcher): Change the warning
10462 that an old GLIBC is used which does not export the capability
10463 bits to be an error.
10464 * doc/extend.texi (target_clones attribute): Document the
10465 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10466 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10467 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10468 the macros defined by GCC if the newer GLIBC is available.
10470 2017-07-12 Jeff Law <law@redhat.com>
10472 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10473 remaining includes slightly.
10474 * config/riscv/riscv-builtins.c: Include profile-count.h.
10476 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10479 * config/avr/avr.c (avr_set_current_function): In diagnostic
10480 messages: Quote keywords and (parts of) identifiers.
10481 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10484 2017-07-12 Carl Love <cel@us.ibm.com>
10486 * config/rs6000/rs6000-c.c: Add support for built-in functions
10487 vector bool char vec_revb (vector bool char);
10488 vector bool short vec_revb (vector short char);
10489 vector bool int vec_revb (vector bool int);
10490 vector bool long long vec_revb (vector bool long long);
10491 * doc/extend.texi: Update the built-in documentation file for the
10492 new built-in functions.
10494 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10496 * config/s390/s390.md: Remove movcc splitter.
10498 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10500 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10501 load/store on condition.
10503 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10506 * config/avr/avr.c (avr_encode_section_info)
10507 [progmem && !TREE_READONLY]: Error if progmem object needs
10510 2017-07-11 Michael Collison <michael.collison@arm.com>
10512 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10515 2017-07-11 Carl Love <cel@us.ibm.com>
10517 * config/rs6000/rs6000-c.c: Add support for builtins
10518 vector unsigned int vec_parity_lsbb (vector signed int);
10519 vector unsigned int vec_parity_lsbb (vector unsigned int);
10520 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10521 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10522 vector unsigned long long vec_parity_lsbb (vector signed long long);
10523 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10524 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10525 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10526 * doc/extend.texi: Update the built-in documentation file for the
10527 new built-in functions.
10529 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10531 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10532 (layout::m_primary_loc): New field.
10533 (layout::layout): Initialize new field. Move location filtering
10534 logic from here to...
10535 (layout::maybe_add_location_range): ...this new method. Add
10536 support for filtering to just the lines already specified by other
10538 (layout::will_show_line_p): New method.
10539 (gcc_rich_location::add_location_if_nearby): New method.
10540 (selftest::test_add_location_if_nearby): New test function.
10541 (selftest::diagnostic_show_locus_c_tests): Call it.
10542 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10545 2017-07-11 Tom de Vries <tom@codesourcery.com>
10547 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10548 (bb_first_real_insn): New function.
10549 (nvptx_single): Add extra initialization of broadcasted condition
10552 2017-07-11 Nathan Sidwell <nathan@acm.org>
10554 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10556 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10558 * doc/extend.texi (AVR Function Attributes): Remove weblink to
10559 Binutils doc as TEXI will mess them up.
10560 * doc/invoke.texi (AVR Options): Same here.
10562 2017-07-11 Daniel Cederman <cederman@gaisler.com>
10564 * config/sparc/sparc.opt (mfix-ut700): New option.
10565 (mfix-gr712rc): Likewise.
10566 (sparc_fix_b2bst): New variable.
10567 * doc/invoke.texi (SPARC options): Document them.
10568 (ARM options): Fix warnings.
10569 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
10570 instructions to prevent sequences that can trigger the store-store
10571 errata for certain LEON3FT processors.
10572 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
10573 (sparc_option_override): Set sparc_fix_b2bst appropriately.
10574 * config/sparc/sparc.md (fix_b2bst): New attribute.
10575 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
10577 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
10580 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
10582 (*rsqrtsf2_sse): Ditto.
10584 (div<mode>3): Macroize insn from divdf3 and divsf3
10585 using MODEF mode iterator.
10587 2017-07-10 Martin Sebor <msebor@redhat.com>
10589 PR tree-optimization/80397
10590 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10591 instead of testing for equality to INTEGER_TYPE.
10593 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10595 * config.gcc: Remove uclibc from arc target spec.
10597 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10599 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10601 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10604 * lto-wrapper.c (remove_option): New function.
10605 (merge_and_complain): Merge PIC/PIE options more realistically.
10607 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10609 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10613 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10614 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10615 * config.in: Regenerate.
10616 * configure: Regenerate.
10617 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10618 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10619 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10620 (TARGET_GASISR_PROLOGUES): ...target mask.
10621 * common/config/avr/avr-common.c
10622 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10623 Set -mgas-isr-prologues.
10624 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10625 INSERT_PASS_BEFORE for it.
10626 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10627 * config/avr/avr.c (avr_option_override)
10628 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10629 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10630 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10631 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10632 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10634 (make_avr_pass_pre_proep): New function.
10635 (emit_push_sfr) <treg>: Add argument to function and use it
10636 instead of TMP_REG.
10637 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10638 and set machine->gasisr.yes.
10639 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10640 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10641 __gcc_isr.n_pushed to .L__stack_usage.
10642 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10643 (avr_asm_final_postscan_insn): ...this new static function.
10644 * config/avr/avr.h (machine_function)
10645 <is_no_gccisr, use_L__stack_usage>: New fields.
10646 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10647 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10648 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10649 (gasisr, *gasisr): New expander and insn.
10650 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10651 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10652 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10654 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10656 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10659 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10661 Move jump-tables out of .text again.
10664 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10665 (ASM_OUTPUT_ADDR_VEC): New function.
10666 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10667 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10668 INSN_ADDRESSes as asm comment.
10669 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10670 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10671 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10672 * config/avr/avr.md (*tablejump): Adjust comment.
10673 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10674 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10676 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10677 (avr_output_addr_vec): New proto.
10678 (avr_log_t) <insn_addresses>: New field.
10680 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10683 * config/i386/i386.c (ix86_function_arg_advance): Set
10684 outgoing_args_on_stack to true if there are outgoing arguments
10686 (ix86_function_arg): Likewise.
10687 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10688 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10689 * config/i386/i386.h (machine_function): Add
10690 outgoing_args_on_stack.
10692 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10694 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10695 supporting pthreds.
10696 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10698 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10700 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10701 (REAL_H): Remove $(MACHMODE_H).
10702 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10704 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10705 $(MACHMODE_H) and double-int.h.
10706 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10708 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10711 2017-07-07 Andrew Pinski <apinski@cavium.com>
10713 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10714 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10716 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10718 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10719 Add warning if GCC was not configured to link against a GLIBC that
10720 exports the hardware capability bits.
10721 (make_resolver_func): Make resolver function private and not a
10722 COMDAT function. Create the name with clone_function_name instead
10723 of make_unique_name.
10726 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10727 correct operand in doing the split.
10729 2017-07-07 Carl Love <cel@us.ibm.com>
10731 * config/rs6000/rs6000-c: Add support for built-in function
10732 vector unsigned short vec_pack_to_short_fp32 (vector float,
10734 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10735 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10736 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10737 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10738 (convert_4f32_8i16): Add define_expand.
10739 * doc/extend.texi: Update the built-in documentation file for the
10740 new built-in function.
10742 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10744 * config/sparc/m8.md: New file.
10745 * config/sparc/sparc.md: Include m8.md.
10747 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10749 * config/sparc/sparc.opt: New option -mvis4b.
10750 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10751 (sparc_option_override): Handle VIS4B.
10752 (enum sparc_builtins): Define
10753 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10754 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10755 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10756 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10757 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10758 (check_constant_argument): New function.
10759 (sparc_vis_init_builtins): Define builtins
10760 __builtin_vis_dictunpack{8,16,32},
10761 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10762 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10763 __builtin_vis_fpcmpde{8,16,32}shl and
10764 __builtin_vis_fpcmpur{8,16,32}shl.
10765 (sparc_expand_builtin): Check that the constant operands to
10766 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10767 constant and in range.
10768 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10770 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10771 (SPARC_IMM5_P): Likewise.
10772 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10773 (enabled): Handle vis4b.
10774 (UNSPEC_DICTUNPACK): New unspec.
10775 (UNSPEC_FPCMPSHL): Likewise.
10776 (UNSPEC_FPUCMPSHL): Likewise.
10777 (UNSPEC_FPCMPDESHL): Likewise.
10778 (UNSPEC_FPCMPURSHL): Likewise.
10779 (cpu_feature): New CPU feature `vis4b'.
10780 (dictunpack{8,16,32}): New insns.
10781 (FPCSMODE): New mode iterator.
10782 (fpcscond): New code iterator.
10783 (fpcsucond): Likewise.
10784 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10785 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10786 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10787 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10788 * config/sparc/constraints.md: Define constraints `q' for unsigned
10789 2-bit integer constants and `t' for unsigned 5-bit integer
10791 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10793 (imm5_operand_dictunpack16): Likewise.
10794 (imm5_operand_dictunpack32): Likewise.
10795 (imm2_operand): Likewise.
10796 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10797 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10798 ditunpack* and fpcmp*shl builtins.
10800 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10802 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10803 * config.in: Add HAVE_AS_SPARC6 define.
10804 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10806 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10808 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10809 (CPP_CPU_SPEC): Handle m8.
10810 (ASM_CPU_SPEC): Likewise.
10811 * config/sparc/sparc-opts.h (enum processor_type): Add
10813 * config/sparc/sparc.c (m8_costs): New struct.
10814 (sparc_option_override): Handle TARGET_CPU_m8.
10815 (sparc32_initialize_trampoline): Likewise.
10816 (sparc64_initialize_trampoline): Likewise.
10817 (sparc_issue_rate): Likewise.
10818 (sparc_register_move_cost): Likewise.
10819 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10820 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10821 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10822 (CPP_CPU_SPEC): Handle M8.
10823 (ASM_CPU_SPEC): Likewise.
10824 (AS_M8_FLAG): Define.
10825 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10826 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10827 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10829 * configure: Regenerate.
10830 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10833 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10835 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10837 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10838 ("*movdi_insn_sp32"): Do not set v3pipe.
10839 ("*movsi_insn"): Likewise.
10840 ("*movdi_insn_sp64"): Likewise.
10841 ("*movsf_insn"): Likewise.
10842 ("*movdf_insn_sp32"): Likewise.
10843 ("*movdf_insn_sp64"): Likewise.
10844 ("*zero_extendsidi2_insn_sp64"): Likewise.
10845 ("*sign_extendsidi2_insn"): Likewise.
10846 ("*mov<VM32:mode>_insn"): Likewise.
10847 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10848 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10849 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10850 ("<vlop:code><VL:mode>3"): Likewise.
10851 ("*not_<vlop:code><VL:mode>3"): Likewise.
10852 ("*nand<VL:mode>_vis"): Likewise.
10853 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10854 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10855 ("one_cmpl<VL:mode>2"): Likewise.
10856 ("faligndata<VM64:mode>_vis"): Likewise.
10857 ("alignaddrsi_vis"): Likewise.
10858 ("alignaddrdi_vis"): Likweise.
10859 ("alignaddrlsi_vis"): Likewise.
10860 ("alignaddrldi_vis"): Likewise.
10861 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10862 ("bmaskdi_vis"): Likewise.
10863 ("bmasksi_vis"): Likewise.
10864 ("bshuffle<VM64:mode>_vis"): Likewise.
10865 ("cmask8<P:mode>_vis"): Likewise.
10866 ("cmask16<P:mode>_vis"): Likewise.
10867 ("cmask32<P:mode>_vis"): Likewise.
10868 ("pdistn<P:mode>_vis"): Likewise.
10869 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10871 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10873 * config/sparc/sparc.md ("subtype"): New insn attribute.
10874 ("*wrgsr_sp64"): Set insn subtype.
10875 ("*rdgsr_sp64"): Likewise.
10876 ("alignaddrsi_vis"): Likewise.
10877 ("alignaddrdi_vis"): Likewise.
10878 ("alignaddrlsi_vis"): Likewise.
10879 ("alignaddrldi_vis"): Likewise.
10880 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10881 ("fexpand_vis"): Likewise.
10882 ("fpmerge_vis"): Likewise.
10883 ("faligndata<VM64:mode>_vis"): Likewise.
10884 ("bshuffle<VM64:mode>_vis"): Likewise.
10885 ("cmask8<P:mode>_vis"): Likewise.
10886 ("cmask16<P:mode>_vis"): Likewise.
10887 ("cmask32<P:mode>_vis"): Likewise.
10888 ("fchksm16_vis"): Likewise.
10889 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10890 ("fmean16_vis"): Likewise.
10891 ("fp<plusminus_insn>64_vis"): Likewise.
10892 ("<plusminus_insn>v8qi3"): Likewise.
10893 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10894 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10895 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10896 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10897 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10898 ("*movqi_insn"): Likewise.
10899 ("*movhi_insn"): Likewise.
10900 ("*movsi_insn"): Likewise.
10901 ("movsi_pic_gotdata_op"): Likewise.
10902 ("*movdi_insn_sp32"): Likewise.
10903 ("*movdi_insn_sp64"): Likewise.
10904 ("movdi_pic_gotdata_op"): Likewise.
10905 ("*movsf_insn"): Likewise.
10906 ("*movdf_insn_sp32"): Likewise.
10907 ("*movdf_insn_sp64"): Likewise.
10908 ("*zero_extendhisi2_insn"): Likewise.
10909 ("*zero_extendqihi2_insn"): Likewise.
10910 ("*zero_extendqisi2_insn"): Likewise.
10911 ("*zero_extendqidi2_insn"): Likewise.
10912 ("*zero_extendhidi2_insn"): Likewise.
10913 ("*zero_extendsidi2_insn_sp64"): Likewise.
10914 ("ldfsr"): Likewise.
10915 ("prefetch_64"): Likewise.
10916 ("prefetch_32"): Likewise.
10917 ("tie_ld32"): Likewise.
10918 ("tie_ld64"): Likewise.
10919 ("*tldo_ldub_sp32"): Likewise.
10920 ("*tldo_ldub1_sp32"): Likewise.
10921 ("*tldo_ldub2_sp32"): Likewise.
10922 ("*tldo_ldub_sp64"): Likewise.
10923 ("*tldo_ldub1_sp64"): Likewise.
10924 ("*tldo_ldub2_sp64"): Likewise.
10925 ("*tldo_ldub3_sp64"): Likewise.
10926 ("*tldo_lduh_sp32"): Likewise.
10927 ("*tldo_lduh1_sp32"): Likewise.
10928 ("*tldo_lduh_sp64"): Likewise.
10929 ("*tldo_lduh1_sp64"): Likewise.
10930 ("*tldo_lduh2_sp64"): Likewise.
10931 ("*tldo_lduw_sp32"): Likewise.
10932 ("*tldo_lduw_sp64"): Likewise.
10933 ("*tldo_lduw1_sp64"): Likewise.
10934 ("*tldo_ldx_sp64"): Likewise.
10935 ("*mov<VM32:mode>_insn"): Likewise.
10936 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10937 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10939 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10941 * config/sparc/sparc.md ("type"): New insn type viscmp.
10942 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10944 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10945 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10946 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10947 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10949 ("n7_vis_logical_11cycle"): Likewise.
10950 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10951 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10952 * config/sparc/niagara.md ("niag_vis"): Likewise.
10953 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10954 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10956 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10958 * config/sparc/sparc.md: New instruction type `bmask'.
10959 (bmaskdi_vis): Use the `bmask' type.
10960 (bmasksi_vis): Likewise.
10961 * config/sparc/ultra3.md (us3_array): Likewise.
10962 * config/sparc/niagara7.md (n7_array): Likewise.
10963 * config/sparc/niagara4.md (n4_array): Likewise.
10964 * config/sparc/niagara2.md (niag2_vis): Likewise.
10965 (niag3_vis): Likewise.
10966 * config/sparc/niagara.md (niag_vis): Likewise.
10968 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10970 * ipa-comdats.c: Remove optimize check from gate.
10971 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10972 for functions not optimized.
10973 (ipa_fn_summary_read): Skip optimize check.
10974 (ipa_fn_summary_write): Likewise.
10975 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10977 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10979 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10980 (check_callers): Give up on caller which is not optimized.
10981 (inline_small_functions): Likewise.
10982 (ipa_inline): Do not give up when not optimizing.
10983 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10984 away unoptimizes cdtors.
10985 (whole_program_function_and_variable_visibility): Do
10986 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10987 * ipa.c (process_references): Do not check optimize.
10988 (symbol_table::remove_unreachable_nodes): Update optimize check.
10989 (set_writeonly_bit): Update optimize check.
10990 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10991 (pass_ipa_single_use::gate): Remove.
10993 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10995 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10996 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10997 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10998 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10999 permute_load, permute_store, adjust_extract, adjust_splat,
11000 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11001 replace_swap_with_copy, dump_swap_insn_table,
11002 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11003 recombine_lvx_pattern, recombine_stvx_pattern,
11004 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11005 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11006 to file rs6000-p8swap.c.
11007 * config/rs6000/rs6000-p8swap.c: New file.
11008 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11009 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11010 and rs6000*-*-* targets.
11012 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11014 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11016 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11018 * lto-wrapper.c (merge_and_complain): Do not merge
11019 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11020 fsigned_zeros, ftrapping_math, fwrapv.
11021 (append_compiler_options): Do not track these options.
11022 (append_linker_options): Likewie
11024 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11026 * cgraphunit.c (cgraph_node::finalize_function): When
11027 !flag_toplevel_reorde set no_reorder flag.
11028 (varpool_node::finalize_decl): Likewise.
11029 (symbol_table::compile): Drop no toplevel reorder path.
11031 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11033 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11034 edges; zero probability is not better than uninitialized.
11036 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11038 * asan.h (asan_sanitize_allocas_p): Declare.
11039 * asan.c (asan_sanitize_allocas_p): New function.
11040 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11041 (handle_builtin_alloca): Likewise.
11042 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11043 if !asan_sanitize_allocas_p.
11044 * params.def (asan-instrument-allocas): Add new option.
11045 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11046 * opts.c (common_handle_option): Disable allocas sanitization for
11049 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11051 * asan.c: Include gimple-fold.h.
11052 (get_last_alloca_addr): New function.
11053 (handle_builtin_stackrestore): Likewise.
11054 (handle_builtin_alloca): Likewise.
11055 (asan_emit_allocas_unpoison): Likewise.
11056 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11057 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11058 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11059 (instrument_builtin_call): Pass gimple iterator to
11060 get_mem_refs_of_builtin_call.
11061 (last_alloca_addr): New global.
11062 * asan.h (asan_emit_allocas_unpoison): Declare.
11063 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11064 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11065 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11066 if function calls alloca.
11067 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11068 * gimple-fold.h (replace_call_with_value): Declare.
11069 * internal-fn.c: Include asan.h.
11070 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11071 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11073 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11075 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11076 (C_SELFTEST_FLAGS): New.
11077 (CPP_SELFTEST_FLAGS): New.
11078 (SELFTEST_DEPS): New, from deps of s-selftest.
11079 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11080 (CPP_SELFTEST_DEPS): New.
11081 (selftest): Add dependency on s-selftest-c++.
11082 (s-selftest): Rename to...
11083 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11084 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11085 than SELFTEST_FLAGS.
11086 (selftest-gdb): Rename to...
11087 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11089 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11090 (selftest-valgrind): Rename to...
11091 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11093 (selftest-valgrind): Reintroduce as an alias for
11094 selftest-c-valgrind.
11095 (s-selftest-c++): New.
11096 (selftest-c++-gdb): New.
11097 (selftest-c++-valgrind): New.
11099 2017-07-06 Olivier Hainque <hainque@adacore.com>
11101 * gcc.c (process_command): When deciding if undefined variables
11102 should be ignored when processing specs, accept "gcc -v" as well.
11104 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11106 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11107 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11109 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11111 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11112 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11113 * config/arm/arm-tables.opt: Regenerate.
11114 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11116 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11118 2017-07-06 Carl Love <cel@us.ibm.com>
11120 * ChangeLog: Clean up from mid air collision
11122 2017-07-06 Carl Love <cel@us.ibm.com>
11124 * config/rs6000/rs6000-c.c: Add support for built-in functions
11125 vector signed int vec_subc (vector signed int, vector signed int);
11126 vector signed __int128 vec_subc (vector signed __int128,
11127 vector signed __int128);
11128 vector unsigned __int128 vec_subc (vector unsigned __int128,
11129 vector unsigned __int128);
11130 vector signed int vec_sube (vector signed int, vector signed int,
11131 vector signed int);
11132 vector unsigned int vec_sube (vector unsigned int,
11133 vector unsigned int,
11134 vector unsigned int);
11135 vector signed __int128 vec_sube (vector signed __int128,
11136 vector signed __int128,
11137 vector signed__int128);
11138 vector unsigned __int128 vec_sube (vector unsigned __int128,
11139 vector unsigned __int128,
11140 vector unsigned __int128);
11141 vector signed int vec_subec (vector signed int, vector signed int,
11142 vector signed int);
11143 vector unsigned int vec_subec (vector unsigned int,
11144 vector unsigned int,
11145 vector unsigned int);
11146 vector signed __int128 vec_subec (vector signed __int128,
11147 vector signed __int128,
11148 vector signed__int128);
11149 vector unsigned __int128 vec_subec (vector unsigned __int128,
11150 vector unsigned __int128,
11151 vector unsigned __int128);
11152 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11153 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11154 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11155 BU_ALTIVEC_OVERLOAD_X definitions.
11156 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11157 * doc/extend.texi: Update the built-in documentation file for the new
11158 built-in functions.
11160 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11163 * diagnostic-show-locus.c (layout::layout): Use start and finish
11164 spelling location for the start and finish of each range.
11165 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11166 Add unused aspect param.
11167 * input.c (expand_location_1): Add "aspect" param, and use it
11168 to access the correct part of the location.
11169 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11171 (expand_location_to_spelling_point): Likewise.
11172 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11173 param, and pass it to expand_location_1.
11175 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11177 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11178 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11179 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11180 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11181 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11182 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11183 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11184 _mm_maskz_getmant_ss): New intrinsics.
11185 (__builtin_ia32_getexpss128_mask): Changed to ...
11186 __builtin_ia32_getexpss128_round ... this.
11187 (__builtin_ia32_getexpsd128_mask): Changed to ...
11188 __builtin_ia32_getexpsd128_round ... this.
11189 * config/i386/i386-builtin-types.def
11190 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11191 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11192 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11193 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11194 __builtin_ia32_getmantss_mask_round): New builtins.
11195 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11196 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11197 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11198 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11199 * config/i386/sse.md
11200 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11201 avx512f_sgetexp<mode><mask_scalar_name>
11202 <round_saeonly_scalar_name> ... this.
11203 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11204 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11205 vgetexp<ssescalarmodesuffix>
11206 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11207 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11208 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11209 avx512f_vgetmant<mode><mask_scalar_name>
11210 <round_saeonly_scalar_name> ... this.
11211 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11212 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11213 vgetmant<ssescalarmodesuffix>
11214 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11215 %0<mask_scalar_operand4>, %1, %2
11216 <round_saeonly_scalar_mask_op4>, %3} ... this.
11217 * config/i386/subst.md (mask_scalar_operand4,
11218 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11219 round_saeonly_scalar_nimm_predicate): New subst attributes.
11221 2017-07-06 Julia Koval <julia.koval@intel.com>
11223 * config/i386/i386.c (ix86_erase_embedded_rounding):
11224 Remove code for old rounding pattern.
11226 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11228 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11230 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11232 * doc/sourcebuild.texi (Test Directives, Variants of
11233 dg-require-support): Add documentation for dg-require-stack-check.
11235 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11237 * config/i386/subst.md (mask_scalar, round_scalar,
11238 round_saeonly_scalar): New meta-templates.
11239 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11240 round_scalar_mask_operand3, round_scalar_mask_op3,
11241 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11242 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11243 round_saeonly_scalar_constraint,
11244 round_saeonly_scalar_prefix): New subst attribute.
11245 * config/i386/sse.md
11246 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11247 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11248 <round_scalar_name> ... this.
11249 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11250 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11251 <round_scalar_name> ... this.
11252 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11253 <sse>_vm<code><mode>3<mask_scalar_name>
11254 <round_saeonly_scalar_name> ... this.
11255 (v<plusminus_mnemonic><ssescalarmodesuffix>
11256 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11257 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11258 v<plusminus_mnemonic><ssescalarmodesuffix>
11259 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11260 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11261 (v<multdiv_mnemonic><ssescalarmodesuffix>
11262 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11263 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11264 v<multdiv_mnemonic><ssescalarmodesuffix>
11265 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11266 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11267 (v<maxmin_float><ssescalarmodesuffix>
11268 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11269 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11270 v<maxmin_float><ssescalarmodesuffix>
11271 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11272 %0<mask_scalar_operand3>, %1, %<iptr>2
11273 <round_saeonly_scalar_mask_op3>} ... this.
11275 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11277 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11278 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11280 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11281 Alan Hayward <alan.hayward@arm.com>
11282 David Sherwood <david.sherwood@arm.com>
11284 * combine.c (simplify_if_then_else): Remove "enum" before
11286 * compare-elim.c (can_eliminate_compare): Likewise.
11287 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11289 (aarch64_lookup_simd_builtin_type): Likewise.
11290 (aarch64_simd_builtin_type): Likewise.
11291 (aarch64_init_simd_builtin_types): Likewise.
11292 (aarch64_simd_expand_args): Likewise.
11293 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11295 (aarch64_reverse_mask): Likewise.
11296 (aarch64_simd_emit_reg_reg_move): Likewise.
11297 (aarch64_gen_adjusted_ldpstp): Likewise.
11298 (aarch64_ccmp_mode_to_code): Likewise.
11299 (aarch64_operands_ok_for_ldpstp): Likewise.
11300 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11301 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11303 (aarch64_min_divisions_for_recip_mul): Likewise.
11304 (aarch64_reassociation_width): Likewise.
11305 (aarch64_get_condition_code_1): Likewise.
11306 (aarch64_simd_emit_reg_reg_move): Likewise.
11307 (aarch64_simd_attr_length_rglist): Likewise.
11308 (aarch64_reverse_mask): Likewise.
11309 (aarch64_operands_ok_for_ldpstp): Likewise.
11310 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11311 (aarch64_gen_adjusted_ldpstp): Likewise.
11312 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11314 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11315 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11316 (arm_lookup_simd_builtin_type): Likewise.
11317 (arm_simd_builtin_type): Likewise.
11318 (arm_init_simd_builtin_types): Likewise.
11319 (arm_expand_builtin_args): Likewise.
11320 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11321 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11322 (ft32_setup_incoming_varargs): Likewise.
11323 (ft32_function_arg): Likewise.
11324 (ft32_function_arg_advance): Likewise.
11325 (ft32_pass_by_reference): Likewise.
11326 (ft32_arg_partial_bytes): Likewise.
11327 (ft32_valid_pointer_mode): Likewise.
11328 (ft32_addr_space_pointer_mode): Likewise.
11329 (ft32_addr_space_legitimate_address_p): Likewise.
11330 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11332 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11333 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11334 (iamcu_alignment): Likewise.
11335 (canonicalize_vector_int_perm): Likewise.
11336 (ix86_noce_conversion_profitable_p): Likewise.
11337 (ix86_mpx_bound_mode): Likewise.
11338 (ix86_operands_ok_for_move_multiple): Likewise.
11339 * config/microblaze/microblaze-protos.h
11340 (microblaze_expand_conditional_branch_reg): Likewise.
11341 * config/microblaze/microblaze.c
11342 (microblaze_expand_conditional_branch_reg): Likewise.
11343 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11345 (rs6000_reassociation_width): Likewise.
11346 (rs6000_invalid_binary_op): Likewise.
11347 (fusion_p9_p): Likewise.
11348 (emit_fusion_p9_load): Likewise.
11349 (emit_fusion_p9_store): Likewise.
11350 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11352 (riscv_hard_regno_mode_ok_p): Likewise.
11353 (riscv_address_insns): Likewise.
11354 (riscv_split_symbol): Likewise.
11355 (riscv_legitimize_move): Likewise.
11356 (riscv_function_value): Likewise.
11357 (riscv_hard_regno_nregs): Likewise.
11358 (riscv_expand_builtin): Likewise.
11359 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11360 (riscv_build_integer): Likewise.
11361 (riscv_split_integer): Likewise.
11362 (riscv_legitimate_constant_p): Likewise.
11363 (riscv_cannot_force_const_mem): Likewise.
11364 (riscv_regno_mode_ok_for_base_p): Likewise.
11365 (riscv_valid_base_register_p): Likewise.
11366 (riscv_valid_offset_p): Likewise.
11367 (riscv_valid_lo_sum_p): Likewise.
11368 (riscv_classify_address): Likewise.
11369 (riscv_legitimate_address_p): Likewise.
11370 (riscv_address_insns): Likewise.
11371 (riscv_load_store_insns): Likewise.
11372 (riscv_force_binary): Likewise.
11373 (riscv_split_symbol): Likewise.
11374 (riscv_force_address): Likewise.
11375 (riscv_legitimize_address): Likewise.
11376 (riscv_move_integer): Likewise.
11377 (riscv_legitimize_const_move): Likewise.
11378 (riscv_legitimize_move): Likewise.
11379 (riscv_address_cost): Likewise.
11380 (riscv_subword): Likewise.
11381 (riscv_output_move): Likewise.
11382 (riscv_canonicalize_int_order_test): Likewise.
11383 (riscv_emit_int_order_test): Likewise.
11384 (riscv_function_arg_boundary): Likewise.
11385 (riscv_pass_mode_in_fpr_p): Likewise.
11386 (riscv_pass_fpr_single): Likewise.
11387 (riscv_pass_fpr_pair): Likewise.
11388 (riscv_get_arg_info): Likewise.
11389 (riscv_function_arg): Likewise.
11390 (riscv_function_arg_advance): Likewise.
11391 (riscv_arg_partial_bytes): Likewise.
11392 (riscv_function_value): Likewise.
11393 (riscv_pass_by_reference): Likewise.
11394 (riscv_setup_incoming_varargs): Likewise.
11395 (riscv_print_operand): Likewise.
11396 (riscv_elf_select_rtx_section): Likewise.
11397 (riscv_save_restore_reg): Likewise.
11398 (riscv_for_each_saved_reg): Likewise.
11399 (riscv_register_move_cost): Likewise.
11400 (riscv_hard_regno_mode_ok_p): Likewise.
11401 (riscv_hard_regno_nregs): Likewise.
11402 (riscv_class_max_nregs): Likewise.
11403 (riscv_memory_move_cost): Likewise.
11404 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11405 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11406 (rl78_addr_space_address_mode): Likewise.
11407 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11409 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11410 (rs6000_reassociation_width): Likewise.
11411 (rs6000_invalid_binary_op): Likewise.
11412 (fusion_p9_p): Likewise.
11413 (emit_fusion_p9_load): Likewise.
11414 (emit_fusion_p9_store): Likewise.
11415 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11416 (ok_for_simple_move_operands): Likewise.
11417 (ok_for_simple_move_strict_operands): Likewise.
11418 (ok_for_simple_arith_logic_operands): Likewise.
11419 (visium_legitimize_reload_address): Likewise.
11420 (visium_select_cc_mode): Likewise.
11421 (output_cbranch): Likewise.
11422 (visium_split_double_move): Likewise.
11423 (visium_expand_copysign): Likewise.
11424 (visium_expand_int_cstore): Likewise.
11425 (visium_expand_fp_cstore): Likewise.
11426 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11427 (visium_function_arg): Likewise.
11428 (visium_function_arg_advance): Likewise.
11429 (visium_libcall_value): Likewise.
11430 (visium_setup_incoming_varargs): Likewise.
11431 (visium_legitimate_constant_p): Likewise.
11432 (visium_legitimate_address_p): Likewise.
11433 (visium_legitimize_address): Likewise.
11434 (visium_secondary_reload): Likewise.
11435 (visium_register_move_cost): Likewise.
11436 (visium_memory_move_cost): Likewise.
11437 (prepare_move_operands): Likewise.
11438 (ok_for_simple_move_operands): Likewise.
11439 (ok_for_simple_move_strict_operands): Likewise.
11440 (ok_for_simple_arith_logic_operands): Likewise.
11441 (visium_function_value_1): Likewise.
11442 (rtx_ok_for_offset_p): Likewise.
11443 (visium_legitimize_reload_address): Likewise.
11444 (visium_split_double_move): Likewise.
11445 (visium_expand_copysign): Likewise.
11446 (visium_expand_int_cstore): Likewise.
11447 (visium_expand_fp_cstore): Likewise.
11448 (visium_split_cstore): Likewise.
11449 (visium_select_cc_mode): Likewise.
11450 (visium_split_cbranch): Likewise.
11451 (output_cbranch): Likewise.
11452 (visium_print_operand_address): Likewise.
11453 * expmed.c (flip_storage_order): Likewise.
11454 * expmed.h (emit_cstore): Likewise.
11455 (flip_storage_order): Likewise.
11456 * genrecog.c (validate_pattern): Likewise.
11457 * hsa-gen.c (gen_hsa_addr): Likewise.
11458 * internal-fn.c (expand_arith_overflow): Likewise.
11459 * ira-color.c (allocno_copy_cost_saving): Likewise.
11460 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11461 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11462 (process_invariant_for_inheritance): Likewise.
11463 * lra-eliminations.c (move_plus_up): Likewise.
11464 * omp-low.c (lower_oacc_reductions): Likewise.
11465 * simplify-rtx.c (simplify_subreg): Likewise.
11466 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11467 (TARGET_CHKP_BOUND_MODE): Likewise..
11468 * targhooks.c (default_chkp_bound_mode): Likewise.
11469 (default_setup_incoming_vararg_bounds): Likewise.
11470 * targhooks.h (default_chkp_bound_mode): Likewise.
11471 (default_setup_incoming_vararg_bounds): Likewise.
11472 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11473 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11474 (have_whole_vector_shift): Likewise.
11475 * tree-vect-stmts.c (vectorizable_load): Likewise.
11476 * doc/tm.texi: Regenerate.
11478 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11480 Graceful degrade if Binutils PR21472 is not available.
11483 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11484 .rodata in flash test fails.
11485 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11486 * confgure: Regenerate.
11487 * config.in: Regenerate.
11488 * config/avr/avr.c (avr_asm_named_section)
11489 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11490 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11491 (avr_asm_init_sections): Same.
11493 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11495 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11496 (fma<VH:mode>4_intrinsic): Likewise.
11497 (*fmsub<VCVTF:mode>4): Likewise.
11498 (*fmsub<VH:mode>4_intrinsic): Likewise.
11500 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11503 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11504 Don't depend on "optimize > 0".
11505 (out_movhi_r_mr, out_movqi_mr_r): Same.
11506 (out_movhi_mr_r, out_movqi_r_mr): Same.
11507 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11508 io_address_operand on "optimize > 0".
11510 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11512 * tree-loop-distribution.c: Add general explanantion on the pass.
11513 (generate_loops_for_partition): Mark distributed loop.
11514 (pg_add_dependence_edges): New parameter. Handle alias data
11515 dependence specially and record it in the parameter if asked.
11516 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11517 (init_partition_graph_vertices, add_partition_graph_edge): New.
11518 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11519 (free_partition_graph_vdata, build_partition_graph): New.
11520 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11521 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11522 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11523 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11524 (version_for_distribution_p, finalize_partitions): New.
11525 (distribute_loop): Handle alias data dependence specially. Factor
11526 out loop fusion code as functions and call these functions.
11528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11530 * tree-loop-distribution.c (classify_partition): New parameter and
11531 better handle reduction statement.
11532 (rdg_build_partitions): Revise comment.
11533 (distribute_loop): Compute statements in all partitions and pass it
11534 to classify_partition.
11536 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11538 * tree-loop-distribution.c (enum partition_type): New.
11539 (struct partition): New field type.
11540 (partition_merge_into): Add parameter. Update partition type.
11541 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11542 (build_rdg_partition_for_vertex): Compute partition type.
11543 (rdg_build_partitions): Dump partition type.
11544 (distribute_loop): Update calls to partition_merge_into.
11546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11548 * tree-loop-distribution.c (struct ddr_hasher): New.
11549 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11551 (classify_partition): Call get_data_dependence.
11552 (pg_add_dependence_edges): Ditto.
11553 (distribute_loop): Release data dependence hash table.
11555 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11557 * tree-loop-distribution.c (ref_base_address): Delete.
11558 (similar_memory_accesses): Rename ...
11559 (share_memory_accesses): ... to this. Check if partitions access
11560 the same memory reference.
11561 (distribute_loop): Call share_memory_accesses.
11563 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11565 * tree-loop-distribution.c (struct partition): New field recording
11566 its data reference.
11567 (partition_alloc, partition_free): Init and release data refs.
11568 (partition_merge_into): Merge data refs.
11569 (build_rdg_partition_for_vertex): Collect data refs for partition.
11570 (pg_add_dependence_edges): Change parameters from vector to bitmap.
11572 (distribute_loop): Remve data refs from vertice data of partition
11575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11577 * tree-loop-distribution.c (params.h): Include header file.
11578 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
11579 (datarefs_vec): New global var.
11580 (create_rdg_vertices): Use datarefs_vec directly.
11581 (free_rdg): Don't free data references.
11582 (build_rdg): Update use. Don't free data references.
11583 (distribute_loop): Compute global variable for data references.
11584 Bail out if there are too many data references.
11586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11588 * tree-loop-distribution.c (loop_nest): New global var.
11589 (build_rdg): Use loop directly, rather than loop nest.
11590 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11592 (distribute_loop): Compute global variable loop nest. Update use.
11594 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11596 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11597 (partition_merge_into): New parameter. Dump reason for fusion.
11598 (distribute_loop): Update use of partition_merge_into.
11600 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11602 * tree-loop-distribution.c (bb_top_order_index): New.
11603 (bb_top_order_index_size, bb_top_order_cmp): New.
11604 (stmts_from_loop): Use topological order.
11605 (pass_loop_distribution::execute): Compute and release topological
11606 order for basic blocks.
11608 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11610 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11613 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11615 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11616 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11617 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11618 * internal-fn.def (LOOP_DIST_ALIAS): New.
11619 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11620 (fold_loop_internal_call): ... this.
11621 (vect_loop_dist_alias_call): New function.
11622 (set_uid_loop_bbs): Call fold_loop_internal_call.
11623 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11626 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11629 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11630 Require dead FLAGS_REG at the beginning of a peephole.
11632 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11635 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11636 arguments in the call to __builtin_ia32_sbb_u32.
11637 (_subborrow_u64): Swap _X and _Y arguments in the call to
11638 __builtin_ia32_sbb_u64.
11640 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11643 * tree-vrp.c (compare_assert_loc): Turn into a function template
11644 with stable template parameter. Only test if a->e is NULL,
11645 !a->e == !b->e has been verified already. Use e == NULL or
11646 e != NULL instead of e or ! e tests. If stable is true, don't use
11647 iterative_hash_expr, on the other side allow a or b or both NULL
11648 and sort the NULLs last.
11649 (process_assert_insertions): Sort using compare_assert_loc<false>
11650 instead of compare_assert_loc, later sort using
11651 compare_assert_loc<true> before calling process_assert_insertions_for
11652 in a loop. Use break instead of continue once seen NULL pointer.
11654 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11656 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11657 Cortex-R7 and Cortex-R8 processors.
11659 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11661 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11662 uninitialized while src is not.
11664 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11666 * common/config/arm/arm-common.c: Adjust include path for
11668 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11669 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11670 (arm-cpu-data.h): Likewise.
11671 (arm-cpu-cdata.h): Likewise.
11672 * config/arm/arm-cpu.h: Delete.
11673 * config/arm/arm-cpu-cdata.h: Delete.
11674 * config/arm/arm-cpu-data.h: Delete.
11676 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11678 * config/arm/arm-cpus.in (cortex-a55): New.
11679 (cortex-a75): Likewise.
11680 (cortex-a75.cortex-a55): Likewise.
11681 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11683 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11684 * config/arm/arm-cpu-cdata.h: Regenerate.
11685 * config/arm/arm-cpu-data.h: Regenerate.
11686 * config/arm/arm-cpu.h: Regenerate.
11687 * config/arm/arm-tables.opt: Regenerate.
11688 * config/arm/arm-tune.md: Regenerate.
11690 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11692 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11694 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11696 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11699 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11701 PR tree-optimization/81292
11702 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11703 full_string_p, also call adjust_related_strinfos if the adjustment
11704 is simple, otherwise invalidate related strinfos.
11706 2017-07-04 Martin Liska <mliska@suse.cz>
11709 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11710 newly created variable as DECL_IGNORED_P.
11712 2017-07-04 Martin Liska <mliska@suse.cz>
11715 * ipa-inline.c (inline_small_functions):
11716 Use xstrdup_for_dump.
11718 2017-07-04 Tom de Vries <tom@codesourcery.com>
11720 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11722 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11725 * config/darwin.c (darwin_function_switched_text_sections):
11728 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11730 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11732 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11734 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11736 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11738 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11739 min_profitable_iters, and th as inclusive lower bounds.
11740 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11741 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11742 for min_profitable_iters and min_profitable_estimate.
11743 (vect_transform_loop): Treat th as an inclusive lower bound.
11744 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11746 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11749 * config/darwin.c (darwin_function_switched_text_sections):
11750 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11751 in two pieces, and suppress the use of buf.
11753 2017-07-03 Nathan Sidwell <nathan@acm.org>
11755 * hash-table.h (hash_table_mod1): Fix indentation.
11757 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11759 PR middle-end/81290
11760 * predict.c (force_edge_cold): Be more careful about propagation
11762 * profile-count.h (profile_probability::guessed,
11763 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11765 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11767 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11769 * doc/invoke.texi (rcpc architecture extension): Document it.
11771 2017-07-03 Richard Biener <rguenther@suse.de>
11773 PR tree-optimization/60510
11774 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11775 the scalar reduction PHI and use it.
11776 (vectorizable_reduction): Properly guard the single_defuse_cycle
11777 path for non-SLP reduction chains where we cannot use it.
11778 Rework reduc_def/index and vector type deduction. Rework
11779 vector operand gathering during reduction op code-gen.
11780 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11781 chains dissolve the chain and leave it to non-SLP reduction
11784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11786 * tree-data-ref.h (dr_alignment): Declare.
11787 * tree-data-ref.c (dr_alignment): New function.
11788 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11789 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11791 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11795 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11796 and base_misalignment fields.
11797 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11798 * tree-data-ref.c: Include builtins.h.
11799 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11800 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11801 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11802 * tree-vect-data-refs.c: Include tree-cfg.h.
11803 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11804 fields instead of calculating an alignment here.
11805 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11806 innermost_loop_behavior fields.
11808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11810 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11812 (DR_STEP_ALIGNMENT): New macro.
11813 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11814 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11815 (create_data_ref): Print it.
11816 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11817 to tell whether the step preserves vector (mis)alignment.
11818 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11819 Move the check for an integer step and generalise to all INTEGER_CST.
11820 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11821 Print the outer step alignment.
11823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11825 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11826 with offset_alignment.
11827 (DR_ALIGNED_TO): Delete.
11828 (DR_OFFSET_ALIGNMENT): New macro.
11829 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11830 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11831 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11832 (create_data_ref): Likewise.
11833 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11834 (vect_analyze_data_refs): Likewise.
11835 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11836 creating dummy innermost behavior.
11838 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11840 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11841 with a "innermost_loop_behavior *" and refeence tree.
11842 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11843 (create_data_ref): Update call accordingly.
11844 * tree-predcom.c (find_looparound_phi): Likewise.
11846 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11848 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11849 fields with dr_wrt_vec_loop.
11850 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11851 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11852 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11853 (vect_dr_behavior): New function.
11854 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11855 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11856 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11857 track whether the step preserves the misalignment.
11858 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11859 Use vect_dr_behavior.
11860 (vect_setup_realignment): Update call accordingly.
11861 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11862 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11863 call to vect_create_addr_base_for_vector_ref.
11864 (vect_create_cond_for_align_checks): Likewise.
11865 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11866 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11867 (vect_recog_mask_conversion_pattern): Likewise.
11868 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11869 (new_stmt_vec_info): Remove redundant zeroing.
11871 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11873 * common/config/arm/arm-common.c (arm_be8_option): New function.
11874 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11875 (ISA_ARMv6): Add isa_bit_be8.
11876 * config/arm/arm.h (arm_be8_option): Add prototype.
11877 (BE8_SPEC_FUNCTION): New define.
11878 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11879 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11880 (mlittle-endian): Similarly.
11881 (mbe8, mbe32): New options.
11882 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11883 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11885 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11887 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11889 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11891 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11892 (cleanup_tree_cfg_bb): Use it.
11893 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11895 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11897 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11900 * loop-doloop.c (add_test): Update profile.
11902 2017-07-03 Martin Liska <mliska@suse.cz>
11905 * sanopt.c (rewrite_usage_of_param): New function.
11906 (sanitize_rewrite_addressable_params): Likewise.
11907 (pass_sanopt::execute): Call rewrite_usage_of_param.
11909 2017-07-03 Richard Biener <rguenther@suse.de>
11911 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11912 back to using VIEW_CONVERT_EXPR.
11914 2017-07-03 Martin Liska <mliska@suse.cz>
11917 * doc/extend.texi: Document when a resolver function is
11918 generated for target_clones.
11920 2017-07-03 Martin Liska <mliska@suse.cz>
11922 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11923 and shadow memory of auto variables which are subject of
11924 use-after-scope sanitization.
11925 (asan_expand_mark_ifn): Add do set only when is_poison.
11927 2016-07-03 Richard Biener <rguenther@suse.de>
11929 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11931 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11932 (vectorizable_reduction): Perform reduction PHI creation when
11933 visiting a reduction PHI and adjust and simplify code generation
11934 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11935 (vect_transform_loop): Visit reduction PHIs.
11936 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11937 defs into the SLP tree.
11938 (vect_build_slp_tree): Reduction defs terminate the recursion.
11939 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11941 (vect_get_vec_defs_for_stmt_copy): Export.
11942 (vect_get_vec_defs): Likewise.
11943 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11945 (vect_get_vec_defs_for_stmt_copy): Declare.
11946 (vect_get_vec_defs): Likewise.
11948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11950 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11951 parameter with a "loop" parameter and use it instead of the
11952 loop containing DR_STMT. Don't check simple_iv when doing
11953 BB analysis. Describe the two analysis modes in the comment.
11955 2017-07-03 Tom de Vries <tom@codesourcery.com>
11957 PR tree-optimization/69468
11958 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11959 (find_same_succ_bb): Handle ignore_edge_flags.
11961 2017-07-03 Tom de Vries <tom@codesourcery.com>
11963 PR tree-optimization/81192
11964 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11966 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11968 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11970 2017-07-03 Tom de Vries <tom@codesourcery.com>
11972 PR tree-optimization/81192
11973 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11974 BB_SAME_SUCC (bb) == NULL.
11976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11978 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11981 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11983 * dumpfile.c: Include profile-count.h
11984 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11986 (insert_cond_bb): Update profile.
11987 * tree-cfg.h (insert_cond_bb): Update prototype.
11988 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11989 * tree-dump.c: Do not include tree-cfg.
11991 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11993 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11995 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11997 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12000 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12002 * tree-complex.c (expand_complex_div_wide): update profile.
12004 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12005 Alan Hayward <alan.hayward@arm.com>
12006 David Sherwood <david.sherwood@arm.com>
12008 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12009 (CORETYPES_H): New define.
12010 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12011 (insn-modes-inline.h, s-modes-inline-h): New rules.
12012 (generated_files): Add insn-modes-inline.h.
12013 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12014 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12015 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12016 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12017 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12018 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12019 (build/gencodes.o, build/genconditions.o): Likewise.
12020 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12021 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12022 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12023 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12024 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12025 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12026 * coretypes.h: Include everything up to real.h for generators.
12027 Include insn-modes.h first. Include wide-int-print.h after
12028 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12029 * machmode.h: Don't include insn-modes.h here.
12030 * function-tests.c: Remove includes of signop.h, machmode.h,
12031 double-int.h and wide-int.h.
12033 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12035 * optc-save-gen.awk: Likewise.
12036 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12037 * godump.c: Remove include of wide-int-print.h.
12038 * pretty-print.h: Likewise.
12039 * wide-int-print.cc: Likewise.
12040 * wide-int.cc: Likewise.
12041 * hash-map-tests.c: Remove include of signop.h.
12042 * hash-set-tests.c: Likewise.
12043 * rtl-tests.c: Likewise.
12044 * mkconfig.sh: Remove include of machmode.h.
12045 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12047 (emit_insn_modes_inline_h): ...this new function. Emit the code
12048 into an insn-modes-inline.h header file, adding appropriate
12049 include guards and end comments.
12050 (emit_insn_modes_c_header): Remove include of machmode.h.
12051 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12053 (main): Handle -i flag and call emit_insn_modes_inline_h when
12056 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12058 * tree-ssa-strlen.c (strinfo): Rename the length field to
12059 nonzero_chars. Add a full_string_p field.
12060 (compare_nonzero_chars, zero_length_string_p): New functions.
12061 (get_addr_stridx): Add an offset_out parameter.
12062 Use compare_nonzero_chars.
12063 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12064 (new_strinfo): Update after above changes to strinfo.
12065 (set_endptr_and_length): Set full_string_p.
12066 (get_string_length): Update after above changes to strinfo.
12067 (unshare_strinfo): Update call to new_strinfo.
12068 (maybe_invalidate): Likewise.
12069 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12070 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12071 as a uhwi instead of an shwi. Update after above changes to
12072 strinfo and new_strinfo.
12073 (zero_length_string): Assert that chainsi contains full strings.
12074 Use zero_length_string_p. Update call to new_strinfo.
12075 (adjust_related_strinfos): Update after above changes to strinfo.
12076 Copy full_string_p from origsi.
12077 (adjust_last_stmt): Use zero_length_string_p.
12078 (handle_builtin_strlen): Update after above changes to strinfo and
12079 new_strinfo. Install the lhs as the string length if the previous
12080 entry didn't describe a full string.
12081 (handle_builtin_strchr): Update after above changes to strinfo
12083 (handle_builtin_strcpy): Likewise.
12084 (handle_builtin_strcat): Likewise.
12085 (handle_builtin_malloc): Likewise.
12086 (handle_pointer_plus): Likewise.
12087 (handle_builtin_memcpy): Likewise. Track nonzero characters
12088 that aren't necessarily followed by a nul terminator.
12089 (handle_char_store): Likewise.
12091 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12093 PR tree-optimization/80769
12094 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12095 for malloc and calloc. Document the new invariant that all related
12096 strinfos have delayed lengths or none do.
12097 (verify_related_strinfos): Move earlier in file.
12098 (set_endptr_and_length): New function, split out from...
12099 (get_string_length): ...here. Also set the lengths of related
12101 (zero_length_string): Assert that chainsi has known (rather than
12103 (adjust_related_strinfos): Likewise.
12105 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12107 PR tree-optimization/81136
12108 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12109 assert that two references with the same misalignment have the same
12110 compile-time misalignment if those compile-time misalignments
12113 2017-07-01 Andi Kleen <ak@linux.intel.com>
12115 * print-tree.c (print_node): Print all attributes.
12117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12119 * cfg.c (scale_bbs_frequencies): New function.
12120 * cfg.h (scale_bbs_frequencies): Declare it.
12121 * cfgloopanal.c (single_likely_exit): Cleanup.
12122 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12124 (scale_loop_profile): Likewise.
12125 (loop_version): Likewise.
12126 (create_empty_loop_on_edge): Update.
12127 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12128 scale_loop_frequencies, scale_loop_profile, loopify,
12129 loop_version): Update prototypes.
12130 * modulo-sched.c (sms_schedule): Update.
12131 * predict.c (unlikely_executed_edge_p): Also check probability.
12132 (probably_never_executed_edge_p): Fix typo.
12133 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12134 * tree-parloops.c (gen_parallel_loop): Update.
12135 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12136 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12137 * tree-ssa-loop-split.c (split_loop): Update.
12138 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12139 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12140 (vect_loop_versioning): Update.
12141 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12143 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12145 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12147 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12149 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12150 to keep profile consistent.
12152 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12154 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12155 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12156 * profile-count.h (max_safe_multiplier): Make unsigned.
12157 (profile_count::guessed_zero): New.
12159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12161 * bb-reorder.c (fix_up_crossing_landing_pad,
12162 fix_crossing_conditional_branches): Use make_single_succ_edge
12163 to keep profile consistent.
12165 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12167 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12170 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12173 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12174 the right scopes, make sure cond_jump isn't preserved between multiple
12175 iterations. Search for fallthru edge whenever there are 3+ edges and
12176 use find_fallthru_edge for it.
12178 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12180 Patch by Alexander Monakov <amonakov@ispras.ru>
12181 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12182 probabilities consistently.
12184 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12186 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12187 emit_cmp_and_jump_insns.
12189 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12192 * tree-inline.c (expand_call_inline): Combine profile statuses.
12194 2017-06-30 Andrew Pinski <apinski@cavium.com>
12196 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12197 fold_stmt returned true.
12199 2017-06-30 Nathan Sidwell <nathan@acm.org>
12201 * ggc.h (empty_string): Delete.
12202 * cfgexpand.c (expand_asm_stmt): Use plain "".
12203 * optabs.c (expand_asm_memory_barrier): Likewise.
12204 * stringpool.c (empty_string): Delete.
12205 (digit_vector, digit_string): Delete.
12206 (ggc_alloc_string): Use plain "", don't optimize single digit
12207 strings. Use ggc_alloc_atomic.
12209 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12211 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12212 comparison set and one other set, use the cost of the non-comparison
12215 2017-06-30 Nathan Sidwell <nathan@acm.org>
12217 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12220 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12222 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12223 loops. Remove now unneeded calls to gimple_switch_set_label() that
12224 just set removed labels to NULL_TREE.
12226 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12228 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12229 (set_range_info): ...here. Only call set_range_info_raw if domain
12231 (set_nonzero_bits): Call set_range_info_raw.
12232 * tree-ssanames.h (set_range_info_raw): New.
12234 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12237 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12238 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12239 of nonimmediate_operand and <store_mask_constraint> instead of m
12240 for the input operand. For V8FI iterator, always split if input
12241 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12242 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12243 <store_mask_predicate> instead of register_operand and
12244 <store_mask_constraint> instead of v for the input operand. Make
12245 sure both operands aren't MEMs for if not <mask_applied>.
12247 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12249 * lto-wrapper.c (copy_file) Close both file descriptors before
12252 2017-06-30 Martin Liska <mliska@suse.cz>
12255 * multiple_target.c (create_dispatcher_calls): Make ifunc
12256 also for function that don't have calls or are not referenced.
12258 2017-06-30 Richard Biener <rguenther@suse.de>
12260 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12261 analyze the first scalar stmt. Move vector type computation
12262 for the BB case here from ...
12263 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12264 live operation processing in the SLP case properly.
12266 2017-06-30 Richard Biener <rguenther@suse.de>
12268 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12270 2017-06-30 Martin Liska <mliska@suse.cz>
12273 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12274 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12276 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12278 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12281 2017-06-30 Richard Biener <rguenther@suse.de>
12283 PR tree-optimization/81249
12284 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12285 condition reduction result to original scalar type.
12287 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12289 * profile-count.h (enum profile_quality): Fix typos and whitespace
12292 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12294 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12295 type for branch probabilities.
12297 2017-06-29 Julian Brown <julian@codesourcery.com>
12298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12300 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12301 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12302 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12303 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12305 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12307 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12308 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12309 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12310 CC usage from generic code to here.
12311 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12312 CC usage into the target macros.
12314 2017-06-29 Maya Rashish <coypu@sdf.org>
12316 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12319 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12321 * arm/arm-builtins.c: Include profile-count.h
12322 * except.c (sjlj_emit_function_enter): Use
12323 profile_probability::unlikely.
12325 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12327 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12328 and tocrel_offset be pointer args rather than implicitly using
12330 (legitimate_constant_pool_address_p, rs6000_emit_move,
12331 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12332 tocrel_offset and use in toc_relative_expr_p call.
12333 (print_operand, print_operand_address): Use static tocrel_base_oac
12334 and tocrel_offset_oac.
12335 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12338 2017-06-29 Maya Rashish <coypu@sdf.org>
12340 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12342 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12344 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12345 objects, take into account only the alignment of 'op0' and 'mode1' if
12348 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12350 * ccmp.c (ccmp_tree_comparison_p): New function.
12351 (ccmp_candidate_p): Update to use above function.
12352 (get_compare_parts): New function.
12353 (expand_ccmp_next): Update to use new functions.
12354 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12356 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12357 take mode as argument.
12358 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12359 * expr.c (expand_expr_real_1): Pass mode as argument.
12361 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12363 * combine.c (combine_instructions): Print insns to dump_file, together
12366 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12368 * asan.c (asan_emit_stack_protection): Update.
12369 (create_cond_insert_point): Update.
12370 * auto-profile.c (afdo_propagate_circuit): Update.
12371 * basic-block.h (struct edge_def): Turn probability to
12372 profile_probability.
12373 (EDGE_FREQUENCY): Update.
12374 * bb-reorder.c (find_traces_1_round): Update.
12375 (better_edge_p): Update.
12376 (sanitize_hot_paths): Update.
12377 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12378 (make_single_succ_edge): Update.
12379 (check_bb_profile): Update.
12380 (dump_edge_info): Update.
12381 (update_bb_profile_for_threading): Update.
12382 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12383 probabilitycount to 0.
12384 * cfgbuild.c (compute_outgoing_frequencies): Update.
12385 * cfgcleanup.c (try_forward_edges): Update.
12386 (outgoing_edges_match): Update.
12387 (try_crossjump_to_edge): Update.
12388 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12389 (expand_gimple_tailcall): Update.
12390 (construct_init_block): Use make_single_succ_edge.
12391 (construct_exit_block): Use make_single_succ_edge.
12392 * cfghooks.c (verify_flow_info): Update.
12393 (redirect_edge_succ_nodup): Update.
12394 (split_edge): Update.
12395 (account_profile_record): Update.
12396 * cfgloopanal.c (single_likely_exit): Update.
12397 * cfgloopmanip.c (scale_loop_profile): Update.
12398 (set_zero_probability): Remove.
12399 (duplicate_loop_to_header_edge): Update.
12400 * cfgloopmanip.h (loop_version): Update prototype.
12401 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12402 (force_nonfallthru_and_redirect): Update.
12403 (update_br_prob_note): Update.
12404 (rtl_verify_edges): Update.
12405 (purge_dead_edges): Update.
12406 (rtl_lv_add_condition_to_bb): Update.
12407 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12408 * cgraphunit.c (init_lowered_empty_function): Update.
12409 (cgraph_node::expand_thunk): Update.
12410 * cilk-common.c: Include profile-count.h
12411 * dojump.c (inv): Remove.
12412 (jumpifnot): Update.
12413 (jumpifnot_1): Update.
12414 (do_jump_1): Update.
12416 (do_jump_by_parts_greater_rtx): Update.
12417 (do_compare_rtx_and_jump): Update.
12418 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12419 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12420 * dwarf2cfi.c: Include profile-count.h
12421 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12422 (sjlj_emit_dispatch_table): Likewise.
12423 * explow.c: Include profile-count.h
12424 * expmed.c (emit_store_flag_force): Update.
12425 (do_cmp_and_jump): Update.
12426 * expr.c (compare_by_pieces_d::generate): Update.
12427 (compare_by_pieces_d::finish_mode): Update.
12428 (emit_block_move_via_loop): Update.
12429 (store_expr_with_bounds): Update.
12430 (store_constructor): Update.
12431 (expand_expr_real_2): Update.
12432 (expand_expr_real_1): Update.
12433 * expr.h (try_casesi, try_tablejump): Update prototypes.
12434 * gimple-pretty-print.c (dump_probability): Update.
12435 (dump_profile): New.
12436 (dump_gimple_label): Update.
12437 (dump_gimple_bb_header): Update.
12438 * graph.c (draw_cfg_node_succ_edges): Update.
12439 * hsa-gen.c (convert_switch_statements): Update.
12440 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12441 (find_if_case_1): Update.
12442 (find_if_case_2): Update.
12443 * internal-fn.c (expand_arith_overflow_result_store): Update.
12444 (expand_addsub_overflow): Update.
12445 (expand_neg_overflow): Update.
12446 (expand_mul_overflow): Update.
12447 (expand_vector_ubsan_overflow): Update.
12448 * ipa-cp.c (good_cloning_opportunity_p): Update.
12449 * ipa-split.c (split_function): Use make_single_succ_edge.
12450 * ipa-utils.c (ipa_merge_profiles): Update.
12451 * loop-doloop.c (add_test): Update.
12452 (doloop_modify): Update.
12453 * loop-unroll.c (compare_and_jump_seq): Update.
12454 (unroll_loop_runtime_iterations): Update.
12455 * lra-constraints.c (lra_inheritance): Update.
12456 * lto-streamer-in.c (input_cfg): Update.
12457 * lto-streamer-out.c (output_cfg): Update.
12458 * mcf.c (adjust_cfg_counts): Update.
12459 * modulo-sched.c (sms_schedule): Update.
12460 * omp-expand.c (expand_omp_for_init_counts): Update.
12461 (extract_omp_for_update_vars): Update.
12462 (expand_omp_ordered_sink): Update.
12463 (expand_omp_for_ordered_loops): Update.
12464 (expand_omp_for_generic): Update.
12465 (expand_omp_for_static_nochunk): Update.
12466 (expand_omp_for_static_chunk): Update.
12467 (expand_cilk_for): Update.
12468 (expand_omp_simd): Update.
12469 (expand_omp_taskloop_for_outer): Update.
12470 (expand_omp_taskloop_for_inner): Update.
12471 * omp-simd-clone.c (simd_clone_adjust): Update.
12472 * optabs.c (expand_doubleword_shift): Update.
12473 (expand_abs): Update.
12474 (emit_cmp_and_jump_insn_1): Update.
12475 (expand_compare_and_swap_loop): Update.
12476 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12477 * predict.c (predictable_edge_p): Update.
12478 (edge_probability_reliable_p): Update.
12479 (set_even_probabilities): Update.
12480 (combine_predictions_for_insn): Update.
12481 (combine_predictions_for_bb): Update.
12482 (propagate_freq): Update.
12483 (estimate_bb_frequencies): Update.
12484 (force_edge_cold): Update.
12485 * profile-count.c (profile_count::dump): Add missing space into dump.
12486 (profile_count::debug): Add newline.
12487 (profile_count::differs_from_p): Explicitly convert to unsigned.
12488 (profile_count::stream_in): Update.
12489 (profile_probability::dump): New member function.
12490 (profile_probability::debug): New member function.
12491 (profile_probability::differs_from_p): New member function.
12492 (profile_probability::differs_lot_from_p): New member function.
12493 (profile_probability::stream_in): New member function.
12494 (profile_probability::stream_out): New member function.
12495 * profile-count.h (profile_count_quality): Rename to ...
12496 (profile_quality): ... this one.
12497 (profile_probability): New.
12498 (profile_count): Update.
12499 * profile.c (compute_branch_probabilities): Update.
12500 * recog.c (peep2_attempt): Update.
12501 * sched-ebb.c (schedule_ebbs): Update.
12502 * sched-rgn.c (find_single_block_region): Update.
12503 (compute_dom_prob_ps): Update.
12504 (schedule_region): Update.
12505 * sel-sched-ir.c (compute_succs_info): Update.
12506 * stmt.c (struct case_node): Update.
12507 (do_jump_if_equal): Update.
12508 (get_outgoing_edge_probs): Update.
12509 (conditional_probability): Update.
12510 (emit_case_dispatch_table): Update.
12511 (expand_case): Update.
12512 (expand_sjlj_dispatch_table): Update.
12513 (emit_case_nodes): Update.
12514 * targhooks.c: Update.
12515 * tracer.c (better_p): Update.
12516 (find_best_successor): Update.
12517 * trans-mem.c (expand_transaction): Update.
12518 * tree-call-cdce.c: Update.
12519 * tree-cfg.c (gimple_split_edge): Upate.
12520 (move_sese_region_to_fn): Upate.
12521 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12522 * tree-eh.c (lower_resx): Upate.
12523 (cleanup_empty_eh_move_lp): Upate.
12524 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12525 * tree-inline.c (copy_edges_for_bb): Update.
12526 (copy_cfg_body): Update.
12527 * tree-parloops.c (gen_parallel_loop): Update.
12528 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12529 (gimple_gen_time_profiler): Update.
12530 * tree-ssa-dce.c (remove_dead_stmt): Update.
12531 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12532 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12533 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12534 (unloop_loops): Update.
12535 (try_peel_loop): Update.
12536 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12537 * tree-ssa-loop-split.c (connect_loops): Update.
12538 (split_loop): Update.
12539 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12540 (hoist_guard): Update.
12541 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12542 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12543 (value_replacement): Update.
12544 * tree-ssa-reassoc.c (branch_fixup): Update.
12545 * tree-ssa-tail-merge.c (replace_block_by): Update.
12546 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12547 (create_edge_and_update_destination_phis): Update.
12548 (compute_path_counts): Update.
12549 (recompute_probabilities): Update.
12550 (update_joiner_offpath_counts): Update.
12551 (freqs_to_counts_path): Update.
12552 (duplicate_thread_path): Update.
12553 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12554 (struct switch_conv_info): Update.
12555 (gen_inbound_check): Update.
12556 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12557 (vect_do_peeling): Update.
12558 (vect_loop_versioning): Update.
12559 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12560 (optimize_mask_stores): Update.
12561 * ubsan.c (ubsan_expand_null_ifn): Update.
12562 * value-prof.c (gimple_divmod_fixed_value): Update.
12563 (gimple_divmod_fixed_value_transform): Update.
12564 (gimple_mod_pow2): Update.
12565 (gimple_mod_pow2_value_transform): Update.
12566 (gimple_mod_subtract): Update.
12567 (gimple_mod_subtract_transform): Update.
12568 (gimple_ic): Update.
12569 (gimple_stringop_fixed_value): Update.
12570 (gimple_stringops_transform): Update.
12571 * value-prof.h: Update.
12573 2017-06-29 Carl Love <cel@us.ibm.com>
12575 * config/rs6000/rs6000-c.c: Add support for built-in functions
12576 vector signed int vec_signed (vector float);
12577 vector signed long long vec_signed (vector double);
12578 vector signed int vec_signed2 (vector double, vector double);
12579 vector signed int vec_signede (vector double);
12580 vector signed int vec_signedo (vector double);
12581 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
12582 instruction generator.
12583 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12584 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12585 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12587 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12588 vunsignede_v2df): Add define_expands.
12589 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12590 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12591 VEC_UNSIGNEDO): Add definitions.
12592 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12593 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12594 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12595 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12596 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12597 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12598 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12599 vec_unsignede, vec_unsignedo): Add builtin defines.
12600 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12602 * doc/extend.texi: Update the built-in documentation file for the
12603 new built-in functions.
12605 2017-06-29 Richard Biener <rguenther@suse.de>
12607 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12608 reduction chains to LOOP_VINFO_REDUCTIONS.
12609 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12610 SLP reductions after processing reduction chains.
12612 2017-06-29 Nathan Sidwell <nathan@acm.org>
12614 * builtins.c (fold_builtin_FUNCTION): Use
12615 lang_hooks.decl_printable_name.
12617 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12619 PR middle-end/81194
12620 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12621 with only one label.
12622 * stmt.c (expand_case): Assert NCASES is greater than one.
12624 2017-06-29 Richard Biener <rguenther@suse.de>
12626 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12628 (group_case_labels): Likewise.
12629 (find_taken_edge): Push sanity checking on val to workers...
12630 (find_taken_edge_cond_expr): ... here
12631 (find_taken_edge_switch_expr): ... and here, handle cases
12632 with just a default label.
12633 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12634 (group_case_labels): Likewise.
12635 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12636 group_case_labels does anything cleanup the CFG again.
12638 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12640 PR tree-optimization/81196
12641 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12642 exit condition comparing two IVs.
12644 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12646 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12647 profile to the dummy entry at the end of the list of architectures.
12648 * config/arm/arm-cpu-cdata.h: Regenerated.
12650 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12651 Michael Collison <michael.collison@arm.com>
12654 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12656 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12657 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12658 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12659 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12660 of shift when the shift amount is masked with constant equal to
12661 the size of the mode.
12662 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12665 2017-06-29 Martin Liska <mliska@suse.cz>
12667 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12669 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12672 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12675 (edge_growth_cache_entry::edge_growth_cache_entry): New
12677 (reset_edge_growth_cache): Update to use constructor.
12679 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12681 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12682 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12683 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12685 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12687 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12688 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12690 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12692 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12693 (*-linux-uclibc*): Add t-uclibc tmake_file.
12694 * config/t-musl: New.
12695 * config/t-uclibc: New.
12697 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12699 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12701 (gen_comm_data): Emit architectural setting of arch_prof.
12702 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12704 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12705 (armv8-m.base, armv8-m.main): Likewise.
12706 * arm-protos.h (arm_build_target): Add profile field.
12707 (arch_option): Likewise.
12708 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12710 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12711 arm_active_target.profile.
12713 2017-06-28 Richard Biener <rguenther@suse.de>
12715 PR middle-end/81227
12716 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12717 TYPE_OVERFLOW_WRAPS.
12718 * match.pd (negate_expr_p): Likewise.
12719 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12720 fold_build2, not fold_binary.
12722 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12724 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12725 Convert memory address to Pmode.
12726 (aarch64_print_operand): Assert MEM operands are always Pmode.
12728 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12731 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12732 Remove redundant if.
12733 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12734 * config/arm/aarch-common-protos.h
12735 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12736 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12738 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12741 * multiple_target.c (create_dispatcher_calls): Set the default
12742 clone to be static, not public.
12744 2017-06-28 Richard Biener <rguenther@suse.de>
12746 * tree-vect-loop.c (vectorizable_reduction): Move special
12747 cond reduction IV var creation ...
12748 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12749 parameter. Use STMT_VINFO_VECTYPE.
12750 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12753 2017-06-28 Martin Liska <mliska@suse.cz>
12756 * ipa-visibility.c (non_local_p): Handle visibility.
12758 2017-06-28 Martin Liska <mliska@suse.cz>
12761 * common.opt: Add IntegerRange to various options.
12762 * opt-functions.awk (integer_range_info): New function.
12763 * optc-gen.awk: Add integer_range_info to cl_options struct.
12764 * opts-common.c (decode_cmdline_option): Handle
12765 CL_ERR_INT_RANGE_ARG.
12766 (cmdline_handle_error): Likewise.
12767 * opts.c (print_filtered_help): Show valid interval in
12768 when --help is provided.
12769 * opts.h (struct cl_option): Add range_min and range_max fields.
12770 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12772 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12774 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12775 (x * C EQ/NE y * C): New transformation.
12777 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12779 * genmultilib (combination_space): Accept '+' in option names.
12781 2017-06-28 Martin Liska <mliska@suse.cz>
12784 * asan.c (instrument_derefs): Bail out inner references
12785 that are hard register variables.
12787 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12790 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12791 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12793 2017-06-28 Richard Biener <rguenther@suse.de>
12795 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12796 (vect_get_slp_defs): Adjust.
12797 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12799 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12801 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12802 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12803 (vectorizable_reduction): Adjust.
12804 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12806 (vect_get_slp_defs): Likewise.
12807 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12808 (vectorizable_bswap): Adjust.
12809 (vectorizable_call): Likewise.
12810 (vectorizable_conversion): Likewise.
12811 (vectorizable_assignment): Likewise.
12812 (vectorizable_shift): Likewise.
12813 (vectorizable_operation): Likewise.
12814 (vectorizable_store): Likewise.
12815 (vectorizable_condition): Likewise.
12816 (vectorizable_comparison): Likewise.
12818 2017-06-28 Michael Collison <michael.collison@arm.com>
12821 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12823 (arm_gen_movmemqi): Removed unused variable 'i'.
12824 Convert 'for' loop into 'while' loop.
12825 (arm_expand_prologue): Remove last unnecessary set of insn.
12826 (thumb_pop): Remove unused variable 'pushed_words'.
12827 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12829 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12831 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12832 * config/s390/s390-protos.h: Add prototype of
12833 s390_rel_address_ok_p.
12834 * config/s390/s390.c (s390_got_symbol): New function.
12835 (s390_rel_address_ok_p): New function.
12836 (legitimize_pic_address): Use s390_rel_address_ok_p.
12837 (s390_load_got): Use s390_got_symbol.
12838 (s390_option_override): Issue error if
12839 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12840 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12842 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12844 2017-06-27 Andrew Pinski <apinski@cavium.com>
12846 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12847 (X * copysign (1.0, X)): New pattern.
12848 (X * copysign (1.0, -X)): New pattern.
12849 (copysign (-1.0, CST)): New pattern.
12851 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12853 * genmultilib (combination_space): Remove variable.
12854 Validate reuse rules against regular expression for any sequence
12855 of multilib options in any order.
12857 2017-06-27 Michael Collison <michael.collison@arm.com>
12859 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12860 call aarch64_split_simd_combine.
12861 * (aarch64_combine_internal<mode>): Delete pattern.
12862 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12863 Allow register and subreg operands.
12865 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12867 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12868 specific need, just fallback on defaults.
12869 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12871 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12872 Olivier Hainque <hainque@adacore.com>
12874 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12876 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12877 targets. Pick a default if no particular attempt applied.
12878 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12881 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12883 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12884 (x86_64-wrs-vxworks7): Likewise.
12886 2017-06-27 Marek Polacek <polacek@redhat.com>
12889 * ubsan.c (instrument_null): Check get_base_address's result for null.
12891 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12893 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12895 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12897 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12898 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12899 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12900 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12901 New function types.
12902 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12903 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12904 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12905 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12906 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12907 BUILT_IN_FEUPDATEENV): New builtins.
12908 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12909 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12910 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12911 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12913 (builtin_structptr_types): Adjust size.
12914 * tree.c (builtin_structptr_types): Add four entries.
12916 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12917 Olivier Hainque <hainque@adacore.com>
12919 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12920 (TLS_SYM): New local macro, forcing reference to __tls__ on
12921 link command lines for VxWorks 7 RTPs, triggering initialization
12923 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12924 OS features TLS support, true for RTPs on VxWorks 7.
12925 * config/vxworks.c (vxworks_override_options): Setup emutls
12928 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12930 * predict.c (test_prediction_value_range): Use -1U instead of -1
12931 to avoid narrowing conversion warning.
12932 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12933 to avoid narrowing conversion warning.
12934 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12936 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12938 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12940 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12941 64bit configurations.
12942 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12943 (SIZE_TYPE): Likewise.
12944 * config/vxworks.c (vxworks_emutls_var_fields): Use
12945 long_unsigned_type_node instead of unsigned_type_node as the offset
12946 field type, which is "pointer" mode in emutls.c.
12948 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12951 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12953 PR middle-end/81207
12954 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12955 gimple_vuse copying separately from gimple_vdef copying.
12957 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12959 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12962 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12963 Olivier Hainque <hainque@adacore.com>
12965 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12966 for all vxworks7 targets.
12967 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12968 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12969 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12970 variations for VX6/VX7 and 32/64bits later on in ...
12971 (VXWORKS_LIB_SPEC): Leverage new macros.
12972 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12973 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12975 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12977 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12978 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12980 2017-06-26 Carl Love <cel@us.ibm.com>
12982 * config/rs6000/rs6000-c.c: Add support for built-in functions
12983 vector bool char vec_reve (vector bool char);
12984 vector signed char vec_reve (vector signed char);
12985 vector unsigned char vec_reve (vector unsigned char);
12986 vector bool int vec_reve (vector bool int);
12987 vector signed int vec_reve (vector signed int);
12988 vector unsigned int vec_reve (vector unsigned int);
12989 vector bool long long vec_reve (vector bool long long);
12990 vector signed long long vec_reve (vector signed long long);
12991 vector unsigned long long vec_reve (vector unsigned long long);
12992 vector bool short vec_reve (vector bool short);
12993 vector signed short vec_reve (vector signed short);
12994 vector double vec_reve (vector double);
12995 vector float vec_reve (vector float);
12996 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12997 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12998 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12999 (altivec_vreve): New pattern.
13000 * config/rs6000/altivec.h (vec_reve): New define.
13001 * doc/extend.texi (vec_rev): Update the built-in documentation file
13002 for the new built-in functions.
13004 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13006 PR tree-optimization/71815
13007 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13009 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13011 (slsr_process_phi): Likewise.
13012 (replace_uncond_cands_and_profitable_phis): Don't replace a
13013 multiply candidate with a stride of 1 (copy or cast).
13014 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13016 (lowest_cost_path): Likewise.
13017 (total_savings): Likewise.
13019 2017-06-26 Richard Biener <rguenther@suse.de>
13022 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13023 Use def_builtin_pure for all gather builtins.
13025 2017-06-26 Richard Biener <rguenther@suse.de>
13027 PR tree-optimization/81203
13028 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13029 non-dominating BBs.
13031 2017-06-26 Marek Polacek <polacek@redhat.com>
13034 * doc/invoke.texi: Document -Wmultistatement-macros.
13036 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13038 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13039 arm_neon_ok_no_float_abi effective target.
13041 2017-06-26 Richard Biener <rguenther@suse.de>
13043 PR tree-optimization/80928
13044 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13045 (copy_bbs): Set BB_DUPLICATED flag early.
13046 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13048 (execute_on_shrinking_pred): Likewise.
13049 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13050 BB_DUPLICATED blocks.
13051 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13052 iterate over all PHIs considering removal of *gsi.
13054 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13056 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13059 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13061 * config/rs6000/rs6000-string.c: (expand_block_clear,
13062 do_load_for_compare, select_block_compare_mode,
13063 compute_current_alignment, expand_block_compare,
13064 expand_strncmp_align_check, expand_strn_compare,
13065 expand_block_move, rs6000_output_load_multiple)
13066 Move functions related to string/block move/compare
13067 to a separate file.
13068 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13069 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13070 for this function which is now used in two files.
13071 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13072 * config.gcc: Add rs6000-string.o to extra_objs for
13073 targets powerpc*-*-* and rs6000*-*-*.
13075 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13078 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13079 32-bit, since indexed is not valid for DImode.
13080 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13081 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13082 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13083 (define_peephole2 for Altivec d-form store): Likewise.
13086 * multiple_target.c (create_dispatcher_calls): Only create the
13087 dispatcher call if the function is the default clone of a
13088 versioned function.
13090 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13092 PR middle-end/80902
13093 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13094 a call, force the call to not be a tail call.
13096 2017-06-23 Jeff Law <law@redhat.com>
13098 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13101 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13103 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13104 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13105 handling for early expansion of vector shifts (sl,sr,sra,rl).
13106 (builtin_function_type): Add vector shift right instructions
13107 to the unsigned argument list.
13109 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13111 rtl-optimizatoin/79286
13112 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13113 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13114 trap. PIC register plus a const unspec without offset can never trap.
13116 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13118 * tree.h (builtin_structptr_type): New type.
13119 (builtin_structptr_types): Declare new array.
13120 * tree.c (builtin_structptr_types): New array.
13121 (free_lang_data, build_common_tree_nodes): Use it.
13123 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13126 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13129 2017-06-22 Matt Turner <mattst88@gmail.com>
13131 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13132 Lake models to skylake case. Assume skylake for unknown
13133 models with clflushopt.
13135 2017-06-22 Jeff Law <law@redhat.com>
13137 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13138 frame sizes that do not satisfy aarch64_uimm12_shift.
13140 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13142 * profile-count.h (apply_probability,
13143 apply_scale, probability_in): Fix checks for zero.
13145 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13147 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13148 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13150 2016-06-22 Richard Biener <rguenther@suse.de>
13152 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13153 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13154 REDUC_MAX_EXPR support.
13155 (vectorizable_reduction): Likewise.
13156 (vect_create_epilog_for_reduction): Likewise.
13158 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13160 * match.pd (A / (1 << B) -> A >> B): New.
13161 * generic-match-head.c: Include optabs-tree.h.
13162 * gimple-match-head.c: Likewise.
13163 * optabs-tree.h (target_supports_op_p): New.
13164 * optabs-tree.c (target_supports_op_p): New.
13166 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13168 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13169 $gcc_cv_ld --help output.
13170 (gcc_cv_ld_demangle): Likewise.
13171 (gcc_cv_ld_eh_frame_hdr): Likewise.
13172 (gcc_cv_ld_pie): Likewise.
13173 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13174 (gcc_cv_ld_buildid): Likewise.
13175 (gcc_cv_ld_sysroot): Likewise.
13176 (ld_bndplt_support): Likewise.
13177 (ld_pushpopstate_support): Likewise.
13178 * configure: Regenerate.
13179 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13181 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13184 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13185 operands indexes to avoid gap between operands and match_dups.
13187 2017-06-21 Andrew Pinski <apinski@cavium.com>
13189 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13190 Increment Arith_shift and Arith_shift_reg by 1.
13191 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13193 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13194 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13195 (aarch64_strip_extend): Add new argument and test for it.
13196 (aarch64_cheap_mult_shift_p): New function.
13197 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13198 add a cost if it is true.
13199 Update calls to aarch64_strip_extend.
13200 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13202 2017-06-21 Andrew Pinski <apinski@cavium.com>
13204 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13206 (thunderxt88): Likewise.
13207 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13208 (thunderx_prefetch_tune): New variable.
13209 (thunderx2t99_prefetch_tune): Update for the correct values.
13210 (thunderxt88_tunings): New variable.
13211 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13212 generic_prefetch_tune.
13213 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13215 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13217 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13218 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13219 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13220 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13221 (aarch64_atomic_cas<mode>, GPI): Likewise.
13223 2017-06-21 Martin Liska <mliska@suse.cz>
13225 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13226 statements on cold and hot labels.
13227 * predict.c (tree_estimate_probability_bb): Remove the
13228 prediction from this place.
13230 2017-06-21 Martin Liska <mliska@suse.cz>
13232 PR tree-optimization/79489
13233 * gimplify.c (maybe_add_early_return_predict_stmt): New
13235 (gimplify_return_expr): Call the function.
13236 * predict.c (tree_estimate_probability_bb): Remove handling
13238 * predict.def: Update comment about early return predictor.
13239 * gimple-predict.h (is_gimple_predict): New function.
13240 * predict.def: Change default value of early return to 66.
13241 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13243 * passes.def: Put pass_strip_predict_hints to the beginning of
13246 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13248 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13249 FUNCTION_DECL declarations.
13250 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13252 (dwaf2out_decl): Likewise.
13253 * godump.c (go_early_global_decl): Skip call to the real debug hook
13254 for FUNCTION_DECL declarations.
13255 * passes.c (rest_of_decl_compilation): Skip call to the
13256 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13257 -fdump-go-spec is passed.
13259 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13261 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13262 Reorder for compactness.
13263 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13264 (def_builtin_pure, def_builtin_pure2): New functions.
13265 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13267 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13269 * match.pd (nop_convert): New predicate.
13270 ((A +- CST1) +- CST2): Allow some NOP conversions.
13272 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13275 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13276 with ctors/dtors if GOVD_SHARED is set.
13278 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13280 * config/aarch64/aarch64.md (movti_aarch64):
13281 Emit mov rather than orr.
13282 (movtf_aarch64): Likewise.
13283 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13284 Emit mov rather than orr.
13286 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13288 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13289 Swap alternatives, make integer dup more expensive.
13291 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13293 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13294 Return true for non-tls symbols.
13296 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13298 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13299 (cortex-a75): Likewise.
13300 (cortex-a75.cortex-a55): Likewise.
13301 * config/aarch64/aarch64-tune.md: Regenerate.
13302 * doc/invoke.texi (-mtune): Document new values for -mtune.
13304 2017-06-21 Tom de Vries <tom@codesourcery.com>
13306 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13307 stack_size feature.
13308 (Effective-Target Keywords, Other attributes): Suggest using
13309 dg-add-options stack_size feature to get stack limit in stack_size
13310 effective target documentation.
13312 2017-06-21 Julian Brown <julian@codesourcery.com>
13313 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13315 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13316 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13317 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13319 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13320 attribute type list for neon_multiply.
13321 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13322 attribute type list for neon_multiply.
13323 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13324 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13325 attribute type list for neon_multiply.
13326 * config/arm/types.md (crypto_pmull): Add.
13327 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13328 attribute type list.
13330 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13332 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13335 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13337 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13338 to make sure not to dereference a NULL cost_classes_ptr pointer.
13340 2017-06-20 Carl Love <cel@us.ibm.com>
13342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13343 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13344 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13346 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13347 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13348 VMULOSW): New enum "unspec" values.
13349 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13350 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13351 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13352 altivec_vmulosw): New patterns.
13353 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13354 VMULOSW): Add definitions.
13356 2017-06-20 Julia Koval <julia.koval@intel.com>
13358 * config/i386/i386.c: Fix rounding expand for new pattern.
13359 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13361 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13363 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13364 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13366 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13368 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13371 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13373 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13374 architecture, then by implementer ID.
13375 * config/aarch64/aarch64-tune.md: Regenerate.
13377 2017-06-20 Richard Biener <rguenther@suse.de>
13379 PR middle-end/81097
13380 * fold-const.c (split_tree): Fold to type before negating.
13382 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13384 * diagnostic-show-locus.c
13385 (selftest::test_fixit_deletion_affecting_newline): New function.
13386 (selftest::diagnostic_show_locus_c_tests): Call it.
13388 2017-06-20 Andreas Schwab <schwab@suse.de>
13391 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13394 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13396 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13397 __ARM_FEATURE_COPROC according to support.
13399 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13401 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13402 Rewritten to avoid overflow for > 32-bit pointers.
13405 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13406 by removing enum keyword.
13407 (ubsan_type_descriptor): Likewise. Formatting fix.
13410 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13411 splitter): Require TARGET_SSE2 in the condition.
13413 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13416 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13417 for doing vector set of SFmode on ISA 3.0.
13418 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13419 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13421 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13422 SFmode value into a V4SF variable that was extracted from another
13423 V4SF variable without converting the element to double precision
13424 and back to single precision vector format.
13425 (vsx_insert_extract_v4sf_p9_2): Likewise.
13427 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13429 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13430 in UWHI to avoid undefined overflow.
13433 * ubsan.h (enum ubsan_encode_value_phase): New.
13434 (ubsan_encode_value): Change second argument to
13435 enum ubsan_encode_value_phase with default value of
13436 UBSAN_ENCODE_VALUE_GENERIC.
13437 * ubsan.c (ubsan_encode_value): Change second argument to
13438 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13439 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13440 create_tmp_var_raw instead of create_tmp_var and use a
13442 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13443 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13444 ubsan_encode_value callers.
13447 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13448 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13449 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13451 2017-06-19 Richard Biener <rguenther@suse.de>
13453 PR middle-end/81118
13454 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13455 estimates if we changed anything.
13457 2017-06-19 Richard Biener <rguenther@suse.de>
13459 PR tree-optimization/80887
13460 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13461 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13462 simplified lookups, then reset mprts_hook.
13463 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13465 (try_to_simplify): Likewise.
13467 2017-06-19 Martin Liska <mliska@suse.cz>
13470 * gimplify.c (gimplify_switch_expr):
13471 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13473 2017-06-19 Martin Liska <mliska@suse.cz>
13475 * doc/install.texi: Document that PGO runs in 4 stages.
13477 2017-06-19 Martin Liska <mliska@suse.cz>
13480 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13481 to dispatcher function name.
13482 * multiple_target.c (replace_function_decl): New function.
13483 (create_dispatcher_calls): Redirect both edges and references.
13485 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13487 * profile-count.c (profile_count::dump): Dump quality.
13488 (profile_count::differs_from_p): Update for unsigned val.
13489 * profile-count.h (profile_count_quality): New enum.
13490 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13492 2017-06-19 Richard Biener <rguenther@suse.de>
13494 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13495 struct function as arg.
13496 (estimate_numbers_of_iterations): Export overload with loop arg.
13497 (free_numbers_of_iterations_estimates_loop): Use an overload of
13498 free_numbers_of_iterations_estimates instead.
13499 * tree-cfg.c (remove_bb): Adjust.
13500 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13501 * tree-parloops.c (gen_parallel_loop): Likewise.
13502 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13504 (tree_unroll_loops_completely): Likewise.
13505 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13506 Use an overload instead and export.
13507 (estimated_loop_iterations): Adjust.
13508 (max_loop_iterations): Likewise.
13509 (likely_max_loop_iterations): Likewise.
13510 (estimate_numbers_of_iterations): Take struct function as arg
13512 (loop_exits_before_overflow): Adjust.
13513 (free_numbers_of_iterations_estimates_loop): Use an overload.
13514 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13515 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13517 2017-06-19 Richard Biener <rguenther@suse.de>
13520 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13521 RANGE_EXPR conservatively.
13523 2017-06-16 Carl Love <cel@us.ibm.com>
13525 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13526 definitions for vec_float, vec_float2, vec_floato,
13527 vec_floate built-ins.
13528 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13529 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13531 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13532 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13533 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13534 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13535 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13536 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13537 vec_floato): Add builtin defines.
13538 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13539 Update the built-in documentation file for the new built-in
13542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13544 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13545 (mthumb): Mark as the negative of -marm.
13547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13549 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13551 (ARM Options, -mtune): Document that this accepts the same
13552 extension options as -mcpu.
13553 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13557 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13558 permitted extensions.
13560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13562 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
13563 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
13564 (armv8-m.main): Add option +nodsp.
13565 * config/arm/arm-cpu-cdata.h: Regenerated.
13567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13569 * config/arm/t-fuchsia: New file.
13570 * config.gcc (arm*-*-fuchsia*): Use it.
13572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13574 * config/arm/t-symbian: Rewrite for new option infrastructure.
13576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13578 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
13579 (MULTILIB_REQUIRED): Likewise.
13581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13583 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13584 (MULTILIB_RESUE): Likewise.
13585 (MULTILIB_MATCHES): Likewise.
13586 (MULTLIB_REQUIRED): Likewise.
13588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13590 * config/arm/t-rtems: Rewrite for new option framework.
13592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13594 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13595 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13596 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13597 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13598 * config/arm/t-multilib: ... here.
13599 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13600 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13601 armv7-a and armv8*-a when A-profile libraries have not been built.
13602 * config/arm/t-rmprofile: Rewrite.
13604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13606 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13607 with a backslash. Remove the backslash after substituting unescaped
13609 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13613 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13614 the driver to pass through the default setting of -mfloat-abi.
13615 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13617 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13618 (all_feat_combs): New rule.
13619 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13621 * config/arm/t-aprofile: Rewrite.
13623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13625 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13626 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13627 fpu is not specified by the user/command-line.
13628 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13629 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13630 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13631 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13632 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13633 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13637 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13638 * config/arm/t-arm-elf: Rewritten.
13640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13642 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13643 have some floating-point instructions.
13644 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13645 (TARGET_MAYBE_HARD_FLOAT): New macro.
13646 * config/arm/arm-builtins.c (arm_init_builtins): Use
13647 TARGET_MAYBE_HARD_FLOAT.
13648 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13652 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13653 (configargs.h): Include it.
13654 (arm_print_hint_for_fpu_option): New function.
13655 (arm_parse_fpu_option): New function.
13656 (candidate_extension): New class.
13657 (arm_canon_for_multilib): New function.
13658 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13659 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13660 (ARCH_CANONICAL_SPECS): New macro.
13661 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13665 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13666 are set after handling multilib fragments. Set target_cpu_default2
13669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13671 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13673 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13674 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13676 (check_arch): Likewise.
13677 * config/arm/arm.c (arm_configure_build_target): Handle
13678 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13679 options in the default.
13681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13683 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13684 when an option is an alias of another.
13685 * config/arm/parsecpu.awk (optalias): New parser token.
13686 (gen_comm_data): Mark non-alias options as such. Emit entries
13687 for extension aliases.
13688 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13689 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13690 (armv6kz, armv6zk, armv6t2): Likewise.
13691 (armv7): Make vfpv3-d16 an alias.
13692 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13694 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13695 Sort in canonical order.
13696 (armv8-a): Sort in canonical order.
13697 (armv8.1-a, armv8.2-a): Likewise.
13698 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13700 (cortex-a9): Sort in canonical order.
13701 * config/arm/arm.c (selftests.h): Include it.
13702 (arm_test_cpu_arch_data): New function.
13703 (arm_run_self_tests): New function.
13704 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13705 (targetm): Move declaration to the end of the file.
13706 * arm-cpu-cdata.h: Regenerated.
13708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13710 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13711 call to target_mode_check describing the type of option passed.
13712 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13713 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13714 arm_parse_cpu_option_name to match parameters against list of
13716 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13717 arm_arch_core_flags data structure.
13718 * config/arm/arm-cpu_cdata.h: Regenerated.
13720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13722 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13724 (arm_print_hint_for_cpu_option): Likewise.
13725 (arm_print_hint_for_arch_option): Likewise.
13726 (arm_parse_cpu_option_name): Likewise.
13727 (arm_parse_arch_option_name): Likewise.
13728 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13729 of entries in the all_fpus list.
13730 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13731 (arm_parse_cpu_option_name): Declare.
13732 (arm_parse_arch_option_name): Declare.
13733 (arm_parse_option_features): Declare.
13734 (arm_intialize_isa): Declare.
13735 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13737 (gen_comm_data): ... here. Make definitions non-static.
13738 * config/arm/arm-cpu-data.h: Regenerated.
13739 * config/arm/arm-cpu-cdata.h: Regenerated.
13741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13743 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13744 (cpu_arch_extension): New structure.
13745 (cpu_arch_option, arch_option, cpu_option): New structures.
13746 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13747 architecture types.
13748 (gen_data): Generate new format data tables.
13749 * config/arm/arm.c (cpu_tune): New structure.
13750 (cpu_option, processors): Delete.
13751 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13752 (arm_print_hint_for_cpu_option): ... this and ...
13753 (arm_print_hint_for_arch_option): ... this.
13754 (arm_parse_arch_cpu_name): Delete. Replace with ...
13755 (arm_parse_cpu_option_name): ... this and ...
13756 (arm_parse_arch_option_name): ... this.
13757 (arm_unrecognized_feature): Change type of target parameter to
13759 (arm_parse_arch_cpu_features): Delete. Replace with ...
13760 (arm_parse_option_features): ... this.
13761 (arm_configure_build_target): Rework to use new configuration data
13763 (arm_print_tune_info): Rework for new configuration data tables.
13764 * config/arm/arm-cpu-data.h: Regenerated.
13765 * config/arm/arm-cpu.h: Regenerated.
13767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13769 * Makefile.in (OBJS): Move sbitmap.o from here ...
13770 (OBJS-libcommon): ... to here.
13772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13774 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13775 (ISA_ALL_CRYPTO): New macro.
13776 (ISA_ALL_SIMD): New macro
13777 (ISA_ALL_FP): New macro.
13778 * config/arm/arm.c (fpu_bitlist): Update initializer.
13779 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13781 (arm9e): Add fpu. Add option for nofp
13782 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13783 (arm926ej-s, arm1026ej-s): Likewise.
13784 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13785 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13786 neon-fp16, neon-vfpv4, nofp and nosimd.
13787 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13788 (cortex-a8): Add fpu. Add option for nofp.
13789 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13790 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13791 (cortex-r4f): Add fpu.
13792 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13793 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13795 (cortex-r8): Likewise.
13796 (cortex-m4): Add fpu. Add option for nofp.
13797 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13798 (cortex-a17.cortex-a7): Likewise.
13799 (cortex-a32): Add fpu. Add options for crypto and nofp.
13800 (cortex-a35, cortex-a53): Likewise.
13801 (cortex-a57): Add fpu. Add option for crypto.
13802 (cortex-a72, cortex-a73): Likewise.
13803 (exynos-m1): Likewise.
13804 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13805 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13806 (cortex-m33): Add fpu. Add option for nofp.
13807 * config/arm/arm-cpu-cdata.h: Regenerated
13808 * config/arm/arm-cpu-data.h: Regenerated.
13810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13812 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13813 (armv5te, armv5tej): Likewise.
13814 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13815 (armv7): Add options fp and vfpv3-d16.
13816 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13817 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13819 (armv7ve): Likewise.
13820 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13821 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13822 (armv8-a): Add nocrypto option.
13823 (armv8.1-a, armv8.2-a): Likewise.
13824 (armv8-m.main): add options fp, fp.dp and nofp.
13826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13828 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13830 (armv8-a+crc): Delete.
13831 (armv8.1-a): Add options simd, crypto and nofp.
13832 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13833 (armv8.2-a+fp16): Delete.
13834 (armv8-m.main): Add option dsp.
13835 (armv8-m.main+dsp): Delete.
13836 (cortex-a8): Add fpu. Add nofp option.
13837 (cortex-a9): Add fpu. Add nofp and nosimd options.
13838 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13839 link to main cpu and architecture data structures.
13840 (gen_comm_data): Only put isa attributes from the main architecture
13842 (option): New statement for architecture and CPU entries.
13843 * arm.c (struct cpu_option): New structure.
13844 (struct processors): Add entry for options.
13845 (arm_unrecognized_feature): New function.
13846 (arm_parse_arch_cpu_name): Ignore any characters after the first
13848 (arm_parse_arch_cpu_feature): New function.
13849 (arm_configure_build_target): Separate out any CPU and architecture
13850 features and parse separately. Don't error out if -mfpu=auto is
13851 used with only an architecture string.
13852 (arm_print_asm_arch_directives): New function.
13853 (arm_file_start): Call it.
13854 * config/arm/arm-cpu-cdata.h: Regenerated.
13855 * config/arm/arm-cpu-data.h: Likewise.
13856 * config/arm/arm-tables.opt: Likewise.
13858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13860 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13861 assembler when it is not -mfpu=auto.
13863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13865 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13866 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13867 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13868 (ASM_CPU_SPEC): Rewrite.
13869 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13870 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13871 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13872 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13873 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13874 copied string is NUL-terminated. Also strip any characters prefixed
13876 (arm_rewrite_selected_arch): New function.
13877 (arm_rewrite_march): New function.
13879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13881 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13882 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13883 (march, mcpu, mtune): Convert to string-based options.
13884 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13885 (arm_parse_arch_cpu_name): New function.
13886 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13887 identify selected architecture or CPU.
13888 (arm_option_save): New function.
13889 (TARGET_OPTION_SAVE): Redefine.
13890 (arm_option_restore): Restore string options.
13891 (arm_option_print): Print string options.
13893 2017-06-16 Martin Sebor <msebor@redhat.com>
13895 PR tree-optimization/80933
13896 PR tree-optimization/80934
13897 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13898 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13899 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13900 (gimple_fold_builtin): Call them.
13902 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13904 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13905 as unlikely; update profile.
13907 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13909 * predict.c (force_edge_cold): Handle declaring edges impossible
13912 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13914 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13916 (try_unroll_loop_completely): Fix reporting.
13918 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13920 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13922 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13925 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13926 if given a non-constant argument for an intrinsic which requires a
13929 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13931 * profile.c (compare_freqs): New function.
13932 (branch_prob): Sort edge list.
13933 (find_spanning_tree): Assume that the list is priority sorted.
13935 2017-06-16 Richard Biener <rguenther@suse.de>
13937 PR tree-optimization/81090
13938 * passes.def (pass_record_bounds): Remove.
13939 * tree-pass.h (make_pass_record_bounds): Likewise.
13940 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13941 make_pass_record_bounds): Likewise.
13942 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13943 not free niter estimates at the beginning but at the end.
13944 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13946 2017-06-16 Richard Biener <rguenther@suse.de>
13948 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13949 initializer to workaround ICE in host GCC 4.8.
13951 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13953 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13955 (clone_inlined_nodes): Update.
13957 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13959 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13960 prefetch settings, and enable prefetching by default at -O3.
13962 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13964 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13965 Set flag_prefetch_loop_arrays according to tuning data.
13967 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13969 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13970 New tune structure.
13971 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13972 [Unrelated to main purpose of the patch] Place the pointer field last
13973 to enable type checking errors when tune structure are wrongly merged.
13974 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13975 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13976 (thunderx2t99_prefetch_tune): New tune constants.
13977 (tune_params *_tunings): Update all tunings (no functional change).
13978 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13979 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13980 from tunings structures.
13982 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13985 * ubsan.c (instrument_null): Add T argument, use it instead
13986 of computing it based on IS_LHS.
13987 (instrument_object_size): Likewise.
13988 (pass_ubsan::execute): Adjust instrument_null and
13989 instrument_object_size callers to pass gimple_get_lhs or
13990 gimple_assign_rhs1 result to it. Use instrument_null instead of
13991 calling get_base_address and instrument_mem_ref. Handle
13992 aggregate call arguments for object-size sanitization.
13994 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13996 PR tree-optimization/81089
13997 * tree-vrp.c (is_masked_range_test): Validate operands of
14000 2017-06-15 Martin Sebor <msebor@redhat.com>
14003 * dumpfile.c (dump_register): Avoid calling memset to initialize
14004 a class with a default ctor.
14005 * gcc.c (struct compiler): Remove const qualification.
14006 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14007 * hash-table.h: Ditto.
14008 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14010 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14011 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14013 * params.h (struct param_info): Make struct members non-const.
14014 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14015 with default initialization.
14016 * vec.h (vec_copy_construct, vec_default_construct): New helper
14018 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14019 with vec_copy_construct.
14020 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14021 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14022 * doc/invoke.texi (-Wclass-memaccess): Document.
14024 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14026 * emit-rtl.h (is_leaf): Update comment about local
14027 register allocator.
14029 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14032 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14033 for a variable to have a section before checking if the section has a
14035 Set section to.persistent if persistent attribute is set.
14036 Warn if .persistent attribute is used on an automatic variable.
14038 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14040 PR rtl-optimization/80474
14041 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14043 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14045 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14048 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14050 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14051 (casesi): Emit bounds checking as RTL.
14052 (casesi_internal_mips16_<mode>): Remove bounds checking.
14054 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14056 * config/xtensa/xtensa.c (xtensa_option_override): Append
14057 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14058 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14059 xtensa_doloop_hooks): Define unconditionally.
14060 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14062 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14063 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14064 for it in xtensa_option_override.
14065 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14066 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14068 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14070 * doc/cppopts.texi: Document '-' special value to -MF.
14072 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14074 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14075 (cortex_a53_fconst): Likewise.
14076 (cortex_a53_fpmul): Likewise.
14077 (cortex_a53_f_load_64): Likewise.
14078 (cortex_a53_f_load_many): Likewise.
14079 (cortex_a53_advsimd_alu): Likewise.
14080 (cortex_a53_advsimd_alu_q): Likewise.
14081 (cortex_a53_advsimd_mul): Likewise.
14082 (cortex_a53_advsimd_mul_q): Likewise.
14083 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14084 Add missing fmul, r2f_cvt and fconst cases.
14086 2017-06-14 Richard Biener <rguenther@suse.de>
14088 PR middle-end/81088
14089 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14091 (fold_binary_loc): When associating do not treat pre-existing
14092 TREE_OVERFLOW on literal constants as a reason to allow
14093 TREE_OVERFLOW on associated literal constants.
14095 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14097 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14098 (MASK_FEATURES): New macro.
14099 * config/sparc/sparc.c (sparc_option_override): Remove the special
14100 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14102 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14104 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14105 a division of 0 if non-call exceptions are enabled.
14107 2017-06-14 Andrew Pinski <apinski@cavium.com>
14108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14111 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14112 Improve vector initialization code gen for only variable case.
14114 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14116 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14118 2017-06-14 Richard Biener <rguenther@suse.de>
14120 PR tree-optimization/81083
14121 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14124 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14126 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14127 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14128 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14129 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14130 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14131 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14133 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14135 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14136 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14138 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14140 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14142 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14144 * config/rs6000/t-rtems: Don't handle SPE.
14146 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14148 * config/rs6000/t-linux: Don't handle SPE.
14150 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14152 * config/rs6000/eabispe.h: Delete file.
14154 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14156 * config/rs6000/t-spe: Delete file.
14158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14160 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14161 (rs6000_legitimate_offset_address_p): Return false for anything in
14162 V2SImode or V2SFmode.
14164 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14166 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14167 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14168 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14170 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14171 (rs6000_legitimate_offset_address_p): Ditto.
14172 (rs6000_emit_move): Ditto.
14173 (rs6000_init_builtins): Remove V4HI_type_node.
14175 2017-06-13 Martin Liska <mliska@suse.cz>
14178 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14179 (gate_asan): Likewise.
14180 * asan.h (asan_no_sanitize_address_p): Remove the function.
14181 (sanitize_flags_p): New function.
14182 * builtins.def: Fix coding style.
14183 * common.opt: Use renamed enum value.
14184 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14185 * doc/extend.texi: Document no_sanitize attribute.
14186 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14187 to SANITIZE_UNDEFINED_NONDEFAULT.
14188 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14189 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14190 Use sanitize_flags_p.
14191 * gimplify.c (gimplify_function_tree): Likewise.
14192 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14193 * opts.c (parse_no_sanitize_attribute): New function.
14194 (common_handle_option): Use renamed enum value.
14195 * opts.h (parse_no_sanitize_attribute): Declare.
14196 * tree.c (sanitize_flags_p): New function.
14197 * tree.h: Declared here.
14198 * tsan.c: Use sanitize_flags_p.
14199 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14200 (instrument_mem_ref): Likewise.
14201 (instrument_bool_enum_load): Likewise.
14202 (do_ubsan_in_current_function): Remove the function.
14203 (pass_ubsan::execute): Use sanitize_flags_p.
14204 * ubsan.h: Remove do_ubsan_in_current_function
14205 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14206 (dump_function_to_file): Use it here.
14208 2017-06-13 Martin Jambor <mjambor@suse.cz>
14210 PR tree-optimization/80803
14211 PR tree-optimization/81063
14212 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14213 (propagate_subaccesses_across_link): Enqueue subtree whenever
14214 necessary instead of relying on the caller.
14216 2017-06-13 Martin Jambor <mjambor@suse.cz>
14218 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14219 that have a first_link.
14220 (sort_and_splice_var_accesses): Do not check first_link before
14222 (subtree_mark_written_and_enqueue): Likewise.
14223 (propagate_all_subaccesses): Likewise and do not stop at first
14224 parent with a first_link.
14226 2017-06-13 Martin Jambor <mjambor@suse.cz>
14228 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14231 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14233 * match.pd: New pattern.
14235 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14237 * tree-vrp.c (is_masked_range_test): New function.
14238 (register_edge_assert_for): Determine ranges for
14241 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14243 PR tree-optimization/67328
14244 * fold-const.c (maskable_range_p): New function.
14245 (build_range_check): Generate bittests if possible.
14247 2017-06-13 Martin Liska <mliska@suse.cz>
14249 * gimple-pretty-print.c (dump_probability): Add new argument.
14250 (dump_edge_probability): Dump both probability and count.
14251 (dump_gimple_label): Likewise.
14252 (dump_gimple_bb_header): Likewise.
14254 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14257 * config/avr/avr-devices.c: Fix indentation.
14258 * config/avr/gen-avr-mmcu-specs.c: Dito.
14260 2017-06-13 Richard Biener <rguenther@suse.de>
14262 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14263 instead get vector type from stmt_info.
14264 (vectorizable_reduction): Adjust. Remove dead code.
14266 2017-06-13 Richard Biener <rguenther@suse.de>
14268 PR middle-end/81065
14269 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14270 case of C * (x * C2 + C3).
14271 (fold_addr_of_array_ref_difference): Properly fold index difference.
14273 2017-06-12 David S. Miller <davem@davemloft.net>
14276 * config/sparc/sparc.md (return expander): Emit frame blockage if
14277 function uses alloca.
14279 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14281 * combine.c (make_field_assignment): Check len rather than the mode
14282 precision when calling force_to_mode.
14284 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14286 Support multilibs and devices that see flash in RAM address range.
14289 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14290 (avr_mcu_t) <flash_pm_offset>: New field.
14291 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14292 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14293 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14294 (AVR_TINY_PM_OFFSET): Remove macro.
14295 * config/avr/avr.opt (-mshort-calls): New option.
14296 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14297 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14298 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14299 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14300 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14301 instead of avr_arch->have_jmp_call.
14302 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14303 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14304 avr_arch->flash_pm_offset to define.
14305 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14306 new field flash_pm_offset. Add entry for avrxmega3.
14307 (avr_texinfo): Add entry for avrxmega3.
14308 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14309 attiny212, attiny214,
14310 attiny412, attiny414, attiny416, attiny417,
14311 attiny814, attiny816, attiny817,
14312 attiny1614, attiny1616, attiny1617,
14313 attiny3214, attiny3216, attiny3217.
14314 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14315 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14316 (avr_print_operand_address) [AVR_TINY]: Same.
14317 (avr_asm_init_sections) <readonly_data_section>: Only patch
14318 callback if avr_arch->flash_pm_offset = 0.
14319 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14320 for rodata if avr_arch->flash_pm_offset != 0.
14321 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14322 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14323 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14324 (m_options): Append opt_rcall.
14325 (m_dirnames): Append dir_rcall.
14326 * config/avr/t-multilib: Regenerate.
14328 * configure.ac [target=avr]: Check whether avrxmega3 default
14329 linker description file works as needed.
14330 * configure: Regenerate.
14331 * doc/avr-mmcu.texi: Regenerate.
14332 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14333 <__AVR_ARCH__>: Document avrxmega3 and 103.
14334 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14335 <__AVR_SHORT_CALLS__>: Document it.
14336 <__AVR_PM_BASE_ADDRESS__>: Document it.
14337 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14338 (AVR Variable Attributes) <progmem>: Document this is
14339 not needed for avrxmega3.
14340 (AVR Named Address Spaces) <__flash>: Dito.
14342 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14344 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14346 2017-06-12 Doug Rupp <rupp@adacore.com>
14348 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14349 Append vxworks-stdint.h to the tm_file list.
14350 * config/vxworks-stdint.h: New file.
14352 2017-06-12 Martin Liska <mliska@suse.cz>
14354 PR tree-optimization/81041
14355 * tree-profile.c (gimple_gen_ic_func_profiler):
14356 Create an extra BB in profile-generate
14357 (gimple_gen_time_profiler): Likewise.
14359 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14361 PR tree-optimization/81003
14362 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14363 (update_range_test): Use it instead of force_gimple_operand_gsi.
14365 2017-06-12 Richard Biener <rguenther@suse.de>
14367 PR tree-optimization/81053
14368 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14369 with backedge value not defined in loop. Simplify def stmt
14372 2017-06-11 Tom de Vries <tom@codesourcery.com>
14375 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14377 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14378 nvptx_cannot_force_const_mem.
14380 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14382 * opts.c (finish_options): Move test for flag_split_stack after
14383 it has been initialized.
14385 2017-06-11 Jason Merrill <jason@redhat.com>
14387 * tree.h (id_equal): New.
14388 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14389 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14390 instead of strcmp of IDENTIFIER_POINTER.
14392 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14394 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14395 (mark_all_inlined_calls_cdtor): Fix formating.
14396 (inline_transform): Rescale profile before inlining.
14398 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14400 * cgraph.h (cgraph_edge::clone): Update prototype.
14401 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14402 (cgraph_node::create_clone): Update.
14403 (cgraph_node::create_version_clone): Update.
14404 * tree-inline.c (copy_bb): Update.
14405 (expand_call_inline): Update.
14407 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14409 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14410 factored out from ...
14411 (rs6000_emit_prologue): ... here.
14413 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14415 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14416 factored out from ...
14417 (rs6000_emit_prologue): ... here.
14419 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14421 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14423 (handle_missing_profiles): Fix computation of tp_first_run.
14424 (counts_to_freqs): Do not touch freqs when count is 0.
14426 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14428 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14431 2017-06-10 Tom de Vries <tom@codesourcery.com>
14433 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14434 attributes): Document signal effective target.
14436 2017-06-10 Tom de Vries <tom@codesourcery.com>
14438 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14439 Document effective target stack_size.
14441 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14443 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14444 to the edit_context if they can be auto-applied.
14446 2017-06-9 Ian Lance Taylor <iant@golang.org>
14448 * opts.c (finish_options): If -fsplit-stack, disable implicit
14449 -forder-blocks-and-partition.
14450 * doc/invoke.texi (Optimize Options): Document that when using
14451 -fsplit-stack -forder-blocks-and-partition is not implicitly
14454 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14456 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14457 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14458 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14459 * builtins.def (abort, trap, unreachable): Declare cold.
14460 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14461 * tree-core.h (ECF_COLD): New.
14462 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14463 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14465 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14467 * predict.c (unlikely_executed_stmt_p): Cleanup.
14469 2017-06-09 Richard Biener <rguenther@suse.de>
14471 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14472 model if the ref is always written to.
14474 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14476 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14478 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14480 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14483 2017-06-09 Tom de Vries <tom@codesourcery.com>
14486 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14487 "target cannot support label values" when encountering LABEL_REF.
14489 2017-06-09 Martin Liska <mliska@suse.cz>
14491 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14492 (gimple_gen_ic_func_profiler): Emit direct comparison
14493 of __gcov_indirect_call_callee with NULL.
14494 (gimple_gen_time_profiler): Change probability from
14495 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14497 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14499 * profile.c (edge_gcov_counts): Turn to pointer.
14500 (compute_branch_probabilities, compute_branch_probabilities): Update.
14501 (branch_prob): Do not clear edge_gcov_count.
14502 * profile.h (edge_gcov_counts): Turn to pointer.
14503 (edge_gcov_count): Update.
14505 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14507 * gimple.h (gimple_check_failed): Mark cold.
14509 2017-06-09 Richard Biener <rguenther@suse.de>
14511 PR tree-optimization/66623
14512 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14513 refactor check_reduction into two parts, properly computing
14514 whether we have to check reduction validity for outer loop
14517 2017-06-09 Richard Biener <rguenther@suse.de>
14519 PR tree-optimization/79483
14520 * graphite-scop-detection.c (order): New global.
14521 (get_order): Compute bb to order mapping that satisfies code
14522 generation constraints.
14523 (cmp_pbbs): New helper.
14524 (build_scops): Start domwalk at entry block, sort generated
14527 2017-06-09 Richard Biener <rguenther@suse.de>
14529 PR middle-end/81007
14530 * ipa-polymorphic-call.c
14531 (ipa_polymorphic_call_context::restrict_to_inner_class):
14532 Skip FIELD_DECLs with error_mark_node type.
14533 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14536 2017-06-09 Martin Liska <mliska@suse.cz>
14538 * predict.c (struct branch_predictor): New struct.
14539 (test_prediction_value_range): New test.
14540 (predict_c_tests): New function.
14541 * selftest-run-tests.c (selftest::run_tests): Run the function.
14542 * selftest.h: Declare new tests.
14544 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14547 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14548 gen_add3_insn did not fail.
14549 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14550 r0, construct that number in a temporary reg and add that reg to r0.
14551 If asked to put the result in r0 as well, fail.
14553 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14555 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14556 for early expansion of vec_eqv.
14558 2017-06-08 Jakub Jelinek <jakub@redhat.com>
14560 PR middle-end/81005
14561 * ubsan.c (instrument_null): Avoid pointless code temporary.
14562 (pass_ubsan::execute): Instrument aggregate arguments of calls.
14564 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
14568 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
14571 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
14574 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14576 * predict.c (unlikely_executed_edge_p): Move ahead.
14577 (probably_never_executed_edge_p): Use it.
14579 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14581 PR middle-end/79988
14582 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14583 gimple_call_builtin_p call.
14585 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14587 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14588 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14589 rtl_check_failed_type2, rtl_check_failed_code1,
14590 rtl_check_failed_code2, rtl_check_failed_code_mode,
14591 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14592 rtvec_check_failed_bounds, rtl_check_failed_flag,
14593 _fatal_insn_not_found, _fatal_insn): Likewise.
14594 * tree.h (tree_contains_struct_check_failed,
14595 tree_check_failed, tree_not_check_failed,
14596 tree_class_check_failed, tree_range_check_failed,
14597 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14598 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14599 tree_operand_check_failed, omp_clause_check_failed,
14600 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14603 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14605 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14606 flag_branch_probabilities.
14607 * ipa-inline.c (edge_badness): Likewise.
14608 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14609 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14610 * predict.c (maybe_hot_frequency_p): Likewise.
14611 (probably_never_executed): Likewise.
14612 * sched-ebb.c (schedule_ebbs): Likewise.
14613 * sched-rgn.c (find_single_block_region): Likewise.
14614 * tracer.c (tail_duplicate): Likewise.
14616 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14618 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14619 longer requires x_flag_profile_use.
14621 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14623 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14624 instead of flag_reorder_blocks_and_partition.
14625 * dbxout.c (dbxout_function_end): Likewise.
14626 * dwarf2out.c (gen_subprogram_die): Likewise.
14627 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14628 * hw-doloop.c (reorg_loops): Likewise.
14629 * varasm.c (assemble_start_function,
14630 assemble_end_function): Likewise.
14631 (decide_function_section): Do not check for
14632 flag_reorder_blocks_and_partition.
14634 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14636 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14638 (chkp_get_hard_register_fake_addr_expr): Ditto.
14639 (chkp_build_addr_expr): Add check for hard reg case.
14640 (chkp_parse_array_and_component_ref): Ditto.
14641 (chkp_find_bounds_1): Ditto.
14642 (chkp_process_stmt): Don't generate bounds store for
14645 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14647 * predict.c (maybe_hot_bb_p): Do not check profile status.
14648 (maybe_hot_edge_p): Likewise.
14649 (probably_never_executed): Check for zero counts even if profile
14651 (unlikely_executed_edge_p): New function.
14652 (unlikely_executed_stmt_p): New function.
14653 (unlikely_executed_bb_p): New function.
14654 (set_even_probabilities): Use unlikely predicates.
14655 (combine_predictions_for_bb): Likewise.
14656 (predict_paths_for_bb): Likewise.
14657 (predict_paths_leading_to_edge): Likewise.
14658 (determine_unlikely_bbs): New function.
14659 (estimate_bb_frequencies): Use it.
14660 (compute_function_frequency): Use zero counts even if profile is
14662 * profile-count.h: Fix typo.
14664 2017-08-08 Julia Koval <julia.koval@intel.com>
14666 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14667 _mm512_mask_cvtsepi16_storeu_epi8,
14668 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14669 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14670 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14671 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14672 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14673 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14674 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14675 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14676 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14677 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14678 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14679 __builtin_ia32_pmovuswb256mem_mask,
14680 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14681 __builtin_ia32_pmovwb512mem_mask): New builtins.
14683 2017-08-08 Julia Koval <julia.koval@intel.com>
14685 PR target/73350,80862
14686 * config/i386/subst.md (round): Fix round pattern.
14687 * config/i386/i386.c (ix86_erase_embedded_rounding):
14688 Fix erasing rounding for the fixed pattern.
14690 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14692 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14694 2017-06-08 Martin Liska <mliska@suse.cz>
14696 PR gcov-profile/80911
14697 * gcov.c (block_info::block_info): New constructor.
14699 2017-06-07 Carl Love <cel@us.ibm.com>
14701 * config/rs6000/rs6000-c: The return type of the following
14702 built-in functions was implemented as int not long long. Fix sign
14703 of return value for the unsigned version of vec_mulo and vec_mule.
14704 vector unsigned long long vec_bperm (vector unsigned long long,
14705 vector unsigned char)
14706 vector signed long long vec_mule (vector signed int,
14708 vector unsigned long long vec_mule (vector unsigned int,
14709 vector unsigned int)
14710 vector signed long long vec_mulo (vector signed int,
14712 vector unsigned long long vec_mulo (vector unsigned int,
14713 vector unsigned int)
14714 * doc/extend.texi: Fix the documentation for the built-in
14717 2017-06-07 Carl Love <cel@us.ibm.com>
14720 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14723 2017-06-07 Carl Love <cel@us.ibm.com>
14725 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14726 support, Generate doublehv for signed int/float for BE case only.
14728 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14730 * doc/invoke.texi (mcx16): Rewrite.
14732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14734 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14735 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14736 *mov<mode>_softfloat, and an anonymous splitter): Use
14737 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14739 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14741 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14743 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14744 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14745 (rs6000_debug_reg_global): Adjust.
14746 (rs6000_init_hard_regno_mode_ok): Adjust.
14747 (rs6000_dbx_register_number): Adjust.
14748 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14749 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14750 Remove SPE_ACC and SPEFSCR.
14751 (REG_ALLOC_ORDER): Ditto.
14752 (FRAME_POINTER_REGNUM): Change to 111.
14753 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14754 (REG_CLASS_NAMES): Ditto.
14755 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14756 (REGISTER_NAMES): Ditto.
14757 (ADDITIONAL_REG_NAMES): Ditto.
14758 (rs6000_reg_names): Ditto.
14759 * config/rs6000/rs6000.md: Renumber some register number
14762 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14764 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14766 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14767 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14769 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14770 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14771 Delete the SPE high registers.
14772 (REG_ALLOC_ORDER): Ditto.
14773 (enum reg_class): Remove SPE_HIGH_REGS.
14774 (REG_CLASS_NAMES): Ditto.
14775 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14776 (REGISTER_NAMES): Ditto.
14777 (rs6000_reg_names): Ditto.
14778 * doc/tm.texi.in: Remove SPE as example.
14779 * doc/tm.texi: Regenerate.
14781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14783 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14784 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14785 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14786 * config/rs6000/rs6000.md (type): Remove "brinc".
14788 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14790 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14791 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14792 * config/rs6000/linuxspe.h: Delete file.
14793 * config/rs6000/rs6000.md: Don't include spe.md.
14794 * config/rs6000/spe.h: Delete file.
14795 * config/rs6000/spe.md: Delete file.
14796 * config/rs6000/t-rs6000: Remove spe.md.
14798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14800 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14801 (reg_or_none500mem_operand): Delete.
14802 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14803 instead of reg_or_none500mem_operand.
14805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14808 handling of SPE flags.
14809 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14811 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14813 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14815 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14816 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14817 paired_divv2sf3): Similar.
14818 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14819 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14820 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14822 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14823 Rename the paired_* instruction patterns.
14824 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14826 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14827 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14828 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14829 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14830 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14831 PAIRED_VECTOR_MODE.
14832 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14833 (spe_func_has_64bit_regs_p): Delete.
14834 (spe_expand_predicate_builtin): Delete.
14835 (spe_expand_evsel_builtin): Delete.
14836 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14837 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14838 (invalid_e500_subreg): Delete.
14839 (rs6000_legitimize_address): Always force_reg op2 as well, for
14840 paired single memory accesses.
14841 (rs6000_member_type_forces_blk): Delete.
14842 (rs6000_spe_function_arg): Delete.
14843 (rs6000_expand_unop_builtin): Delete SPE handling.
14844 (rs6000_expand_binop_builtin): Ditto.
14845 (spe_expand_stv_builtin): Delete.
14846 (bdesc_2arg_spe): Delete.
14847 (spe_expand_builtin): Delete.
14848 (spe_expand_predicate_builtin): Delete.
14849 (spe_expand_evsel_builtin): Delete.
14850 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14851 (spe_init_builtins): Delete.
14852 (spe_func_has_64bit_regs_p): Delete.
14853 (savres_routine_name): Delete "info" parameter. Adjust callers.
14854 (rs6000_emit_stack_reset): Ditto.
14855 (rs6000_dwarf_register_span): Delete.
14856 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14857 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14858 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14860 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14862 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14863 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14864 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14865 mulv2sf3, divv2sf3): Delete expanders.
14867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14869 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14871 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14873 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14874 * config/rs6000/rs6000.c: Ditto.
14876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14878 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14879 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14880 comparison_operator.
14882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14884 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14885 * config/rs6000/rs6000.opt: Ditto.
14886 * config/rs6000/t-rtems: Ditto.
14888 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14890 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14891 TARGET_E500_SINGLE by 0, simplify.
14892 * config/rs6000/rs6000.c: Ditto.
14893 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14894 (spe_build_register_parallel): Delete.
14895 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14896 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14897 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14898 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14899 (E500_CONVERT): Delete.
14900 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14904 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14905 * config/rs6000/dfp.md: Ditto.
14906 (negdd2, *negdd2_fpr): Merge.
14907 (absdd2, *absdd2_fpr): Merge.
14908 (negtd2, *negtd2_fpr): Merge.
14909 (abstd2, *abstd2_fpr): Merge.
14910 * config/rs6000/e500.h: Delete file.
14911 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14912 TARGET_FPRS by 1 and simplify.
14913 * config/rs6000/rs6000-c.c: Ditto.
14914 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14915 TARGET_DF_SPE by 0.
14916 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14918 * config/rs6000/rs6000.md: Ditto.
14919 (floatdidf2, *floatdidf2_fpr): Merge.
14920 (move_from_CR_gt_bit): Delete.
14921 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14922 (E500_CR_IOR_COMPARE): Delete.
14923 (All patterns that require !TARGET_FPRS): Delete.
14924 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14926 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14928 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14930 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14932 * graphds.c (add_edge): Intitialize edge's attached data.
14933 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14934 pointer parameter. Call pointed function on each edge during
14935 graph traversing. Skip traversing the edge when the function
14937 (graphds_dfs, graphds_scc): Ditto.
14938 (for_each_edge): New parameter. Pass the new parameter to callback
14940 * graphds.h (skip_edge_callback): New function pointer type.
14941 (graphds_dfs, graphds_scc): New function pointer parameter.
14942 (graphds_edge_callback, for_each_edge): New parameter.
14944 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14946 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14947 out code checking if runtime alias check is possible to below ...
14948 Call the new function.
14949 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14950 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14952 2017-06-07 Marek Polacek <polacek@redhat.com>
14955 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14956 TYPE_OVERFLOW_WRAPS check.
14958 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14960 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14961 if versioning is required.
14962 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14963 peeling with the check for versioning.
14965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14967 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14968 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14969 Set true to new parameter if new ssa variable is defined.
14970 (vect_gen_vector_loop_niters): Refactor. Set range information
14971 for the new vector loop bound variable.
14972 (vect_do_peeling): Ditto.
14974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14976 * tree-affine.c (ssa.h): Include header file.
14977 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14978 has wrapping overflow behavior.
14980 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14982 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14984 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14986 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14987 (tree_to_aff_combination): ... here.
14989 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14991 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14992 reg_pressure model function.
14993 (ivopts_global_cost_for_size): Delete.
14994 (determine_set_costs, iv_ca_recount_cost): Call new model function
14995 ivopts_estimate_reg_pressure.
14997 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14999 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15000 expensive than udiv. Remove floating point cases from mod.
15002 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15004 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15005 Increase idiv cost.
15007 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15009 * config/aarch64/aarch64.md
15010 (copysignsf3): Fix mask generation.
15012 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15014 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15016 (class dump_manager): Add register_dumps method.
15017 * dumpfile.c: Include langhooks.h.
15018 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15019 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15020 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15021 (dump_manager::dump_register): Start with 512 entries instead of 32.
15022 (dump_manager::register_dumps): New method.
15023 * toplev.c (general_init): Instead of invoking register_dumps
15024 langhook, invoke register_dumps method on the dump manager.
15025 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15028 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15030 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15031 Say that # requires an associated define_split to exist, and that
15032 the define_split must be suitable for use after register allocation.
15034 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15036 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15037 (compute_outgoing_frequencies): Also initialize zero counts.
15038 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15039 around loops; preserve more of profile when nothing changes.
15041 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15043 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15045 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15046 * config/arm/arm-cpu-cdata.h: Regenerate.
15047 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15048 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15049 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15050 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15051 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15053 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15055 * config/arm/t-rmprofile: Likewise.
15056 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15058 2017-06-06 David S. Miller <davem@davemloft.net>
15061 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15062 blockage if function uses alloca.
15064 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15066 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15067 New "uid" fields to hold pretty-print IDs of group and ref.
15068 Memory references are now identified as <group_id>:<ref_id>
15069 instead of using [random] addresses.
15070 (dump_mem_details): Simplify, no functional change.
15071 (dump_mem_ref): Simplify and make output more concise.
15072 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15073 (find_or_create_group): Initialize group uid.
15074 (record_ref): Initialize ref uid. Improve debug output.
15075 (prune_group_by_reuse, should_issue_prefetch_p,)
15076 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15077 (mark_nontemporal_store, determine_loop_nest_reuse):
15078 Improve debug output.
15080 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15082 * dbgcnt.def (prefetch): New debug counter.
15083 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15084 (schedule_prefetches): Stop issueing prefetches if debug counter
15087 2017-06-06 Tom de Vries <tom@codesourcery.com>
15089 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15090 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15092 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15094 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15095 Use aarch64_reg_or_zero predicate for operand 4.
15096 (aarch64_compare_and_swap<mode> define_insn_and_split):
15097 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15098 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15100 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15102 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15103 (arm_compute_save_core_reg_mask): This.
15104 (thumb1_compute_save_reg_mask): Rename into ...
15105 (thumb1_compute_save_core_reg_mask): This.
15106 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15107 (arm_compute_frame_layout): Likewise.
15109 2017-06-06 Richard Biener <rguenther@suse.de>
15111 PR tree-optimization/80974
15112 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15113 keep or clear leaders SSA info.
15115 2017-06-06 Tom de Vries <tom@codesourcery.com>
15117 * config/nvptx/nvptx.c (split_mode_p): New function.
15118 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15120 2017-06-06 Tom de Vries <tom@codesourcery.com>
15122 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15124 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15127 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15130 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15132 * shrink-wrap.c (handle_simple_exit): Update profile.
15133 (try_shrink_wrapping): Upate profile.
15135 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15137 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15138 (tree_guess_outgoing_edge_probabilities): New.
15139 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15140 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15142 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15144 * ipa-split.c (split_function): Initialize return bb profile.
15146 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15148 * profile.c (compute_branch_probabilities): Also initialize
15149 EXIT_BLOCK profile.
15151 2017-06-06 Richard Biener <rguenther@suse.de>
15153 PR tree-optimization/80928
15154 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15155 (vect_analyze_loop_operations): Properly guard analysis for
15157 (vect_transform_loop): Likewise.
15158 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15159 (vect_model_induction_cost): Do not cost for pure SLP.
15160 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15161 of induction in inner loop vectorization.
15162 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15163 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15164 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15166 (vect_analyze_slp_cost_1): Cost induction.
15167 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15168 (vect_get_slp_vect_defs): Likewise.
15169 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15170 (vect_transform_stmt): Handle SLP reductions.
15171 * tree-vectorizer.h (vectorizable_induction): Adjust.
15173 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15175 * config/rs6000/rs6000.c (make_resolver_func): Update
15176 init_lowered_empty_function call.
15178 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15180 * doc/invoke.texi: Document the -fprofile-abs-path option.
15181 * common.opt (fprofile-abs-path): New option.
15182 * gcov-io.h (gcov_write_filename): Declare.
15183 * gcov-io.c (gcov_write_filename): New function.
15184 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15185 * profile.c (output_location): Likewise.
15187 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15189 * shring-wrap.c: Revert accidental commit.
15191 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15193 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15195 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15197 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15199 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15200 profile in callgraph edge.
15201 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15202 (apply_scale): Likewise.
15203 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15204 Also scale profile when inlining function with zero profile.
15205 (initialize_cfun): Update exit block profile even when it is zero.
15206 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15207 when profile is read.
15209 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15211 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15212 (CLONE_*): New constants to define the processors we can generate
15213 code for with the target_clone attribute.
15214 (rs6000_clone_map): New array to identify which clone processors
15215 the current program is running on.
15216 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15217 target_clone attribute.
15218 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15219 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15220 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15221 (cpu_expand_builtin): Add support for target_clone attribute.
15222 (rs6000_valid_attribute_p): Allow "default" attribute.
15223 (get_decl_name): New debug function to simplify printing the
15224 current function name in debugging statements.
15225 (rs6000_clone_priority): New functions to support the target_clone
15226 attribute, and be able to generate code to switch between ISA 2.05
15227 through ISA 3.0 (power6 through power9).
15228 (rs6000_compare_version_priority): Likewise.
15229 (rs6000_get_function_versions_dispatcher): Likewise.
15230 (make_resolver_func): Likewise.
15231 (add_condition_to_bb): Likewise.
15232 (dispatch_function_versions): Likewise.
15233 (rs6000_generate_version_dispatcher_body): Likewise.
15234 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15235 (fusion_gpr_load_p): Fix a spacing issue.
15236 * doc/extend.texi (Common Function Attributes): Document that the
15237 PowerPC supports the target_clone attribute.
15239 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15241 * config/arm/arm.h: explain F symbol found in description of ARM
15242 register allocation in its legend.
15244 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15246 * config/mips/frame-header-opt.c: Include profile-count.h.
15247 * config/riscv/riscv.c: Include profile-count.h
15249 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15251 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15253 (sm_set_flag_if_changed): Add bbs field.
15254 (execute_sm_if_changed_flag_set): Pass BBS.
15255 (execute_sm): Update.
15257 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15259 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15262 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15264 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15265 (peephole2): New peephole2 to emit the above.
15266 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15268 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15270 * config/aarch64/aarch64.c (define_peephole2 above
15271 *sub_<shift>_<mode>): New peephole.
15273 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15275 * config/i386/i386.c (make_resolver_func): Update.
15276 * Makefile.in: Add profile-count.h and profile-count.o
15277 * auto-profile.c (afdo_indirect_call): Update to new API.
15278 (afdo_set_bb_count): Update.
15279 (afdo_propagate_edge): Update.
15280 (afdo_propagate_circuit): Update.
15281 (afdo_calculate_branch_prob): Update.
15282 (afdo_annotate_cfg): Update.
15283 * basic-block.h: Include profile-count.h
15284 (struct edge_def): Turn count to profile_count.
15285 (struct basic_block_def): Likewie.
15286 (REG_BR_PROB_BASE): Move to profile-count.h
15287 (RDIV): Move to profile-count.h
15288 * bb-reorder.c (max_entry_count): Turn to profile_count.
15289 (find_traces): Update.
15290 (rotate_loop):Update.
15291 (connect_traces):Update.
15292 (sanitize_hot_paths):Update.
15293 * bt-load.c (migrate_btr_defs): Update.
15294 * cfg.c (RDIV): Remove.
15295 (init_flow): Use alloc_block.
15296 (alloc_block): Uninitialize count.
15297 (unchecked_make_edge): Uninitialize count.
15298 (check_bb_profile): Update.
15299 (dump_edge_info): Update.
15300 (dump_bb_info): Update.
15301 (update_bb_profile_for_threading): Update.
15302 (scale_bbs_frequencies_int): Update.
15303 (scale_bbs_frequencies_gcov_type): Update.
15304 (scale_bbs_frequencies_profile_count): New.
15305 * cfg.h (update_bb_profile_for_threading): Update.
15306 (scale_bbs_frequencies_profile_count): Declare.
15307 * cfgbuild.c (compute_outgoing_frequencies): Update.
15308 (find_many_sub_basic_blocks): Update.
15309 * cfgcleanup.c (try_forward_edges): Update.
15310 (try_crossjump_to_edge): Update.
15311 * cfgexpand.c (expand_gimple_tailcall): Update.
15312 (construct_exit_block): Update.
15313 * cfghooks.c (verify_flow_info): Update.
15314 (dump_bb_for_graph): Update.
15315 (split_edge): Update.
15316 (make_forwarder_block): Update.
15317 (duplicate_block): Update.
15318 (account_profile_record): Update.
15319 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15320 (get_estimated_loop_iterations): Update.
15321 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15322 (single_likely_exit): Update.
15323 * cfgloopmanip.c (scale_loop_profile): Update.
15325 (set_zero_probability): Update.
15326 (lv_adjust_loop_entry_edge): Update.
15327 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15328 (purge_dead_edges): Update.
15329 (rtl_account_profile_record): Update.
15330 * cgraph.c (cgraph_node::create): Uninitialize count.
15331 (symbol_table::create_edge): Uninitialize count.
15332 (cgraph_update_edges_for_call_stmt_node): Update.
15333 (cgraph_edge::dump_edge_flags): Update.
15334 (cgraph_node::dump): Update.
15335 (cgraph_edge::maybe_hot_p): Update.
15336 * cgraph.h: Include profile-count.h
15337 (create_clone), create_edge, create_indirect_edge): Update.
15338 (cgraph_node): Turn count to profile_count.
15339 (cgraph_edge0: Likewise.
15340 (make_speculative, clone): Update.
15341 (create_edge): Update.
15342 (init_lowered_empty_function): Update.
15343 * cgraphclones.c (cgraph_edge::clone): Update.
15344 (duplicate_thunk_for_node): Update.
15345 (cgraph_node::create_clone): Update.
15346 * cgraphunit.c (cgraph_node::analyze): Update.
15347 (cgraph_node::expand_thunk): Update.
15348 * final.c (dump_basic_block_info): Update.
15349 * gimple-streamer-in.c (input_bb): Update.
15350 * gimple-streamer-out.c (output_bb): Update.
15351 * graphite.c (print_global_statistics): Update.
15352 (print_graphite_scop_statistics): Update.
15353 * hsa-brig.c: Include basic-block.h.
15354 * hsa-dump.c: Include basic-block.h.
15355 * hsa-gen.c (T sum_slice): Update.
15356 (convert_switch_statements):Update.
15357 * hsa-regalloc.c: Include basic-block.h.
15358 * ipa-chkp.c (chkp_produce_thunks): Update.
15359 * ipa-cp.c (struct caller_statistics): Update.
15360 (init_caller_stats): Update.
15361 (gather_caller_stats): Update.
15362 (ipcp_cloning_candidate_p): Update.
15363 (good_cloning_opportunity_p): Update.
15364 (get_info_about_necessary_edges): Update.
15365 (dump_profile_updates): Update.
15366 (update_profiling_info): Update.
15367 (update_specialized_profile): Update.
15368 (perhaps_add_new_callers): Update.
15369 (decide_about_value): Update.
15370 (ipa_cp_c_finalize): Update.
15371 * ipa-devirt.c (struct odr_type_warn_count): Update.
15372 (struct decl_warn_count): Update.
15373 (struct final_warning_record): Update.
15374 (possible_polymorphic_call_targets): Update.
15375 (ipa_devirt): Update.
15376 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15377 * ipa-icf.c (sem_function::merge): Update.
15378 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15379 * ipa-inline.c (compute_uninlined_call_time): Update.
15380 (compute_inlined_call_time): Update.
15381 (want_inline_small_function_p): Update.
15382 (want_inline_self_recursive_call_p): Update.
15383 (edge_badness): Update.
15384 (lookup_recursive_calls): Update.
15385 (recursive_inlining): Update.
15386 (inline_small_functions): Update.
15387 (dump_overall_stats): Update.
15388 (dump_inline_stats): Update.
15389 * ipa-profile.c (ipa_profile_generate_summary): Update.
15390 (ipa_propagate_frequency): Update.
15391 (ipa_profile): Update.
15392 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15393 * ipa-utils.c (ipa_merge_profiles): Update.
15394 * loop-doloop.c (doloop_modify): Update.
15395 * loop-unroll.c (report_unroll): Update.
15396 (unroll_loop_runtime_iterations): Update.
15397 * lto-cgraph.c (lto_output_edge): Update.
15398 (lto_output_node): Update.
15399 (input_node): Update.
15400 (input_edge): Update.
15401 (merge_profile_summaries): Update.
15402 * lto-streamer-in.c (input_cfg): Update.
15403 * lto-streamer-out.c (output_cfg): Update.
15404 * mcf.c (create_fixup_graph): Update.
15405 (adjust_cfg_counts): Update.
15406 (sum_edge_counts): Update.
15407 * modulo-sched.c (sms_schedule): Update.
15408 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15409 * predict.c (maybe_hot_count_p): Update.
15410 (probably_never_executed): Update.
15411 (dump_prediction): Update.
15412 (combine_predictions_for_bb): Update.
15413 (propagate_freq): Update.
15414 (handle_missing_profiles): Update.
15415 (counts_to_freqs): Update.
15416 (rebuild_frequencies): Update.
15417 (force_edge_cold): Update.
15418 * predict.h: Include profile-count.h
15419 (maybe_hot_count_p, counts_to_freqs): UPdate.
15420 * print-rtl-function.c: Do not include cfg.h
15421 * print-rtl.c: Include basic-block.h
15422 * profile-count.c: New file.
15423 * profile-count.h: New file.
15424 * profile.c (is_edge_inconsistent): Update.
15425 (correct_negative_edge_counts): Update.
15426 (is_inconsistent): Update.
15427 (set_bb_counts): Update.
15428 (read_profile_edge_counts): Update.
15429 (compute_frequency_overlap): Update.
15430 (compute_branch_probabilities): Update; Initialize and deinitialize
15432 (branch_prob): Update.
15433 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15434 (edge_gcov_count): New.
15435 (bb_gcov_count): New.
15436 * shrink-wrap.c (try_shrink_wrapping): Update.
15437 * tracer.c (better_p): Update.
15438 * trans-mem.c (expand_transaction): Update.
15439 (ipa_tm_insert_irr_call): Update.
15440 (ipa_tm_insert_gettmclone_call): Update.
15441 * tree-call-cdce.c: Update.
15442 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15443 (gimple_duplicate_sese_tail): Update.
15444 (gimple_account_profile_record): Update.
15445 (execute_fixup_cfg): Update.
15446 * tree-inline.c (copy_bb): Update.
15447 (copy_edges_for_bb): Update.
15448 (initialize_cfun): Update.
15449 (freqs_to_counts): Update.
15450 (copy_cfg_body): Update.
15451 (expand_call_inline): Update.
15452 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15453 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15454 (try_unroll_loop_completely): Update.
15455 (try_peel_loop): Update.
15456 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15457 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15458 * tree-ssa-loop-split.c (connect_loops): Update.
15459 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15460 * tree-ssa-reassoc.c (branch_fixup): Update.
15461 * tree-ssa-tail-merge.c (replace_block_by): Update.
15462 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15463 (compute_path_counts): Update.
15464 (update_profile): Update.
15465 (recompute_probabilities): Update.
15466 (update_joiner_offpath_counts): Update.
15467 (estimated_freqs_path): Update.
15468 (freqs_to_counts_path): Update.
15469 (clear_counts_path): Update.
15470 (ssa_fix_duplicate_block_edges): Update.
15471 (duplicate_thread_path): Update.
15472 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15473 (struct switch_conv_info): Update.
15474 * tree-tailcall.c (decrease_profile): Update.
15475 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15476 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15477 * value-prof.c (check_counter): Update.
15478 (gimple_divmod_fixed_value): Update.
15479 (gimple_mod_pow2): Update.
15480 (gimple_mod_subtract): Update.
15481 (gimple_ic_transform): Update.
15482 (gimple_stringop_fixed_value): Update.
15483 * value-prof.h (gimple_ic): Update.
15485 2017-06-02 Carl Love <cel@us.ibm.com>
15487 * config/rs6000/rs6000-c: Add support for built-in functions
15488 vector double vec_doublee (vector signed int);
15489 vector double vec_doublee (vector unsigned int);
15490 vector double vec_doublee (vector float);
15491 vector double vec_doubleh (vector signed int);
15492 vector double vec_doubleh (vector unsigned int);
15493 vector double vec_doubleh (vector float);
15494 vector double vec_doublel (vector signed int);
15495 vector double vec_doublel (vector unsigned int);
15496 vector double vec_doublel (vector float);
15497 vector double vec_doubleo (vector signed int);
15498 vector double vec_doubleo (vector unsigned int);
15499 vector double vec_doubleo (vector float);.
15500 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15501 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15503 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15504 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15505 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15507 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15508 vec_doublel, vec_doubleh.
15509 * doc/extend.texi: Update the built-in documentation file for the
15510 new built-in functions.
15512 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15515 * ipa-inline-analysis.c (free_growth_caches): Set
15516 edge_removal_hook_holder to NULL after removing it.
15518 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15520 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15521 comparision with zero.
15523 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15525 for early expansion of vec_min and vec_max builtins.
15526 (builtin_function_type): Add min/max unsigned variants to those
15527 identified as having unsigned arguments.
15529 2017-06-02 Olivier Hainque <hainque@adacore.com>
15531 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15533 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15535 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15536 Use VALL_F16 iterator rather than VALL.
15538 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15540 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15541 Emit CBNZ inside loop when doing a strong exchange and comparing
15542 against zero. Generate the CC flags after the loop.
15544 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15546 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15547 (dl_section_ref): New.
15548 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15549 On AIX, append an expression to subtract the size of the
15550 section length to dl_section_ref.
15552 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15554 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15555 for early expansion of vector absolute builtins.
15557 2017-06-02 Richard Biener <rguenther@suse.de>
15559 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
15560 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
15562 2017-06-02 Richard Biener <rguenther@suse.de>
15564 PR tree-optimization/80948
15565 * tree-tailcall.c (find_tail_calls): Track stmts to move in
15566 stmt order as well.
15568 2017-06-02 Richard Biener <rguenther@suse.de>
15570 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
15572 * tree-vect-stmts.c (process_use): Do not mark backedge defs
15573 for inductions as relevant.
15575 2017-06-02 Richard Biener <rguenther@suse.de>
15577 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
15578 (vectorizable_induction): ... this. Remove dead code.
15580 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
15582 * builtins. (expand_builtin_alloca): Remove second parameter and
15583 infer its value from the first parameter instead.
15584 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15586 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15588 PR rtl-optimization/80903
15589 * loop-doloop.c (add_test): Unshare sequence.
15591 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15593 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15595 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15597 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15599 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15600 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15601 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15602 (xlogue_layout::get_stub_rtx): Make static.
15603 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15604 (xlogue_layout::compute_stub_managed_regs): Rename to...
15605 (xlogue_layout::count_stub_managed_regs): ...this.
15606 (xlogue_layout::is_stub_managed_reg): New function.
15607 (xlogue_layout::m_stub_names): Rename to...
15608 (xlogue_layout::s_stub_names): ...this, make static.
15609 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15610 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15611 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15612 xlogue_layout::s_stub_names): Instantiate statics.
15613 (stub_managed_regs): Remove.
15614 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15615 (disable_call_ms2sysv_xlogues): Rename to...
15616 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15617 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15619 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15620 change after reload_completed.
15621 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15623 (ix86_expand_prologue): Likewise.
15624 (ix86_expand_epilogue): Likewise.
15625 (ix86_expand_split_stack_prologue): Likewise.
15626 (ix86_compute_frame_layout): Remove frame parameter ...
15627 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15628 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15630 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15631 (ix86_frame): Move from here ...
15632 * config/i386/i386.h (ix86_frame): ... to here.
15633 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15634 complete ix86_frame data structure instead. Remove some_ld_name.
15636 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15638 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15639 symbols that hold a DECL_VALUE_EXPR.
15641 2017-06-01 Martin Jambor <mjambor@suse.cz>
15643 PR tree-optimization/80898
15644 * tree-sra.c (process_subtree_disqualification): Removed.
15645 (disqualify_candidate): Do not acll
15646 process_subtree_disqualification.
15647 (subtree_mark_written_and_enqueue): New function.
15648 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15649 RHS has been disqualified and re-queue LHS if necessary. Apart
15650 from that, ignore disqualified RHS.
15652 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15654 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15655 address fetch for z10 or later.
15657 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15659 * config/arc/arc.md (tst_movb): Add guard when splitting.
15661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15663 * config/arc/arc.c (arc_can_eliminate): Test against
15664 arc_frame_pointer_needed.
15666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15668 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15669 to prevent store reordering.
15670 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15671 (type): Add block type.
15672 (stack_tie): Define special instruction to be used in
15675 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15677 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15678 constraint. It is not valid for the pattern.
15679 (noncommutative_binary_comparison): Likewise.
15681 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15683 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15686 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15688 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15689 be used by the reg-alloc.
15691 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15693 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15694 reg-alloc when having mul64 or mul32x16 instructions.
15695 (mulsidi3): Likewise.
15696 (umulsidi3): Likewise.
15697 (mulsi32x16): New pattern.
15698 (mulsi64): Likewise.
15699 (mulsidi64): Likewise.
15700 (umulsidi64): Likewise.
15701 (MUL32x16_REG): Define.
15702 (mul64_600): Use MUL32x16_REG.
15703 (mac64_600): Likewise.
15704 (umul64_600): Likewise.
15705 (umac64_600): Likewise.
15707 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15709 * config/arc/arc.md (mulsi3_700): Make it commutative.
15711 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15713 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15715 (*sign_extendsidi2_insn): Likewise for movstosw.
15717 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15719 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15720 the type of the input discriminant value. Convert the
15721 discriminant value of signedness vary.
15723 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15725 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15726 Add to -Wall section.
15728 2017-06-01 Richard Biener <rguenther@suse.de>
15730 PR middle-end/66313
15731 * fold-const.c (fold_plusminus_mult_expr): If the factored
15732 factor may be zero use a wrapping type for the inner operation.
15733 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15734 and handle moved defs.
15735 (process_assignment): Properly guard the unary op case. Return a
15736 tri-state indicating that moving the stmt before the call may allow
15737 to continue. Pass through to_move.
15738 (find_tail_calls): Handle moving unrelated defs before
15741 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15744 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15745 splitter result in the canonical way.
15747 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15749 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15750 also for 32bit target. Update insn attributes.
15751 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15753 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15755 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15756 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15757 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15758 (_mm_maskz_min_ss): New intrinsics.
15760 2017-05-31 Martin Liska <mliska@suse.cz>
15762 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15763 Change comment style to one we normally use.
15764 (vectorizable_reduction): Likewise.
15765 (vectorizable_induction): Likewise.
15766 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15767 (vectorizable_call): Likewise.
15768 (vectorizable_simd_clone_call): Likewise.
15769 (vectorizable_conversion): Likewise.
15770 (vectorizable_assignment): Likewise.
15771 (vectorizable_shift): Likewise.
15772 (vectorizable_operation): Likewise.
15773 (vectorizable_store): Likewise.
15774 (vectorizable_load): Likewise.
15775 * tree-vectorizer.h: Likewise.
15777 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15779 * passes.c (emergency_dump_function): New.
15780 * tree-pass.h (emergency_dump_function): Declare.
15781 * plugin.c (plugins_internal_error_function): Remove.
15782 * plugin.h (plugins_internal_error_function): Remove declaration.
15783 * toplev.c (internal_error_function): New static function. Use it...
15784 (general_init): ...here.
15786 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15788 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15789 (arc_rtx_costs): Add costs for new patterns.
15790 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15791 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15793 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15795 * tree-ssa-strlen.c (get_next_strinfo): New function.
15796 (get_stridx_plus_constant): Use it.
15797 (zero_length_string): Likewise.
15798 (adjust_related_strinfos): Likewise.
15799 (adjust_last_stmt): Likewise.
15801 2017-05-31 Richard Biener <rguenther@suse.de>
15804 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15805 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15807 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15809 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15810 loop_vinfo argument and use of dependence distance vectors.
15811 Check instead whether the two references differ only in their
15812 initial value and assume that they have the same alignment if the
15813 difference is a multiple of the vector alignment.
15814 (vect_analyze_data_refs_alignment): Update call accordingly.
15816 2017-05-31 Martin Liska <mliska@suse.cz>
15819 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15821 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15823 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15824 (create_intersect_range_checks): Move from ...
15825 * tree-data-ref.c (create_intersect_range_checks_index)
15826 (create_intersect_range_checks): ... to here.
15827 (create_runtime_alias_checks): New function factored from ...
15828 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15829 here. Call above function.
15830 * tree-data-ref.h (create_runtime_alias_checks): New function.
15832 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15834 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15835 segment length for dr_b and compute it in wide_int.
15837 2017-05-31 Richard Biener <rguenther@suse.de>
15839 PR tree-optimization/80906
15840 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15841 and pass through iv_map.
15842 (copy_bb_and_scalar_dependences): Adjust.
15843 (translate_pending_phi_nodes): Likewise.
15844 (copy_loop_close_phi_args): Handle code-generating IVs instead
15847 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15849 * diagnostic-color.c (color_dict): Add "type-diff".
15850 (parse_gcc_colors): Update comment.
15851 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15852 -fdiagnostics-show-template-tree and -fno-elide-type.
15853 (GCC_COLORS): Add type-diff to example.
15855 (-fdiagnostics-show-template-tree): New.
15856 (-fno-elide-type): New.
15857 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15858 the pp_format_decoder callback. Call any m_format_postprocessor's
15860 (pretty_printer::pretty_printer): Initialize
15861 m_format_postprocessor.
15862 (pretty_printer::~pretty_printer): Delete any
15863 m_format_postprocessor.
15864 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15865 (class format_postprocessor): New class.
15866 (struct pretty_printer::format_decoder): Document the new parameters.
15867 (struct pretty_printer::m_format_postprocessor): New field.
15868 * tree-diagnostic.c (default_tree_printer): Update for new
15869 bool and const char ** params.
15870 * tree-diagnostic.h (default_tree_printer): Likewise.
15872 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15874 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15875 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15877 rs6000_gen_cell_microcode code.
15878 (rs6000_final_prescan_insn): Delete.
15879 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15880 "warn-cell-microcode" entries.
15881 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15882 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15883 throughout. Change cc_reg_not_micro_cr0_operand to
15884 cc_reg_not_cr0_operand throughout.
15885 (*extendhi<mode>2_noload): Delete.
15886 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15887 (mwarn-cell-microcode): Delete.
15888 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15889 -mgen-cell-microcode and -mwarn-cell-microcode.
15891 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15894 * config/i386/constraints.md (Yd): New constraint.
15896 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15897 and (?Yd, r) alternatives. Update insn attributes.
15898 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15899 and (?*Yd, r) alternatives. Update insn attributes.
15900 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15902 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15904 * gimplify.c (gimplify_modify_expr): Don't create a
15905 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15908 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15910 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15912 2017-05-30 Richard Biener <rguenther@suse.de>
15914 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15915 and reduc_def fields.
15916 (STMT_VINFO_REDUC_TYPE): New define.
15917 (STMT_VINFO_REDUC_DEF): Likewise.
15918 (vect_force_simple_reduction): Adjust prototype.
15919 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15920 (vect_is_simple_reduction): Remove check_reduction argument.
15921 (vect_force_simple_reduction): Adjust and set
15922 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15923 (vectorizable_reduction): Do not re-do reduction analysis
15924 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15925 * tree-parloops.c (gather_scalar_reductions): Adjust.
15927 2017-05-30 Richard Biener <rguenther@suse.de>
15929 PR middle-end/80901
15930 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15933 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15935 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15936 Introduce unknown_misalignment parameter and remove vf.
15937 (vect_peeling_hash_get_lowest_cost):
15938 Pass unknown_misalignment parameter.
15939 (vect_enhance_data_refs_alignment):
15940 Fix unsupportable data ref treatment.
15942 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15944 * tree-vect-data-refs.c (vect_get_data_access_cost):
15945 Workaround for SLP handling.
15946 (vect_enhance_data_refs_alignment):
15947 Compute costs for doing no peeling at all, compare to the best
15948 peeling costs so far and avoid peeling if cheaper.
15950 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15952 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15953 Return peeling info and set costs to zero for unlimited cost
15955 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15956 with unknown misalignment. Compute and costs for unknown
15957 misalignment, compare them to the costs for known misalignment
15958 and choose the cheapest for peeling.
15960 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15962 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15963 (vect_get_peeling_costs_all_drs): Create function.
15964 (vect_peeling_hash_get_lowest_cost):
15965 Use vect_get_peeling_costs_all_drs.
15966 (vect_peeling_supportable): Create function.
15967 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15969 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15971 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15972 DR_HAS_NEGATIVE_STEP.
15973 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15974 (vect_enhance_data_refs_alignment): Use.
15975 (vect_duplicate_ssa_name_ptr_info): Use.
15976 * tree-vectorizer.h (dr_misalignment): Use.
15977 (known_alignment_for_access_p): Use.
15979 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15982 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15984 (has_section_name): New function.
15986 2017-05-30 Martin Liska <mliska@suse.cz>
15989 * auto-profile.c (get_function_decl_from_block): Fix
15992 2017-05-30 Richard Biener <rguenther@suse.de>
15994 PR middle-end/80876
15995 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15997 2017-05-30 Martin Liska <mliska@suse.cz>
15999 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16000 * dumpfile.h (struct dump_file_info): Remove ctors.
16002 2017-05-30 Martin Liska <mliska@suse.cz>
16004 * predict.def: Fix GNU coding style.
16006 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16008 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16009 Mark 'to' argument with ATTRIBUTE_UNUSED.
16011 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16013 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16014 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16015 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16018 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16020 * doc/install.texi (Options specification): Restore entry of
16021 --enable-sjlj-exceptions.
16023 2017-05-27 Michael Eager <eager@eagercon.com>
16026 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16028 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16030 * config/microblaze/microblaze.h
16031 (FIXED_REGISTERS): Update in macro.
16032 (CALL_USED_REGISTERS): Update in macro.
16034 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16036 * doc/install.texi: Add links to macOS binary distributions.
16038 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16042 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16044 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16046 2017-05-26 Martin Liska <mliska@suse.cz>
16048 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16050 2017-05-26 Martin Liska <mliska@suse.cz>
16052 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16053 always leading ';; '.
16054 (dump_bb_info): Likewise.
16055 (brief_dump_cfg): Likewise.
16056 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16057 * dumpfile.c: Remove usage of TDF_VERBOSE.
16058 * dumpfile.h (enum dump_kind): Likewise.
16059 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16060 * print-tree.c (debug_verbose): Remove.
16061 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16062 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16063 * tree-diagnostic.c (default_tree_printer): Replace
16064 TDF_DIAGNOSTIC with TDF_SLIM.
16066 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16068 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16069 in parameter loop, rather than loop_vinfo.
16070 (create_intersect_range_checks): Ditto.
16071 (vect_create_cond_for_alias_checks): Update call to above functions.
16073 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16075 PR tree-optimization/80815
16076 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16077 for merging runtime alias checks. Handle negative DR_STEPs.
16079 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16081 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16083 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16084 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16085 out code pruning runtime alias checks.
16086 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16087 factored out from above.
16088 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16090 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16092 (prune_runtime_alias_test_list): New decalaration.
16094 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16096 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16097 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16098 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16099 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16100 (operator==, comp_dr_with_seg_len_pair): Ditto.
16101 (vect_prune_runtime_alias_test_list): Ditto.
16103 2017-05-26 Martin Liska <mliska@suse.cz>
16106 * params.def: Bound partial-inlining-entry-probability param.
16108 2017-05-26 Marek Polacek <polacek@redhat.com>
16111 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16114 2017-05-26 Richard Biener <rguenther@suse.de>
16116 PR tree-optimization/80842
16117 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16120 2017-05-26 Richard Biener <rguenther@suse.de>
16122 PR tree-optimization/80844
16123 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16125 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16127 * doc/md.texi (Machine Constraints): Update x86 family
16128 machine constraints section to match 'config/i386/constraints.md'.
16130 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16132 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16134 2017-05-25 Nathan Sidwell <nathan@acm.org>
16136 * doc/invoke.texi (--enable-languages): Update documentation.
16138 2017-05-25 Martin Liska <mliska@suse.cz>
16140 * dumpfile.c: Add TDF_FOLDING.
16141 * dumpfile.h (enum dump_kind): Likewise.
16142 * genmatch.c (dt_simplify::gen_1): Use it.
16144 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16146 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16148 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16150 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16151 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16153 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16155 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16156 * match.pd (X == C): Rewrite it here.
16157 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16158 with_certain_nonzero_bits2): New predicates.
16159 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16161 2017-05-24 Nathan Sidwell <nathan@acm.org>
16163 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16166 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16169 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16171 * config/powerpcspe: New port. Files are copied from the rs6000
16172 port, with "rs6000" in filenames replaced by "powerpcspe".
16174 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16176 PR rtl-optimization/80754
16177 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16179 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16181 * config/sparc/sparc.md (length): Return the correct value for -mflat
16182 sibcalls to match output_sibcall.
16184 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16188 * config/rs6000/rs6000.c (struct machine_function): Add new field
16190 (rs6000_get_separate_components): Init that field, use it.
16191 (rs6000_components_for_bb): Use the field.
16193 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16195 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16197 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16199 PR middle-end/80823
16200 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16202 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16205 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16206 address against address_operand predicate.
16207 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16209 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16211 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16212 parameters passed indirectly.
16214 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16216 * config/i386/i386.md (*movdi_internal): Remove SSE4
16217 alternative 18 (?r, *v). Update insn attributes.
16218 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16219 Update insn attributes.
16220 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16221 Update insn attributes.
16222 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16223 alternative 1 (r, v). Remove isa attribute.
16224 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16225 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16226 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16228 2017-05-23 Tom de Vries <tom@codesourcery.com>
16230 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16233 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16235 * cgraphunit.c (symbol_table::process_new_functions): Update.
16236 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16237 (inline_generate_summary): Rename to ...
16238 (ipa_fn_summary_generate): ... this one.
16239 (inline_read_summary): Rename to ...
16240 (ipa_fn_summary_read): ... this one.
16241 (inline_write_summary): Rename to ...
16242 (ipa_fn_summary_write): ... this one.
16243 (inline_free_summary): Rename to ...
16244 (ipa_free_fn_summary): ... this one.
16245 (pass_data_local_fn_summary, pass_local_fn_summary,
16246 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16247 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16248 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16249 make_pass_ipa_fn_summary): New.
16250 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16251 inline_write_summary, inline_free_summary): Remove.
16252 (ipa_free_fn_summary) : New.
16253 * ipa-inline.c (ipa_inline): Update.
16254 (pass_ipa_inline): Do not generate summaries.
16255 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16257 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16258 and add pass_ipa_fn_summary.
16259 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16261 (make_pass_inline_parameters): Remove.
16263 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16265 * omp-low.c (struct omp_context): Remove "default_kind" member.
16268 * omp-offload.c (execute_oacc_device_lower): Remove the
16269 parallelism dimensions function attributes for unparallelized
16270 OpenACC kernels constructs.
16272 2017-05-23 Martin Liska <mliska@suse.cz>
16274 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16276 (cgraph_edge::make_speculative): Likewise.
16277 (cgraph_edge::resolve_speculation): Likewise.
16278 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16279 (cgraph_node::dump): Likewise.
16280 * cgraph.h: Likewise.
16281 * cgraphunit.c (analyze_functions): Likewise.
16282 (symbol_table::compile): Likewise.
16283 * ipa-cp.c (print_all_lattices): Likewise.
16284 (determine_versionability): Likewise.
16285 (initialize_node_lattices): Likewise.
16286 (ipcp_verify_propagated_values): Likewise.
16287 (estimate_local_effects): Likewise.
16288 (update_profiling_info): Likewise.
16289 (create_specialized_node): Likewise.
16290 (perhaps_add_new_callers): Likewise.
16291 (decide_about_value): Likewise.
16292 (decide_whether_version_node): Likewise.
16293 (identify_dead_nodes): Likewise.
16294 (ipcp_store_bits_results): Likewise.
16295 * ipa-devirt.c (dump_targets): Likewise.
16296 (ipa_devirt): Likewise.
16297 * ipa-icf.c (sem_item::dump): Likewise.
16298 (sem_function::equals): Likewise.
16299 (sem_variable::equals): Likewise.
16300 (sem_item_optimizer::read_section): Likewise.
16301 (sem_item_optimizer::execute): Likewise.
16302 (congruence_class::dump): Likewise.
16303 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16304 (dump_inline_summary): Likewise.
16305 (estimate_node_size_and_time): Likewise.
16306 (inline_analyze_function): Likewise.
16307 * ipa-inline-transform.c (inline_call): Likewise.
16308 * ipa-inline.c (report_inline_failed_reason): Likewise.
16309 (want_early_inline_function_p): Likewise.
16310 (edge_badness): Likewise.
16311 (update_edge_key): Likewise.
16312 (inline_small_functions): Likewise.
16313 * ipa-profile.c (ipa_profile): Likewise.
16314 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16315 (ipa_make_edge_direct_to_target): Likewise.
16316 (remove_described_reference): Likewise.
16317 (ipa_impossible_devirt_target): Likewise.
16318 (propagate_controlled_uses): Likewise.
16319 (ipa_print_node_params): Likewise.
16320 (ipcp_transform_function): Likewise.
16321 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16322 (propagate_pure_const): Likewise.
16323 * ipa-reference.c (generate_summary): Likewise.
16324 (read_write_all_from_decl): Likewise.
16325 (propagate): Likewise.
16326 (ipa_reference_read_optimization_summary): Likewise.
16327 * ipa-utils.c (ipa_merge_profiles): Likewise.
16328 * ipa.c (walk_polymorphic_call_targets): Likewise.
16329 (symbol_table::remove_unreachable_nodes): Likewise.
16330 (ipa_single_use): Likewise.
16331 * passes.c (execute_todo): Likewise.
16332 * predict.c (drop_profile): Likewise.
16333 * symtab.c (symtab_node::get_dump_name): New function.
16334 (symtab_node::dump_name): Likewise.
16335 (symtab_node::dump_asm_name): Likewise.
16336 (symtab_node::dump_references): Likewise.
16337 (symtab_node::dump_referring): Likewise.
16338 (symtab_node::dump_base): Likewise.
16339 (symtab_node::debug_symtab): Likewise.
16340 * tree-sra.c (convert_callers_for_node): Likewise.
16341 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16342 * value-prof.c (init_node_map): Likewise.
16344 2017-05-23 Martin Liska <mliska@suse.cz>
16346 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16347 and symtab_node::debug_symtab to symbol_table::debug.
16348 * cgraphunit.c (analyze_functions): Use the renamed function.
16349 (symbol_table::compile): Likewise.
16350 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16351 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16352 * passes.c (execute_todo): Likewise.
16353 * symtab.c (symbol_table::dump): New function.
16354 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16356 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16358 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16359 that nonconst implies exec.
16361 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16363 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16364 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16365 (inline_edge_summary_vec): Turn into ...
16366 (ipa_call_summaries): ... this one.
16367 (redirect_to_unreachable, edge_set_predicate,
16368 evaluate_properties_for_edge, inline_summary_alloc,
16369 reset_ipa_call_summary, reset_inline_summary,
16370 inline_summary_t::duplicate): Update.
16371 (inline_edge_duplication_hook): Turn to ...
16372 (ipa_call_summary_t::duplicate): ... this one.
16373 (inline_edge_removal_hook): Turn to ...
16374 (ipa_call_summary_t::remove): ... this one.
16375 (dump_inline_edge_summary): Turn to ...
16376 (dump_ipa_call_summary): ... this one.
16377 (estimate_function_body_sizes): Update.
16378 (inline_update_callee_summaries): Update.
16379 (remap_edge_change_prob): Update.
16380 (remap_edge_summaries): Update.
16381 (inline_merge_summary): Update.
16382 (do_estimate_edge_time): Update.
16383 (inline_generate_summary): Update.
16384 (inline_read_section): Update.
16385 (inline_read_summary): Update.
16386 (inline_free_summary): Update.
16387 * ipa-inline.c (can_inline_edge_p): Update.
16388 (compute_inlined_call_time): Update.
16389 (want_inline_small_function_p): Update.
16390 (edge_badness): Update.
16391 (early_inliner): Update.
16392 * ipa-inline.h (inline_edge_summary): Turn to ...
16393 (ipa_call_summary): ... this one.
16394 (ipa_call_summary_t): New class.
16395 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16396 (ipa_call_summaries): New.
16397 (inline_edge_summary): Remove.
16398 (estimate_edge_growth): Update.
16399 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16400 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16401 * ipa-split.c (execute_split_functions): Update.
16402 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16404 2017-05-23 Tom de Vries <tom@codesourcery.com>
16406 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16407 attributes): Document rdrand effective target.
16409 2017-05-23 Tom de Vries <tom@codesourcery.com>
16411 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16412 attributes): Sort alphabetically.
16414 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16416 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16418 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16421 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16422 V2DF/V2DI splat into two separate patterns, one that handles
16423 registers, and the other that only handles memory. Drop support
16424 for splatting from a GPR on ISA 2.07 and then splitting the
16425 splat into direct move and splat.
16426 (vsx_splat_<mode>_reg): Likewise.
16427 (vsx_splat_<mode>_mem): Likewise.
16429 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16431 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16433 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16435 PR middle-end/80809
16436 * omp-low.c (finish_taskreg_remap): New function.
16437 (finish_taskreg_scan): If unit size of ctx->record_type
16438 is non-constant, unshare the size expression and replace
16439 decls in it with possible outer var refs.
16441 PR middle-end/80809
16442 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16443 GOVD_SHARED rather than GOVD_PRIVATE with it.
16444 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16445 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16447 PR middle-end/80853
16448 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16449 as last argument to build_outer_var_ref for pointer bases of array
16450 section reductions.
16452 2017-05-19 Martin Sebor <msebor@redhat.com>
16454 * print-tree.c (print_node): Print DECL_READ_P flag.
16456 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16458 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16459 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16460 * cgraph.c: Likewise.
16461 * cgraphunit.c: Likewise.
16462 * gengtype.c: Likewise.
16463 * ipa-cp.c: Likewise.
16464 * ipa-devirt.c: Likewise.
16465 * ipa-icf.c: Likewise.
16466 * ipa-predicate.c: Likewise.
16467 * ipa-profile.c: Likewise.
16468 * ipa-prop.c: Likewise.
16469 * ipa-split.c: Likewise.
16471 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16472 edge_predicate_pool, dump_inline_hints,
16473 inline_summary::account_size_time, redirect_to_unreachable,
16474 edge_set_predicate, set_hint_predicate,
16475 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16476 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16477 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16478 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16479 ipa_call_summary_t::remove, initialize_growth_caches,
16480 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16481 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16482 mark_modified, unmodified_parm_1, unmodified_parm,
16483 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16484 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16485 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16486 will_be_nonconstant_predicate, record_modified_bb_info,
16487 get_minimal_bb, record_modified, param_change_prob,
16488 phi_result_unknown_predicate, predicate_for_phi_result,
16489 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16490 estimate_function_body_sizes, compute_inline_parameters,
16491 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16492 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16493 inline_update_callee_summaries, remap_edge_change_prob,
16494 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16495 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16496 inline_analyze_function, inline_summary_t::insert,
16497 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16498 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16499 inline_free_summary): Move to ipa-fnsummary.h
16500 (predicate_t): Remove.
16501 * ipa-fnsummary.c: New file.
16502 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16503 (enum inline_hints_vals, inline_hints, agg_position_info,
16504 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16505 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16506 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16507 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16508 inline_read_summary, inline_write_summary, inline_free_summary,
16509 inline_analyze_function, initialize_inline_failed,
16510 inline_merge_summary, inline_update_overall_summary,
16511 compute_inline_parameters): Move to ipa-fnsummary.h
16512 * ipa-fnsummary.h: New file.
16513 * ipa-inline-transform.h: Include ipa-inline.h.
16514 * ipa-inline.c: LIkewise.
16516 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16518 * ipa-inline.c (edge_badness): Use inlined_time instead of
16519 inline_summaries->get.
16521 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16523 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16525 2017-05-22 Nathan Sidwell <nathan@acm.org>
16527 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16528 (fdump-lang): Document 'raw' option.
16529 * dumpfile.h (TDI_tu): Delete.
16530 * dumpfile.c (dump_files): Remove translation-unit.
16531 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16533 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16535 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16536 command option from $(AWK) call.
16537 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16538 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16539 [FORMAT]: Remove handling of variable.
16540 * config/avr/t-multilib: Regenerate.
16542 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16544 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16546 (dump_inline_summary): Do not print self_time.
16547 (estimate_function_body_sizes): Do not set self_time.
16548 (compute_inline_parameters): Likewise.
16549 (inline_read_section, inline_write_summary): Do not stream self_time.
16550 * ipa-inline.h (inline_summary): Drop self_time.
16552 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16554 * ipa-inline-analysis.c (account_size_time): Rename to ...
16555 (inline_summary::account_size_time): ... this one.
16556 (reset_ipa_call_summary): Turn to ...
16557 (ipa_call_summary::reset): ... this one.
16558 (reset_inline_summary): Turn to ...
16559 (inline_summary::reset): ... this one.
16560 (inline_summary_t::remove): Update.
16561 (inline_summary_t::duplicate): Update.
16562 (ipa_call_summary_t::remove): Update.
16563 (dump_inline_summary): Update.
16564 (estimate_function_body_sizes): Update.
16565 (compute_inline_parameters): Update.
16566 (estimate_node_size_and_time): Update.
16567 (inline_merge_summary): Update.
16568 (inline_update_overall_summary): Update.
16569 (inline_read_section): Update.
16570 (inline_write_summary): Update.
16571 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
16572 add account_size_time and reset member functions.
16573 (ipa_call_summary): Add reset function.
16574 * ipa-predicate.h (predicate::operator &): Constify.
16576 2017-05-22 Richard Biener <rguenther@suse.de>
16578 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
16580 2017-05-19 Jason Merrill <jason@redhat.com>
16582 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16584 2017-05-19 Marek Polacek <polacek@redhat.com>
16587 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16588 TYPE_OVERFLOW_WRAPS checks.
16590 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16592 * tree-core.h (enum omp_clause_default_kind): Add
16593 "OMP_CLAUSE_DEFAULT_PRESENT".
16594 * tree-pretty-print.c (dump_omp_clause): Handle it.
16595 * gimplify.c (enum gimplify_omp_var_data): Add
16596 "GOVD_MAP_FORCE_PRESENT".
16597 (gimplify_adjust_omp_clauses_1): Map it to
16598 "GOMP_MAP_FORCE_PRESENT".
16599 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16601 * gimplify.c (oacc_default_clause): Clarify.
16603 2017-05-19 Nathan Sidwell <nathan@acm.org>
16605 LANG_HOOK_REGISTER_DUMPS
16606 * toplev.c (general_init): Call register dump lang hook.
16607 * doc/invoke.texi: Document -fdump-lang option family.
16608 * dumpfile.c (dump_files): Remove class dump here.
16609 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16610 * dumpfile.h (tree_dump_index): Remove TDI_class.
16611 * langhooks-def.h (lhd_register_dumps): Declare.
16612 (LANG_HOOKS_REGISTER_DUMPS): Define.
16613 (LANG_HOOKS_INITIALIZER): Add it.
16614 * langhooks.c (lhd_register_dumps): Define.
16615 * langhooks.h (struct lang_hooks): Add register_dumps.
16617 2017-05-19 Nathan Sidwell <nathan@acm.org>
16619 * context.h (context::set_passes): New.
16620 * context.c (context::context): Do not create pass manager.
16621 * toplev.c (general_init): Create pass manager here.
16623 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16625 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16626 use this splitter if two add or or instructions would also work for
16627 the constant we want to generate.
16629 2017-05-19 Richard Biener <rguenther@suse.de>
16632 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16633 predicate evaluation.
16635 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16637 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16639 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16642 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16644 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16645 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16646 (inline_edge_summary_vec): Turn into ...
16647 (ipa_call_summaries): ... this one.
16648 (redirect_to_unreachable, edge_set_predicate,
16649 evaluate_properties_for_edge, inline_summary_alloc,
16650 reset_ipa_call_summary, reset_inline_summary,
16651 inline_summary_t::duplicate): Update.
16652 (inline_edge_duplication_hook): Turn to ...
16653 (ipa_call_summary_t::duplicate): ... this one.
16654 (inline_edge_removal_hook): Turn to ...
16655 (ipa_call_summary_t::remove): ... this one.
16656 (dump_inline_edge_summary): Turn to ...
16657 (dump_ipa_call_summary): ... this one.
16658 (estimate_function_body_sizes): Update.
16659 (inline_update_callee_summaries): Update.
16660 (remap_edge_change_prob): Update.
16661 (remap_edge_summaries): Update.
16662 (inline_merge_summary): Update.
16663 (do_estimate_edge_time): Update.
16664 (inline_generate_summary): Update.
16665 (inline_read_section): Update.
16666 (inline_read_summary): Update.
16667 (inline_free_summary): Update.
16668 * ipa-inline.c (can_inline_edge_p): Update.
16669 (compute_inlined_call_time): Update.
16670 (want_inline_small_function_p): Update.
16671 (edge_badness): Update.
16672 (early_inliner): Update.
16673 * ipa-inline.h (inline_edge_summary): Turn to ...
16674 (ipa_call_summary): ... this one.
16675 (ipa_call_summary_t): New class.
16676 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16677 (ipa_call_summaries): New.
16678 (inline_edge_summary): Remove.
16679 (estimate_edge_growth): Update.
16680 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16681 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16682 * ipa-split.c (execute_split_functions): Update.
16683 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16685 2017-05-19 Richard Biener <rguenther@suse.de>
16687 PR middle-end/80764
16688 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16690 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16692 * config/rs6000/rs6000.c (struct machine_function): Add field
16693 fpr_is_wrapped_separately.
16694 (rs6000_get_separate_components): Use 64 components. Handle the
16695 new FPR components.
16696 (rs6000_components_for_bb): Handle the FPR components.
16697 (rs6000_emit_prologue_components): Handle the FPR components.
16698 (rs6000_emit_epilogue_components): Handle the FPR components.
16699 (rs6000_set_handled_components): Handle the FPR components.
16700 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16701 that are already separately shrink-wrapped.
16702 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16703 that are already separately shrink-wrapped.
16705 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16708 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16711 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16712 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16713 cases where the register allocator uses a move and an offsettable
16714 memory operation to/from a FPR register on ISA 2.06/2.07.
16715 (define_peephole2 for Altivec d-form store): Likewise.
16717 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16720 * config/i386/mmx.md (*mov<mode>_internal): Enable
16721 alternatives 11, 12, 13 and 14 also for 32bit targets.
16722 Remove alternatives 15, 16, 17 and 18.
16723 * config/i386/sse.md (vec_concatv2di): Change
16724 alternative (!x, *y) to (x, ?!*Yn).
16726 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16728 * dumpfile.h (enum dump_kind): Remove stray comma.
16730 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16732 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16733 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16734 predicate::num_conditions
16735 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16736 (CHANGED): turn into predicate::changed.
16737 (agg_position_info): Move to ipa-predicate.h
16738 (add_condition, predicate::add_clause, predicate::operator &=,
16739 predicate::or_with, predicate::evaluate, predicate::probability,
16740 dump_condition, dump_clause, predicate::dump,
16741 predicate::remap_after_duplication, predicate::remap_after_inlining,
16742 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16743 (evaluate_conditions_for_known_args): Update.
16744 (set_cond_stmt_execution_predicate): Update.
16745 * ipa-inline.h: Include ipa-predicate.h
16746 (condition, inline_param_summary, conditions, agg_position_info,
16747 predicate): Move to ipa-predicate.h
16748 * ipa-predicate.c: New file.
16749 * ipa-predicate.h: New file.
16751 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16753 * final.c (leaf_function_p): Check we are not in a sequence.
16755 2017-05-18 Martin Liska <mliska@suse.cz>
16757 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16758 * dumpfile.c (dump_register): Use new enum dump_kind.
16759 (get_dump_file_name): Likewise.
16760 (dump_enable_all): Likewise.
16761 (dump_switch_p_1): Likewise.
16762 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16763 * dumpfile.h (enum dump_kind): New enum type.
16764 (struct dump_file_info): Create constructor and
16765 format fields and comments.
16766 * passes.c (pass_manager::register_one_dump_file):
16768 * statistics.c (statistics_early_init): Likewise.
16769 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16770 TDF_TREE with TDF_SLIM.
16771 (gather_memory_references_ref): Likewise.
16773 2017-05-18 Martin Liska <mliska@suse.cz>
16775 * vec.h (struct vnull): Use it.
16777 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16779 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16780 (true_predicate, false_predicate, true_predicate_p,
16781 false_predicate_p): Remove.
16782 (single_cond_predicate, not_inlined_predicate): Turn to member function
16784 (add_condition): Update.
16785 (add_clause): Turn to...
16786 (predicate::add_clause): ... this one; update; allow passing NULL
16788 (and_predicates): Turn to ...
16789 (predicate::operator &=): ... this one.
16790 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16791 (or_predicates): Turn to ...
16792 (predicate::or_with): ... this one.
16793 (evaluate_predicate): Turn to ...
16794 (predicate::evaluate): ... this one.
16795 (predicate_probability): Turn to ...
16796 (predicate::probability): ... this one.
16797 (dump_condition): Update.
16798 (dump_predicate): Turn to ...
16799 (predicate::dump): ... this one.
16800 (account_size_time): Update.
16801 (edge_set_predicate): Update.
16802 (set_hint_predicate): UPdate.
16803 (evaluate_conditions_for_known_args): Update.
16804 (evaluate_properties_for_edge): Update.
16805 (remap_predicate_after_duplication): Turn to...
16806 (predicate::remap_after_duplication): ... this one.
16807 (remap_hint_predicate_after_duplication): Update.
16808 (inline_summary_t::duplicate): UPdate.
16809 (dump_inline_edge_summary): Update.
16810 (dump_inline_summary): Update.
16811 (set_cond_stmt_execution_predicate): Update.
16812 (set_switch_stmt_execution_predicate): Update.
16813 (compute_bb_predicates): Update.
16814 (will_be_nonconstant_expr_predicate): Update.
16815 (will_be_nonconstant_predicate): Update.
16816 (phi_result_unknown_predicate): Update.
16817 (predicate_for_phi_result): Update.
16818 (array_index_predicate): Update.
16819 (estimate_function_body_sizes): Update.
16820 (estimate_node_size_and_time): Update.
16821 (estimate_ipcp_clone_size_and_time): Update.
16822 (remap_predicate): Rename to ...
16823 (predicate::remap_after_inlining): ... this one.
16824 (remap_hint_predicate): Update.
16825 (inline_merge_summary): Update.
16826 (inline_update_overall_summary): Update.
16827 (estimate_size_after_inlining): Update.
16828 (read_predicate): Rename to ...
16829 (predicate::stream_in): ... this one.
16830 (read_inline_edge_summary): Update.
16831 (write_predicate): Rename to ...
16832 (predicate::stream_out): ... this one.
16833 (write_inline_edge_summary): Update.
16834 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16835 (clause_t): Turn to uint32_t
16836 (predicate): Turn to class; implement constructor and operators
16838 (size_time_entry): Update.
16839 (inline_summary): Update.
16840 (inline_edge_summary): Update.
16842 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16844 * fold-const.c (fold_binary_loc): Move transformation...
16845 * match.pd (C - X CMP X): ... here.
16847 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16849 * config/sparc/sparc.c (sparc_option_override): Set function
16850 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16851 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16853 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16855 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16857 2017-05-18 Marek Polacek <polacek@redhat.com>
16860 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16861 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16862 gimple_assign_load_p.
16864 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16866 PR middle-end/80692
16867 * real.c (do_compare): Give decimal_do_compare preference over
16868 comparing just the signs.
16870 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16872 * doc/md.texi (Canonicalization of Instructions): Describe the
16873 canonical form of instructions that inherently set a condition
16876 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16878 PR middle-end/80775
16879 * tree-cfg.c: Move deletion of unreachable case statements to after
16880 the merging of consecutive case labels.
16882 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16884 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16885 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16886 restoring of callee-saved registers.
16888 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16890 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16891 * config/visium/visium.c (single_set_and_flags): Likewise.
16892 * config/visium/visium.md (Substitutions): Likewise.
16894 2017-05-17 Martin Liska <mliska@suse.cz>
16896 * cfg.c: Introduce dump_flags_t type and
16897 use it instead of int type.
16899 * cfghooks.c: Likewise.
16900 * cfghooks.h (struct cfg_hooks): Likewise.
16901 * cfgrtl.c: Likewise.
16902 * cfgrtl.h: Likewise.
16903 * cgraph.c (cgraph_node::get_body): Likewise.
16904 * coretypes.h: Likewise.
16905 * domwalk.c: Likewise.
16906 * domwalk.h: Likewise.
16907 * dumpfile.c (struct dump_option_value_info): Likewise.
16908 (dump_enable_all): Likewise.
16909 (dump_switch_p_1): Likewise.
16910 (opt_info_switch_p): Likewise.
16911 * dumpfile.h (enum tree_dump_index): Likewise.
16912 (struct dump_file_info): Likewise.
16913 * genemit.c: Likewise.
16914 * generic-match-head.c: Likewise.
16915 * gengtype.c (open_base_files): Likewise.
16916 * gimple-pretty-print.c: Likewise.
16917 * gimple-pretty-print.h: Likewise.
16918 * graph.c (print_graph_cfg): Likewise.
16919 * graphite-scop-detection.c (dot_all_sese): Likewise.
16920 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16921 * loop-unroll.c (report_unroll): Likewise.
16922 * passes.c (pass_manager::register_one_dump_file): Likewise.
16923 * print-tree.c: Likewise.
16924 * statistics.c: Likewise.
16925 * tree-cfg.c: Likewise.
16926 * tree-cfg.h: Likewise.
16927 * tree-dfa.c: Likewise.
16928 * tree-dfa.h: Likewise.
16929 * tree-dump.c (dump_function): Likewise.
16930 * tree-dump.h (struct dump_info): Likewise.
16931 * tree-pretty-print.c: Likewise.
16932 * tree-pretty-print.h: Likewise.
16933 * tree-ssa-live.c: Likewise.
16934 * tree-ssa-live.h: Likewise.
16935 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16936 * tree-vect-loop.c: Likewise.
16937 * tree-vect-slp.c: Likewise.
16939 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16940 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16942 PR tree-optimization/80457
16943 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16944 of all arguments to a statement as scalar_to_vec operations.
16945 (vectorizable_call): Adjust call to vect_model_simple_cost for
16947 (vectorizable_conversion): Likewise.
16948 (vectorizable_assignment): Likewise.
16949 (vectorizable_shift): Likewise.
16950 (vectorizable_operation): Likewise.
16951 (vectorizable_comparison): Likewise.
16952 (vect_is_simple_cond): Record the def types for operands.
16953 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16954 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16955 for statement argument count.
16957 2017-05-16 Carl Love <cel@us.ibm.com>
16959 * config/rs6000/rs6000-c: Add support for built-in functions
16960 vector unsigned long long vec_bperm (vector unsigned long long,
16961 vector unsigned char)
16962 vector signed long long vec_mule (vector signed int,
16964 vector unsigned long long vec_mule (vector unsigned int,
16965 vector unsigned int)
16966 vector signed long long vec_mulo (vector signed int,
16968 vector unsigned long long vec_mulo (vector unsigned int,
16969 vector unsigned int)
16970 vector signed char vec_sldw (vector signed char,
16971 vector signed char,
16973 vector unsigned char vec_sldw (vector unsigned char,
16974 vector unsigned char,
16976 vector signed short vec_sldw (vector signed short,
16977 vector signed short,
16979 vector unsigned short vec_sldw (vector unsigned short,
16980 vector unsigned short,
16982 vector signed int vec_sldw (vector signed int,
16985 vector unsigned int vec_sldw (vector unsigned int,
16986 vector unsigned int,
16988 vector signed long long vec_sldw (vector signed long long,
16989 vector signed long long,
16991 vector unsigned long long vec_sldw (vector unsigned long long,
16992 vector unsigned long long,
16994 * config/rs6000/rs6000-c: Add support for built-in functions
16995 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16996 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16997 * doc/extend.texi: Update the built-in documentation for the
16998 new built-in functions.
17000 2017-05-16 Marek Polacek <polacek@redhat.com>
17004 * tree.c (save_expr): Don't fold the expression.
17006 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17008 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17009 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17010 and (?*y,m). Update insn attributes.
17012 2017-05-16 Martin Liska <mliska@suse.cz>
17014 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17015 flags argument of print_gimple_stmt, print_gimple_expr,
17016 print_generic_stmt and print_generic_expr.
17017 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17018 * coretypes.h: Likewise.
17019 * except.c (dump_eh_tree): Likewise.
17020 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17021 * gimple-pretty-print.h: Likewise.
17022 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17023 (backprop::push_to_worklist): Likewise.
17024 (backprop::pop_from_worklist): Likewise.
17025 (backprop::process_use): Likewise.
17026 (backprop::intersect_uses): Likewise.
17027 (note_replacement): Likewise.
17028 * gimple-ssa-store-merging.c
17029 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17030 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17031 (pass_store_merging::execute): Likewise.
17032 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17033 (ssa_base_cand_dump_callback): Likewise.
17034 (dump_incr_vec): Likewise.
17035 (replace_refs): Likewise.
17036 (replace_mult_candidate): Likewise.
17037 (create_add_on_incoming_edge): Likewise.
17038 (create_phi_basis): Likewise.
17039 (insert_initializers): Likewise.
17040 (all_phi_incrs_profitable): Likewise.
17041 (introduce_cast_before_cand): Likewise.
17042 (replace_one_candidate): Likewise.
17043 * gimplify.c (gimplify_expr): Likewise.
17044 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17045 (set_rename): Likewise.
17046 (rename_uses): Likewise.
17047 (copy_loop_phi_nodes): Likewise.
17048 (add_close_phis_to_merge_points): Likewise.
17049 (copy_loop_close_phi_args): Likewise.
17050 (copy_cond_phi_args): Likewise.
17051 (graphite_copy_stmts_from_block): Likewise.
17052 (translate_pending_phi_nodes): Likewise.
17053 * graphite-poly.c (print_pdr): Likewise.
17054 (dump_gbb_cases): Likewise.
17055 (dump_gbb_conditions): Likewise.
17056 (print_scop_params): Likewise.
17057 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17058 (build_cross_bb_scalars_use): Likewise.
17059 (gather_bbs::before_dom_children): Likewise.
17060 * hsa-dump.c (dump_hsa_immed): Likewise.
17061 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17062 (get_replacement_map): Likewise.
17063 * ipa-inline-analysis.c (dump_condition): Likewise.
17064 (estimate_function_body_sizes): Likewise.
17065 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17066 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17067 * ipa-prop.c (ipa_dump_param): Likewise.
17068 (ipa_print_node_jump_functions_for_edge): Likewise.
17069 (ipa_modify_call_arguments): Likewise.
17070 (ipa_modify_expr): Likewise.
17071 (ipa_dump_param_adjustments): Likewise.
17072 (ipa_dump_agg_replacement_values): Likewise.
17073 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17074 * ipa-pure-const.c (check_stmt): Likewise.
17075 (pass_nothrow::execute): Likewise.
17076 * ipa-split.c (execute_split_functions): Likewise.
17077 * omp-offload.c (dump_oacc_loop_part): Likewise.
17078 (dump_oacc_loop): Likewise.
17079 * trans-mem.c (tm_log_emit): Likewise.
17080 (tm_memopt_accumulate_memops): Likewise.
17081 (dump_tm_memopt_set): Likewise.
17082 (dump_tm_memopt_transform): Likewise.
17083 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17084 (print_loop): Likewise.
17085 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17086 (chkp_gather_checks_info): Likewise.
17087 (chkp_get_check_result): Likewise.
17088 (chkp_remove_check_if_pass): Likewise.
17089 (chkp_use_outer_bounds_if_possible): Likewise.
17090 (chkp_reduce_bounds_lifetime): Likewise.
17091 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17092 (chkp_mark_completed_bounds): Likewise.
17093 (chkp_register_incomplete_bounds): Likewise.
17094 (chkp_mark_invalid_bounds): Likewise.
17095 (chkp_maybe_copy_and_register_bounds): Likewise.
17096 (chkp_build_returned_bound): Likewise.
17097 (chkp_get_bound_for_parm): Likewise.
17098 (chkp_build_bndldx): Likewise.
17099 (chkp_get_bounds_by_definition): Likewise.
17100 (chkp_generate_extern_var_bounds): Likewise.
17101 (chkp_get_bounds_for_decl_addr): Likewise.
17102 * tree-chrec.c (chrec_apply): Likewise.
17103 * tree-data-ref.c (dump_data_reference): Likewise.
17104 (dump_subscript): Likewise.
17105 (dump_data_dependence_relation): Likewise.
17106 (analyze_overlapping_iterations): Likewise.
17107 * tree-inline.c (expand_call_inline): Likewise.
17108 (tree_function_versioning): Likewise.
17109 * tree-into-ssa.c (dump_defs_stack): Likewise.
17110 (dump_currdefs): Likewise.
17111 (dump_names_replaced_by): Likewise.
17112 (dump_update_ssa): Likewise.
17113 (update_ssa): Likewise.
17114 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17115 * tree-parloops.c (build_new_reduction): Likewise.
17116 (try_create_reduction_list): Likewise.
17117 (ref_conflicts_with_region): Likewise.
17118 (oacc_entry_exit_ok_1): Likewise.
17119 (oacc_entry_exit_single_gang): Likewise.
17120 * tree-pretty-print.h: Likewise.
17121 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17122 (get_scalar_evolution): Likewise.
17123 (add_to_evolution): Likewise.
17124 (get_loop_exit_condition): Likewise.
17125 (analyze_evolution_in_loop): Likewise.
17126 (analyze_initial_condition): Likewise.
17127 (analyze_scalar_evolution): Likewise.
17128 (instantiate_scev): Likewise.
17129 (number_of_latch_executions): Likewise.
17130 (gather_chrec_stats): Likewise.
17131 (final_value_replacement_loop): Likewise.
17132 (scev_const_prop): Likewise.
17133 * tree-sra.c (dump_access): Likewise.
17134 (disqualify_candidate): Likewise.
17135 (create_access): Likewise.
17136 (reject): Likewise.
17137 (maybe_add_sra_candidate): Likewise.
17138 (create_access_replacement): Likewise.
17139 (analyze_access_subtree): Likewise.
17140 (analyze_all_variable_accesses): Likewise.
17141 (sra_modify_assign): Likewise.
17142 (initialize_constant_pool_replacements): Likewise.
17143 (find_param_candidates): Likewise.
17144 (decide_one_param_reduction): Likewise.
17145 (replace_removed_params_ssa_names): Likewise.
17146 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17147 * tree-ssa-copy.c (dump_copy_of): Likewise.
17148 (copy_prop_visit_cond_stmt): Likewise.
17149 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17150 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17151 (record_equivalences_from_stmt): Likewise.
17152 * tree-ssa-dse.c (compute_trims): Likewise.
17153 (delete_dead_call): Likewise.
17154 (delete_dead_assignment): Likewise.
17155 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17156 (forward_propagate_into_cond): Likewise.
17157 (pass_forwprop::execute): Likewise.
17158 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17159 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17161 (move_computations_worker): Likewise.
17162 (execute_sm): Likewise.
17163 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17164 (remove_exits_and_undefined_stmts): Likewise.
17165 (remove_redundant_iv_tests): Likewise.
17166 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17167 (adjust_iv_update_pos): Likewise.
17168 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17169 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17170 (value_replacement): Likewise.
17171 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17172 * tree-ssa-pre.c (print_pre_expr): Likewise.
17173 (get_representative_for): Likewise.
17174 (create_expression_by_pieces): Likewise.
17175 (insert_into_preds_of_block): Likewise.
17176 (eliminate_insert): Likewise.
17177 (eliminate_dom_walker::before_dom_children): Likewise.
17178 (eliminate): Likewise.
17179 (remove_dead_inserted_code): Likewise.
17180 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17181 * tree-ssa-reassoc.c (get_rank): Likewise.
17182 (eliminate_duplicate_pair): Likewise.
17183 (eliminate_plus_minus_pair): Likewise.
17184 (eliminate_not_pairs): Likewise.
17185 (undistribute_ops_list): Likewise.
17186 (eliminate_redundant_comparison): Likewise.
17187 (update_range_test): Likewise.
17188 (optimize_range_tests_var_bound): Likewise.
17189 (optimize_vec_cond_expr): Likewise.
17190 (rewrite_expr_tree): Likewise.
17191 (rewrite_expr_tree_parallel): Likewise.
17192 (linearize_expr): Likewise.
17193 (break_up_subtract): Likewise.
17194 (linearize_expr_tree): Likewise.
17195 (attempt_builtin_powi): Likewise.
17196 (attempt_builtin_copysign): Likewise.
17197 (transform_stmt_to_copy): Likewise.
17198 (transform_stmt_to_multiply): Likewise.
17199 (dump_ops_vector): Likewise.
17200 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17201 (print_scc): Likewise.
17202 (set_ssa_val_to): Likewise.
17203 (visit_reference_op_store): Likewise.
17204 (visit_use): Likewise.
17205 (sccvn_dom_walker::before_dom_children): Likewise.
17206 (run_scc_vn): Likewise.
17207 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17209 (expr_hash_elt::print): Likewise.
17210 (const_and_copies::pop_to_marker): Likewise.
17211 (const_and_copies::record_const_or_copy_raw): Likewise.
17212 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17213 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17214 (dump_predicates): Likewise.
17215 (find_uninit_use): Likewise.
17216 (warn_uninitialized_phi): Likewise.
17217 (pass_late_warn_uninitialized::execute): Likewise.
17218 * tree-ssa.c (verify_vssa): Likewise.
17219 (verify_ssa): Likewise.
17220 (maybe_optimize_var): Likewise.
17221 * tree-vrp.c (dump_value_range): Likewise.
17222 (dump_all_value_ranges): Likewise.
17223 (dump_asserts_for): Likewise.
17224 (register_edge_assert_for_2): Likewise.
17225 (vrp_visit_cond_stmt): Likewise.
17226 (vrp_visit_switch_stmt): Likewise.
17227 (vrp_visit_stmt): Likewise.
17228 (vrp_visit_phi_node): Likewise.
17229 (simplify_cond_using_ranges_1): Likewise.
17230 (fold_predicate_in): Likewise.
17231 (evrp_dom_walker::before_dom_children): Likewise.
17232 (evrp_dom_walker::push_value_range): Likewise.
17233 (evrp_dom_walker::pop_value_range): Likewise.
17234 (execute_early_vrp): Likewise.
17236 2017-05-16 Richard Biener <rguenther@suse.de>
17238 * dwarf2out.c (loc_list_from_tree_1): Do not create
17239 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17241 2017-05-16 Richard Biener <rguenther@suse.de>
17243 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17245 (note_variable_value_in_expr): If we resolved the decl ref
17246 do not push to the stack.
17248 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17250 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17251 operations in fast-math mode.
17252 (vaddq_f16): Likewise.
17253 (vmul_f16): Likewise.
17254 (vmulq_f16): Likewise.
17255 (vsub_f16): Likewise.
17256 (vsubq_f16): Likewise.
17257 * config/arm/neon.md (add<mode>3): New.
17259 (fma:<VH:mode>3): New. Also remove outdated comment.
17262 2017-05-16 Martin Liska <mliska@suse.cz>
17266 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17267 (odr_types_equivalent_p): Likewise.
17269 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17271 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17273 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17276 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17277 non-interunit SSE move alternatives with '?'.
17278 (zero-extendsidi peephole2): New peephole to skip intermediate
17279 general register in SSE zero-extend sequence.
17281 2017-05-15 Jeff Law <law@redhat.com>
17283 * reorg.c (relax_delay_slots): Create a new variable to hold
17284 the temporary target rather than clobbering TARGET_LABEL.
17286 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17287 missing argument to extract_bit_field call.
17288 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17290 2017-05-15 Martin Liska <mliska@suse.cz>
17293 * gcc.c (process_command): Do not allow empty argument of -o option.
17295 2017-05-15 Renlin Li <renlin.li@arm.com>
17297 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17298 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17299 * config/aarch64/constraints.md (Usf): Add long call check.
17300 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17301 (call_value): Likewise.
17302 (sibcall): Likewise.
17303 (sibcall_value): Likewise.
17305 (call_value_insn): New.
17306 (sibcall_insn): Update rtx pattern.
17307 (sibcall_value_insn): Likewise.
17308 (call_internal): Remove.
17309 (call_value_internal): Likewise.
17310 (sibcall_internal): Likewise.
17311 (sibcall_value_internal): Likewise.
17312 (call_reg): Likewise.
17313 (call_symbol): Likewise.
17314 (call_value_reg): Likewise.
17315 (call_value_symbol): Likewise.
17317 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17320 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17322 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17324 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17325 compatible with CCGOCmode and with CCZmode.
17327 2017-05-14 Martin Sebor <msebor@redhat.com>
17329 PR middle-end/77671
17330 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17331 (gimple_fold_builtin_snprintf): Same.
17332 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17333 (gimple_fold_builtin_snprintf): Same.
17334 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17335 of character types.
17336 (is_call_safe): New function.
17337 (try_substitute_return_value): Call it.
17338 (try_simplify_call): New function.
17339 (pass_sprintf_length::handle_gimple_call): Call it.
17341 2017-05-14 Martin Sebor <msebor@redhat.com>
17343 PR middle-end/80669
17344 * builtins.c (expand_builtin_stpncpy): Simplify.
17346 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17348 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17349 * config/i386/i386.h
17350 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17351 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17352 (struct machine_function): Add new members call_ms2sysv,
17353 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17354 (struct machine_frame_state): New fields sp_realigned and
17355 sp_realigned_offset.
17356 * config/i386/i386.c
17357 (enum xlogue_stub): New enum.
17358 (enum xlogue_stub_sets): New enum.
17359 (class xlogue_layout): New class.
17360 (struct ix86_frame): New fields stack_realign_allocate_offset,
17361 stack_realign_offset and outlined_save_offset. Modify comments to
17362 detail stack layout when using out-of-line stubs.
17363 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17364 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17365 -mcall-ms2sysv-xlogues.
17366 (stub_managed_regs): New static variable.
17367 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17368 registers managed by out-of-line stub.
17369 (disable_call_ms2sysv_xlogues): New function.
17370 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17371 m->call_ms2sysv when appropriate and compute frame layout for
17373 (sp_valid_at, fp_valid_at): New inline functions.
17374 (choose_basereg): New function.
17375 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17377 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17378 Use align parameter of choose_baseaddr to generated aligned SSE movs
17380 (pro_epilogue_adjust_stack): Modify to track
17381 machine_frame_state::sp_realigned.
17382 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17383 (ix86_nsaved_sseregs): Likewise.
17384 (ix86_emit_save_regs): Likewise.
17385 (ix86_emit_save_regs_using_mov): Likewise.
17386 (ix86_emit_save_sse_regs_using_mov): Likewise.
17387 (get_scratch_register_on_entry): Likewise.
17388 (gen_frame_set): New function.
17389 (gen_frame_load): Likewise.
17390 (gen_frame_store): Likewise.
17391 (emit_outlined_ms2sysv_save): Likewise.
17392 (emit_outlined_ms2sysv_restore): Likewise.
17393 (ix86_expand_prologue): Modify stack re-alignment code and call
17394 emit_outlined_ms2sysv_save when appropriate.
17395 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17396 parameter rtx_insn *insn, which allows the function to be used to only
17397 generate the notes.
17398 (ix86_expand_epilogue): Modify validity checks of frame and stack
17399 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17400 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17401 * config/i386/predicates.md
17402 (save_multiple): New predicate.
17403 (restore_multiple): Likewise.
17404 * config/i386/sse.md
17405 (save_multiple<mode>): New pattern.
17406 (save_multiple_realign<mode>): Likewise.
17407 (restore_multiple<mode>): Likewise.
17408 (restore_multiple_and_return<mode>): Likewise.
17409 (restore_multiple_leave_return<mode>): Likewise.
17410 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17412 2017-05-14 Julia Koval <julia.koval@intel.com>
17414 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17415 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17416 (__builtin_ia32_xsetbv): New builtins.
17417 * config/i386/i386.c (ix86_expand_special_args_builtin):
17419 (ix86_expand_builtin): Special expand for new intrinsics.
17420 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17421 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17422 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17426 * cfganal.c (inverted_post_order_compute): Change argument type
17428 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17429 * df-core.c (rest_of_handle_df_initialize): Adjust.
17430 (rest_of_handle_df_finish): Likewise.
17431 (df_analyze_1): Likewise.
17432 (df_analyze): Likewise.
17433 (loop_inverted_post_order_compute): Change argument to be a vec *.
17434 (df_analyze_loop): Adjust.
17435 (df_get_n_blocks): Likewise.
17436 (df_get_postorder): Likewise.
17437 * df.h (struct df_d): Change field to be a vec.
17438 * lcm.c (compute_laterin): Adjust.
17439 (compute_available): Likewise.
17440 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17441 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17442 * tree-ssa-pre.c (compute_antic): Likewise.
17444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17446 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17447 (depth_first_search::depth_first_search): Change structure init
17448 function to this constructor.
17449 (depth_first_search::add_bb): Rename function to this member.
17450 (depth_first_search::execute): Likewise.
17451 (flow_dfs_compute_reverse_finish): Adjust.
17453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17455 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17456 (longest_simple_path): Likewise.
17457 * shrink-wrap.c (spread_components): Likewise.
17458 (disqualify_problematic_components): Likewise.
17459 (emit_common_heads_for_components): Likewise.
17460 (emit_common_tails_for_components): Likewise.
17461 (insert_prologue_epilogue_for_components): Likewise.
17463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17465 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17468 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17470 * df-core.c (df_set_blocks): Start using auto_bitmap.
17471 (df_compact_blocks): Likewise.
17472 * df-problems.c (df_rd_confluence_n): Likewise.
17473 * df-scan.c (df_insn_rescan_all): Likewise.
17474 (df_process_deferred_rescans): Likewise.
17475 (df_update_entry_block_defs): Likewise.
17476 (df_update_exit_block_uses): Likewise.
17477 (df_entry_block_bitmap_verify): Likewise.
17478 (df_exit_block_bitmap_verify): Likewise.
17479 (df_scan_verify): Likewise.
17480 * lra-constraints.c (lra_constraints): Likewise.
17481 (undo_optional_reloads): Likewise.
17482 (lra_undo_inheritance): Likewise.
17483 * lra-remat.c (calculate_gen_cands): Likewise.
17484 (do_remat): Likewise.
17485 * lra-spills.c (assign_spill_hard_regs): Likewise.
17486 (spill_pseudos): Likewise.
17487 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17488 (bitmap_set_subtract_values): Likewise.
17490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17492 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17493 management with auto_bitmap.
17494 (fix_inter_tick): Likewise.
17495 (fix_recovery_deps): Likewise.
17496 * ira.c (add_store_equivs): Likewise.
17497 (find_moveable_pseudos): Likewise.
17498 (split_live_ranges_for_shrink_wrap): Likewise.
17499 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17500 (rtx_reuse_manager::seen_def_p): Likewise.
17501 (rtx_reuse_manager::set_seen_def): Likewise.
17502 * print-rtl.h (class rtx_reuse_manager): Likewise.
17504 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17506 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17508 (migrate_btr_def): Likewise.
17509 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17510 * df-core.c (loop_post_order_compute): Likewise.
17511 (loop_inverted_post_order_compute): Likewise.
17512 * hsa-common.h: Likewise.
17513 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17514 * init-regs.c (initialize_uninitialized_regs): Likewise.
17515 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17516 (inline_small_functions): Likewise.
17517 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17518 * ira.c (combine_and_move_insns): Likewise.
17519 (build_insn_chain): Likewise.
17520 * loop-invariant.c (find_invariants): Likewise.
17521 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17522 * predict.c (tree_predict_by_opcode): Likewise.
17523 (predict_paths_leading_to): Likewise.
17524 (predict_paths_leading_to_edge): Likewise.
17525 (estimate_loops_at_level): Likewise.
17526 (estimate_loops): Likewise.
17527 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17528 (spread_components): Likewise.
17529 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17530 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17531 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17532 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17533 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17534 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17535 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17536 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17537 (mark_threaded_blocks): Likewise.
17538 (thread_through_all_blocks): Likewise.
17539 * tree-ssa.c (verify_ssa): Likewise.
17540 (execute_update_addresses_taken): Likewise.
17541 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17543 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17545 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17547 (post_order_compute): Likewise.
17548 (inverted_post_order_compute): Likewise.
17549 (pre_and_rev_post_order_compute_fn): Likewise.
17551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17553 * genrecog.c (int_set::int_set): Explicitly construct our
17554 auto_vec base class.
17555 * vec.h (auto_vec::auto_vec): New constructor.
17557 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17559 * bitmap.h (class auto_bitmap): New constructor taking
17560 bitmap_obstack * argument.
17562 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17564 * bitmap.h (class auto_bitmap): Change type of m_bits to
17565 bitmap_head, and adjust ctor / dtor and member operators.
17567 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
17569 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
17570 when returned register mode doesn't match original mode.
17572 2017-05-12 Jeff Law <law@redhat.com>
17573 Jakub Jelinek <jakub@redhat.com>
17575 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
17576 we look for cc setter after the compare-elim changes.
17577 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
17578 within the vector to match what compare-elim now expects.
17579 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
17580 (xorsi3_flags, one_cmplsi2_flags): Likewise.
17582 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17583 after the compare-elim changes.
17584 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17585 the vector to match what compare-elim now expects.
17586 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17587 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17588 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17589 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17590 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17592 * config/visium/visium.c (single_set_and_flags): Fix where
17593 we look for cc setter after the compare-elim changes.
17594 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17595 with the vector to match what compare-elim now expects.
17596 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17597 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17598 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17599 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17600 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17601 (neg<mode>2_insn_set_overflow): Likewise.
17603 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17605 PR middle-end/79794
17606 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17607 maybe_expand_insn call, set ops[0].target. If still set after call,
17608 set alt_rtl. Add extra arg to recursive calls.
17609 (extract_bit_field): Add alt_rtl argument. Pass to
17611 * expmed.h (extract_bit_field): Fix prototype.
17612 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17613 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17614 to extract_bit_field_calls.
17615 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17616 Pass alt_rtl to extract_bit_field calls.
17617 * calls.c (store_unaligned_arguments_into_psuedos)
17618 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17619 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17621 * optabs.h (struct expand_operand): Add target bitfield.
17623 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17625 * compare-elim.c (try_eliminate_compare): Canonicalize
17626 operation with embedded compare to
17627 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17628 (set (reg) (operation)].
17630 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17632 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17635 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17636 cost of adding a carry flag for ADC instruction.
17637 [case MINUS]: Ignore the cost of subtracting a carry flag
17638 for SBB instruction.
17640 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17642 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17644 * config/rs6000/bmiintrin.h: New file.
17645 * config/rs6000/bmi2intrin.h: New file.
17646 * config/rs6000/x86intrin.h: New file.
17648 2017-05-12 Jeff Law <law@redhat.com>
17650 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17653 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17655 PR middle-end/80707
17656 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17658 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17660 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17661 early expansion of vector divide builtins.
17662 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17663 builtins identified as having unsigned arguments.
17665 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17667 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17668 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17669 expansion of vector logical operations (and, andc, or, xor,
17672 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17674 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17675 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17677 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17679 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17680 early GIMPLE expansion of vector multiplies.
17682 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17684 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17685 TARGET_HAVE_MOVT conditional.
17686 (movt splitter): Likewise.
17688 2017-05-12 Richard Biener <rguenther@suse.de>
17690 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17691 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17692 Fold all stmts not inplace.
17694 2017-05-12 Richard Biener <rguenther@suse.de>
17696 PR tree-optimization/80713
17697 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17698 inserted_exprs bit for not removed stmts.
17700 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17702 PR middle-end/69921
17703 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17704 parallelized" attribute for parallelized OpenACC kernels.
17705 * omp-offload.c (execute_oacc_device_lower): Use it.
17707 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17708 Set "oacc kernels" attribute.
17709 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17710 parameter. Adjust all users.
17711 (oacc_fn_attrib_kernels_p): Remove function.
17712 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17713 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17714 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17715 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17716 assert "oacc kernels" attribute is set.
17718 2017-05-11 Carl Love <cel@us.ibm.com>
17720 * config/rs6000/rs6000-c: Add support for built-in functions
17721 vector unsigned char vec_popcnt (vector signed char)
17722 vector unsigned char vec_popcnt (vector unsigned char)
17723 vector unsigned short vec_popcnt (vector signed short)
17724 vector unsigned short vec_popcnt (vector unsigned short)
17725 vector unsigned int vec_popcnt (vector signed int)
17726 vector unsigned int vec_popcnt (vector unsigned int)
17727 vector unsigned long long vec_popcnt (vector signed long long)
17728 vector unsigned long long vec_popcnt (vector unsigned long long)
17729 vector signed long long vec_slo (vector signed long long,
17730 vector signed char)
17731 vector signed long long vec_slo (vector signed long long,
17732 vector unsigned char)
17733 vector unsigned long long vec_slo (vector unsigned long long,
17734 vector signed char)
17735 vector unsigned long long vec_slo (vector unsigned long long,
17736 vector unsigned char)
17737 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17738 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17739 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17740 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17741 * doc/extend.texi: Update the built-in documentation file for the
17742 new built-in functions.
17744 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17746 * attribs.h (sorted_attr_string): Move machine independent
17747 functions for target clone support from the i386 port to common
17748 code. Rename ix86_function_versions to common_function_versions.
17749 Rename make_name to make_unique_name.
17750 (common_function_versions): Likewise.
17751 (make_unique_name): Likewise.
17752 (make_dispatcher_decl): Likewise.
17753 (is_function_default_version): Likewise.
17754 * attribs.c (attr_strcmp): Likewise.
17755 (sorted_attr_string): Likewise.
17756 (common_function_versions): Likewise.
17757 (make_unique_name): Likewise.
17758 (make_dispatcher_decl): Likewise.
17759 (is_function_default_version): Likewise.
17760 * config/i386/i386.c (attr_strcmp): Likewise.
17761 (sorted_attr_string): Likewise.
17762 (ix86_function_versions): Likewise.
17763 (make_name): Likewise.
17764 (make_dispatcher_decl): Likewise.
17765 (is_function_default_version): Likewise.
17766 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17768 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17771 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17772 Account for direct move costs for vec_construct of integer
17775 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17778 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17779 (UNSPEC_STX_ATOMIC): Ditto.
17780 (loaddi_via_sse): New insn.
17781 (storedi_via_sse): Ditto.
17782 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17783 Update corresponding peephole2 patterns.
17784 (atomic_storedi_fpu): Ditto.
17786 2017-05-11 Julia Koval <julia.koval@intel.com>
17788 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17789 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17791 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17792 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17793 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17795 2017-05-11 Nathan Sidwell <nathan@acm.org>
17797 * graphite-poly.c: Include dumpfile.h.
17799 * dumpfle.h (dump_function): Declare here ...
17800 * tree-dump.h (dump_function): ... not here.
17801 * dumpfile.c: #include tree-cfg.h.
17802 (dump_function): Move here from ...
17803 * tree-dump.c (dump_function): ... here.
17804 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17805 * graphite-poly.c: Don't include tree-dump.h.
17806 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17807 * print-tree.c: Likewise.
17808 * stor-layout.c: Likewise.
17809 * tree-nested.c: Likewise.
17811 * dumpfile.c (dump_start): Use TDF_FLAGS.
17812 (dump_enable_all): Fix TDF_KIND check thinko.
17814 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17816 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17817 array entries to represent two legal parameterizations of the
17818 overloaded __builtin_cmpb function, as represented by the
17819 P6_OV_BUILTIN_CMPB constant.
17820 (altivec_resolve_overloaded_builtin): Add special case handling
17821 for the __builtin_cmpb function, as represented by the
17822 P6_OV_BUILTIN_CMPB constant.
17823 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17824 (BU_P6_64BIT_2): New macro.
17825 (BU_P6_OVERLOAD_2): New macro
17826 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17827 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17828 (CMPB): Add overload support to represent both 32-bit and 64-bit
17829 compare-bytes function.
17830 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17831 support for TARGET_CMPB.
17832 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17833 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17834 documentation of the __builtin_cmpb overloaded built-in function.
17836 2017-05-11 Richard Biener <rguenther@suse.de>
17838 PR tree-optimization/80705
17839 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17840 bases are not vectorizable.
17842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17844 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17845 when counting register pressure.
17847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17849 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17850 (iv_ca_more_deps): Renamed to ...
17851 (iv_ca_compare_deps): ... this.
17852 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17856 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17858 (determine_group_iv_costs): ... here.
17859 (find_inv_vars_cb): Record inv var if it's not recorded before.
17861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17863 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17864 (get_shiftadd_cost): Ditto.
17866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17868 * tree-ssa-address.c: Include header file.
17869 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17871 (add_to_parts): Refactor.
17872 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17873 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17878 PR tree-optimization/53090
17879 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17881 (extract_cond_operands): Detect condition with IV on both sides
17882 and return COMP_IV_EXPR_2.
17883 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17884 (rewrite_use_compare): Simplify by removing call to function
17885 extract_cond_operands.
17887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17889 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17890 (extract_cond_operands): Detect condition comparing against non-
17891 invariant bound and return appropriate enum value.
17892 (find_interesting_uses_cond): Update use of extract_cond_operands.
17893 Handle its return value accordingly.
17894 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17898 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17899 nonlinear iv_use computation in loop invariant sensitive way.
17901 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17903 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17904 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17908 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17909 (dump_cand): Support iv_cand.inv_exprs.
17910 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17912 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17915 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17917 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17919 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17920 as local function. Include necessary header files.
17921 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17925 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17929 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17930 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17931 RSHIFT_EXPR and BIT_NOT_EXPR.
17933 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17935 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17936 (adjust_setup_cost): New parameter supporting round up adjustment.
17937 (struct address_cost_data): Delete.
17938 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17939 (split_address_cost, ptr_difference_cost): Delete.
17940 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17941 (struct ainc_cost_data): New struct.
17942 (get_address_cost_ainc): New function.
17943 (get_address_cost, get_computation_cost): Reimplement.
17944 (determine_group_iv_cost_address): Record inv_expr for all uses of
17946 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17947 (iv_ca_has_deps): Reimplemented to ...
17948 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17950 (iv_ca_extend): Call iv_ca_more_deps.
17952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17954 * tree-ssa-address.c (struct mem_address): Move to header file.
17955 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17956 * tree-ssa-address.h (struct mem_address): Move from C file.
17957 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17959 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17961 * tree-affine.h (aff_combination_type): New interface.
17962 (aff_combination_zero_p): Remove static.
17963 (aff_combination_const_p): New interface.
17964 (aff_combination_singleton_var_p): New interfaces.
17966 2017-05-11 Richard Biener <rguenther@suse.de>
17968 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17969 Skip unreachable blocks and destinations.
17970 (eliminate): Move stmt removal and fixup ...
17971 (fini_eliminate): ... here. Skip inserted exprs.
17972 (pass_pre::execute): Move fini_pre after fini_eliminate.
17973 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17974 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17975 PRE to get rid of dead code that has invalid SSA form and
17976 split critical edges again.
17978 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17980 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17982 2017-05-11 Richard Biener <rguenther@suse.de>
17984 * passes.c (execute_function_todo): Verify loops if they are
17985 said to be up-to-date.
17986 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17987 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17989 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17992 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17993 handle calling assemble_external ourself.
17996 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17997 modes with zero size. Enhance comment.
17999 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18001 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18002 built-ins for vec_xl and vec_xst with short and char pointer
18005 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18007 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18008 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18009 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18010 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18011 (_mm_maskz_min_round_ss): New intrinsics.
18012 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18013 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18014 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18015 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18016 (__builtin_ia32_minss_mask_round): New builtins.
18017 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18018 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18019 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18021 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18022 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18024 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18027 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18029 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18030 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18031 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18032 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18033 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18034 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18035 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18036 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18037 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18038 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18039 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18040 (__builtin_ia32_mulss_mask_round): New builtins.
18041 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18042 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18043 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18045 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18046 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18048 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18051 2017-05-10 Julia Koval <julia.koval@intel.com>
18053 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18054 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18055 (_mm256_setr_m128i): New intrinsics.
18057 2017-05-10 Julia Koval <julia.koval@intel.com>
18059 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18060 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18061 (_mm_maskz_rcp14_ss): New intrinsics.
18062 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18063 (__builtin_ia32_rcp14ss_mask): New builtins.
18064 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18066 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18068 PR tree-optimization/51513
18069 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18070 (assert_unreachable_fallthru_edge_p): Use it.
18071 (group_case_labels_stmt): Likewise.
18072 * tree-cfg.h: Prototype it.
18073 * stmt.c: Include cfghooks.h and tree-cfg.h.
18074 (emit_case_dispatch_table) <gap_label>: New local variable.
18075 Use it to fill dispatch table gaps.
18076 Test for default_label before updating probabilities.
18077 (expand_case) <default_label>: Remove unneeded initialization.
18078 Test for unreachable default case statement and remove its edge.
18079 Set default_label accordingly.
18080 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18082 2017-05-10 Carl Love <cel@us.ibm.com>
18084 * config/rs6000/rs6000-c: Add support for built-in functions
18085 vector signed char vec_neg (vector signed char)
18086 vector signed short int vec_neg (vector short int)
18087 vector signed int vec_neg (vector signed int)
18088 vector signed long long vec_neg (vector signed long long)
18089 vector float vec_neg (vector float)
18090 vector double vec_neg (vector double)
18091 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18093 * config/rs6000/altivec.h: Add define for vec_neg
18094 * doc/extend.texi: Update the built-in documentation for the
18095 new built-in functions.
18097 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18099 PR tree-optimization/77644
18100 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18102 2017-05-10 Nathan Sidwell <nathan@acm.org>
18104 * dumpfile.h (TDI_lang_all): New.
18105 (TDF_KIND): New. Renumber others
18106 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18108 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18110 (get_dump_file_name): Adjust suffix generation.
18111 (dump_enable_all): Use TDF_KIND.
18112 * doc/invoke.texi (-fdump-lang-all): Document.
18114 * dumpfile.h: Tabify.
18116 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18119 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18120 Move member access before delete.
18122 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18124 * tree-inline.c (expand_call_inline): Split block at stmt
18127 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18130 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18131 are now unused after splitting mov{sf,sd}_hardfloat.
18132 (f32_lr2): Likewise.
18133 (f32_lm): Likewise.
18134 (f32_lm2): Likewise.
18135 (f32_li): Likewise.
18136 (f32_li2): Likewise.
18137 (f32_lv): Likewise.
18138 (f32_sr): Likewise.
18139 (f32_sr2): Likewise.
18140 (f32_sm): Likewise.
18141 (f32_sm2): Likewise.
18142 (f32_si): Likewise.
18143 (f32_si2): Likewise.
18144 (f32_sv): Likewise.
18145 (f32_dm): Likewise.
18146 (f32_vsx): Likewise.
18147 (f32_av): Likewise.
18148 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18149 For movsf, order stores so the VSX stores occur before the GPR
18150 store which encourages the register allocator to use a traditional
18151 FPR instead of a GPR. For movsd, order the stores so that the GPR
18152 store comes before the VSX stores to allow the power6 to work.
18153 This is due to the power6 not having a 32-bit integer store
18154 instruction from a FPR.
18155 (movsf_hardfloat): Likewise.
18156 (movsd_hardfloat): Likewise.
18158 2017-05-09 Martin Sebor <msebor@redhat.com>
18160 PR translation/80280
18161 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18164 PR translation/80280
18165 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18166 data member added in r247778.
18167 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18169 2017-05-09 Nathan Sidwell <nathan@acm.org>
18171 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18173 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18176 2017-05-09 Marek Polacek <polacek@redhat.com>
18178 * doc/invoke.texi: Fix typo.
18180 2017-05-09 Richard Biener <rguenther@suse.de>
18182 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18183 (vrp_val_is_min): Likewise.
18184 (set_value_range_to_value): Likewise.
18185 (set_value_range_to_nonnegative): Likewise.
18186 (gimple_assign_nonzero_p): Likewise.
18187 (gimple_stmt_nonzero_p): Likewise.
18188 (vrp_int_const_binop): Likewise. Remove unreachable case.
18189 (adjust_range_with_scev): Adjust comments.
18190 (compare_range_with_value): Likewise.
18191 (extract_range_from_phi_node): Likewise.
18192 (test_for_singularity): Likewise.
18194 2017-05-09 Richard Biener <rguenther@suse.de>
18196 * tree-vrp.c (get_single_symbol): Add assert that we don't
18197 get overflowed constants as invariant part.
18198 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18199 checks. Use wi::cmp instead of recursing for integer constants.
18200 (compare_values): Just ignore whether we assumed undefined
18201 overflow instead of failing the compare.
18202 (extract_range_for_var_from_comparison_expr): Add comment before the
18203 TREE_NO_WARNING sets.
18204 (test_for_singularity): Likewise.
18205 (extract_range_from_comparison): Do not disable optimization
18206 when we assumed undefined overflow.
18207 (extract_range_basic): Remove init of unused var.
18209 2017-05-09 Richard Biener <rguenther@suse.de>
18211 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18212 (extract_range_from_multiplicative_op_1): Adjust.
18213 (extract_range_from_binary_expr_1): Use int_const_binop.
18215 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18218 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18219 rs6000_store_data_bypass_p in seven define_bypass directives and
18220 in several comments.
18221 * config/rs6000/rs6000-protos.h: Add prototype for
18222 rs6000_store_data_bypass_p function.
18223 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18224 function implements slightly different (rs6000-specific) semantics
18225 than store_data_bypass_p, returning false rather than aborting
18226 with assertion error when arguments do not satisfy the
18227 requirements of store data bypass.
18228 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18229 rs6000_store_data_bypass_p.
18231 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18233 * config/xtensa/xtensa-protos.h
18234 (xtensa_initial_elimination_offset): New declaration.
18235 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18236 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18237 macro definition, add case for FRAME_POINTER_REGNUM when
18238 FRAME_GROWS_DOWNWARD.
18239 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18240 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18241 xtensa_initial_elimination_offset.
18243 2017-05-08 Nathan Sidwell <nathan@acm.org>
18245 * doc/invoke.texi: Alphabetize -fdump options.
18247 2017-05-08 Martin Sebor <msebor@redhat.com>
18249 PR translation/80280
18250 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18252 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18254 * target.def (compute_frame_layout): New optional target hook.
18255 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18256 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18257 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18259 * reload1.c (verify_initial_elim_offsets): Likewise.
18260 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18261 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18262 (arm_get_frame_offsets): Split up into this ...
18263 (arm_compute_frame_layout): ... and this function.
18265 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18267 * config/aarch64/constraints.md (Usa): New constraint.
18268 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18270 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18272 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18273 with_multilib_list after it has been checked.
18275 2017-05-08 Richard Biener <rguenther@suse.de>
18277 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18278 (bitmap_set_subtract_values): Likewise.
18280 2017-05-08 Richard Biener <rguenther@suse.de>
18282 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18283 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18285 (gimple_stmt_nonzero_warnv_p): Rename to ...
18286 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18288 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18289 (extract_range_basic): Adjust, do not disable propagation on
18290 strict overflow sensitive simplification.
18291 (vrp_visit_cond_stmt): Likewise.
18293 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18295 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18296 body size unconditionally.
18298 2017-05-07 Jeff Law <law@redhat.com>
18301 2017-05-06 Jeff Law <law@redhat.com>
18302 PR tree-optimization/78496
18303 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18306 PR tree-optimization/78496
18307 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18308 (simplify_stmt_using_ranges): Call it.
18309 (vrp_dom_walker::before_dom_children): Extract equivalences
18310 from an ASSERT_EXPR with an equality comparison against a
18313 2017-05-06 Jeff Law <law@redhat.com>
18315 PR tree-optimization/78496
18316 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18319 PR tree-optimization/78496
18320 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18321 (simplify_stmt_using_ranges): Call it.
18322 (vrp_dom_walker::before_dom_children): Extract equivalences
18323 from an ASSERT_EXPR with an equality comparison against a
18326 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18328 * lra-constraints.c (lra_copy_reg_equiv): New function.
18329 (split_reg): Use it to copy equivalence information from the
18330 original register to the spill register.
18332 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18334 PR rtl-optimization/75964
18335 * simplify-rtx.c (simplify_const_relational_operation): Remove
18336 invalid handling of comparisons of integer ABS.
18338 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18340 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18341 initialize to zero.
18342 (init_regs): Remove declaration.
18343 (function_arg_advance_32): Initialize error_p as boolean variable.
18345 2017-05-05 Nathan Sidwell <nathan@acm.org>
18347 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18348 lines. Use for (;;).
18350 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18352 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18353 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18354 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18355 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18356 VF=2 that require versioning.
18358 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18360 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18363 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18365 * diagnostic.h (diagnostic_override_option_index): Convert from
18366 macro to inline function.
18368 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18370 * diagnostic.c (last_module_changed_p): New function.
18371 (set_last_module): New function.
18372 (diagnostic_report_current_module): Convert macro usage to
18373 the above functions.
18374 * diagnostic.h (diagnostic_context::last_module): Strengthen
18375 from const line_map * to const line_map_ordinary *.
18376 (diagnostic_last_module_changed): Delete macro.
18377 (diagnostic_set_last_module): Delete macro.
18379 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18381 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18382 with diagnostic_report_diagnostic.
18383 (diagnostic_n_impl_richloc): Likewise.
18384 * diagnostic.h (report_diagnostic): Delete macro.
18385 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18386 with diagnostic_report_diagnostic.
18387 * substring-locations.c (format_warning_va): Likewise.
18389 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18391 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18392 save/restor of format_spec. Move option-printing code to...
18393 (print_option_information): ...this new function, and
18394 reimplement by simply printing to the pretty_printer,
18395 rather than appending to the format string.
18397 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18399 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18400 handling logic into...
18401 (update_effective_level_from_pragmas): ...this new function.
18403 2017-05-04 Andrew Waterman <andrew@sifive.com>
18405 * config/riscv/riscv.opt (mstrict-align): New option.
18406 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18407 (SLOW_UNALIGNED_ACCESS): Define.
18408 (riscv_slow_unaligned_access): Declare.
18409 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18411 (riscv_slow_unaligned_access): New variable.
18412 (rocket_tune_info): Set slow_unaligned_access to true.
18413 (optimize_size_tune_info): Set slow_unaligned_access to false.
18414 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18415 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18416 (riscv_option_override): Set riscv_slow_unaligned_access.
18417 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18419 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18421 * config/riscv/riscv.md: Unify indentation.
18423 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18428 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18430 (extendsi<mode>2): Add support for doing sign extension via
18431 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18432 don't have ISA 3.0 instructions.
18433 (extendsi<mode>2 splitter): Likewise.
18434 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18435 generate the normal insns since SImode can now go in vector
18436 registers. Disallow the special UNSPECs needed for previous
18437 machines to hide SImode being used. Add new insns
18438 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18439 (fix_trunc<mode>si2_stfiwx): Likewise.
18440 (fix_trunc<mode>si2_internal): Likewise.
18441 (fixuns_trunc<mode>si2): Likewise.
18442 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18443 (fctiw<u>z_<mode>_smallint): Likewise.
18444 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18445 of floating point to 32-bit integer from doing a direct move to
18446 the GPR registers to do a store.
18447 (fctiwz_<mode>): Break long line.
18449 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18451 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18452 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18453 (addr_list, addr_offset_valid_p): New.
18454 (split_address_groups): Check offset validity with above function.
18455 (gt-tree-ssa-loop-ivopts.h): Include header file.
18457 2017-05-05 Nathan Sidwell <nathan@acm.org>
18459 * config.gcc (arm*-*-*): Add missing 'fi'.
18461 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18463 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18464 included in -fopt-info does not matter.
18465 * doc/optinfo.texi (-fopt-info): Fix description of default
18466 behavour. Explicitly say order of options included in -fopt-info
18469 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18471 * config.gcc: Allow combinations of aprofile and rmprofile values for
18472 --with-multilib-list.
18473 * config/arm/t-multilib: New file.
18474 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18475 variables. Remove setting of ISA and floating-point ABI in
18476 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18477 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18478 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18479 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18481 * config/arm/t-rmprofile: Likewise except for the matches changes.
18482 * doc/install.texi (--with-multilib-list): Document the combination of
18483 aprofile and rmprofile values and warn about pitfalls in doing that.
18485 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18487 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18488 (movdi_aarch64): Likewise.
18490 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18492 PR tree-optimization/80632
18493 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18495 (build_arrays): Initialize it for virtual phis.
18496 (fix_phi_nodes): Use it for virtual phis.
18498 PR tree-optimization/80558
18499 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18500 [x, y] op z into [x op, y op z] for op & or | if conditions
18503 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18504 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18507 * config/arm/arm.md (use_literal_pool): Remove.
18508 (64-bit immediate split): No longer takes cost into consideration
18509 if arm_disable_literal_pool is enabled.
18510 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18511 used when arm_disable_literal_pool is enabled.
18512 (arm_max_const_double_inline_cost): Remove use of
18513 arm_disable_literal_pool.
18514 (push_minipool_fix): Add assert.
18515 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18516 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18517 (no_literal_pool_sf_immediate): New.
18519 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18521 PR tree-optimization/80613
18522 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18523 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18525 2017-05-05 Richard Biener <rguenther@suse.de>
18527 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18529 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18531 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18532 of this flag from insn conditions due to removal from r247495.
18534 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18536 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18538 (arm_early_store_addr_dep_ptr): Likewise.
18539 * config/arm/aarch-common-protos.h
18540 (arm_early_load_addr_dep_ptr): Add prototype.
18541 (arm_early_store_addr_dep_ptr): Likewise.
18542 * config/arm/cortex-a53.md: Add new bypasses.
18544 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18546 * tree.c (next_type_uid): Change type to unsigned.
18547 (type_hash_canon): Decrement back next_type_uid if
18548 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18549 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18552 2017-05-04 Martin Sebor <msebor@redhat.com>
18554 * builtins.c: Fix a trivial typo in a comment.
18556 PR middle-end/79234
18557 * builtins.c (check_sizes): Adjust to handle reading past the end.
18558 Avoid printing excessive upper bound of ranges. Use %E to print
18559 tree nodes instead of converting them to %wu.
18560 (expand_builtin_memchr): New function.
18561 (compute_dest_size): Rename...
18562 (compute_objsize): ...to this.
18563 (expand_builtin_memcpy): Adjust.
18564 (expand_builtin_mempcpy): Adjust.
18565 (expand_builtin_strcat): Adjust.
18566 (expand_builtin_strcpy): Adjust.
18567 (check_strncat_sizes): Adjust.
18568 (expand_builtin_strncat): Adjust.
18569 (expand_builtin_strncpy): Adjust and simplify.
18570 (expand_builtin_memset): Adjust.
18571 (expand_builtin_bzero): Adjust.
18572 (expand_builtin_memcmp): Adjust.
18573 (expand_builtin): Handle memcmp.
18574 (maybe_emit_chk_warning): Check strncat just once.
18576 2017-05-04 Martin Sebor <msebor@redhat.com>
18578 PR preprocessor/79214
18579 PR middle-end/79222
18580 PR middle-end/79223
18581 * builtins.c (check_sizes): Add inlining context and issue
18582 warnings even when -Wno-system-headers is set.
18583 (check_strncat_sizes): Same.
18584 (expand_builtin_strncat): Same.
18585 (expand_builtin_memmove): New function.
18586 (expand_builtin_stpncpy): Same.
18587 (expand_builtin): Handle memmove and stpncpy.
18589 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18591 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18592 which is not used any more.
18594 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18596 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18598 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18600 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18601 (cortexa53_tunings): Likewise.
18602 (cortexa57_tunings): Likewise.
18603 (cortexa72_tunings): Likewise.
18604 (cortexa73_tunings): Likewise.
18606 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18608 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18609 Set loop alignment to 8.
18611 2017-05-04 Martin Sebor <msebor@redhat.com>
18613 PR translation/80280
18614 * builtins.c (expand_builtin_object_size): Add missing quoting to
18615 %D and like directives.
18616 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18617 (hsa_type_for_tree_type): Same.
18618 (verify_function_arguments): Same.
18619 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18620 * varasm.c (get_section): Same.
18623 2017-05-04 Martin Sebor <msebor@redhat.com>
18625 PR translation/80280
18626 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18628 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18630 * config/aarch64/aarch64.c (generic_addrcost_table):
18631 Change HI/TI mode setting.
18633 2017-05-04 Martin Jambor <mjambor@suse.cz>
18635 PR tree-optimization/80622
18636 * tree-sra.c (comes_initialized_p): New function.
18637 (build_accesses_from_assign): Only set write lazily when
18638 comes_initialized_p is false.
18639 (analyze_access_subtree): Use comes_initialized_p.
18640 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18641 instead of testing for PARM_DECL.
18643 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18645 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18646 constraint on operand 0 to allow more general addressing modes.
18647 Adjust output template.
18648 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18650 * config/aarch64/aarch64-protos.h
18651 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18652 * config/aarch64/constraints.md (Dp): New address constraint.
18653 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18656 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18658 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18659 update use of estimate_ipcp_clone_size_and_time.
18660 (estimate_local_effects): Update use of
18661 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18662 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18663 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18664 Return nonspecialized time.
18666 2017-05-04 Richard Biener <rguenther@suse.de>
18668 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18669 for the last VUSE which def dominates the PHI. Directly call
18671 (get_continuation_for_phi_1): Remove.
18673 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18675 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18676 to explain the use of truncating division. Cap the number of
18677 iterations to the maximum given by nb_iterations_upper_bound,
18680 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18682 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18683 * configure: Regenerate.
18684 * config.in: Regenerate.
18685 * config/i386/driver-mingw32.c: new file.
18686 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18687 * config.host: Link driver-mingw32.o on MinGW host.
18688 * doc/install.texi: Document new --enable-mingw-wildcard configure
18691 2017-05-04 Marek Polacek <polacek@redhat.com>
18693 PR tree-optimization/80612
18694 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18696 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18697 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18699 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18700 (movt splitter): Likewise.
18701 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18702 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18703 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18704 block for Thumb-1 with MOVT.
18705 (thumb2_legitimate_address_p): Move code block ...
18706 (can_avoid_literal_pool_for_label_p): ... into this new function.
18707 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18709 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18710 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18711 "M-profile targets with the MOVT instruction".
18713 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18715 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18716 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18717 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18719 2017-05-04 Martin Liska <mliska@suse.cz>
18721 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18722 variable cond_code.
18724 2017-05-04 Richard Biener <rguenther@suse.de>
18726 * tree.c (array_at_struct_end_p): Handle arrays at struct
18727 end with flexarrays more conservatively. Refactor and treat
18728 arrays of arrays or aggregates more strict. Fix
18729 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18730 * tree.c (array_at_struct_end_p): Adjust prototype.
18731 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18732 * gimple-fold.c (get_range_strlen): Likewise.
18733 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18735 2017-05-04 Richard Biener <rguenther@suse.de>
18737 PR tree-optimization/31130
18738 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18740 (supports_overflow_infinity): Likewise.
18741 (is_negative_overflow_infinity): Likewise.
18742 (is_positive_overflow_infinity): Likewise.
18743 (is_overflow_infinity): Likewise.
18744 (stmt_overflow_infinity): Likewise.
18745 (overflow_infinity_range_p): Likewise.
18746 (usable_range_p): Remove as always returning true.
18747 (make_overflow_infinity): Remove.
18748 (negative_overflow_infinity): Likewise.
18749 (positive_overflow_infinity): Likewise.
18750 (avoid_overflow_infinity): Likewise.
18751 (set_value_range): Adjust accordingly.
18752 (set_value_range_to_nonnegative): Likewise, remove now unused
18753 overflow_infinity arg.
18754 (vrp_operand_equal_p): Adjust.
18755 (update_value_range): Likewise.
18756 (range_int_cst_singleton_p): Likewise.
18757 (operand_less_p): Likewise.
18758 (compare_values_warnv): Likewise.
18759 (extract_range_for_var_from_comparison_expr): Likewise.
18760 (vrp_int_const_binop): Likewise.
18761 (zero_nonzero_bits_from_vr): Likewise.
18762 (extract_range_from_multiplicative_op_1): Likewise.
18763 (extract_range_from_binary_expr_1): Likewise.
18764 (extract_range_from_unary_expr): Likewise.
18765 (extract_range_from_comparison): Likewise.
18766 (extract_range_basic): Likewise.
18767 (adjust_range_with_scev): Likewise.
18768 (compare_ranges): Likewise.
18769 (compare_range_with_value): Likewise.
18770 (dump_value_range): Likewise.
18771 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18773 (simplify_cond_using_ranges): Adjust.
18775 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18777 * brig-builtins.def: Added a builtin for class_f64.
18778 * builtin-types.def: Added a builtin type needed by class_f64.
18780 2017-05-03 Jason Merrill <jason@redhat.com>
18782 * timevar.def: Add TV_CONSTEXPR.
18784 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18786 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18788 2017-05-03 Martin Jambor <mjambor@suse.cz>
18790 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18791 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18792 * ipa-cp.c (ipcp_propagate_stage): Do not call
18793 ipa_update_after_lto_read.
18794 * ipa-inline.c (ipa_inline): Likewise.
18796 2017-05-03 Martin Jambor <mjambor@suse.cz>
18798 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18799 tag. Added a default constructor and a destructor.
18800 (ipa_edge_args_sum_t): New class;
18801 (ipa_edge_args_sum): Declare.
18802 (ipa_edge_args_vector): Remove declaration.
18803 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18804 (ipa_free_edge_args_substructures): Remove declaration.
18805 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18806 (ipa_edge_args_info_available_for_edge_p): Likewise.
18807 * ipa-prop.c (ipa_edge_args_vector): Removed.
18808 (edge_removal_hook_holder): Likewise.
18809 (edge_duplication_hook_holder): Likewise.
18810 (ipa_edge_args_sum): New variable.
18811 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18812 ipa_edge_args_vector.
18813 (ipa_free_edge_args_substructures): Likewise.
18814 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18815 ipa_edge_args_vector.
18816 (ipa_edge_removal_hook): Turned into method
18817 ipa_edge_args_sum_t::remove.
18818 (ipa_edge_duplication_hook): Turned into method
18819 ipa_edge_args_sum_t::duplicate.
18820 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18821 registering edge hooks.
18822 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18823 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18824 ipa_edge_args_sum instead of ipa_edge_args_vector.
18825 * ipa-profile.c (ipa_profile): Likewise.
18827 2017-05-03 Martin Jambor <mjambor@suse.cz>
18829 * symbol-summary.h (function_summary): New method exists.
18830 (function_summary::symtab_removal): Deallocate through release.
18831 (call_summary): New class.
18832 (gt_ggc_mx): New overload.
18833 (gt_pch_nx): Likewise.
18834 (gt_pch_nx): Likewise.
18836 2017-05-03 Jeff Law <law@redhat.com>
18838 PR tree-optimization/78496
18839 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18840 from simplify_cond_using_ranges. Split off code to walk
18841 backwards through casts into ...
18842 (simplify_cond_using_ranges_2): New function.
18843 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18844 (execute_vrp): After identifying jump threads, call
18845 simplify_cond_using_ranges_2.
18847 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18850 * ipa-inline.h (inline_summary): Add ctor.
18851 (create_ggc): Do not use ggc_cleared_alloc.
18853 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18854 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18856 * gcc.c (handle_braces): Support escaping in switch matching
18858 * doc/invoke.texi (Spec Files): Document it.
18859 Remove superfluous @code markup in items.
18861 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18863 * diagnostic-show-locus.c (struct column_range): New struct.
18864 (get_affected_columns): New function.
18865 (get_printed_columns): New function.
18866 (struct correction): New struct.
18867 (correction::ensure_capacity): New function.
18868 (correction::ensure_terminated): New function.
18869 (struct line_corrections): New struct.
18870 (line_corrections::~line_corrections): New dtor.
18871 (line_corrections::add_hint): New function.
18872 (layout::print_trailing_fixits): Reimplement in terms of the new
18874 (selftest::test_overlapped_fixit_printing): New function.
18875 (selftest::diagnostic_show_locus_c_tests): Call it.
18877 2017-05-03 Nathan Sidwell <nathan@acm.org>
18879 Canonicalize canonical type hashing
18880 * tree.h (type_hash_canon_hash): Declare.
18881 * tree.c (type_hash_list, attribute_hash_list): Move into
18882 type_hash_canon_hash.
18883 (build_type_attribute_qual_variant): Break out hash code calc into
18884 type_hash_canon_hash.
18885 (type_hash_canon_hash): New. Generic type hash computation.
18886 (build_range_type_1, build_array_type_1, build_function_type,
18887 build_method_type_directly, build_offset_type, build_complex_type,
18888 make_vector_type): Call it.
18890 2017-05-03 Richard Biener <rguenther@suse.de>
18892 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18893 When all DRs have unknown misaligned do not always peel
18894 when there is a store but apply the same costing model as if
18895 there were only loads.
18897 2017-05-03 Richard Biener <rguenther@suse.de>
18900 PR tree-optimization/80492
18901 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18902 compare_base_decls returning dont-know properly.
18904 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18906 * config/arm/iterators.md (CCSI): New mode iterator.
18907 (arch): New mode attribute.
18908 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18909 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18910 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18911 code iterator for success result mode.
18912 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18913 the corresponding new insn generators.
18915 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18918 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18919 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18921 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18923 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18924 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18925 (DDR_A): Wrap DDR argument in brackets.
18926 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18927 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18928 (DDR_REVERSED_P): Likewise.
18930 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18932 PR tree-optimization/79472
18933 * tree-switch-conversion.c (struct switch_conv_info): Add
18934 contiguous_range and default_case_nonstandard fields.
18935 (collect_switch_conv_info): Compute contiguous_range and
18936 default_case_nonstandard fields, don't clear final_bb if
18937 contiguous_range and only the default case doesn't have the required
18939 (check_all_empty_except_final): Set default_case_nonstandard instead
18940 of failing if contiguous_range and the default case doesn't have empty
18942 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18943 and only the default case doesn't have the required constants. Skip
18945 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18946 if default_case_nonstandard.
18947 (build_constructors): Build constant 1 just once. Assert that default
18948 values aren't inserted in between cases if contiguous_range. Skip
18950 (build_arrays): Skip virtual phis.
18951 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18952 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18953 Handle virtual phis.
18954 (gen_inbound_check): Handle default_case_nonstandard case.
18955 (process_switch): Adjust check_final_bb caller. Call
18956 gather_default_values with the first non-default case instead of
18957 default case if default_case_nonstandard.
18959 2017-05-02 Nathan Sidwell <nathan@acm.org>
18961 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18962 check. Fix formatting.
18964 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18966 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18967 errors when comparing specialized and unspecialized times.
18969 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18971 * diagnostic-show-locus.c
18972 (layout::should_print_annotation_line_p): Make private.
18973 (layout::print_annotation_line): Make private.
18974 (layout::annotation_line_showed_range_p): Make private.
18975 (layout::show_ruler): Make private.
18976 (layout::print_source_line): Make private. Pass in line and
18977 line_width, rather than calling location_get_source_line. Drop
18979 (layout::print_leading_fixits): New method.
18980 (layout::print_any_fixits): Rename to...
18981 (layout::print_trailing_fixits): ...this, and make private.
18982 Don't print newline fixits.
18983 (diagnostic_show_locus): Move logic for printing one row into...
18984 (layout::print_line): ...this new function. Move the
18985 location_get_source_line call and error-handling from
18986 print_source_line to here. Call print_leading_fixits, and rename
18987 print_any_fixits to print_trailing_fixits.
18988 (selftest::test_fixit_insert_containing_newline): Update now that
18989 newlines are partially supported.
18990 (selftest::test_fixit_insert_containing_newline_2): New test.
18991 (selftest::test_fixit_replace_containing_newline): Update comments.
18992 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18993 * edit-context.c (class added_line): New class.
18994 (class edited_line): Describe newline handling in comment.
18995 (edited_line::actually_edited_p): New method.
18996 (edited_line::print_content): Delete redundant decl.
18997 (edited_line::m_predecessors): New field.
18998 (edited_file::print_content): Call edited_line::print_content.
18999 (edited_file::print_diff): Update to support newlines.
19000 (edited_file::print_diff_hunk): Likewise.
19001 (edited_file::print_run_of_changed_lines): New function.
19002 (edited_file::print_diff_line): Convert to...
19003 (print_diff_line): ...this.
19004 (edited_file::get_effective_line_count): New function.
19005 (edited_line::edited_line): Initialize new field m_predecessors.
19006 (edited_line::~edited_line): Clean up m_predecessors.
19007 (edited_line::apply_fixit): Handle newlines.
19008 (edited_line::get_effective_line_count): New function.
19009 (edited_line::print_content): New function.
19010 (edited_line::print_diff_lines): New function.
19011 (selftest::test_applying_fixits_insert_containing_newline): New
19013 (selftest::test_applying_fixits_replace_containing_newline): New
19015 (selftest::insert_line): New function.
19016 (selftest::test_applying_fixits_multiple_lines): Add example of
19018 (selftest::edit_context_c_tests): Call the new tests.
19020 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19022 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19023 parameter cand. Update dump information.
19024 (get_computation_cost): Update uses.
19026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19028 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19029 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19030 (get_computation_at, rewrite_use_address): Update use of
19031 get_computation_aff.
19033 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19035 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19036 (get_computation): Delete.
19037 (get_computation_cost): Implement like get_computation_cost_at.
19038 Use get_computation_at.
19039 (get_computation_cost_at): Delete.
19040 (rewrite_use_nonlinear_expr): Use get_computation_at.
19041 (rewrite_use_compare, remove_unused_ivs): Ditto.
19043 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19045 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19049 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19050 (ivopts_global_cost_for_size): Rename parameter and update uses.
19051 (iv_ca_recount_cost): Update uses.
19052 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19053 candidates seperately in n_invs and n_cands.
19054 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19056 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19058 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19059 (find_inv_vars_cb): New.
19060 (find_depends): Renamed to ...
19061 (find_inv_vars): ... this.
19062 (add_candidate_1, force_var_cost): Call find_inv_vars.
19063 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19065 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19067 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19068 inv_vars. Add inv_exprs.
19069 (struct iv_cand): Rename depends_on to inv_vars.
19070 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19071 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19072 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19073 (dump_cand): Dump inv_vars.
19074 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19075 (record_invariant, find_depends, add_candidate_1): Ditto.
19076 (set_group_iv_cost, force_var_cost): Ditto.
19077 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19078 (get_computation_cost_at, get_computation_cost): Ditto.
19079 (determine_group_iv_cost_generic): Ditto.
19080 (determine_group_iv_cost_address): Ditto.
19081 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19082 (determine_group_iv_costs): Ditto.
19083 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19084 (iv_ca_set_remove_invariants): Renamed to ...
19085 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19086 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19087 (iv_ca_set_add_invariants): Renamed to ...
19088 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19089 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19090 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19091 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19092 (create_new_ivs): Remove useless dump.
19094 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19096 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19098 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19099 (iv_ca_set_no_cp, create_new_iv): Ditto.
19101 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19103 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19105 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19107 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19108 function tree_check2.
19110 2017-05-02 Martin Liska <mliska@suse.cz>
19112 * doc/gcov.texi: Add missing preposition.
19113 * gcov.c (function_info::function_info): Properly fill up
19114 all member variables.
19116 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19118 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19120 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19122 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19124 2017-05-02 Martin Liska <mliska@suse.cz>
19127 * lto-streamer-in.c (lto_read_tree_1): Remove
19128 LTO_STREAMER_DEBUG.
19129 * lto-streamer.c (struct tree_hash_entry): Likewise.
19130 (struct tree_entry_hasher): Likewise.
19131 (tree_entry_hasher::hash): Likewise.
19132 (tree_entry_hasher::equal): Likewise.
19133 (lto_streamer_init): Likewise.
19134 (lto_orig_address_map): Likewise.
19135 (lto_orig_address_get): Likewise.
19136 (lto_orig_address_remove): Likewise.
19137 * lto-streamer.h: Likewise.
19138 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19139 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19141 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19143 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19144 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19145 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19146 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19147 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19148 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19149 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19150 (mm_maskz_sub_ss): New intrinsics.
19151 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19152 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19153 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19154 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19155 (__builtin_ia32_subss_mask_round): New builtins.
19156 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19157 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19158 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19160 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19161 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19163 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19166 2017-05-02 Martin Jambor <mjambor@suse.cz>
19168 PR tree-optimization/78687
19169 * tree-sra.c (access): New field parent.
19170 (process_subtree_disqualification): New function.
19171 (disqualify_candidate): Call it.
19172 (build_accesses_from_assign): Reset write flag if creating an
19174 (build_access_subtree): Fill in parent field and also prpagate
19175 down grp_write flag.
19176 (create_artificial_child_access): New parameter set_grp_write, set
19177 grp_write to its value.
19178 (propagate_subaccesses_across_link): Also propagate grp_write flag
19180 (propagate_all_subaccesses): Push the closest parent back to work
19181 queue if add_access_to_work_queue returned true.
19183 2017-05-02 Richard Biener <rguenther@suse.de>
19185 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19186 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19187 -fstrict-overflow documentation.
19188 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19189 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19190 flag_strict_overflow.
19191 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19192 * lto-opts.c (lto_write_options): Do not stream it.
19193 * lto-wrapper.c (merge_and_complain): Do not handle it.
19194 * opts.c (default_options_table): Do not set -fstrict-overflow.
19195 (finish_options): Likewise do not clear it when sanitizing.
19196 * simplify-rtx.c (simplify_const_relational_operation): Do not
19197 test flag_strict_overflow.
19199 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19201 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19202 using enabled attribute.
19203 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19204 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19205 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19206 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19207 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19208 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19209 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19210 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19211 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19212 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19214 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19216 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19218 2017-05-02 Richard Biener <rguenther@suse.de>
19220 PR tree-optimization/80591
19222 2017-04-10 Richard Biener <rguenther@suse.de>
19224 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19227 2017-05-02 Richard Biener <rguenther@suse.de>
19229 PR tree-optimization/80549
19230 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19231 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19232 headers if they do not have a preheader.
19234 2017-05-02 Martin Liska <mliska@suse.cz>
19237 * common.opt: Fix typo.
19238 * doc/invoke.texi: Likewise.
19240 2017-05-01 Jan Beulich <jbeulich@suse.com>
19242 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19243 swapping, add (x,x,m,x,n) alternative.
19245 2017-05-01 Nathan Sidwell <nathan@acm.org>
19247 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19248 unnecessary unadjusted_alignment check.
19250 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19253 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19255 * gimplify.c (gimplify_cilk_detach): New function.
19256 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19257 * tree-core.h: Document EXPR_CILK_SPAWN.
19258 * tree.h (EXPR_CILK_SPAWN): Define.
19260 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19262 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19263 to use new fixit_hint representation, using the "replace" logic.
19264 (get_line_span_for_fixit_hint): Likewise.
19265 (layout::print_any_fixits): Likewise.
19266 (selftest::test_one_liner_many_fixits): Rename to...
19267 (selftest::test_one_liner_many_fixits_1): ...this, and update
19268 comment and expected output to reflect that the multiple fix-it
19269 hints are now consolidated into one insertion.
19270 (selftest::test_one_liner_many_fixits_2): New test.
19271 (selftest::test_diagnostic_show_locus_one_liner): Update for
19273 (selftest::test_fixit_consolidation): Update for fix-it API
19275 * diagnostic.c (print_parseable_fixits): Likewise.
19276 * edit-context.c (edited_line::m_line_events): Convert from
19277 auto_vec <line_event *> to auto_vec <line_event>.
19278 (class line_event): Convert from abstract base class to a concrete
19279 class, taking over the role of replace_event.
19280 (class insert_event): Delete.
19281 (class replace_event): Rename to class line_event. Convert to
19283 (edit_context::add_fixits): Reimplement.
19284 (edit_context::apply_insert): Delete.
19285 (edit_context::apply_replace): Rename to...
19286 (edit_context::apply_fixit): ...this. Convert to half-open range.
19287 (edited_file::apply_insert): Delete.
19288 (edited_file::apply_replace): Rename to...
19289 (edited_file::apply_fixit): ...this.
19290 (edited_line::~edited_line): Drop deletion of events.
19291 (edited_line::apply_insert): Delete.
19292 (edited_line::apply_replace): Rename to...
19293 (edited_line::apply_fixit): ...this. Convert to half-open range.
19294 Update for change to type of m_line_events.
19295 * edit-context.h (edit_context::apply_insert): Delete.
19296 (edit_context::apply_replace): Rename to...
19297 (edit_context::apply_fixit): ...this.
19299 2017-05-01 Martin Sebor <msebor@redhat.com>
19301 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19304 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19307 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19308 __get_cpuid_max returns 0.
19309 (__get_cpuid_count): Ditto.
19311 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19313 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19314 replacement expression is another instance of one of its arguments.
19316 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19319 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19320 check for stack push/pop autoinc.
19321 * config/i386/i386.c (ix86_agi_dependent): Return false
19322 if the only reason why modified_in_p returned true is that
19323 addr is SP based and set_insn is a push or pop.
19325 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19327 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19330 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19333 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19334 (account_size_time): Use two predicates - exec_pred and
19336 (evaluate_conditions_for_known_args): Compute both clause and
19338 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19339 (inline_summary_t::duplicate): Update.
19340 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19342 (compute_inline_parameters): Likewise.
19343 (estimate_edge_size_and_time): Update caluclation of time.
19344 (estimate_node_size_and_time): Compute both time and nonspecialized
19346 (estimate_ipcp_clone_size_and_time): Update.
19347 (inline_merge_summary): Update.
19348 (do_estimate_edge_time): Update.
19349 (do_estimate_edge_size): Update.
19350 (do_estimate_edge_hints): Update.
19351 (inline_read_section, inline_write_summary): Stream both new predicates.
19352 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19354 (compute_inlined_call_time): Cleanup.
19355 (big_speedup_p): Update.
19356 (edge_badness): Update.
19357 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19358 (size_time_entry): Replace predicate by exec_predicate and
19359 nonconst_predicate.
19360 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19361 (estimate_edge_time): Return also nonspec_time.
19362 (reset_edge_growth_cache): Update.
19364 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19366 PR rtl-optimization/80491
19367 * ifcvt.c (noce_process_if_block): When looking for x setter
19368 with missing else_bb, don't check only the insn right before
19369 cond_earliest, but look for the last insn that x is modified in
19370 within the same bb.
19372 PR rtl-optimization/80491
19373 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19375 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19377 PR tree-optimization/80487
19378 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19380 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19382 PR tree-optimization/79697
19383 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19384 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19385 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19387 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19388 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19390 2017-04-28 Martin Sebor <msebor@redhat.com>
19392 PR tree-optimization/80523
19393 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19394 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19396 (maybe_warn, format_directive, parse_directive): Use new functions.
19397 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19399 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19401 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19403 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19405 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19406 target_header_dir): Set correctly.
19407 * configure: Regenerated.
19408 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19409 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19410 instead of SYSTEM_HEADER_DIR.
19412 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19414 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19415 (estimate_local_effects): Likewise.
19416 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19417 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19418 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19419 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19420 do_estimate_edge_time, estimate_edge_time): Likewise.
19421 * ipa-inline-analysis.c (estimate_node_size_and_time,
19422 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19423 (estimate_time_after_inlining): Remove.
19425 2017-04-28 Martin Liska <mliska@suse.cz>
19427 * doc/gcov.texi: Enhance documentation of gcov.
19429 2017-04-28 Martin Liska <mliska@suse.cz>
19431 * doc/gcov.texi: Sort options in alphabetic order.
19432 * doc/gcov-dump.texi: Likewise.
19433 * doc/gcov-tool.texi: Likewise.
19434 * gcov.c (print_usage): Likewise.
19435 * gcov-dump.c (print_usage): Likewise.
19436 * gcov-tool.c (print_merge_usage_message): Likewise.
19437 (print_rewrite_usage_message): Likewise.
19438 (print_overlap_usage_message): Likewise.
19440 2017-04-28 Martin Liska <mliska@suse.cz>
19442 PR gcov-profile/53915
19443 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19445 2017-04-28 Martin Liska <mliska@suse.cz>
19447 PR gcov-profile/79891
19448 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19449 is marked by compiler as living on a line.
19450 (get_cycles_count): Remove usage of the union.
19451 (output_intermediate_file): Likewise.
19452 (find_source): Fix GNU coding style.
19453 (accumulate_line_counts): Remove old non-all block mode.
19454 (output_lines): Remove usage of the union.
19455 * profile.c (output_location): Include all BBs, even if
19456 belonging to a same line (and file) as a previous BB.
19458 2017-04-28 Martin Liska <mliska@suse.cz>
19460 * gcov.c (process_args): Handle new argument 'w'.
19461 (read_graph_file): Assign ID to BBs.
19462 (output_branch_count): Display BB # if verbose flag is set.
19463 (output_lines): Likewise for arcs.
19464 (print_usage): Add '--verbose' option help.
19465 * doc/gcov.texi: Document --verbose (-w) option.
19467 2017-04-28 Martin Liska <mliska@suse.cz>
19469 * gcov.c (struct block_location_info): New struct.
19470 (process_file): Fill up the new structure.
19471 (read_graph_file): Replace usage of encoding by the newly added
19473 (add_line_counts): Likewise.
19474 (accumulate_line_counts): Remove usage of the union.
19475 (function_info::function_info): New function.
19476 (function_info::~function_info): Likewise.
19477 (process_file): Call delete instead of release_function.
19478 (release_function): Release the function.
19479 (release_structures): Call delete instead of release_function.
19480 (solve_flow_graph): Replace usage of num_blocks.
19481 (find_exception_blocks): Likewise.
19482 (output_lines): Fix GNU coding style.
19484 2017-04-28 Martin Liska <mliska@suse.cz>
19487 * coverage.c (coverage_remove_note_file): New function.
19488 * coverage.h: Declare the function.
19489 * toplev.c (finalize): Clean if an error has been seen.
19491 2017-04-28 Martin Liska <mliska@suse.cz>
19493 PR gcov-profile/80031
19494 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19495 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19496 * gcov.c (read_graph_file): Read just number of blocks.
19497 * profile.c (branch_prob): Do not stream 0 flags per a basic
19500 2017-04-28 Martin Liska <mliska@suse.cz>
19502 * gcov-dump.c (tag_*): Add new argument to declarations.
19503 (dump_gcov_file): Likewise.
19504 (tag_blocks): Add and use new argument depth.
19505 (tag_arcs): Likewise.
19506 (tag_lines): Likewise.
19507 (tag_counters): Likewise.
19508 (tag_summary): Likewise.
19509 (dump_working_sets): Use depth to do a proper indentation.
19511 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19514 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19515 * symtab.c (symtab_node::debug_symtab): Move definition here.
19517 2017-04-28 Richard Biener <rguenther@suse.de>
19519 * lto-streamer.h (LTO_major_version): Bump to 7.
19521 2017-04-28 Richard Biener <rguenther@suse.de>
19523 * tree-vrp.c (assert_info): New struct.
19524 (add_assert_info): New helper.
19525 (register_edge_assert_for_2): Refactor to add asserts to a vector
19527 (register_edge_assert_for_1): Likewise.
19528 (register_edge_assert_for): Likewise.
19529 (finish_register_edge_assert_for): New helper actually registering
19530 asserts where live on edge.
19531 (find_conditional_asserts): Adjust.
19532 (find_switch_asserts): Likewise.
19533 (evrp_dom_walker::try_find_new_range): Generalize.
19534 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19536 2017-04-27 Marek Polacek <polacek@redhat.com>
19539 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19540 arg10 and arg11 to itype.
19542 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19544 * doc/extend.texi (Object Size Checking): Improve grammar.
19546 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19549 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19550 that the logic for permitting reciprocal estimates matches that
19553 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19556 * tree.c (type_cache_hasher::equal): Only compare
19557 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19558 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
19559 non-aggregate element types.
19560 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
19561 about the flag on ARRAY_TYPEs in the comment, formatting fix.
19563 2017-04-27 Richard Biener <rguenther@suse.de>
19565 PR middle-end/80533
19566 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
19567 stripping ARRAY_REFs from MEM_EXPR make sure we're not
19568 keeping a reference to a trailing array.
19570 2017-04-27 Richard Biener <rguenther@suse.de>
19572 PR middle-end/80539
19573 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
19574 being in loop-closed SSA form conservatively.
19575 (chrec_fold_multiply_poly_poly): Likewise.
19577 2017-04-27 Tamar Christina <tamar.christina@arm.com>
19579 PR middle-end/79665
19580 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
19581 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19583 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19586 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19587 (aarch64_function_arg_alignment): Return unsigned int again, but still
19588 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19589 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19590 Don't emit -Wpsabi note.
19591 (aarch64_function_arg_boundary): Likewise.
19592 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19595 2017-04-26 Nathan Sidwell <nathan@acm.org>
19597 * tree.h (crc32_unsigned_n): Declare.
19598 (crc32_unsigned, crc32_unsigned): Make inline.
19599 * tree.c (crc32_unsigned_bits): Replace with ...
19600 (crc32_unsigned_n): ... this.
19601 (crc32_unsigned, crc32_byte): Remove.
19602 (crc32_string): Remove unnecessary braces.
19604 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19606 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19607 * ipa-inline-analysis.c (MAX_TIME): Remove.
19608 (account_size_time): Use sreal for time.
19609 (dump_inline_summary): Update.
19610 (estimate_function_body_sizes): Update.
19611 (estimate_edge_size_and_time): Update.
19612 (estimate_calls_size_and_time): Update.
19613 (estimate_node_size_and_time): Update.
19614 (inline_merge_summary): Update.
19615 (inline_update_overall_summary): Update.
19616 (estimate_time_after_inlining): Update.
19617 (inline_read_section): Update.
19618 (inline_write_summary): Update.
19619 * ipa-inline.c (compute_uninlined_call_time): Update.
19620 (compute_inlined_call_time): Update.
19621 (recursive_inlining): Update.
19622 (inline_small_functions): Update.
19623 (dump_overall_stats): Update.
19624 * ipa-inline.h: Include sreal.h.
19625 (size_time_entry): Turn time to sreal.
19626 (inline_summary): Turn self_time nad time to sreal.
19628 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19630 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19632 (sreal::stream_out, sreal::stream_in): New.
19633 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19635 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19637 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19640 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19643 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19645 (dimode_scalar_chain::compute_convert_gain): Ditto.
19646 (dimode_scalar_chain::make_vector_copies): Ditto.
19647 (dimode_scalar_chain::convert_reg): Ditto.
19648 (dimode_scalar_chain::convert_insn): Ditto.
19649 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19650 (VI248_AVX512BW_1): New mode iterator.
19651 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19652 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19655 2017-04-25 Martin Sebor <msebor@redhat.com>
19657 PR tree-optimization/80497
19658 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19659 constants are representable in HOST_WIDE_INT.
19660 (parse_directive): Ditto.
19662 2017-04-25 Martin Sebor <msebor@redhat.com>
19665 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19666 (new_zero_array): Adjust signature.
19667 (dom_info::dom_init): Used unsigned rather that size_t.
19668 (dom_info::dom_info): Same.
19670 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19671 Jakub Jelinek <jakub@redhat.com>
19674 * config/arm/arm.c: Include gimple.h.
19675 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19676 returns negative, increment ncrn only if it returned positive.
19677 (arm_needs_doubleword_align): Return int instead of bool,
19678 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19679 members, but if there is any such non-FIELD_DECL
19680 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19681 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19682 returns negative, increment nregs only if it returned positive.
19683 (arm_setup_incoming_varargs): Likewise.
19684 (arm_function_arg_boundary): Emit -Wpsabi note if
19685 arm_needs_doubleword_align returns negative, return
19686 DOUBLEWORD_ALIGNMENT only if it returned positive.
19688 2017-04-25 Marek Polacek <polacek@redhat.com>
19691 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19692 first argument to type.
19694 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19697 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19698 type checks to test for compatibility instead of equality.
19700 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19701 Jakub Jelinek <jakub@redhat.com>
19704 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19706 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19707 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19708 the alignment computation, but return their maximum in warn_alignment.
19709 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19710 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19712 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19713 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19716 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19718 * config/arc/simdext.md (dmpyh): Fix typo.
19720 2017-04-25 Richard Biener <rguenther@suse.de>
19722 PR tree-optimization/80492
19723 * alias.c (compare_base_decls): Handle registers with asm
19724 specification conservatively.
19725 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19726 compare_base_decls returning dont-know properly.
19728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19730 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19731 (legitimate_offset_address_p): New function.
19732 (arc_legitimate_address_p): Use above function.
19734 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19736 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19738 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19740 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19741 ACCH registers whenever they are available.
19743 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19745 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19746 double regs fix when not used.
19748 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19750 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19752 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19753 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19755 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19757 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19758 when emitting short ADD instructions.
19760 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19762 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19764 (cmpsi_cc_c_insn): Likewise.
19765 (cbranchsi4_scratch): Compute proper instruction length using
19766 compact_hreg_operand.
19767 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19769 2017-04-25 Richard Biener <rguenther@suse.de>
19771 PR middle-end/80509
19772 * passes.c (pass_manager::pass_manager): Initialize
19773 m_name_to_pass_map.
19775 2017-04-25 Richard Biener <rguenther@suse.de>
19777 PR tree-optimization/79201
19778 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19780 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19783 * config/s390/vector.md: Split MEM->GPR vector moves for
19784 non-s_operand addresses.
19786 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19789 * config/s390/predicates.md (reload_const_wide_int_operand): New
19791 * config/s390/s390.md ("movti"): Remove d/P alternative.
19792 ("movti_bigconst"): New pattern definition.
19794 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19797 * s390-protos.h (s390_expand_cs_hqi): Removed.
19798 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19799 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19800 modes as well as CCZ1mode and CCZmode.
19801 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19802 signature of s390_emit_compare_and_swap.
19803 (s390_expand_cs_hqi): Likewise, make static.
19804 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19805 compare-and-swap, in some cases.
19806 (s390_expand_cs): Wrapper function.
19807 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19809 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19810 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19811 patterns for small and large integers. Forbid symref memory operands.
19812 Move expander to s390.c. Require cc register.
19813 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19814 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19815 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19816 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19817 symref memory operands. Remove CC mode and call s390_match_ccmode
19819 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19821 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19823 * config/s390/s390.md (define_peephole2): New peephole to help
19824 combining the load-and-test pattern with volatile memory.
19826 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19828 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19829 with CCZmode for TARGET_Z196.
19831 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19833 PR rtl-optimization/80501
19834 * combine.c (make_compound_operation_int): Set subreg_code to SET
19835 even for AND with mask of the sign bit of mode.
19837 PR rtl-optimization/80500
19838 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19839 sum's initial value.
19841 2017-04-25 Julian Brown <julian@codesourcery.com>
19842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19844 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19846 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19848 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19850 2017-04-25 Julian Brown <julian@codesourcery.com>
19851 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19853 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19854 (thunderx2t99_sha): New Reservation.
19856 2017-04-25 Julian Brown <julian@codesourcery.com>
19857 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19859 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19860 type for 1-element load.
19862 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19864 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19866 2017-04-24 Martin Jambor <mjambor@suse.cz>
19868 PR tree-optimization/80293
19869 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19870 char arrays not totally scalarizable if it is false.
19871 (analyze_all_variable_accesses): Pass correct value in the new
19872 parameter. Add a statistics counter.
19874 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19876 PR middle-end/79931
19877 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19879 2017-04-24 Richard Biener <rguenther@suse.de>
19881 PR tree-optimization/80494
19882 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19883 out for complex types.
19885 2017-04-24 Richard Biener <rguenther@suse.de>
19887 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19888 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19889 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19890 (DFS): Adjust and never fail.
19891 (sccvn_dom_walker::fail): Remove.
19892 (sccvn_dom_walker::before_dom_children): Adjust.
19893 (run_scc_vn): Likewise and never fail.
19894 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19895 (pass_fre::execute): Likewise.
19897 2017-04-24 Richard Biener <rguenther@suse.de>
19899 PR tree-optimization/79725
19900 * tree-ssa-sink.c (statement_sink_location): Return whether
19901 failure reason was zero uses. Move that check later.
19902 (sink_code_in_bb): Deal with zero uses by removing the stmt
19905 2017-04-24 Richard Biener <rguenther@suse.de>
19908 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19909 pointer-based references.
19911 2017-04-24 Richard Biener <rguenther@suse.de>
19914 * pass_manager.h (pass_manager::operator new): Remove.
19915 (pass_manager::operator delete): Likewise.
19916 * passes.c (pass_manager::operator new): Remove.
19917 (pass_manager::operator delete): Likewise.
19918 (pass_manager::pass_manager): Zero individual pass members.
19920 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19923 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19924 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19925 Check "XEXP (src, 1)" operand here.
19926 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19927 Check "XEXP (src, 1)" operand here.
19928 (dimode_scalar_chain::make_vector_copies): Detect count register
19929 of a shift instruction. Zero extend count register from QImode
19930 to DImode to satisfy vector shift pattern count operand predicate.
19931 Substitute vector shift count operand with a DImode copy.
19932 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19935 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19937 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19938 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19939 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19940 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19941 (UNSPEC_NOREX_MEM): Remove definition.
19943 2017-04-21 Richard Biener <rguenther@suse.de>
19945 PR tree-optimization/79547
19946 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19947 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19948 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19949 without any constraints.
19951 2017-04-21 Richard Biener <rguenther@suse.de>
19953 PR tree-optimization/78847
19954 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19956 2017-04-21 Richard Biener <rguenther@suse.de>
19958 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19959 (build_distinct_type_copy): Likewise.
19960 (build_variant_type_copy): Likewise.
19961 * tree.c (build_qualified_type): Pass down mem-stat info.
19962 (build_distinct_type_copy): Likewise.
19963 (build_variant_type_copy): Likewise.
19965 2017-04-21 Richard Biener <rguenther@suse.de>
19967 PR tree-optimization/80237
19968 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19970 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19971 for a simplified result.
19973 2016-04-21 Richard Biener <rguenther@suse.de>
19975 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19976 sth as strict as a simple_iv but a chrec without symbols and an
19977 operand defined in the loop we are peeling (and not some subloop).
19978 (propagate_constants_for_unrolling): Propagate all constants.
19980 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19983 * config/i386/i386.c (print_reg): Remove assert for disalowed
19984 regno values, call output_operand_lossage instead.
19986 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19989 * config/i386/constraints.md (Yc): New register constraint.
19990 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19991 Use Yc constraint for alternative 2 of operand 0. Remove
19992 preferred_for_speed attribute.
19994 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19996 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19997 lastprivate clauses in SIMT case.
19999 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20001 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20003 2017-04-20 Richard Biener <rguenther@suse.de>
20005 PR tree-optimization/57796
20006 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20007 as N scalar stores.
20008 (vect_model_load_cost): Cost gathers as N scalar loads.
20010 2017-04-20 Richard Biener <rguenther@suse.de>
20012 * ggc-page.c (ggc_allocated_p): Rename to ...
20013 (safe_lookup_page_table_entry): ... this and return the lookup
20015 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20017 2017-04-20 Richard Biener <rguenther@suse.de>
20019 PR tree-optimization/80453
20020 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20021 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20022 from the conditions.
20023 (vn_phi_eq): Pass them down.
20024 (vn_phi_lookup): Record them.
20025 (vn_phi_insert): Likewise.
20027 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20029 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20030 uninitialized variable warning to avoid buffer overrun.
20032 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20035 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20036 is suppressed for '{ 0 }' in C.
20038 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20040 * BASE-VER: Set to 8.0.0.
20042 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20044 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20045 priority .init_array and .fini_array section with SECTION_NOTYPE
20048 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20050 PR middle-end/80423
20051 * tree.h (build_array_type): Add typeless_storage default argument.
20052 * tree.c (type_cache_hasher::equal): Also compare
20053 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20054 (build_array_type): Add typeless_storage argument, set
20055 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20057 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20058 (build_array_type): Likewise. Add typeless_storage argument.
20060 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20061 Jakub Jelinek <jakub@redhat.com>
20063 PR tree-optimization/80426
20064 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20065 operation on symbolic operands, also compute the overflow for the
20066 invariant part when the operation degenerates into a negation.
20068 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20071 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20072 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20075 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20077 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20080 * config/avr/avr.c (tree.h): Include it.
20081 (cgraph.h): Include it.
20082 (avr_encode_section_info): Don't warn for uninitialized progmem
20083 variable if it's just an alias.
20085 2017-04-19 Richard Biener <rguenther@suse.de>
20088 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20089 when needed by AutoPGO.
20091 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20094 * doc/lto.texi: Remove an extra 'that'.
20096 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20098 PR rtl-optimization/80429
20099 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20100 are only used in debug insns.
20102 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20103 Vladimir Makarov <vmakarov@redhat.com>
20105 * config/sparc/predicates.md (input_operand): Add comment. Return
20106 true for any memory operand when LRA is in progress.
20107 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20109 2017-04-18 Jeff Law <law@redhat.com>
20112 * mips.md ({return,simple_return}_internal): Do not overwrite
20115 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20117 PR tree-optimization/80443
20118 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20119 instead of adding 1, subtract -1 and similarly instead of subtracting
20122 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20124 PR rtl-optimization/80357
20125 * haifa-sched.c (tmp_bitmap): New variable.
20126 (model_recompute): Handle duplicate use records.
20127 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20128 (free_global_sched_pressure_data): Free it.
20130 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20133 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20134 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20135 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20136 instead of SYSTEM_HEADER_DIR.
20138 2017-04-18 Jeff Law <law@redhat.com>
20140 PR middle-end/80422
20141 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20142 predecessors after walking up the insn chain.
20144 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20147 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20148 sizetype type into debug info.
20150 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20153 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20154 unneeded test for TARGET_UPPER_REGS_SF.
20155 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20157 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20160 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20161 instead of gsi_after_labels.
20163 2017-04-18 Jeff Law <law@redhat.com>
20165 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20169 2017-04-13 Jeff Law <law@redhat.com>
20170 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20171 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20173 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20176 * config/avr/avr.c (intl.h): Include it.
20177 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20179 2017-04-18 Martin Liska <mliska@suse.cz>
20181 PR gcov-profile/78783
20182 * gcov-tool.c (gcov_output_files): Validate that destination
20183 file is either removed by the tool or by a user.
20185 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20186 Guy Benyei <guybe@mellanox.com>
20188 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20189 block, and do not negate it, the stored id is already negative.
20191 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20193 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20195 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20198 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20199 masks of options that should be turned off if the VSX vector
20200 options are turned off.
20201 (OTHER_P8_VECTOR_MASKS): Likewise.
20202 (OTHER_VSX_VECTOR_MASKS): Likewise.
20203 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20204 rs6000_disable_incompatible_switches to validate no type switches
20206 (rs6000_incompatible_switch): New function to disallow turning on
20207 other vector options if -mno-vsx, -mno-power8-vector, or
20208 -mno-power9-vector are specified.
20210 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20212 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20214 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20216 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20217 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20218 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20219 (ARG_POINTER_CFA_OFFSET): Likewise.
20221 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20223 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20224 conditions to take advantage of various optimizations.
20226 2017-04-13 Jeff Law <law@redhat.com>
20228 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20229 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20230 (zero_extendsidi2_dext): Likewise.
20232 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20235 * fold-const.c (fold_ternary_loc): Revert
20236 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20239 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20241 PR rtl-optimization/80343
20242 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20243 new scratch pseudo.
20245 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20248 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20249 to ubsan_encode_value.
20251 2017-04-13 Jeff Law <law@redhat.com>
20253 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20254 appearing in DEBUG_INSNs.
20256 2017-04-13 Martin Liska <mliska@suse.cz>
20258 PR gcov-profile/80413
20259 * gcov-io.c (gcov_write_string): Copy to buffer just when
20260 allocated size is greater than zero.
20262 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20265 * dwarf2out.c (decls_for_scope): Ignore declarations of
20266 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20268 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20271 * ipa-visibility.c (non_local_p): Fix typos.
20272 (localize_node): When localizing symbol in same comdat group,
20273 dissolve the group only when we know external symbols are going
20275 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20277 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20279 PR tree-optimization/79390
20280 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20281 order does not result in usable sequence, retry with reversed operand
20287 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20288 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20289 op0 instead of fold_convert_loc (loc, type, arg0).
20291 2017-04-12 Jeff Law <law@redhat.com>
20293 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20294 has a delay slot in the generated code.
20296 * config/cris/cris.md (cris_preferred_reload_class): Return
20297 GENNONACR_REGS rather than GENERAL_REGS.
20299 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20302 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20303 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20304 signedness of the result type.
20306 2017-04-12 Richard Biener <rguenther@suse.de>
20307 Jeff Law <law@redhat.com>
20309 PR tree-optimization/80359
20310 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20311 trim stores to TARGET_MEM_REFs.
20313 2017-04-12 Richard Biener <rguenther@suse.de>
20315 PR tree-optimization/79390
20316 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20317 threading case even more.
20319 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20322 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20323 for quad_address_p for TImode, instead of just not indexed_address.
20325 2017-04-12 Richard Biener <rguenther@suse.de>
20326 Bernd Edlinger <bernd.edlinger@hotmail.de>
20328 PR middle-end/79671
20329 * alias.c (component_uses_parent_alias_set_from): Handle
20330 TYPE_TYPELESS_STORAGE.
20331 (get_alias_set): Likewise.
20332 * tree-core.h (tree_type_common): Add typeless_storage flag.
20333 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20334 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20335 for types containing members with TYPE_TYPELESS_STORAGE.
20336 (place_field): Likewise.
20337 (layout_type): Likewise for ARRAY_TYPE.
20338 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20339 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20340 TYPE_TYPELESS_STORAGE.
20341 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20343 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20346 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20347 first argument to type.
20349 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20353 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20354 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20355 (rs6000_expand_binop_builtin): Likewise.
20356 (rs6000_expand_ternop_builtin): Likewise; also add missing
20357 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20358 vshasigma built-ins.
20359 * doc/extend.texi: Document that vec_xxpermdi's third argument
20360 must be a constant.
20362 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20364 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20365 Use shift_const cost parameter when calculating gain of STV shifts.
20367 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20369 PR rtl-optimization/70478
20370 * lra-constraints.c (process_alt_operands): Check memory for
20371 disfavoring memory insn operand.
20373 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20375 PR middle-end/80100
20376 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20377 left shift in unsigned HOST_WIDE_INT type.
20379 PR rtl-optimization/80385
20380 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20381 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20384 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20385 if they have any depend clauses.
20387 2017-04-11 Martin Liska <mliska@suse.cz>
20390 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20391 * ipa-split.c (split_function): Create a local comdat symbol
20392 if caller is in a comdat group.
20394 2017-04-11 Martin Liska <mliska@suse.cz>
20397 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20400 2017-04-11 Martin Sebor <msebor@redhat.com>
20402 PR middle-end/80364
20403 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20404 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20406 (directive::set_width, directive::set_precision, format_character):
20408 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20411 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20414 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20415 conflict, set target->arch_name instead of target->cpu_name.
20417 2017-04-11 Richard Biener <rguenther@suse.de>
20419 PR tree-optimization/80374
20420 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20421 build_zero_cst, remove fold_convertible_p check again.
20423 2017-04-11 Martin Liska <mliska@suse.cz>
20426 * ubsan.c (instrument_object_size): Do not instrument register
20429 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20432 * config/i386/i386-builtin-types.def
20433 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20434 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20435 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20436 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20437 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20438 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20439 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20440 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20441 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20442 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20443 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20444 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20445 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20446 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20447 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20448 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20449 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20450 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20451 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20452 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20453 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20454 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20455 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20456 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20457 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20458 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20459 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20460 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20461 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20462 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20463 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20464 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20465 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20466 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20467 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20468 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20469 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20470 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20471 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20472 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20473 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20474 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20475 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20476 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20477 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20478 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20480 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20481 flag to second_arg_count, handle 4 argument function type _COUNT
20482 aliases, handle second_arg_count on second argument rather than last.
20484 2017-04-10 Jeff Law <law@redhat.com>
20486 PR tree-optimization/80374
20487 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20488 record anything if we can not convert integer_zero_node to the
20491 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20494 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20495 Enhance special handling given to the TARGET_P9_MINMAX option in
20496 relation to certain other options.
20498 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20500 PR tree-optimization/80153
20501 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20502 remove POINTER_PLUS_EXPR's base part directly, rather than through
20505 2017-04-10 Richard Biener <rguenther@suse.de>
20506 Bin Cheng <bin.cheng@arm.com>
20508 PR tree-optimization/80153
20509 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20510 the first element of pointer type aff_tree. Build result expr in
20512 (add_elt_to_tree): Convert to type unconditionally. Remove other
20513 fold_convert calls.
20514 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20515 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20517 2017-04-10 Richard Biener <rguenther@suse.de>
20519 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20522 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20524 PR rtl-optimization/70478
20525 * lra-constraints.c (curr_small_class_check): New.
20526 (update_and_check_small_class_inputs): New.
20527 (process_alt_operands): Update curr_small_class_check. Disfavor
20528 alternative insn memory operands. Check available regs for small
20531 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20534 * config/mips/mips.opt (-mvirt): Update description.
20535 * doc/invoke.texi (-mvirt): Likewise.
20537 2017-04-10 Richard Biener <rguenther@suse.de>
20539 PR middle-end/80362
20540 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20541 looking for NEGATE_EXPR in -A / -B to A / B folding.
20543 2017-04-10 Martin Liska <mliska@suse.cz>
20545 PR gcov-profile/80224
20546 * gcov.c (print_usage): Fix usage string.
20547 (get_gcov_intermediate_filename): Remove.
20548 (output_gcov_file): Use both for normal and intermediate format.
20549 (generate_results): Do not initialize special file for
20550 intermediate format.
20552 2017-04-10 Richard Biener <rguenther@suse.de>
20554 PR tree-optimization/80304
20555 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20558 2017-04-10 Nathan Sidwell <nathan@acm.org>
20561 * config/rs6000/rs6000.c (rs6000_vector_type): New.
20562 (rs6000_init_builtins): Use it.
20564 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20566 * config/arm/arm.md (<mrc>): Add mode to SET source.
20567 (<mrrc>): Likewise.
20569 2017-04-10 Richard Biener <rguenther@suse.de>
20571 PR middle-end/80344
20572 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
20574 2017-04-10 Jakub Jelinek <jakub@redhat.com>
20577 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
20578 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
20579 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
20580 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
20581 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
20582 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20583 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20584 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20585 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20586 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20587 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20588 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20589 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20590 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20591 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20592 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20593 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20594 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20595 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20596 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20597 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20598 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20599 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20601 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20603 PR rtl-optimization/70478
20604 * lra-constraints.c: Reverse the last patch.
20606 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20608 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20609 Add comment for WCHAR_T.
20611 2017-04-08 Martin Liska <mliska@suse.cz>
20614 2017-04-07 Martin Liska <mliska@suse.cz>
20617 * ipa-split.c (split_function): Add function part to a same comdat
20620 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20623 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20625 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20627 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20628 (vec_pairing): ...to this.
20629 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20630 (rs6000_sched_init): Adjust for name change.
20631 (struct rs6000_sched_context): Likewise.
20632 (rs6000_init_sched_context): Likewise.
20633 (rs6000_set_sched_context): Likewise.
20635 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20641 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20643 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20644 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20645 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20647 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20649 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20651 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20653 PR rtl-optimization/70703
20654 * ira-color.c (update_conflict_hard_regno_costs): Use
20655 int64_t instead of HOST_WIDE_INT.
20657 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20659 PR rtl-optimization/70478
20660 * lra-constraints.c (process_alt_operands): Disfavor alternative
20661 insn memory operands.
20663 2017-04-07 Jeff Law <law@redhat.com>
20665 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20666 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20668 2017-04-07 Martin Liska <mliska@suse.cz>
20671 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20672 Show error message instead of an ICE.
20674 2017-04-07 Martin Liska <mliska@suse.cz>
20677 * ipa-split.c (split_function): Add function part to a same comdat
20680 2017-04-07 Richard Biener <rguenther@suse.de>
20682 PR middle-end/80341
20683 * tree.c (get_unwidened): Also handle ! for_type case for
20685 * convert.c (do_narrow): Split out from ...
20686 (convert_to_integer_1): ... here. Do not pass final truncation
20687 type to get_unwidened for TRUNC_DIV_EXPR.
20689 2017-04-07 Richard Biener <rguenther@suse.de>
20691 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20693 (aff_combination_const): Adjust.
20694 (aff_combination_scale): Likewise.
20695 (aff_combination_add_elt): Likewise.
20696 (aff_combination_add_cst): Likewise.
20697 (aff_combination_convert): Likewise.
20698 (add_elt_to_tree): Likewise. Remove unused argument.
20699 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20701 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20703 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20705 * config/arm/arm.c (arm_default_short_enums): Use
20706 ARM_DEFAULT_SHORT_ENUMS.
20707 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20709 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20712 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20713 members with redundant out-of-class redeclaration.
20715 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20718 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20719 * config/i386/i386.md (*zero_extendsidi2):
20720 Add (?*x,*x) and (?*v,*v) alternatives.
20722 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20725 * config/i386/i386.c (ix86_expand_builtin)
20726 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20727 mode from insn data. Convert operands to insn operand mode.
20728 Copy operands that don't satisfy insn predicate to a register.
20730 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20732 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20735 2017-04-06 Richard Biener <rguenther@suse.de>
20737 PR tree-optimization/80334
20738 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20739 preserve alignment of accesses.
20741 2017-04-06 Richard Biener <rguenther@suse.de>
20743 PR tree-optimization/80262
20744 * tree-sra.c (build_ref_for_offset): Preserve address-space
20746 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20747 Drop useless address-space information on MEM_REF offsets.
20749 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20751 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20753 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20755 PR rtl-optimization/70703
20756 * ira-color.c (update_conflict_hard_regno_costs): Use
20757 HOST_WIDE_INT instead of long.
20759 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20762 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20763 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20765 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20766 for x86_64 target. Handle -m3dnowa option.
20768 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20770 PR rtl-optimization/70703
20771 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20772 (update_conflict_hard_regno_costs): Use long instead of unsigned
20773 arithmetic for cost calculation.
20775 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20776 Bernd Edlinger <bernd.edlinger@hotmail.de>
20779 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20782 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20785 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20786 ptr_mode with Pmode throughout.
20787 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20788 into probe_stack_range and use DImode.
20790 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20793 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20794 call_eh_return is true.
20796 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20798 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20799 Initialize last_match_fntype_index.
20801 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20804 * tree-nvr.c: Include internal-fn.h.
20805 (pass_return_slot::execute): Ignore internal calls without
20808 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20809 Richard Biener <rguenther@suse.de>
20812 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20813 captures used multiple times, except for the last use.
20814 * generic-match-head.c: Include gimplify.h.
20816 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20818 PR tree-optimization/79390
20819 * target.h (struct noce_if_info): Declare.
20820 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20821 * target.def (noce_conversion_profitable_p): New target hook.
20822 * ifcvt.h (struct noce_if_info): New type, moved from ...
20823 * ifcvt.c (struct noce_if_info): ... here.
20824 (noce_conversion_profitable_p): Renamed to ...
20825 (default_noce_conversion_profitable_p): ... this. No longer
20827 (noce_try_store_flag_constants, noce_try_addcc,
20828 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20829 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20830 instead of noce_conversion_profitable_p.
20831 * config/i386/i386.c: Include ifcvt.h.
20832 (ix86_option_override_internal): Don't override
20833 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20834 (ix86_noce_conversion_profitable_p): New function.
20835 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20836 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20837 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20838 * doc/tm.texi: Regenerated.
20840 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20842 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20845 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20848 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20849 instructions for small multiply cores.
20851 2017-04-04 Jeff Law <law@redhat.com>
20853 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20855 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20856 that are not set by the loop over the elements.
20858 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20861 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20862 int mode, convert_modes it to mode as unsigned, otherwise use
20863 lowpart_subreg to mode rather than SImode.
20864 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20865 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20866 Use DImode instead of SImode for the shift count operand.
20867 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20870 2017-04-04 Richard Biener <rguenther@suse.de>
20872 PR middle-end/80281
20873 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20874 arithmetic done for the negate or the plus. Simplify.
20875 (A - (-B) -> A + B): Likewise.
20876 * fold-const.c (split_tree): Make sure to not negate pointers.
20878 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20880 PR rtl-optimization/60818
20881 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20882 a compare of comparisons with the thing compared if this results
20883 in a different machine mode.
20885 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20887 * alias.c (base_alias_check): Fix typo in comment.
20888 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20889 * cgraphunit.c (symbol_table::compile): Likewise.
20890 * collect2.c (maybe_run_lto_and_relink): Likewise.
20891 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20892 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20893 * config/avr/avr.c (avr_map_op_t): Likewise.
20894 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20895 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20896 * config/epiphany/epiphany.md (movcc): Likewise.
20897 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20898 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20900 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20901 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20902 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20903 * config/sh/sh.c (sh_rtx_costs): Likewise.
20904 * fold-const.c (fold_truth_andor): Likewise.
20905 * genautomata.c (collapse_flag): Likewise.
20906 * gengtype.h (struct type::u::s): Likewise.
20907 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20908 * input.c (FORMAT_AMOUNT): Likewise.
20909 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20910 (known_aggs_to_agg_replacement_list): Likewise.
20911 * ipa-inline-analysis.c: Likewise.
20912 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20913 * ipa-polymorphic-call.c
20914 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20915 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20916 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20918 * modulo-sched.c (apply_reg_moves): Likewise.
20919 * omp-expand.c (build_omp_regions_1): Likewise.
20920 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20921 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20922 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20923 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20924 * value-prof.c: Likewise.
20925 * var-tracking.c (val_reset): Likewise.
20927 2017-04-03 Richard Biener <rguenther@suse.de>
20929 PR tree-optimization/80275
20930 * fold-const.c (split_address_to_core_and_offset): Handle
20933 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20935 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20936 descriptors is at least equal to that of functions.
20938 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20940 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20942 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20945 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20946 (mov<IMOD4:mode>): New expander.
20947 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20949 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20951 PR rtl-optimization/79405
20952 * fwprop.c (propagations_left): New variable.
20953 (forward_propagate_into): Decrement it.
20954 (fwprop_init): Initialize it.
20955 (fw_prop): If the variable has reached zero, stop propagating.
20956 (fwprop_addr): Ditto.
20958 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20961 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20962 a FUNCTION_DECL, pass it as decl instead of origin to
20965 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20967 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20970 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20973 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20974 TARGET_VSX_SMALL_INTEGER.
20976 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20978 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20979 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20981 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20983 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20984 extraction from odd-numbered MSA register.
20986 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20988 PR middle-end/80173
20989 * expmed.c (store_bit_field_1): Don't attempt to create
20990 a word subreg out of hard registers wider than word if they
20991 have HARD_REGNO_NREGS of 1 for their mode.
20993 PR middle-end/80163
20994 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20995 conversions to integer types wider than word and pointer.
20998 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20999 (rtx_equal_for_cselib_p): Pass 0 to it.
21000 * cselib.c (cselib_hasher::equal): Likewise.
21001 (rtx_equal_for_cselib_1): Add depth argument. If depth
21002 is 128, don't look up VALUE locs and punt. Increment
21003 depth in recursive calls when walking VALUE locs.
21005 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21007 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21008 (make_gcov_file_name): Use the canonical path name for generating
21010 (read_line): Fix handling of files with ascii null bytes.
21012 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21014 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21015 to initialise a vector register instead
21016 of using a const_int.
21018 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21020 PR translation/80189
21021 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21022 diagnostic messages.
21024 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21027 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21028 (dfp_diex_<mode>): Update mode of operand 1.
21029 * doc/extend.texi (dxex, dxexq): Document change to return type.
21030 (diex, diexq): Document change to argument type.
21032 2017-03-30 Martin Jambor <mjambor@suse.cz>
21035 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21036 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21037 it reflects the signature changes performed at the callee side.
21038 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21039 to cgraph_build_function_type_skip_args.
21040 (build_function_decl_skip_args): Adjust call to the above function.
21042 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21045 * config/i386/sse.md
21046 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21047 register as dest whenever it is a MEM not rtx_equal_p to the
21048 corresponding dup operand, and when forcing into reg move the
21049 reg into the memory afterwards.
21050 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21051 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21052 for the force_reg mode.
21053 (avx512vl_vextractf128<mode>): Use register as dest either
21054 always when a MEM, or when it is a MEM not rtx_equal_p to the
21055 corresponding dup operand, or even not when it is a CONST_VECTOR
21056 depending on the mode and lo vs. hi.
21057 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21059 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21060 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21061 Likewise. Require that operands[2] is even.
21062 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21063 Remove extraneous parens. Require that operands[2] is a multiple
21065 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21066 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21067 disallow memory then.
21069 2017-03-30 Richard Biener <rguenther@suse.de>
21071 PR tree-optimization/77498
21072 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21073 to non-constants over backedges.
21075 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21077 PR rtl-optimization/80233
21078 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21079 as last_combined_insn. Do not test for BARRIER_P separately.
21081 2017-03-29 Andreas Schwab <schwab@suse.de>
21084 * calls.c (prepare_call_address): Convert funexp to Pmode before
21085 copying to temp reg.
21087 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21089 PR tree-optimization/80158
21090 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21091 Handle possible future case of more than one alternate
21093 (replace_rhs_if_not_dup): Likewise.
21094 (replace_one_candidate): Likewise.
21096 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21098 PR rtl-optimization/80193
21099 * ira.c (ira): Do not check allocation for LRA.
21101 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21103 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21104 (nvptx_output_simt_exit): Declare.
21105 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21106 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21107 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21108 (nvptx_declare_function_name): ...here. Emit declaration of local
21109 memory space buffer for omp_simt_enter insn.
21110 (nvptx_output_unisimt_switch): New.
21111 (nvptx_output_softstack_switch): New.
21112 (nvptx_output_simt_enter): New.
21113 (nvptx_output_simt_exit): New.
21114 * config/nvptx/nvptx.h (struct machine_function): New fields
21115 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21116 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21117 (UNSPECV_SIMT_EXIT): Ditto.
21118 (omp_simt_enter_insn): New insn.
21119 (omp_simt_enter): New expansion.
21120 (omp_simt_exit): New insn.
21121 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21123 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21124 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21125 (expand_GOMP_SIMT_EXIT): New.
21126 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21127 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21128 (GOMP_SIMT_EXIT): Ditto.
21129 * target-insns.def (omp_simt_enter): New insn.
21130 (omp_simt_exit): Ditto.
21131 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21133 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21134 (lower_rec_input_clauses): Likewise.
21135 (lower_lastprivate_clauses): Handle SIMT privatization.
21137 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21138 (ompdevlow_adjust_simt_enter): New.
21139 (find_simtpriv_var_op): New.
21140 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21141 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21143 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21144 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21145 (copy_decl_for_dup_finish): Ditto.
21147 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21149 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21152 * config/i386/i386.c (ix86_option_override_internal): Always
21153 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21155 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21157 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21159 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21161 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21162 mark new edge's irreducible flag accordign to it.
21163 (vect_do_peeling): Check loop preheader edge's irreducible flag
21164 and pass it to function slpeel_add_loop_guard.
21166 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21168 PR tree-optimization/80218
21169 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21170 Update block frequencies and counts.
21172 2017-03-28 Richard Biener <rguenther@suse.de>
21174 PR tree-optimization/78644
21175 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21176 of a simplification result we may not use it at all.
21178 2017-03-28 Richard Biener <rguenther@suse.de>
21181 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21182 without arguments, generate default definition of a SSA name.
21184 2017-03-28 Richard Biener <rguenther@suse.de>
21186 PR middle-end/80222
21187 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21188 TYPE_REF_CAN_ALIAS_ALL references.
21189 * fold-const.c (fold_indirect_ref_1): Likewise.
21191 2017-03-28 Martin Liska <mliska@suse.cz>
21194 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21195 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21197 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21198 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21200 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21201 (EXTRA_SPECS): Define.
21202 (SUBTARGET_EXTRA_SPECS): Likewise.
21203 (SUBTARGET_CPP_SPEC): Likewise.
21204 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21205 SUBTARGET_EXTRA_SPECS.
21206 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21208 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21210 * config/arc/simdext.md (vst64_insn): Update pattern.
21211 (vld32wh_insn): Likewise.
21212 (vld32wl_insn): Likewise.
21213 (vld64_insn): Likewise.
21214 (vld32_insn): Likewise.
21216 2017-03-28 Marek Polacek <polacek@redhat.com>
21219 * fold-const.c (fold_comparison): Use protected_set_expr_location
21220 instead of SET_EXPR_LOCATION.
21222 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21224 * tree.c (add_expr): Avoid name lookup warning.
21226 2017-03-27 Jeff Law <law@redhat.com>
21228 PR tree-optimization/80216
21229 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21230 function name. Limit recursion depth.
21231 (record_temporary_equivalences): Corresponding changes.
21233 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21235 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21238 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21241 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21243 * cfgcleanup.c (reg_note_cfa_p): New array.
21244 (insns_have_identical_cfa_notes): New function.
21245 (old_insns_match_p): Don't cross-jump in between /f
21246 and non-/f instructions. If both i1 and i2 are frame related,
21247 verify all CFA notes, their order and content.
21249 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21252 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21253 HImode and SImode with zero extend to DImode to one insn.
21254 (bswap<mode>2_extenddi): Likewise.
21255 (bswapsi2_extenddi): Likewise.
21256 (bswaphi2_extendsi): Likewise.
21257 (bswaphi2): Combine bswap HImode and SImode into one insn.
21258 Separate memory insns from swapping register.
21259 (bswapsi2): Likewise.
21260 (bswap<mode>2): Likewise.
21261 (bswaphi2_internal): Delete, no longer used.
21262 (bswapsi2_internal): Likewise.
21263 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21264 store, and gpr<-gpr swap insns.
21265 (bswap<mode>2_store): Likewise.
21266 (bswaphi2_reg): Register only splitter, combine with the splitter.
21267 (bswaphi2 splitter): Likewise.
21268 (bswapsi2_reg): Likewise.
21269 (bswapsi2 splitter): Likewise.
21270 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21271 the insns into load, store, and register/register insns.
21272 (bswapdi2_ldbrx): Likewise.
21273 (bswapdi2_load): Likewise.
21274 (bswapdi2_store): Likewise.
21275 (bswapdi2_reg): Likewise.
21277 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21279 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21280 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21282 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21285 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21288 special handling for target option conflicts between dform
21289 options (-mpower9-dform, -mpower9-dform-vector,
21290 -mpower9-dform-scalar) and -mno-direct-move.
21292 2017-03-27 Richard Biener <rguenther@suse.de>
21294 PR tree-optimization/80181
21295 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21297 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21299 * config/arc/predicates.md (move_double_src_operand): Replace the
21300 call to move_double_src_operand with a call to address_operand.
21302 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21304 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21305 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21306 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21308 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21310 * config/arc/predicates.md (long_immediate_loadstore_operand):
21311 Consider scaled addresses cases.
21313 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21315 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21316 restored when in interrupt.
21317 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21318 doesn't have delay slot.
21320 2017-03-27 Richard Biener <rguenther@suse.de>
21323 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21324 inlined thunk clones.
21326 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21329 * asan.c (instrument_derefs): Copy over last operand from
21330 original COMPONENT_REF to the new COMPONENT_REF with
21331 DECL_BIT_FIELD_REPRESENTATIVE.
21332 * ubsan.c (instrument_object_size): Likewise.
21334 2017-03-27 Richard Biener <rguenther@suse.de>
21336 PR tree-optimization/80170
21337 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21338 sure DR/SCEV didnt fold in constants we do not see when looking
21339 at the reference base alignment.
21341 2017-03-27 Richard Biener <rguenther@suse.de>
21343 PR middle-end/80171
21344 * gimple-fold.c (fold_ctor_reference): Properly guard against
21345 NULL return value from canonicalize_constructor_val.
21347 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21350 * config/i386/i386.c (ix86_expand_builtin)
21351 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21352 flags reg setting and flags reg using instructions.
21353 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21354 clobbering instructions to zero extend op2.
21356 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21358 * doc/install.texi (Configuration) <--with-aix-soname>:
21359 Update link to AIX ld.
21361 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21363 PR rtl-optimization/80160
21364 PR rtl-optimization/80159
21365 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21366 reg_alternate_class into account.
21368 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21371 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21372 to consider in curr_insn_transform.
21374 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21376 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21377 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21378 and emit_mode_inner.
21380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21382 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21383 argument to the overloaded builtin variants. Use the new flag to
21384 deprecate certain builtin variants.
21385 * config/s390/s390-builtin-types.def: Add new builtin types.
21386 * config/s390/s390-builtins.h: Support new flags field for
21387 overloaded builtins.
21388 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21389 (s390_macro_to_expand): Enable vector float data type.
21390 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21391 builtins by incrementing the __VEC__ version number.
21392 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21394 (s390_resolve_overloaded_builtin): Emit error messages depending
21395 on the builtin flags.
21396 * config/s390/s390.c (s390_expand_builtin): Support additional
21397 flags argument. Change error message to match the messages
21398 emitted in s390-c.c.
21399 * config/s390/s390.md: New UNSPEC_* constants.
21400 (op_type): Add new instruction types.
21401 * config/s390/vecintrin.h: Add new builtins and test data class
21403 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21404 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21405 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21406 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21407 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21408 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21410 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21411 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21412 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21413 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21415 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21416 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21417 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21418 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21420 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21421 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21422 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21423 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21424 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21425 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21426 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21428 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21429 ("vec_scatter_element<V_HW_4:mode>_DI")
21430 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21431 ("vec_fpint<mode>", "vflls")
21432 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21433 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21434 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21435 ("*vec_cmphe<mode>_cc"): ... these.
21437 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21438 mode constant instead of magic value.
21440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21442 * config/s390/s390.c (s390_expand_vec_compare): Support other
21443 vector floating point modes than just V2DF.
21444 (s390_expand_vcond): Likewise.
21445 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21446 (s390_cannot_change_mode_class): Prevent mode changes between TF
21447 and V1TF in vector registers.
21448 * config/s390/s390.md (DF, SF): New mode attributes.
21449 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21450 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21451 SFmode support for VRs.
21452 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21454 (VFT, VF_HW): New mode iterators.
21455 (vw, sdx): New mode attributes.
21456 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21457 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21458 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21459 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21460 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21461 also the new vector floating point modes. Renaming to ...
21463 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21464 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21465 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21466 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21467 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21468 ("vec_unordered<mode>"): ... these.
21470 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21471 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21472 ("*vec_extendv2df"): New insn definitions.
21474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21476 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21477 ("mulditi3_2", "*muldi3_sign"): New patterns.
21478 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21479 rename the pattern definition.
21481 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21483 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21485 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21489 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21490 instruction if possible.
21491 * config/s390/vector.md (vec_halfnumelts): New mode
21493 ("*vec_vllezlf<mode>"): New pattern.
21495 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21497 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21498 ("popcountv4si2", "popcountv2di2"): Rename to ...
21499 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21500 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21502 ("popcount<mode>2_vxe"): New pattern.
21504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21506 * common/config/s390/s390-common.c (processor_flags_table): Add
21508 * config.gcc: Add arch12.
21509 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21510 Default to arch12 for unknown CPU model numbers.
21511 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21512 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21513 PROCESSOR_max sanity check.
21514 * config/s390/s390-opts.h (enum processor_type): Add
21516 * config/s390/s390.c (processor_table): Add arch12.
21517 (s390_expand_builtin): Add check for B_VXE flag.
21518 (s390_issue_rate): Add PROCESSOR_ARCH12.
21519 (s390_get_sched_attrmask): Likewise.
21520 (s390_get_unit_mask): Likewise.
21521 (s390_sched_score): Enable z13 scheduling for arch12.
21522 (s390_sched_reorder): Likewise.
21523 (s390_sched_variable_issue): Likewise.
21524 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21526 (s390_tune_attr): Use z13 scheduling also for arch12.
21527 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21528 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21529 (TARGET_VXE_P): New macros.
21530 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21531 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21532 * config/s390/s390.opt: Add arch12 as processor_type.
21534 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21536 * config/s390/s390.md
21537 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21538 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21539 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21541 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21542 Rename expanders to ...
21544 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21545 ("fixuns_truncdddi2_emu"): ... these.
21547 ("fixuns_trunc<mode>si2_emu"): New expander.
21549 ("*fixuns_truncdfdi2_z13"): Rename to ...
21550 ("*fixuns_truncdfdi2_vx"): ... this.
21552 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21554 * config/s390/2964.md: Remove the single element vector compare
21555 instructions which are no longer used.
21556 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21558 (s390_canonicalize_comparison): Remove handling of DFmode
21560 (s390_expand_vec_compare_scalar): Remove function.
21561 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
21562 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
21564 ("*cmp<mode>_ccs"): Add wfcdb instruction.
21566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21568 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
21570 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
21571 will anyway by matched by mov<mode>_64dfp.
21573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21575 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
21576 vlef/vstef. Add missing operand to vleif.
21578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21580 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
21581 pair for all vector types with 64 bit elements.
21582 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21583 * config/s390/vector.md (V_HW_64): ... here.
21584 (V_128_NOSINGLE): New mode iterator.
21585 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21586 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21587 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21588 ("*vec_load_pairv2di"): Change to ...
21589 ("*vec_load_pair<mode>"): ... this one.
21591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21593 * config/s390/constraints.md: Add comments.
21594 (jKK): Reject element sizes > 8 bytes.
21595 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21597 * config/s390/s390.md: Add the s_operand checks formerly in
21598 s390_split_ok_p to various splitters where they are still
21600 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21601 for 128 bit vectors. Plus two splitters.
21603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21605 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21611 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21612 error if the boundary argument is not constant.
21614 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21616 PR rtl-optimization/80112
21617 * loop-doloop.c (doloop_condition_get): Don't check condition
21618 if cmp isn't SET with IF_THEN_ELSE src.
21620 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21622 PR tree-optimization/80158
21623 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21624 replacing a candidate statement, also replace it for the
21625 candidate's alternate interpretation.
21626 (replace_rhs_if_not_dup): Likewise.
21627 (replace_one_candidate): Likewise.
21629 2017-03-24 Richard Biener <rguenther@suse.de>
21631 PR tree-optimization/80167
21632 * graphite-isl-ast-to-gimple.c
21633 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21635 (translate_isl_ast_to_gimple::get_rename): Likewise.
21637 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21639 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21640 handling of certain combinations of target options, including the
21641 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21642 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21644 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21647 * config/arm/arm.md (*load_multiple): Add reload_completed to
21648 matching condition.
21650 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21651 Richard Biener <rguenth@suse.de>
21653 PR tree-optimization/79908
21654 PR tree-optimization/80136
21655 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21656 been cast away, gimplify_and_add suffices.
21658 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21660 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21662 2017-03-23 Richard Biener <rguenther@suse.de>
21664 PR tree-optimization/80032
21665 * gimplify.c (gimple_push_cleanup): Forced unconditional
21666 cleanups still have to go to the conditional_cleanups
21669 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21671 PR tree-optimization/80072
21672 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21674 (next_operand_entry_id): Change type to unsigned int.
21675 (sort_by_operand_rank): Make sure to return the right return value
21676 even if unsigned fields are bigger than INT_MAX.
21677 (struct oecount): Change cnt and id type to unsigned int.
21678 (oecount_hasher::equal): Formatting fix.
21679 (oecount_cmp): Make sure to return the right return value
21680 even if unsigned fields are bigger than INT_MAX.
21681 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21684 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21685 TREE_READONLY on result if writing it more than once.
21688 * doc/invoke.texi (-fsanitize=thread): Document that with
21689 -fnon-call-exceptions atomics are not able to throw
21693 * tsan.c: Include tree-eh.h.
21694 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21695 maybe_clean_or_replace_eh_stmt where needed.
21696 (instrument_memory_accesses): Add cfg_changed argument.
21697 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21698 if it returned true.
21699 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21701 PR rtl-optimization/63191
21702 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21703 wrapper function, moved the whole old content into ...
21704 (ix86_delegitimize_address_1): ... this. New inline function.
21705 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21706 true as last argument instead of ix86_delegitimize_address.
21708 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21710 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21711 cortexa57_branch_cost.
21713 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21715 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21717 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21720 * doc/md.texi (Constraints): Document wA constraint.
21721 * config/rs6000/constraints.md (wA): New.
21722 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21723 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21724 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21725 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21727 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21730 * gimplify.c (is_oacc_declared): New function.
21731 (oacc_default_clause): Use it to set default flags for acc declared
21732 variables inside parallel regions.
21733 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21734 declared variables.
21735 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21736 declare attribute to any decl as necessary.
21738 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21741 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21742 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21743 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21744 (arm_arch_lpae): This.
21745 * config/arm/arm.c (arm_arch7ve): Rename into ...
21746 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21747 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21750 2017-03-22 Martin Liska <mliska@suse.cz>
21753 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21754 error message instead of an ICE.
21756 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21758 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21760 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21762 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21764 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21767 2017-03-21 Martin Sebor <msebor@redhat.com>
21769 * doc/extend.texi: Use "cannot" instead of "can't."
21770 * doc/hostconfig.texi: Same.
21771 * doc/install.texi: Same.
21772 * doc/invoke.texi: Same.
21773 * doc/loop.texi: Same.
21774 * doc/md.texi: Same.
21775 * doc/objc.texi: Same.
21776 * doc/rtl.texi: Same.
21777 * doc/tm.texi: Same.
21778 * doc/tm.texi.in: Same.
21779 * doc/trouble.texi: Same.
21781 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21784 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21785 (collect_checksum_attributes): Set it.
21786 (die_checksum_ordered): Use it.
21788 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21790 PR tree-optimization/79908
21791 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21792 change: For a VA_ARG whose LHS has been cast away, use
21793 force_gimple_operand to construct the side effects.
21795 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21797 PR translation/80001
21798 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21799 more amenable to translation.
21800 (oacc_loop_auto_partitions): Likewise.
21802 2017-03-21 Marek Polacek <polacek@redhat.com>
21803 Martin Sebor <msebor@redhat.com>
21805 PR tree-optimization/80109
21806 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21807 on INTEGRAL_TYPE_P.
21809 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21810 Segher Boessenkool <segher@kernel.crashing.org>
21813 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21814 check reg_used_between_p between insn and one of succ or succ2
21815 depending on if succ is artificial insn not inserted into insn
21818 2017-03-21 Martin Liska <mliska@suse.cz>
21820 PR gcov-profile/80081
21821 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21822 * doc/gcc.texi: Include gcov-dump stuff.
21823 * doc/gcov-dump.texi: New file.
21825 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21827 PR rtl-optimization/79150
21828 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21829 conditional jump, if the jump is the last insn of the loop.
21831 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21832 Richard Biener <rguenth@suse.de>
21834 PR tree-optimization/79908
21835 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21836 been cast away, use force_gimple_operand to construct the side
21839 2017-03-21 Martin Liska <mliska@suse.cz>
21841 PR libfortran/79956
21842 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21845 2017-03-21 Brad Spengler <spender@grsecurity.net>
21848 * plugin.c (htab_hash_plugin): New function.
21849 (add_new_plugin): Use it and adjust.
21850 (parse_plugin_arg_opt): Adjust.
21851 (init_one_plugin): Likewise.
21853 2017-03-21 Richard Biener <rguenther@suse.de>
21855 PR tree-optimization/80032
21856 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21857 if set force the cleanup to happen unconditionally.
21858 (gimplify_target_expr): Push inserted clobbers with force_uncond
21859 to avoid them being removed by control-dependent DCE.
21861 2017-03-21 Richard Biener <rguenther@suse.de>
21863 PR tree-optimization/80122
21864 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21865 va_arg_pack_len when the inlined call stmt requires pack
21867 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21869 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21872 * tsan.c (instrument_builtin_call): If the memory model argument
21873 is not a constant, assume it is valid.
21876 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21879 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21881 PR rtl-optimization/79910
21882 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21883 if its dest is used by an insn before I2 (other than the combined
21884 insns themselves, which are properly handled already).
21886 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21889 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21891 * combine.c (record_used_regs): New static function.
21892 (try_combine): Handle situations where there is an additional
21893 instruction between I2 and I3 which needs to have a LOG_LINK
21897 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21899 * combine.c (try_combine): Delete redundant i1 test. Call
21900 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21902 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21905 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21906 alternatives 13/14.
21908 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21910 PR tree-optimization/80054
21911 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21912 the optimization if a PHI or any of its arguments is not dominated
21913 by the candidate's basis. Use gphi* rather than gimple* as
21915 (replace_profitable_candidates): Clean up a gimple* variable that
21916 should be a gphi* variable.
21918 2017-03-20 Martin Sebor <msebor@redhat.com>
21921 * doc/extend.texi (attribute constructor): Document present limitation.
21923 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21926 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21927 __POWER9_VECTOR__ #ifdef control, change template definition to
21928 use Power9-specific built-in function.
21929 (vec_any_eq): Likewise.
21930 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21931 to control outcomes from this test.
21932 (vector_ae_<mode>p): For VEC_F modes, likewise.
21934 2017-03-20 Ian Lance Taylor <iant@google.com>
21936 * config/i386/i386.c (ix86_function_regparm): Save an extra
21937 register for -fsplit-stack with DECL_STATIC_CHAIN.
21939 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21942 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21943 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21945 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21947 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21949 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21952 2017-03-20 Marek Polacek <polacek@redhat.com>
21955 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21957 2017-03-20 Richard Biener <rguenther@suse.de>
21959 PR tree-optimization/80113
21960 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21961 allocate extra SSA name for PHI def.
21962 (add_close_phis_to_outer_loops): Likewise.
21963 (add_close_phis_to_merge_points): Likewise.
21964 (copy_loop_close_phi_args): Likewise.
21965 (copy_cond_phi_nodes): Likewise.
21967 2017-03-20 Martin Liska <mliska@suse.cz>
21969 PR middle-end/79753
21970 * tree-chkp.c (chkp_build_returned_bound): Do not build
21971 returned bounds for a LHS that's not a BOUNDED_P type.
21973 2017-03-20 Martin Liska <mliska@suse.cz>
21977 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21978 COMPLEX_CST and VECTOR_CST.
21980 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21983 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21984 target operand. A new splitter adds the clobber statement in case
21985 the target operand is dead anyway.
21987 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21989 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21990 to age-old versions of binutils and glibc.
21992 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21994 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21996 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21998 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22000 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22002 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22003 requirement for binutils 2.13.
22005 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22007 * combine.c (try_combine): Delete redundant i1 test. Call
22008 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22010 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22012 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22013 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22015 <riscv64-*-elf>: Re-arrange section
22016 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22017 <riscv32-*-linux>: Likewise.
22018 <riscv64-*-elf>: Likewise
22019 <riscv64-*-linux>: Likewise.
22021 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22024 * aarch64.opt(verbose-cost-dump): Fix typo.
22026 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22029 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22030 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22032 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22034 * reload.c (find_reloads): When reloading a nonoffsettable address,
22035 use RELOAD_OTHER for it and its address reloads.
22037 PR rtl-optimization/79910
22038 * combine.c (record_used_regs): New static function.
22039 (try_combine): Handle situations where there is an additional
22040 instruction between I2 and I3 which needs to have a LOG_LINK
22043 2017-03-17 Jeff Law <law@redhat.com>
22045 PR tree-optimization/71437
22046 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22047 conditional in the hash table first.
22048 (vrp_dom_walker::before_dom_children): Extract condition from
22049 ASSERT_EXPR. Record condition, its inverion and any implied
22050 conditions as well.
22052 2017-03-17 Marek Polacek <polacek@redhat.com>
22053 Markus Trippelsdorf <markus@trippelsdorf.de>
22055 PR tree-optimization/80079
22056 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22059 2017-03-17 Richard Biener <rguenther@suse.de>
22061 PR middle-end/80075
22062 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22063 Properly verify the LHS before the RHS possibly claims to be
22065 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22068 2017-03-17 Martin Jambor <mjambor@suse.cz>
22070 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22071 (List of -O2 options): Likewise.
22072 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22075 2017-03-17 Tom de Vries <tom@codesourcery.com>
22077 * gcov-dump.c (print_usage): Print bug_report_url.
22079 2017-03-17 Richard Biener <rguenther@suse.de>
22081 PR middle-end/80050
22082 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22083 (parser::peek): Likewise.
22085 2017-03-17 Richard Biener <rguenther@suse.de>
22087 PR tree-optimization/80048
22088 * sese.c (free_sese_info): Properly release rename_map and
22089 copied_bb_map elements.
22091 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22093 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22094 Add linked-list forward and backlinks. Insert on
22095 construction, remove on destruction.
22096 (class pass_store_merging): Add m_stores_head field.
22097 (pass_store_merging::terminate_and_process_all_chains):
22098 Iterate over m_stores_head list.
22099 (pass_store_merging::terminate_all_aliasing_chains):
22101 (pass_store_merging::execute): Check for debug stmts first.
22102 Push new chains onto the m_stores_head stack.
22104 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22107 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22108 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22109 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22111 2017-03-16 Jeff Law <law@redhat.com>
22113 PR tree-optimization/71437
22114 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22115 member function. Implementation moved into after_dom_children
22116 member function and into the threader's thread_outgoing_edges
22118 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22119 some code into new thread_outgoing_edges.
22120 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22121 definition. Simplify marker handling (do it here). Assume we always
22122 have the available expression and the const/copies tables.
22123 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22125 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22126 * tree-vrp.c (equiv_stack): No longer file scoped.
22127 (vrp_dom_walker): New class.
22128 (vrp_dom_walker::before_dom_children): New member function.
22129 (vrp_dom_walker::after_dom_children): Likewise.
22130 (identify_jump_threads): Setup domwalker. Use it rather than
22131 walking edges in a random order by hand. Simplify setup/finalization.
22132 (finalize_jump_threads): Remove.
22133 (vrp_finalize): Do not call identify_jump_threads here.
22134 (execute_vrp): Do it here instead and call thread_through_all_blocks
22137 PR tree-optimization/71437
22138 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22140 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22142 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22143 (dom_opt_dom_walker::thread_across_edge): Remove
22144 handle_dominating_asserts argument. All callers changed.
22145 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22146 changes. Remove calls to lhs_of_dominating_assert. Other
22147 uses of handle_dominating_asserts turn into unconditional code
22148 (simplify_control_stmt_condition_1): Likewise.
22149 (simplify_control_stmt_condition): Likewise.
22150 (thread_through_normal_block, thread_across_edge): Likewise.
22151 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22152 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22153 object if it is not an SSA_NAME.
22154 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22155 before calling into the VRP specific simplifiers.
22156 (identify_jump_threads): Remove handle_dominating_asserts
22159 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22162 * tree-diagnostic.c (default_tree_printer): No longer static.
22163 * tree-diagnostic.h (default_tree_printer): New prototype.
22165 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22167 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22168 Change ins into fmov.
22170 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22172 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22173 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22174 Use h_con constraint for operand 1.
22175 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22176 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22178 2017-03-15 Jeff Law <law@redhat.com>
22180 PR tree-optimization/71437
22181 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22182 (record_temporary_equivalences): Use it.
22184 PR tree-optimization/71437
22185 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22186 tree-ssa-scopedtables.
22187 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22188 (record_conditions, record_cond, vuse_eq): Likewise.
22189 (record_edge_info): Adjust to API tweak of record_conditions.
22190 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22191 (record_temporary_equivalences, optimize_stmt): Likewise.
22192 (eliminate_redundant_computations): Likewise.
22193 (record_equivalences_from_stmt): Likewise.
22194 * tree-ssa-scopedtables.c: Include options.h and params.h.
22195 (vuse_eq): New function, moved from tree-ssa-dom.c
22196 (build_and_record_new_cond): Likewise.
22197 (record_conditions): Likewise. Accept vector of conditions rather
22198 than edge_equivalence structure for first argument.
22199 for the first argument.
22200 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22201 from tree-ssa-dom.c.
22202 (avail_exprs_stack::record_cond): Likewise.
22203 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22204 from tree-ssa-dom.c.
22205 (avail_exprs_stack): Add new member functions lookup_avail_expr
22207 (record_conditions): Declare.
22209 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22212 * lra-constraints.c (process_alt_operands): Increase reject for
22213 reloading an input/output operand.
22215 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22218 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22219 insns to convert from signed/unsigned char/short to IEEE 128-bit
22221 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22223 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22226 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22227 subreg of inner mode for values already in registers.
22229 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22231 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22232 iteration reg is used after the loop.
22234 2017-03-14 Martin Sebor <msebor@redhat.com>
22236 PR tree-optimization/79800
22237 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22238 precision in negative-positive range.
22239 (format_floating): Call non-const overload with adjusted precision.
22241 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22244 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22247 2017-03-14 Martin Sebor <msebor@redhat.com>
22249 PR middle-end/80020
22250 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22251 * builtins.def (aligned_alloc): Use it.
22254 * Makefile.in (GTFILES): Add calls.c.
22255 * calls.c: Include "gt-calls.h".
22257 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22259 PR rtl-optimization/79728
22260 * regs.h (struct target_regs): New field
22261 x_contains_allocatable_regs_of_mode.
22262 (contains_allocatable_regs_of_mode): New macro.
22263 * reginfo.c (init_reg_sets_1): Initialize it, and change
22264 contains_reg_of_mode so it includes global regs as well.
22265 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22266 rather than contains_regs_of_mode.
22268 2017-03-14 Martin Liska <mliska@suse.cz>
22270 * doc/invoke.texi: Document options that can't be combined with
22271 -fcheck-pointer-bounds.
22273 2017-03-14 Martin Liska <mliska@suse.cz>
22275 PR middle-end/79831
22276 * doc/invoke.texi (-Wchkp): Document the option.
22278 2017-03-14 Martin Liska <mliska@suse.cz>
22280 * Makefile.in: Install gcov-dump.
22282 2017-03-14 Martin Liska <mliska@suse.cz>
22284 * multiple_target.c (expand_target_clones): Bail out for
22285 an invalid attribute.
22287 2017-03-14 Richard Biener <rguenther@suse.de>
22289 * alias.c (struct alias_set_entry): Pack properly.
22290 * cfgloop.h (struct loop): Likewise.
22291 * cse.c (struct set): Likewise.
22292 * ipa-utils.c (struct searchc_env): Likewise.
22293 * loop-invariant.c (struct invariant): Likewise.
22294 * lra-remat.c (struct cand): Likewise.
22295 * recog.c (struct change_t): Likewise.
22296 * rtl.h (struct address_info): Likewise.
22297 * symbol-summary.h (function_summary): Likewise.
22298 * tree-loop-distribution.c (struct partition): Likewise.
22299 * tree-object-size.c (struct object_size_info): Likewise.
22300 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22301 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22302 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22303 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22304 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22305 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22306 (struct _stmt_vec_info): Likewise.
22308 2017-03-14 Martin Liska <mliska@suse.cz>
22311 * multiple_target.c (create_dispatcher_calls): Check that
22312 a target can create a function dispatcher.
22314 2017-03-14 Martin Liska <mliska@suse.cz>
22317 * multiple_target.c (expand_target_clones): Drop local.local
22318 flag for default implementation.
22320 2017-03-14 Richard Biener <rguenther@suse.de>
22322 PR tree-optimization/80030
22323 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22325 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22327 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22328 gcc_fallthrough() instead of __attribute__((fallthrough));
22330 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22332 * doc/gcc.texi: Remove "up" link to (DIR).
22333 * doc/gccint.texi: Ditto.
22335 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22337 * doc/install.texi (Specific) <avr>: Remove reference to
22340 2017-03-13 Jeff Law <law@redhat.com>
22342 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22343 attribute rather than comments.
22345 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22346 match_scratch operand is highest.
22348 2017-03-13 Martin Liska <mliska@suse.cz>
22350 PR middle-end/78339
22351 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22352 is a CHKP clone, use original declaration.
22354 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22356 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22357 (arc_conditional_register_usage): Use a different allocation order
22358 when optimizing for size.
22359 * common/config/arc/arc-common.c (arc_option_optimization_table):
22360 Section anchors default on when optimizing for size.
22362 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22364 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22366 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22368 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22369 * config/arc/arc.md (cpu_facility): Add cd variant.
22370 (*movqi_insn): Add code density variant.
22371 (*movhi_insn): Likewise.
22372 (*movqi_insn): Likewise.
22373 (*addsi3_mixed): Likewise.
22374 (subsi3_insn): Likewise.
22376 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22378 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22380 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22382 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22383 expressions with MINUS and UNARY ops.
22385 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22388 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22390 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22391 between vec_select and vector argument.
22392 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22393 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22394 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22395 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22396 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22397 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22399 2017-03-13 Richard Biener <rguenther@suse.de>
22402 * params.def (vect-max-peeling-for-alignment): Fix typo.
22404 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22406 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22407 issue that only occurred with binutils below 2.18.
22409 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22411 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22412 refer to binutils 2.11/2.12 minimum.
22414 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22416 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22417 ftp.kernel.org and simplify binutils requirement.
22419 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22421 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22423 (Optimize Options): Ditto. Also remove redundancy.
22425 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22427 PR translation/79848
22428 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22430 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22431 to G_ to avoid double translation.
22433 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22435 PR translation/79923
22436 * auto-profile.c (get_combined_location): Convert leading
22437 character of diagnostics to lower case and remove trailing period.
22438 (read_profile): Likewise for various diagnostics.
22439 * config/arm/arm.c (arm_option_override): Remove trailing period
22440 from various diagnostics.
22441 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22442 (msp430_expand_delay_cycles): Likewise.
22444 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22447 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22448 full command-line argument, rather than just "str".
22449 (aarch64_validate_march): Likewise.
22450 (aarch64_validate_mtune): Likewise.
22452 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22454 PR rtl-optimization/78911
22455 * lra-assigns.c (must_not_spill_p): New function.
22456 (spill_for): Use it.
22458 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22460 PR tree-optimization/79981
22461 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22462 ATOMIC_COMPARE_EXCHANGE ifn result.
22463 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22464 IFN_ATOMIC_COMPARE_EXCHANGE.
22466 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22469 * opts.c (parse_sanitizer_options): Add missing question mark to
22470 "did you mean" message.
22472 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22474 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22476 (VMULEUH_UNS): Likewise.
22477 (VMULOUB_UNS): Likewise.
22478 (VMULOUH_UNS): Likewise.
22479 * config/rs6000/rs6000.c (builtin_function_type): Remove
22480 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22482 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22485 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22486 x with result of extra_parsing_for_operand_code_0.
22487 (function_reader::extra_parsing_for_operand_code_0): Convert
22488 return type from void to rtx, returning x. When reading
22489 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22490 larger size containing struct block_symbol.
22492 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22494 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22495 -mfloat128-hardware without -m64.
22497 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22500 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22501 entries to the case statement that marks unsigned arguments to
22502 overloaded functions.
22504 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22507 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22509 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22512 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22513 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22515 2017-03-10 Martin Liska <mliska@suse.cz>
22519 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22520 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22523 2017-03-10 Olivier Hainque <hainque@adacore.com>
22525 * tree-switch-conversion (array_value_type): Start by resetting
22526 candidate type to it's main variant.
22528 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22530 PR rtl-optimization/79909
22531 * combine.c (try_combine): Use simplify_replace_rtx on individual
22532 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22533 of the whole CALL_INSN_FUNCTION_USAGE.
22535 PR tree-optimization/79972
22536 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22537 get_range_info on SSA_NAMEs. Formatting fixes.
22539 2017-03-10 Richard Biener <rguenther@suse.de>
22540 Jakub Jelinek <jakub@redhat.com>
22542 PR tree-optimization/77975
22543 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22544 edge to be constant.
22545 (get_val_for): For constant x return it. Formatting fix.
22546 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22547 would use the same bases as the current one.
22549 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22551 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22552 instead of vec_select for V1TImode.
22553 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22555 (VSX_LE_128): Add V1TI to this mode iterator.
22556 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22557 (*vsx_le_perm_store_<mode>): Likewise.
22558 (pre-reload splitter for VSX stores): Likewise.
22559 (post-reload splitter for VSX stores): Likewise.
22560 (*vsx_xxpermdi2_le_<mode>): Likewise.
22561 (*vsx_lxvd2x2_le_<mode>): Likewise.
22562 (*vsx_stxvd2x2_le_<mode>): Likewise.
22564 2017-03-09 Michael Eager <eager@eagercon.com>
22566 Correct failures with --enable-checking=yes,rtl.
22568 * config/microblaze/microblaze.c (microblaze_expand_shift):
22569 Replace GET_CODE test with CONST_INT_P and INTVAL test with
22570 test for const0_rtx.
22571 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
22572 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
22574 2017-03-09 Richard Biener <rguenther@suse.de>
22576 PR tree-optimization/79977
22577 * graphite-scop-detection.c (scop_detection::merge_sese):
22578 Handle the case of extra exits to blocks dominating the entry.
22580 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
22582 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22585 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22587 PR rtl-optimization/79949
22588 * lra-constraints.c (process_alt_operands): Check memory when
22589 trying to predict a cycle. Print about the overall increase.
22591 2017-03-09 Richard Biener <rguenther@suse.de>
22593 PR middle-end/79971
22594 * gimple-expr.c (useless_type_conversion_p): Preserve
22595 TYPE_SATURATING for fixed-point types.
22597 2017-03-09 Richard Biener <rguenther@suse.de>
22600 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22601 alignment of BLKmode params.
22603 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22606 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22607 (VALL_NO_V2Q): Likewise.
22609 * config/aarch64/aarch64-simd.md
22610 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22612 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22613 VALL_NO_V2Q mode iterator.
22614 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22616 2017-03-09 Martin Liska <mliska@suse.cz>
22618 PR tree-optimization/79631
22619 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22620 tree_int_cst_sign_bit just for INTEGER constants.
22622 2017-03-09 Martin Liska <mliska@suse.cz>
22626 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22629 2017-03-09 Marek Polacek <polacek@redhat.com>
22632 * tree.c (inchash::add_expr): Handle TREE_VEC.
22634 2017-03-09 Martin Liska <mliska@suse.cz>
22637 (chkp_narrow_size_and_offset): New function.
22638 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22639 (void chkp_parse_bit_field_ref): New function.
22640 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22641 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22643 2017-03-09 Martin Liska <mliska@suse.cz>
22646 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22647 (chkp_find_bounds_1): Remove gcc_unreachable.
22649 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22652 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22653 BUILT_IN_SYNC*, determine the access type from the size suffix and
22654 always build a MEM_REF with that type. Handle forgotten
22655 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22658 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22659 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22660 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22661 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22662 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22663 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22664 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22665 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22666 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22667 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22668 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22669 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22670 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22671 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22672 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22673 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22674 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22675 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22676 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22677 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22678 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22679 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22680 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22681 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22682 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22683 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22684 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22685 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22686 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22687 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22688 definitions outside of __OPTIMIZE__ guarded section.
22691 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22692 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22693 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22694 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22697 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22699 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22700 ("vfenez<mode>"): Add missing constraints.
22702 2017-03-08 Martin Sebor <msebor@redhat.com>
22705 * config/nds32/nds32.c (nds32_option_override):
22706 Fix misspelled diagnostic.
22708 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22711 * gimplify.c (gimplify_omp_for): Replace index var in outer
22712 taskloop statement with an artificial variable and add
22713 OMP_CLAUSE_PRIVATE clause for it.
22715 2017-03-08 Richard Biener <rguenther@suse.de>
22717 PR tree-optimization/79955
22718 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22719 for accesses that are completely outside of the variable.
22721 2017-03-08 Andrew Haley <aph@redhat.com>
22723 PR tree-optimization/79943
22724 * tree-ssa-loop-split.c (compute_new_first_bound): When
22725 calculating the new upper bound, (END-BEG) should be added, not
22728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22730 * config/avr/avr.md (setmemhi): Make sure match_dup
22731 operand number comes before match_scratch.
22733 2017-03-08 Richard Biener <rguenther@suse.de>
22735 PR tree-optimization/79920
22736 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22737 with ncopies == 1 to ...
22738 (vect_transform_slp_perm_load): ... here. Properly compute
22739 all element loads by iterating VF times over the group. Do
22740 not handle ncopies (computed in a broken way) in
22741 vect_create_mask_and_perm.
22743 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22746 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22747 is a uniform vector, use uniform_vector_p return value instead of
22748 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22750 2017-03-07 Marek Polacek <polacek@redhat.com>
22752 PR middle-end/79809
22753 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22754 (alloca_call_type): Likewise.
22756 2017-03-07 Martin Liska <mliska@suse.cz>
22758 * gcov.c (process_args): Put comment to correct location.
22760 2017-03-07 Martin Liska <mliska@suse.cz>
22762 PR middle-end/68270
22763 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22764 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22765 (chkp_narrow_bounds_for_field): Likewise.
22766 (chkp_parse_array_and_component_ref): Pass one more argument to
22769 2017-03-07 Richard Biener <rguenther@suse.de>
22771 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22774 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22776 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22777 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22779 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22782 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22783 to end of description.
22784 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22786 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22788 PR rtl-optimization/79901
22789 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22791 (*avx512f_<code><mode>3<mask_name>): ... this.
22792 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22793 iterator instead of VI8_AVX2_AVX512BW.
22795 PR rtl-optimization/79901
22796 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22797 min/max expander, expand it using expand_vec_cond_expr.
22800 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22803 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22806 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22807 to void * for PCH reasons.
22808 * dwarf2out.c (output_loc_operands, output_die): Cast
22809 v.val_vec.array to unsigned char *.
22811 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22814 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22817 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22819 PR rtl-optimization/79571
22820 * lra-constraints.c (process_alt_operands): Calculate static
22821 reject and subtract it from overall when only addresses will be
22824 2017-03-06 Julia Koval <julia.koval@intel.com>
22827 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22828 incoming stack boundary to 128 for 64-bit targets.
22830 2017-03-06 Richard Biener <rguenther@suse.de>
22832 PR tree-optimization/79894
22833 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22834 to NULL after folding it.
22836 2017-03-06 Richard Biener <rguenther@suse.de>
22838 PR tree-optimization/79824
22839 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22840 check disabling peeling for gaps.
22842 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22844 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22845 attributes): Document gettimeofday.
22847 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22849 * config/s390/s390.c (s390_option_override_internal): Set
22850 PARAM_MIN_VECT_LOOP_BOUND
22852 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22854 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22855 * config/s390/s390.md: Likewise.
22857 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22860 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22861 (<avx2_avx512>_perm<mode>): Rename to ...
22862 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22864 (<avx512>_perm<mode>_mask): Rename to ...
22865 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22867 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22868 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22869 instead of VI8F_256_512.
22870 (avx512f_perm<mode>): New define_expand.
22871 (avx512f_perm<mode>_mask): Likewise.
22872 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22873 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22875 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22877 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22878 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22880 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22882 2017-03-06 Martin Liska <mliska@suse.cz>
22885 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22886 when having a SSA NAME w/o VAR_DECL assigned to it.
22888 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22890 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22891 msa_dpsub_<su>_d): Fix MODE for vec_select.
22893 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22895 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22897 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22899 2017-03-06 Richard Biener <rguenther@suse.de>
22901 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22902 * plugin.c (register_plugin_info): Likewise.
22903 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22905 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22907 * config/i386/sse.md (sse_storehps, sse_storelps,
22908 avx_<castmode><avxsizesuffix>_<castmode>,
22909 avx512f_<castmode><avxsizesuffix>_<castmode>,
22910 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22911 in condition that at least one operand is not a MEM.
22913 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22915 PR middle-end/79805
22916 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22917 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22919 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22920 gimple_call_nothrow_p flag based on whether original builtin can throw.
22921 If it can, emit following stmts on the fallthrough edge.
22922 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22923 don't create new bb if inserting just debug stmts on the edge, try to
22924 insert them on the fallthru bb or just reset debug stmts.
22926 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22929 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22930 restore recog_data (including the operand rtxes inside it) around
22931 the call to get_insn_template.
22933 2017-03-03 Martin Sebor <msebor@redhat.com>
22935 PR tree-optimization/79699
22936 * context.c (context::~context): Free MPFR caches to avoid
22937 a memory leak on program exit.
22939 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22941 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22942 Use wide_int::ulow () instead of .elt (0).
22944 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22946 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22947 (*pushxf): Limit oF constraint to 32bit targets and add oC
22948 constraint for 64bit targets.
22949 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22950 (*pushdf): Change rmF constraint to rmC.
22952 2017-03-03 Martin Liska <mliska@suse.cz>
22954 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22955 Remove unused variable.
22957 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22960 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22961 is a memory operand, increase num_memory.
22962 (ix86_expand_args_builtin): Likewise.
22964 2017-03-03 Jan Hubicka <jh@suse.cz>
22967 * ipa-devirt.c (maybe_record_node): Properly handle
22968 __cxa_pure_virtual visibility.
22970 2017-03-03 Martin Liska <mliska@suse.cz>
22972 PR tree-optimization/79803
22973 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22975 (pass_loop_prefetch::execute): Disabled optimization if an
22976 assumption about L1 cache size is not met.
22978 2017-03-03 Martin Liska <mliska@suse.cz>
22980 PR rtl-optimization/79574
22981 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22982 (hash_scan_set): Likewise.
22983 (dump_hash_table): Likewise.
22984 (hoist_code): Likewise.
22986 2017-03-03 Richard Biener <rguenther@suse.de>
22988 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22989 in place of uhigh ().
22990 (fixed_convert_from_real): Likewise.
22992 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22995 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22997 2017-03-03 Richard Biener <rguenther@suse.de>
22999 PR middle-end/79818
23000 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23001 TYPE_OVERFLOW_UNDEFINED check.
23003 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23005 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23007 (vector_ae_<mode>_p): Likewise.
23008 (vector_nez_<mode>_p): Likewise.
23009 (vector_ne_v2di_p): Likewise.
23010 (vector_ae_v2di_p): Likewise.
23011 (vector_ne_<mode>_p): Likewise.
23012 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23014 (vsx_tsqrt<mode>2_fe): Likewise.
23016 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23019 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23021 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23023 PR rtl-optimization/79780
23024 * cprop.c (one_cprop_pass): When second and further conditional trap
23025 in a single basic block is turned into an unconditional trap, turn it
23026 into a deleted note to avoid RTL verification failures.
23028 2017-03-02 Richard Biener <rguenther@suse.de>
23030 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23032 2017-03-02 Richard Biener <rguenther@suse.de>
23034 PR tree-optimization/79345
23036 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23037 param and abort the walk, returning -1 if it is hit.
23038 (walk_aliased_vdefs): Take a limit param and pass it on.
23039 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23040 defaulting to 0 and return a signed int.
23041 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23042 (check_defs): New helper.
23043 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23044 about uninitialized memory.
23045 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23046 bogus uninitialized warning.
23047 (fixed_convert_from_real): Likewise.
23049 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23051 PR tree-optimization/66768
23052 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23053 iv_use if base object can't be determined.
23055 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23057 PR tree-optimization/79345
23058 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23059 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23060 (get_pattern_stats): Initialize it.
23061 * genemit.c (gen_expand): Verify match_scratch numbers come after
23062 match_operand/match_dup numbers.
23063 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23064 match_scratch numbers.
23065 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23067 * config/s390/s390.md (trunctdsd2): Likewise.
23069 2017-03-02 Richard Biener <rguenther@suse.de>
23071 * wide-int.h (wide_int_storage::operator=): Implement in terms
23074 2017-03-02 Richard Biener <rguenther@suse.de>
23076 PR tree-optimization/79777
23077 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23078 the to insert expression to sth existing.
23080 2017-03-01 Martin Sebor <msebor@redhat.com>
23082 PR middle-end/79692
23083 * gimple-ssa-sprintf.c
23084 (directive::known_width_and_precision): New function.
23085 (format_integer): Use it.
23086 (get_mpfr_format_length): Consider the full range of precision
23087 when computing %g output with the # flag. Set the likely byte
23088 count to 3 rather than 1 when precision is indeterminate.
23089 (format_floating): Correct the lower bound of precision.
23091 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23093 * doc/invoke.texi: Document default code model for 64-bit Linux.
23095 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23098 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23099 udiv rather than div since input pattern is unsigned.
23101 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23103 * config/i386/i386.c (print_reg): Warn for values of
23104 unsupported size in integer register.
23106 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23109 * config/rs6000/predicates.md (current_file_function_operand): Do
23110 not allow self calls to be local if the function is replaceable.
23112 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23115 * config/rs6000/altivec.h (vec_ctz and others): Change the
23116 preprocessor macro that controls conditional compilation from
23117 _ARCH_PWR9 to __POWER9_VECTOR__.
23118 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23119 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23120 control (instead of _ARCH_PWR9 control) so that template
23121 definition uses power9-specific function.
23122 (vec_any_eq): Likewise.
23123 (vec_all_ne): Change macro definition to use a power9-specific
23124 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23125 _ARCH_PWR9 control).
23126 (vec_any_eq) Likewise.
23127 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23128 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23129 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23130 support for xvcmpnedp instruction.
23131 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23132 macro expansion so that Power9 implementation of vec_all_ne does
23133 not use the AltiVec predicate framework.
23134 (VCMPNEH_P): Likewise.
23135 (VCMPNEW_P): Likewise.
23136 (VCMPNED_P): Likewise.
23137 (VCMPNEFP_P): Likewise.
23138 (VCMPNEDP_P): Likewise.
23139 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23140 implementation of vec_any_eq to not use AltiVec predicate
23142 (VCMPAEH_P): Likewise.
23143 (VCMPAEW_P): Likewise.
23144 (VCMPAED_P): Likewise.
23145 (VCMPAEFP_P): Likewise.
23146 (VCMPAEDP_P): Likewise.
23147 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23148 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23149 not use the AltiVec predicate framework.
23150 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23151 of vec_any_eq to not use AltiVec predicate framework.
23152 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23153 support for predefined __POWER9_VECTOR__ macro to indicate that
23154 Power9 instruction selection is enabled.
23155 (altivec_overloaded_builtins): Remove extraneous
23156 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23157 function argument types RS6000_BTI_bool_V16QI and
23158 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23159 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23160 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23161 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23162 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23163 Power9 for implementations of vec_cmpne. Change the signature for
23164 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23165 (representing vec_all_ne) to remove the previously described first
23166 argument of type RS6000_BTI_INTSI, as this was an artifact of
23167 reliance on the AltiVec predicate framework, which is no longer
23168 used in the implementation of these functions. Add
23169 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23170 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23171 since, unlike the AltiVec predicate framework implementation, we
23172 do not share function descriptors between vec_alle and vec_anyeq.
23173 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23174 set of modes that receive special treatment even when
23175 TARGET_P9_VECTOR is true. The special treatment emits code that
23176 does not depend on Power9 instructions.
23177 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23178 define_expand to not rely on AltiVec predicate framework.
23179 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23181 (vector_ne_v2di_p): Change this define_expand to not rely on
23182 AltiVec predicate framework.
23183 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23185 (vector_ne_<mode>_p): Change this define_expand to not rely on
23186 AltiVec predicate framework.
23187 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23189 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23190 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23191 define_insn pattern.
23192 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23193 define_insn pattern because the xvcmpne<VSs>. instruction is not
23195 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23196 instruction is not supported.
23198 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23200 * config/nvptx/nvptx.c: Include intl.h.
23202 2017-03-01 Martin Jambor <mjambor@suse.cz>
23205 * ipa-prop.h (ipa_bits): Removed field known.
23206 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23207 to pointers. Adjusted their comments to warn about their sharing.
23208 (ipcp_transformation_summary): Change bits to a vector of pointers.
23209 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23210 (ipa_get_ipa_bits_for_value): Declare.
23211 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23212 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23213 (ipa_bits_hash_table): Likewise.
23214 (ipa_vr_ggc_hash_traits): Likewise.
23215 (ipa_vr_hash_table): Likewise.
23216 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23217 being pointers and vr_known being removed.
23218 (ipa_set_jf_unknown): Likewise.
23219 (ipa_get_ipa_bits_for_value): New function.
23220 (ipa_set_jfunc_bits): Likewise.
23221 (ipa_get_value_range): New overloaded functions.
23222 (ipa_set_jfunc_vr): Likewise.
23223 (ipa_compute_jump_functions_for_edge): Use the above functions to
23224 construct bits and vr parts of jump functions.
23225 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23226 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23228 (ipcp_grow_transformations_if_necessary): Also allocate
23229 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23231 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23232 them. Fix too long lines.
23233 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23234 vr_known being removed.
23235 (ipa_read_jump_function): Use new setter functions to construct bits
23236 and vr parts of jump functions or set them to NULL.
23237 (write_ipcp_transformation_info): Adjust for bits being pointers.
23238 (read_ipcp_transformation_info): Likewise.
23239 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23241 Include gt-ipa-prop.h.
23242 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23244 (ipcp_store_bits_results): Likewise.
23245 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23246 Do not write to existing jump functions but use a temporary instead.
23248 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23251 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23252 attempt to use its first operand as BIT_FIELD_REF base.
23254 2017-03-01 Richard Biener <rguenther@suse.de>
23256 PR middle-end/79721
23257 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23258 interpolating formula in wrapping arithmetic.
23259 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23261 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23263 PR tree-optimization/79734
23264 * tree-vect-generic.c (expand_vector_condition): Optimize
23265 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23266 Handle VEC_COND_EXPR where comparison has different inner width from
23267 type's inner width.
23269 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23271 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23272 markup, and similar issues. Remove @opindex entries for things
23273 that aren't options. Add missing -mmpy-option entries.
23275 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23277 PR tree-optimization/79737
23278 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23279 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23280 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23281 instead of byte_size. Formatting fix.
23282 (shift_bytes_in_array_right): Formatting fix.
23284 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23287 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23288 condition on optimize for the leaf function test.
23290 2017-02-28 Martin Liska <mliska@suse.cz>
23293 * read-rtl-function.c (function_reader::handle_unknown_directive):
23294 Bail out when one uses -flto.
23296 2017-02-28 Martin Liska <mliska@suse.cz>
23298 * common.opt: Replace space with tabular for options of <number>
23300 * config/i386/i386.opt: Show <number> value for
23301 -mlarge-data-threshold.
23302 * opts.c (print_filtered_help): Do not display number in hexadecimal
23305 2017-02-28 Martin Liska <mliska@suse.cz>
23307 * common.opt: Fix --help=option -Q for options which are of
23310 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23312 * config/i386/i386.c (print_reg): Error out for values
23313 of 8-bit size in invalid integer register.
23315 2017-02-28 Martin Sebor <msebor@redhat.com>
23317 PR tree-optimization/79691
23318 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23320 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23323 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23324 gcc_unreachable with output_operand_lossage.
23326 2017-02-28 Richard Biener <rguenther@suse.de>
23328 PR tree-optimization/79740
23329 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23331 (visit_nary_op): Insert the nary into the hashtable if we
23332 pattern-matched sth.
23333 * tree-ssa-pre.c (eliminate_insert): Robustify.
23335 2017-02-28 Richard Biener <rguenther@suse.de>
23337 PR middle-end/79731
23338 * fold-const.c (decode_field_reference): Reject out-of-bound
23341 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23343 * config/i386/i386.c: Include intl.h.
23344 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23345 instead of just cond ? "..." : "...".
23346 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23347 * coverage.c (read_counts_file): Likewise.
23348 * omp-offload.c: Include intl.h.
23349 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23350 of just cond ? "..." : "...".
23351 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23352 of just cond ? "..." : "...".
23354 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23357 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23359 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23360 'tune for' CPU name.
23361 * config/arm/arm-cpu-data.h: Regenerated.
23363 2017-02-28 Richard Biener <rguenther@suse.de>
23365 PR tree-optimization/79732
23366 * tree-inline.c (expand_call_inline): Do not shadow var.
23368 2017-02-28 Richard Biener <rguenther@suse.de>
23370 PR tree-optimization/79723
23371 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23372 address-space properly.
23374 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23376 * doc/optinfo.texi (Optimization groups): Fix option used for
23378 * doc/invoke.texi (-fopt-info): Document "omp".
23379 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23380 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23381 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23382 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23383 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23385 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23387 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23388 OPTGROUP_OMP with "omp".
23390 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23393 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23394 for arithmetic shift of unsigned V2DI.
23396 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23398 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23399 arc/linux.h headers.
23400 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23401 (LINK_SPEC): Likewise.
23402 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23403 (EXTRA_SPECS): Likewise.
23404 (STARTFILE_SPEC): Likewise.
23405 (ENDFILE_SPEC): Likewise.
23406 (LIB_SPEC): Likewise.
23407 (TARGET_SDATA_DEFAULT): Likewise.
23408 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23409 (MULTILIB_DEFAULTS): Likewise.
23410 (DWARF2_UNWIND_INFO): Likewise.
23411 * config/arc/big.h: New file.
23412 * config/arc/elf.h: Likewise.
23413 * config/arc/linux.h: Likewise.
23414 * config/arc/t-uClibc: Remove.
23416 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23418 PR tree-optimization/77536
23419 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23420 (tree_transform_and_unroll_loop): Use above function to compute the
23421 estimated niter of unrolled loop and use it when scaling profile.
23422 Also use count info rather than frequency if it's non-zero.
23423 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23424 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23425 (vect_transform_loop): Call above function.
23427 2017-02-27 Richard Biener <rguenther@suse.de>
23429 PR tree-optimization/45397
23430 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23431 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23432 (visit_nary_op): Add pattern matching for CSEing sign-changed
23433 or truncated operations with wider ones.
23435 2017-02-27 Richard Biener <rguenther@suse.de>
23437 PR tree-optimization/79690
23438 * tree-vect-stmts.c (vectorizable_store): Use vector type
23439 built from the DR with address-space.
23441 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23443 * doc/invoke.texi (Optimize Options): Refine the description
23444 of asan-use-after-return.
23446 2017-02-25 Alan Modra <amodra@gmail.com>
23448 PR rtl-optimization/79584
23449 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23450 base, not ad->base_term, the reg within base. Remove assertion
23451 that ad->base == ad->base_term. Replace gen_int_mode using
23452 bogus mode with const0_rtx.
23454 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23456 PR middle-end/79396
23457 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23458 FMA_EXPR like tcc_binary or tcc_unary.
23460 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23463 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23465 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23466 (output_loc_operands): Handle DW_OP_call_ref and
23467 DW_OP_GNU_variable_value.
23468 (struct variable_value_struct): New type.
23469 (struct variable_value_hasher): Likewise.
23470 (variable_value_hash): New variable.
23471 (string_types): Remove.
23472 (copy_loc_descr): New function.
23473 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23474 (prepend_loc_descr_to_each): New function.
23475 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23476 instead of add_loc_descr_to_each if the first argument is single
23477 location list and the second has multiple.
23478 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23479 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23480 when looking for variable value which doesn't have other location info.
23481 (loc_list_from_tree): Formatting fix.
23482 (gen_array_type_die): Simplify DW_AT_string_length handling.
23483 (adjust_string_types): Remove.
23484 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23485 string_types. Call resolve_variable_values.
23486 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23487 (resolve_addr_in_expr): Likewise. Add A argument.
23488 (copy_deref_exprloc): Remove deref argument. Adjust for the
23489 original expression being DW_OP_GNU_variable_value with optionally
23490 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23491 optionally after it.
23492 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23493 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23494 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23495 (variable_value_hasher::hash, variable_value_hasher::equal): New
23497 (resolve_variable_value_in_expr, resolve_variable_value,
23498 resolve_variable_values, note_variable_value_in_expr,
23499 note_variable_value): New functions.
23500 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23503 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23506 * opts.h (handle_generated_option): Add GENERATED_P argument.
23507 * opts-common.c (handle_option): Adjust function comment.
23508 (handle_generated_option): Add GENERATED_P argument, pass it to
23510 (control_warning_option): Pass false to handle_generated_option
23512 * opts.c (maybe_default_option): Pass true to handle_generated_option
23514 * optc-gen.awk: Likewise.
23516 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23518 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23519 a REG, look at the REG it is a SUBREG of.
23520 (splitter for cmpeqsi_t): Ditto.
23522 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23524 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23525 the special USEs with the pattern of the insn, not the insn itself.
23527 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23530 * doc/invoke.texi: Document -mload-store-pairs.
23532 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23533 Sandra Loosemore <sandra@codesourcery.com>
23535 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23536 argument isn't a CONST_INT.
23537 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23538 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23539 (nios2_validate_compare): Bypass alternate compare logic if *op2
23540 is not a CONST_INT.
23541 (ldstwm_operation_p): Return false if first_base is not a REG or
23542 if first_offset is not a CONST_INT.
23544 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23546 * config/cris/cris.md: Use correct operand in a define_peephole2.
23548 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23550 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23552 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23554 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23555 this_insn if it is an INSN or JUMP_INSN.
23556 (force_offsettable): Look at base, not at addr.
23557 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23558 on things that aren't necessarily CONST_INTs.
23560 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
23562 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
23563 -mfpmath=sse is the default also for x86-32 targets with SSE2
23564 instruction set when @option{-ffast-math} is enabled
23566 2017-02-24 Jeff Law <law@redhat.com>
23568 PR rtl-optimizatoin/79286
23569 * ira.c (update_equiv_regs): Drop may_trap_p exception to
23572 2017-02-24 Richard Biener <rguenther@suse.de>
23574 PR tree-optimization/79389
23575 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
23578 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
23580 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
23581 function comment to reflect reality.
23582 (loop_exits_before_overflow): Fix typo in function description.
23584 2017-02-24 Richard Biener <rguenther@suse.de>
23586 PR tree-optimization/79389
23587 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23588 properly that a threading opportunity exists. Detect conditional
23589 copy/constant propagation opportunities.
23591 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23593 * config/visium/visium.md (type): Add trap.
23594 (b): New mode attribute.
23595 (*btst): Rename into...
23596 (*btst<mode>): ...this and adjust.
23597 (*cbranchsi4_btst_insn): Rename into...
23598 (*cbranch<mode>4_btst_insn): ...this and adjust.
23599 (trap): New define_insn.
23601 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23603 PR tree-optimization/79389
23604 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23605 (noce_reversed_cond_code): New function.
23606 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23607 reversed_comparison_code. Formatting fix.
23608 (noce_try_store_flag): Test rev_cond != NULL in addition to
23609 reversed_comparison_code.
23610 (noce_try_store_flag_constants): Likewise.
23611 (noce_try_store_flag_mask): Likewise.
23612 (noce_try_addcc): Use rev_cond if non-NULL instead of
23613 reversed_comparison_code.
23614 (noce_try_cmove_arith): Likewise. Formatting fixes.
23615 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23616 (noce_find_if_block): Initialize rev_cond.
23617 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23618 instead of false as last argument never attempt to reverse it
23621 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23623 PR tree-optimization/79663
23624 * tree-predcom.c (combine_chains): Process refs in reverse order
23625 only for ZERO length chains, and add explaining comment.
23627 2017-02-23 Jeff Law <law@redhat.com>
23629 PR tree-optimization/79578
23630 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23631 in call to operand_equal_p.
23633 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23636 * config/i386/cpuid.h: Fix another undefined behavior.
23638 2017-02-23 Richard Biener <rguenther@suse.de>
23640 PR tree-optimization/79683
23641 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23642 vector types for data-refs.
23644 2017-02-23 Martin Liska <mliska@suse.cz>
23646 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23648 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23650 PR middle-end/79665
23651 * internal-fn.c (get_range_pos_neg): Moved to ...
23652 * tree.c (get_range_pos_neg): ... here. No longer static.
23653 * tree.h (get_range_pos_neg): New prototype.
23654 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23655 are known to be in between 0 and signed maximum inclusive, try to
23656 expand both unsigned and signed divmod and use the cheaper one from
23659 2017-02-22 Jeff Law <law@redhat.com>
23661 PR tree-optimization/79578
23662 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23663 to compare base operands.
23665 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23668 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23669 gpc_reg_operand instead of fpr_reg_operand.
23671 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23673 * config/mips/mips.c (mips_return_in_memory): Force FP
23674 vector types to be returned in memory for o32 ABI.
23676 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23678 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23679 instead of DW_TAG_member for static data member declarations and don't
23680 set no_linkage_name for static inline data members.
23681 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23684 2017-02-22 Martin Liska <mliska@suse.cz>
23686 * doc/invoke.texi: Replace inequality signs with square brackets
23689 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23691 PR tree-optimization/68644
23692 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23694 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23697 * lra-constraints.c (simplify_operand_subreg): Handle
23698 WORD_REGISTER_OPERATIONS targets.
23700 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23703 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23704 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23705 elimination by swapping fld*.
23707 2017-02-22 Richard Biener <rguenther@suse.de>
23709 PR tree-optimization/79673
23710 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23711 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23712 irrelevant address-space qualifiers and avoiding a
23713 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23715 2017-02-22 Richard Biener <rguenther@suse.de>
23717 PR tree-optimization/79666
23718 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23719 to not symbolically negate if that may introduce undefined
23722 2017-02-22 Martin Liska <mliska@suse.cz>
23725 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23726 * data-streamer-out.c (streamer_write_gcov_count_stream):
23728 * value-prof.c (stream_out_histogram_value): Make assert more
23729 precise based on type of counter.
23731 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23734 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23735 Use nonimmediate_operand instead of memory_operand for operand 1.
23736 (float-extend standard_x87sse_constant_load splitter): Ditto.
23738 2017-02-21 Jeff Law <law@redhat.com>
23740 PR tree-optimization/79621
23741 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23742 blocks with edges to themselves.
23744 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23747 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23748 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23749 Use gimple_call_builtin_p.
23752 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23753 on temporarily removed DEBUG_INSNs.
23755 PR tree-optimization/79649
23756 * tree-loop-distribution.c (classify_partition): Give up on
23757 non-generic address space loads/stores.
23759 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23761 * doc/loop.texi (Loop manipulation): Remove nonexistent
23762 tree_ssa_loop_version from the documentation.
23763 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23765 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23768 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23769 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23770 * config/rs6000/rs6000.c: Include except.h.
23771 (rs6000_expand_split_stack_prologue): Call
23772 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23774 2017-02-21 Martin Jambor <mjambor@suse.cz>
23777 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23778 have been analyzed.
23780 2017-02-21 Martin Jambor <mjambor@suse.cz>
23782 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23783 for backward compatibility only.
23784 * doc/invoke.texi (Option Summary): Remove all references to
23785 -fipa-cp-alignment.
23787 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23791 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23793 * lra-constraints.c (curr_insn_transform): Handle
23794 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23796 2017-02-21 Martin Liska <mliska@suse.cz>
23798 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23800 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23802 PR translation/79638
23803 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23805 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23808 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23809 (arm_function_ok_for_sibcall): Return false for an indirect call by
23810 descriptor if all the argument registers are used.
23811 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23812 alignment of the function.
23814 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23816 PR tree-optimization/61441
23817 * simplify-rtx.c (simplify_const_unary_operation): For
23818 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23819 the sNaN unmodified.
23821 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23823 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23824 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23825 instead of SYSTEM_HEADER_DIR.
23827 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23828 Martin Liška <mliska@suse.cz>
23830 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23831 Fix typos and grammar, use active voice, and clarify.
23833 2017-02-20 Marek Polacek <polacek@redhat.com>
23835 PR middle-end/79537
23836 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23839 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23841 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23844 * config/i386/i386.c (ix86_expand_builtin): Handle
23845 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23846 ix86_builtins_isa[fcode].isa as a requirement of those
23847 flags and any other flag in the bitmask.
23848 (ix86_init_mmx_sse_builtins): Use 0 instead of
23849 ~OPTION_MASK_ISA_64BIT as mask.
23850 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23851 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23852 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23853 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23855 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23858 * lra-constraints.c (split_reg): Check requested split mode
23859 is supported by the register.
23861 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23863 * lra-constraints.c (simplify_operand_subreg): Remove early
23866 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23869 * lra-constraints.c (curr_insn_transform): Tighten condition
23870 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23872 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23875 * lra-constraints.c (curr_insn_transform): Handle
23876 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23878 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23881 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23883 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23885 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23888 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23891 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23893 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23896 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23898 * final.c (last_columnnum, override_columnnum): New variables.
23899 (final_start_function): Set last_columnnum, pass it to begin_prologue
23900 hook and pass 0 to dwarf2out_begin_prologue.
23901 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23902 to source_line debug hook.
23903 (notice_source_line): Compute last_columnnum and for debug_column_info
23904 return true on column changes.
23905 * debug.h (struct gcc_debug_hooks): Add column argument to
23906 source_line and begin_prologue hooks.
23907 (debug_nothing_int_charstar_int_bool): Remove prototype.
23908 (debug_nothing_int_int_charstar,
23909 debug_nothing_int_int_charstar_int_bool): New prototypes.
23910 (dwarf2out_begin_prologue): Add column argument.
23911 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23912 begin_prologue hooks.
23913 (debug_nothing_int_charstar_int_bool): Remove.
23914 (debug_nothing_int_int_charstar,
23915 debug_nothing_int_int_charstar_int_bool): New functions.
23916 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23917 through to dwarf2out_source_line.
23918 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23919 (dwarf2out_source_line): Add column argument, emit it if requested.
23920 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23922 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23923 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23924 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23925 through to dwarf2out_begin_prologue.
23926 (vmsdbgout_source_line): Add column argument, pass it through to
23927 dwarf2out_source_line.
23928 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23929 dbxout_source_line caller.
23930 (dbxout_source_line): Add column argument.
23932 * common.opt (gno-column-info, gcolumn-info): New options.
23933 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23934 (check_die): Also test for multiple DW_AT_decl_column attributes.
23935 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23936 DW_AT_decl_column if requested.
23937 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23939 (gen_variable_die): Likewise.
23940 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23941 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23944 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23945 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23946 (ix86_handle_option): Handle OPT_m3dnowa.
23947 * doc/invoke.texi (-m3dnowa): Document.
23948 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23949 -m3dnowa instead of -m3dnow -march=athlon.
23952 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23953 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23955 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23958 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23959 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23960 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23961 generator for vsx_xxpermdi_<mode>_be.
23962 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23963 force big-endian semantics.
23964 (vsx_xxpermdi_<mode>_be): New define_expand with same
23965 implementation as previous version of vsx_xxpermdi_<mode>.
23967 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23969 PR tree-optimization/79327
23970 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23971 variable, its initialization and use.
23973 2017-02-17 Julia Koval <julia.koval@intel.com>
23975 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23976 (OPTION_MASK_ISA_PKU_UNSET): New.
23977 (ix86_handle_option): Handle -mrdpid.
23978 * config/i386/cpuid.h (bit_RDPID): New.
23979 * config/i386/driver-i386.c (host_detect_local_cpu):
23980 Detect RDPID feature.
23981 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23982 * config/i386/i386-c.c (ix86_target_macros_internal):
23984 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23985 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23986 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23987 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23988 * config/i386/i386.md (define_insn "rdpid"): New.
23989 * config/i386/i386.opt Add -mrdpid.
23990 * config/i386/immintrin.h (_rdpid_u32): New.
23992 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23994 PR rtl-optimization/79541
23995 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23996 instead of transforming it into USE.
23998 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24000 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24001 If HONOR_SNANS (SFmode) force the input to a register.
24002 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24003 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24004 an frsp or similar insn.
24006 2017-02-17 Martin Liska <mliska@suse.cz>
24008 PR rtl-optimization/79577
24009 * params.def (selsched-max-sched-times): Increase minimum to 1.
24011 2017-02-17 Martin Liska <mliska@suse.cz>
24013 PR rtl-optimization/79574
24014 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24016 2017-02-17 Martin Liska <mliska@suse.cz>
24018 PR tree-optimization/79529
24019 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24020 ssa_defined_default_def_p to handle cases which are implicitly
24022 * tree-ssa.c (ssa_defined_default_def_p): New function.
24023 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24024 which are implicitly defined.
24025 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24027 2017-02-17 Richard Biener <rguenther@suse.de>
24029 PR middle-end/79576
24030 * params.def (max-ssa-name-query-depth): Limit to 10.
24032 2017-02-17 Richard Biener <rguenther@suse.de>
24034 PR tree-optimization/79552
24035 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24038 2017-02-17 Richard Biener <rguenther@suse.de>
24041 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24043 2017-02-17 Marek Polacek <polacek@redhat.com>
24045 PR middle-end/79536
24046 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24047 (fold_negate_expr): New wrapper.
24049 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24051 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24052 Correct terminology and de-emphasize pre-standard behavior.
24054 2017-02-16 Alan Modra <amodra@gmail.com>
24056 PR rtl-optimization/79286
24057 * ira.c (def_dominates_uses): New function.
24058 (update_equiv_regs): Don't create an equivalence for insns that
24059 may trap where the register def does not dominate the use.
24061 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24063 PR rtl-optimization/78127
24064 * lra.c (lra): Call lra_eliminate before finish the loop after
24067 2017-02-16 Richard Biener <rguenther@suse.de>
24069 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24071 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24072 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24073 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24074 (isl_val_int_from_wi): New function.
24075 (extract_affine_gmp): Rename to ...
24076 (extract_affine_wi): ... this, take a widest_int.
24077 (extract_affine_int): Just wrap extract_affine_wi.
24078 (add_param_constraints): Use isl_val_int_from_wi.
24079 (add_loop_constraints): Likewise, and extract_affine_wi.
24081 2017-02-15 Jeff Law <law@redhat.com>
24083 PR middle-end/79521
24084 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24085 ira_init_register_move_cost_if_necessary.
24087 2017-02-15 Martin Sebor <msebor@redhat.com>
24089 PR middle-end/32003
24090 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24091 removed in a prior commit.
24093 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24095 PR tree-optimization/79347
24096 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24097 counters during peeling.
24099 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24101 * Makefile.in (site.exp): Remove "set ISLVER".
24103 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24106 * real.c (real_from_integer): Call real_convert even for decimal.
24108 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24111 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24113 2017-02-14 Andrew Pinski <apinski@cavium.com>
24115 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24116 cores and change the partno/implementer to be correct.
24117 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24118 the 'B" as the implementer.
24119 * config/aarch64/aarch64-tune.md: Regenerate.
24121 2017-02-14 Carl Love <cel@us.ibm.com>
24123 * config/rs6000/rs6000.c: Add case statement entry to make the
24124 xvcvuxdsp built-in argument unsigned.
24125 * config/rs6000/vsx.md: Fix the source and return operand types so they
24126 match the instruction definitions from the ISA document. Fix typo
24127 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24130 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24133 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24134 member early_clobber_alts.
24135 * lra-lives.c (reg_early_clobber_p): New.
24136 (process_bb_lives): Use it.
24137 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24138 (debug_operand_data): Initialize early_clobber_alts.
24139 (setup_operand_alternative): Set up early_clobber_alts.
24140 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24141 alternatives to new_insn_reg.
24142 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24144 (lra_update_insn_regno_info): Pass the new arg.
24146 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24148 PR middle-end/79505
24149 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24150 (new_oacc_loop_raw): Don't clear already cleared fields.
24153 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24154 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24155 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24157 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24160 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24162 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24165 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24166 the extra instruction to the right place to store 128-bit constant
24169 2017-02-14 Martin Sebor <msebor@redhat.com>
24171 PR middle-end/79448
24172 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24173 warning for strings of unknown length.
24175 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24177 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24179 2017-02-14 Jeff Law <law@redhat.com>
24182 * ira-costs.c (scan_one_insn): Initialize register move costs
24183 for pseudos seen in USE/CLOBBER insns.
24185 PR tree-optimization/79095
24186 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24187 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24188 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24189 if the operands are known to be not equal, then the resulting range
24191 (intersect_ranges): If the new range is ~[0,0] and the old range is
24192 wide, then prefer ~[0,0].
24193 * tree-vrp.c (overflow_comparison_p_1): New function.
24194 (overflow_comparison_p): New function.
24195 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24196 if NAME is used in an overflow test.
24197 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24198 overflow check that can be expressed as an equality test, then adjust
24199 ops to be that equality test.
24201 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24203 * config/s390/s390-builtin-types.def: Remove flags argument.
24204 * config/s390/s390.c (s390_init_builtins): Likewise.
24206 2017-02-14 Martin Liska <mliska@suse.cz>
24208 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24209 vector. Fix trailing white spaces.
24211 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24213 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24216 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24218 PR rtl-optimization/68664
24219 * config/arm/arm.c (arm_sched_can_speculate_insn):
24220 New function. Declare prototype.
24221 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24223 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24225 PR rtl-optimization/68664
24226 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24228 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24230 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24232 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24233 max skip bytes for function, loop and jump.
24235 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24237 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24238 ABS_EXPR for gimple dump.
24240 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24243 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24245 PR tree-optimization/79408
24246 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24247 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24248 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24249 also if rhs1 is INTEGER_CST.
24251 2017-02-14 Richard Biener <rguenther@suse.de>
24253 PR middle-end/79432
24254 * tree-into-ssa.c (insert_phi_nodes): When the function can
24255 have abnormal edges rewrite SSA names with broken use-def
24256 dominance out of SSA and register them for PHI insertion.
24258 2017-02-13 Martin Sebor <msebor@redhat.com>
24260 PR middle-end/79496
24261 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24262 clearing info.nowrite flag when snprintf size argument is a range.
24264 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24266 * cprop.c (cprop_jump): Add missing space in string literal.
24267 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24268 (get_constraint_for_component_ref): Likewise.
24269 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24270 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24271 * lra-constraints.c (process_alt_operands): Likewise.
24272 * ipa-inline.c (inline_small_functions): Likewise.
24273 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24274 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24275 * trans-mem.c (diagnose_tm_1_op): Likewise.
24276 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24277 (grid_parallel_clauses_gridifiable): Likewise.
24279 * config/nvptx/mkoffload.c (process): Add space in between
24282 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24283 "MOD4_SSE_REGS" and "ALL_REGS".
24285 * spellcheck.c (test_data): Add , in between "foo" and "food".
24287 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24290 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24291 boundary crossing check and subsequent code generation agree.
24293 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24295 * config/aarch64/aarch64.c (has_memory_op): Delete.
24296 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24299 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24301 PR rtl-optimization/79388
24302 PR rtl-optimization/79450
24303 * combine.c (distribute_notes): When removing TEM_INSN for which
24304 corresponding dest has last value recorded, invalidate that last
24307 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24309 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24310 of explicit '@'. Add missing assembly comment marker on branch costs
24313 2017-02-13 Nathan Sidwell <nathan@acm.org>
24315 * gengtype-lex.l (<in_struct>): Add '/'.
24317 2017-02-13 Martin Liska <mliska@suse.cz>
24320 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24322 2017-02-13 Richard Biener <rguenther@suse.de>
24324 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24326 * configure: Re-generate.
24327 * config.in: Likewise.
24328 * graphite-dependences.c: Simplify as if
24329 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24330 * graphite-isl-ast-to-gimple.c: Likewise.
24331 * graphite-optimize-isl.c: Likewise.
24332 * graphite-poly.c: Likewise.
24333 * graphite-sese-to-poly.c: Likewise.
24334 * graphite.h: Likewise.
24335 * toplev.c: Include isl/version.h and use isl_version () for
24336 printing the ISL version.
24337 * doc/install.texi: Update ISL requirement.
24339 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24341 * doc/standards.texi (Standards): Update reference to
24344 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24346 * doc/extend.texi (Named Address Spaces): sourceware.org now
24348 * doc/install.texi (Binaries): Ditto.
24351 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24353 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24354 standard terminology "stringize"/"stringizing" throughout.
24355 * doc/cppinternals.texi: Likewise.
24357 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24359 * doc/extend.texi: Fix some spelling mistakes and typos.
24360 * doc/invoke.texi: Likewise.
24362 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24365 * params.def (inline-min-speedup) Change from 10 to 8.
24367 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24369 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24372 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24375 * ipa-inline-analysis.c (get_minimal_bb): New function.
24376 (record_modified): Use it.
24377 (remap_edge_change_prob): Handle also ancestor functions.
24379 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24381 * doc/contrib.texi (Contributors): Remove broken link into
24382 the Mauve CVS repository.
24384 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24386 PR middle-end/79454
24387 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24388 result computation whenever lhs doesn't have vector mode, not
24389 just when it has BLKmode.
24391 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24393 * doc/makefile.texi (profiledbootstrap): Refer to the
24394 installation instructions only in textual form.
24396 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24399 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24401 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24403 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24404 (Specific): Update mingw-w64 reference.
24406 (Specific): Remove broken link to Renesas RX processor.
24408 2017-02-10 Richard Biener <rguenther@suse.de>
24410 * toplev.c (process_options): Do not mention obsolete graphite
24411 options when printing sorry message about missing graphite support.
24412 Mention -floop-nest-optimize.
24414 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24416 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24417 (vtst_p16): Likewise.
24418 (vtstq_p8): Likewise.
24419 (vtstq_p16): Likewise.
24421 (vtstq_p64): Likewise.
24422 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24423 (vset_lane_p64): New.
24424 (vsetq_lane_p64): New.
24426 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24428 PR tree-optimization/79411
24429 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24430 stmt operands are SSA_NAMEs used in abnormal phis.
24431 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24434 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24437 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24440 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24442 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24444 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24446 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24450 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24451 not arbitrary TREE_CONSTANT.
24454 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24455 "omp declare target link" attribute unless is_global_var.
24456 * omp-offload.c (find_link_var_op): Likewise.
24458 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24459 Chung-Lin Tang <cltang@codesourcery.com>
24461 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24463 (gimplify_adjust_omp_clauses): Don't delete TILE.
24464 (gimplify_omp_for): Deal with TILE.
24465 * internal-fn.c (expand_GOACC_TILE): New function.
24466 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24468 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24469 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24471 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24472 avoid DIV for outermost collapse var.
24473 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24474 Remove out of date comments, fix whitespace.
24475 * omp-general.c (omp_extract_for_data): Deal with tiling.
24476 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24477 adjust OLF_DIM_BASE value.
24478 (struct omp_for_data): Add tiling field.
24479 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24480 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24481 for auto loops. Remove default auto determining, moved to
24482 oacc_loop_fixed_partitions.
24483 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24484 stmts, add e_mask field.
24485 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24486 (oacc_thread_numbers): Use oacc_dim_call.
24487 (oacc_xform_tile): New.
24488 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24489 (finish_oacc_loop): Adjust for ifns vector.
24490 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24491 add case for GOACC_TILE fns.
24492 (oacc_loop_xform_loop): Delete.
24493 (oacc_loop_process): Iterate over call list directly, and add
24494 handling for GOACC_TILE fns.
24495 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24497 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24498 vector partitioning to outer loops. Assign 2 partitions to loops
24499 when available. Add TILE handling.
24500 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24501 (execite_oacc_device_lower): Process GOACC_TILE fns,
24502 ignore unknown specs.
24503 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24504 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24505 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24507 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24509 * configure.ac (ACX_BUGURL): Update.
24510 * configure: Regenerate.
24512 2017-02-09 Richard Biener <rguenther@suse.de>
24514 PR tree-optimization/69823
24515 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24516 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24518 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24520 * config/arc/arc-c.def: Add __NPS400__ definition.
24521 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24522 (TARGET_NPS400): Define.
24524 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24526 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24528 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24529 pointer, arch_info.
24530 (arc_cpu_types): Fill the arch_info field with a pointer into the
24531 arc_arch_types table.
24532 (arc_selected_cpu): Declare.
24533 * config/arc/arc.c (arc_selected_cpu): Make global.
24534 (arc_selected_arch): Delete.
24535 (arc_base_cpu): Delete.
24536 (arc_override_options): Remove references to deleted variables,
24537 update access to arch information.
24538 (ARC_OPT): Update access to arch information.
24539 (ARC_OPTX): Likewise.
24540 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24541 (TARGET_ARC600): Update access to arch information.
24542 (TARGET_ARC601): Likewise.
24543 (TARGET_ARC700): Likewise.
24544 (TARGET_EM): Likewise.
24545 (TARGET_HS): Likewise.
24546 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24549 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24552 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24553 condition/operands for integer GE/LE/GEU/LEU operations.
24555 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24557 PR translation/79397
24558 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
24561 2017-02-08 Martin Jambor <mjambor@suse.cz>
24564 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
24565 whether allocation happened.
24566 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
24567 nothing was allocated.
24569 2017-02-08 Jakub Jelinek <jakub@redhat.com>
24571 PR tree-optimization/79408
24572 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
24573 constant, but SSA_NAME with a known integer range, use the minimum
24574 of that range instead of op1 to determine if modulo can be replaced
24575 with its first operand.
24577 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24579 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
24581 2017-02-08 Richard Biener <rguenther@suse.de>
24583 PR tree-optimization/71824
24584 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24585 Check all loops contained in the merged region.
24587 2017-02-07 Andrew Pinski <apinski@cavium.com>
24589 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24591 2017-02-07 Andrew Pinski <apinski@cavium.com>
24593 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24594 (thunderxt88): Likewise.
24595 (thunderxt81): Disable LSE and change v8.1 to v8.
24596 (thunderxt83): Likewise.
24598 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24599 Richard Biener <rguenther@suse.de>
24601 PR middle-end/79399
24602 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24603 type from int to size_t.
24604 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24606 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24608 PR rtl-optimization/79386
24609 * cprop.c (bypass_conditional_jumps): Initialize
24610 bypass_last_basic_block already before splitting bbs after
24611 unconditional traps...
24612 (bypass_conditional_jumps): ... rather than here.
24615 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24616 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24617 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24618 fix -masm=intel patterns.
24620 2017-02-07 Richard Biener <rguenther@suse.de>
24622 PR tree-optimization/79256
24623 PR middle-end/79278
24624 * builtins.c (get_object_alignment_2): Use min_align_of_type
24625 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24626 and ADJUST_FIELD_ALIGN.
24628 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24630 * doc/tm.texi: Regenerate.
24631 * stor-layout.c (layout_decl): Adjust.
24632 (update_alignment_for_field): Likewise.
24633 (place_field): Likewise.
24634 (min_align_of_type): Likewise.
24635 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24636 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24637 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24638 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24639 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24640 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24641 * config/i386/i386.c (x86_field_alignment): Likewise.
24642 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24643 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24644 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24645 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24646 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24647 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24651 2017-01-30 Richard Biener <rguenther@suse.de>
24653 PR tree-optimization/79256
24654 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24655 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24658 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24660 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24661 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24662 builtins to SImode and emit a zero-extend, if necessary.
24664 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24666 * docs/invoke.texi (RISC-V Options): Alphabetize.
24668 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24670 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24673 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24675 * config/riscv/riscv.c: New file.
24676 * common/config/riscv/riscv-common.c: Likewise.
24677 * config.gcc: Likewise.
24678 * config/riscv/constraints.md: Likewise.
24679 * config/riscv/elf.h: Likewise.
24680 * config/riscv/generic.md: Likewise.
24681 * config/riscv/linux.h: Likewise.
24682 * config/riscv/multilib-generator: Likewise.
24683 * config/riscv/peephole.md: Likewise.
24684 * config/riscv/pic.md: Likewise.
24685 * config/riscv/predicates.md: Likewise.
24686 * config/riscv/riscv-builtins.c: Likewise.
24687 * config/riscv/riscv-c.c: Likewise.
24688 * config/riscv/riscv-ftypes.def: Likewise.
24689 * config/riscv/riscv-modes.def: Likewise.
24690 * config/riscv/riscv-opts.h: Likewise.
24691 * config/riscv/riscv-protos.h: Likewise.
24692 * config/riscv/riscv.h: Likewise.
24693 * config/riscv/riscv.md: Likewise.
24694 * config/riscv/riscv.opt: Likewise.
24695 * config/riscv/sync.md: Likewise.
24696 * config/riscv/t-elf-multilib: Likewise.
24697 * config/riscv/t-linux: Likewise.
24698 * config/riscv/t-linux-multilib: Likewise.
24699 * config/riscv/t-riscv: Likewise.
24700 * configure.ac: Likewise.
24701 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24702 Waterman as RISC-V maintainers.
24703 * doc/install.texi: Add RISC-V entries.
24704 * doc/invoke.texi: Add RISC-V options section.
24705 * doc/md.texi: Add RISC-V constraints section.
24706 * configure: Regenerated.
24708 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24711 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24712 false values to be constant vectors with all 0 or all 1 bits set.
24713 (vcondu<mode><mode>): Likewise.
24714 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24716 (fpmask_comparison_operator): Update comment.
24717 (vecint_comparison_operator): New predicate.
24718 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24719 vector conditionals when the true and false values are constant
24720 vectors with all 0 bits or all 1 bits set.
24722 2017-02-06 Martin Sebor <msebor@redhat.com>
24724 PR tree-optimization/79376
24725 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24727 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24729 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24730 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24731 to simplify split condition.
24733 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24735 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24736 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24739 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24741 PR rtl-optimization/68664
24742 * target.def (can_speculate_insn): New hook.
24743 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24744 * doc/tm.texi: Regenerate.
24745 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24746 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24747 (rs6000_sched_can_speculate_insn): New function.
24749 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24751 PR tree-optimization/79284
24752 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24753 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24754 vectorizable_mask_load_store, vectorizable_operation,
24755 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24756 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24757 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24758 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24759 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24760 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24761 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24762 is_gimple_assign (stmt). Replace another such test with
24763 is_gimple_assign (stmt).
24765 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24768 * config/avr/avr.c (rtl-iter.h): Include it.
24769 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24770 (avr_legitimate_combined_insn): ...and implementation.
24772 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24774 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24775 * config/s390/s390.c (s390_const_operand_ok)
24776 (s390_canonicalize_comparison, s390_extract_part)
24777 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24778 (s390_contiguous_bitmask_p, s390_rtx_costs)
24779 (legitimize_pic_address): Likewise.
24780 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24781 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24782 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24783 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24784 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24786 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24788 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24789 REGNO($0) == REGNO($1).
24791 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24793 * config/s390/linux.h(SIZE_TYPE): Add comment.
24795 2017-02-06 Julian Brown <julian@codesourcery.com>
24796 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24797 Virendra Pathak <virendra.pathak@broadcom.com>
24799 * config/aarch64/aarch64-cores.def: Change the scheduler
24801 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24802 * config/aarch64/thunderx2t99.md: New file.
24804 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24806 * doc/standards.texi (Go Language): Update link to language
24809 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24811 * tree-eh.c (lower_resx): Sanitize profile.
24812 (cleanup_empty_eh_move_lp): Likewise.
24814 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24817 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24819 * cfgloopmanip.h (loop_version): Update prototype.
24820 * modulo-sched.c (sms_schedule): Update call of loop_version.
24821 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24822 * tree-parloops.c (gen_parallel_loop): Likewise.
24823 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24824 * tree-ssa-loop-split.c (split_loop): Likewise.
24825 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24826 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24828 2017-02-05 Martin Liska <mliska@suse.cz>
24831 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24833 (print_operand_address): Initialize a struct to zero.
24835 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24837 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24838 garbage collector only in textual form.
24840 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24842 * doc/extend.texi (x86 specific memory model extensions for
24843 transactional memory): Simplify a phrase.
24845 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24848 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24849 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24850 (atomic_storedi_1): Likewise.
24852 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24854 PR tree-optimization/79338
24855 * tree-parloops.c (gather_scalar_reductions): Don't call
24856 vect_analyze_loop_form for loop->inner before destroying loop's
24859 2017-02-03 Martin Sebor <msebor@redhat.com>
24861 PR tree-optimization/79327
24862 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24863 when precision has resulted in leading zeros.
24864 (format_integer): Adjust the likely counter to assume an unknown
24865 argument that may be zero is non-zero.
24867 2017-02-03 Jason Merrill <jason@redhat.com>
24870 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24871 avoid copying non-taken branch.
24873 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24875 PR tree-optimization/79340
24876 * tree-vect-loop.c (vectorizable_reduction): Release
24877 vec_defs elements after safe_splicing them into other vectors.
24880 PR tree-optimization/79327
24881 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24882 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24884 (format_integer): Use wide_int_to_tree instead of build_int_cst
24885 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24886 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24887 of shortest and longest sequence.
24889 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24891 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24892 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24894 2017-02-03 Walter Lee <walt@tilera.com>
24897 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24898 after initial stackframe link reg save.
24899 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24901 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24904 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24905 wu for stxssp alternative.
24907 2017-02-03 Martin Sebor <msebor@redhat.com>
24909 PR tree-optimization/79352
24910 * gimple-fold.c (get_range_strlen): Add argument.
24911 (get_range_strlen): Change return type to bool.
24912 (get_maxval_strlen): Pass in a dummy argument.
24913 * gimple-fold.h (get_range_strlen): Change return type to bool.
24914 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24915 * tree.h (array_at_struct_end_p): Add argument.
24916 * tree.c (array_at_struct_end_p): Handle it.
24918 2017-02-03 Martin Liska <mliska@suse.cz>
24921 * multiple_target.c (create_dispatcher_calls): Redirect edge
24922 from a caller of a dispatcher.
24923 (expand_target_clones): Make the clones local.
24924 (ipa_target_clone): Do both target clones and resolvers.
24925 (ipa_dispatcher_calls): Remove the pass.
24926 (pass_dispatcher_calls::gate): Likewise.
24927 (make_pass_dispatcher_calls): Likewise.
24928 * passes.def (pass_target_clone): Put as very first IPA early
24931 2017-02-03 Martin Liska <mliska@suse.cz>
24933 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24934 in case of a function with ifunc attribute.
24936 2017-02-03 Martin Liska <mliska@suse.cz>
24938 * cgraph.c (cgraph_node::dump): Dump function version info.
24939 * symtab.c (symtab_node::dump_base): Add missing new line.
24941 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24943 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24944 (ifcombine_ifandif): Use it.
24946 2017-02-03 Martin Liska <mliska@suse.cz>
24948 * doc/invoke.texi: Document default value for
24949 use-after-scope-direct-emission-threshold.
24951 2017-02-03 Martin Liska <mliska@suse.cz>
24953 PR tree-optimization/79339
24954 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24955 (format_floating): Likewise.
24957 2017-02-03 Martin Liska <mliska@suse.cz>
24960 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24962 (ipa_node_params_t::remove): Likewise.
24963 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24964 initialization from removed ipa_node_params_t::insert.
24965 (ipa_node_params::~ipa_node_params): Move from removed
24966 ipa_node_params_t::release.
24967 * symbol-summary.h (symbol_summary::m_released): New member.
24968 Do not release a summary twice. Do not allow to call finalizer
24969 for types of a summary that live in GGC memory.
24971 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24973 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24976 2017-02-02 Martin Sebor <msebor@redhat.com>
24978 PR middle-end/79275
24979 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24980 (format_string): Tighten up the range of output for non-constant
24981 strings and correct the expected range for wide non-constant strings.
24983 2017-02-02 Martin Sebor <msebor@redhat.com>
24985 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24987 PR middle-end/32003
24988 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24990 (-fdump-tree-@var): Add to index and document how to come up
24991 with pass-specific option and dump file names.
24992 (-fdump-passes): Clarify where to look for output.
24994 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24996 PR middle-end/77445
24997 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24998 statistics of the analyzed path; allow threading for speed when
24999 any of BBs along the path are optimized for speed.
25001 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25003 PR middle-end/78468
25004 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25005 settings of the virtual registers.
25008 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25010 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25011 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25014 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25016 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25017 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25019 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25021 * config/s390/s390.md: Add missing comments with the expanded
25023 * config/s390/vector.md: Likewise.
25024 * config/s390/vx-builtins.md: Likewise.
25026 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25029 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25030 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25031 conditions on a single line.
25033 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25035 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25036 __S390_VX__ to __VX__.
25038 2017-02-01 Andrew Pinski <apinski@cavium.com>
25040 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25041 stmt_info to record_stmt_cost.
25042 (vect_get_known_peeling_cost): Pass stmt_info if known to
25044 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25045 cpu_vector_cost field into
25046 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25047 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25048 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25049 splitting of scalar_stmt_cost and vec_stmt_cost.
25050 (thunderx_vector_cost): Likewise.
25051 (cortexa57_vector_cost): LIkewise.
25052 (exynosm1_vector_cost): Likewise.
25053 (xgene1_vector_cost): Likewise.
25054 (thunderx2t99_vector_cost): Improve after the splitting of the two
25056 (aarch64_builtin_vectorization_cost): Update for the splitting of
25057 scalar_stmt_cost and vec_stmt_cost.
25059 2017-02-01 Torvald Riegel <triegel@redhat.com>
25060 Richard Henderson <rth@redhat.com>
25062 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25063 conditional on existance of a fast atomic load.
25064 * optabs-query.c (can_atomic_load_p): New function.
25065 * optabs-query.h (can_atomic_load_p): Declare it.
25066 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25067 no fast atomic load is available for the particular size of access.
25068 (expand_atomic_compare_and_swap): Likewise.
25069 (expand_atomic_load): Likewise.
25070 (expand_atomic_store): Likewise.
25071 (expand_atomic_fetch_op): Likewise.
25072 * testsuite/lib/target-supports.exp
25073 (check_effective_target_sync_int_128): Remove x86 because it provides
25074 no fast atomic load.
25075 (check_effective_target_sync_int_128_runtime): Likewise.
25077 2017-02-01 Richard Biener <rguenther@suse.de>
25079 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25080 (graphite_transform_loops): Provide opt-info for optimized nests.
25081 * tree-parloop.c (parallelize_loops): Provide opt-info for
25082 parallelized loops.
25084 2017-02-01 Richard Biener <rguenther@suse.de>
25086 PR middle-end/79315
25087 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25088 was not set before.
25090 2017-02-01 Richard Biener <rguenther@suse.de>
25092 PR tree-optimization/71824
25093 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25094 Verify the loops are valid in the merged SESE region.
25095 (scop_detection::can_represent_loop_1): Check analyzing the
25096 evolution of the number of iterations in the region succeeds.
25098 2017-01-31 Ian Lance Taylor <iant@golang.org>
25100 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25101 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25103 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25105 PR preprocessor/79210
25106 * input.c (get_substring_ranges_for_loc): Replace line_width
25107 assertion with error-handling.
25109 2017-01-31 Richard Biener <rguenther@suse.de>
25111 PR tree-optimization/77318
25112 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25113 (create_pw_aff_from_tree): Take loop parameter.
25114 (add_condition_to_pbb): Pass loop of the condition to
25115 create_pw_aff_from_tree.
25117 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25119 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25120 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25122 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25126 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25128 (convert_int_to_float128): Likewise.
25129 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25130 (convert_int_to_float128): Likewise.
25131 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25132 (UNSPEC_IEEE128_CONVERT): Likewise.
25133 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25134 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25135 Use local variables for IBM extended format.
25136 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25137 (fix_trunc<mode>si2_fprs): Likewise.
25138 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25139 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25140 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25141 to know that we can now have integers of all sizes in vector
25143 (fix<uns>_<mode>di2_hw): Likewise.
25144 (float<uns>_<mode>si2_hw): Likewise.
25145 (fix_<mode>si2_hw): Likewise.
25146 (fixuns_<mode>si2_hw): Likewise.
25147 (float<uns>_<mode>di2_hw): Likewise.
25148 (float_<mode>di2_hw): Likewise.
25149 (float_<mode>si2_hw): Likewise.
25150 (floatuns_<mode>di2_hw): Likewise.
25151 (floatuns_<mode>si2_hw): Likewise.
25152 (xscvqp<su>wz_<mode>): Delete, no longer used.
25153 (xscvqp<su>dz_<mode>): Likewise.
25154 (xscv<su>dqp_<mode>): Likewise.
25155 (ieee128_mfvsrd_64bit): Likewise.
25156 (ieee128_mfvsrd_32bit): Likewise.
25157 (ieee128_mfvsrwz): Likewise.
25158 (ieee128_mtvsrw): Likewise.
25159 (ieee128_mtvsrd_64bit): Likewise.
25160 (ieee128_mtvsrd_32bit): Likewise.
25162 2017-01-31 Martin Liska <mliska@suse.cz>
25165 * ipa-prop.c (ipa_free_all_node_params): Call release method
25166 instead of ~sumbol_summary to not to trigger double times
25169 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25171 PR tree-optimization/71691
25172 * bitmap.h (class auto_bitmap): New.
25173 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25174 is_maybe_undefined instead of ssa_undefined_value_p.
25176 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25178 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25179 __S390_ARCH_LEVEL__ to __ARCH__.
25181 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25183 PR tree-optimization/79267
25184 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25185 if should_remove_lhs_p is true.
25187 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25190 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25191 (add_alignment_attribute): New.
25192 (base_type_die): Add alignment attribute.
25193 (subrange_type_die): Likewise.
25194 (modified_type_die): Likewise.
25195 (gen_array_type_die): Likewise.
25196 (gen_descr_array_type_die: Likewise.
25197 (gen_enumeration_type_die): Likewise.
25198 (gen_subprogram_die): Likewise.
25199 (gen_variable_die): Likewise.
25200 (gen_field_die): Likewise.
25201 (gen_ptr_to_mbr_type_die): Likewise.
25202 (gen_struct_or_union_type_die): Likewise.
25203 (gen_subroutine_type_die): Likewise.
25204 (gen_typedef_die): Likewise.
25205 (base_type_cmp): Compare alignment attribute.
25207 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25210 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25211 (setb_unsigned) New pattern for setb with CCUNS.
25212 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25213 subfc./subfe sequence to avoid overflow problems. Generate a
25214 shorter sequence with cmpld/setb for power9.
25215 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25216 for generating subfc. instruction.
25217 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25218 now uses this instruction.
25220 2017-01-30 Ian Lance Taylor <iant@google.com>
25223 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25224 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25226 2017-01-30 Martin Sebor <msebor@redhat.com>
25228 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25229 Move constant to the right of a relational operator.
25230 (get_mpfr_format_length, format_character, format_string): Ditto.
25231 (should_warn_p, maybe_warn): Same.
25233 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25235 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25238 * asan.c (get_translation_unit_decl): Remove function.
25239 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25241 2017-01-30 Martin Liska <mliska@suse.cz>
25243 PR gcov-profile/79259
25244 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25245 -fprofile-generate.
25247 2017-01-30 Martin Liska <mliska@suse.cz>
25250 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25251 Initialize variables with NULL value.
25253 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25256 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25258 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25260 2017-01-30 Richard Biener <rguenther@suse.de>
25262 PR tree-optimization/79276
25263 * tree-vrp.c (process_assert_insertions): Properly adjust common
25264 when removing a duplicate.
25266 * gcc.dg/torture/pr79276.c: New testcase.
25268 2017-01-30 Richard Biener <rguenther@suse.de>
25270 PR tree-optimization/79256
25271 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25272 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25274 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25276 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25279 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25280 ("*r<noxa>sbg_<mode>_sll_bitmask")
25281 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25282 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25283 Use contiguous_bitmask_nowrap_operand.
25285 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25288 * config/rs6000/altivec.h (vec_xl): Revise #define.
25289 (vec_xst): Likewise.
25291 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25293 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25295 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25297 PR rtl-optimization/79194
25298 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25299 traps before call to bypass_conditional_jumps.
25301 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25303 PR tree-optimization/71374
25304 * lra-constraints.c (check_conflict_input_operands): New.
25305 (match_reload): Use it.
25307 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25310 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25311 account to calculate conflict_set.
25313 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25315 PR rtl-optimization/78559
25316 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25317 other_insn in combine.
25319 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25321 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25322 uint16_type_node for BT_UINT16.
25324 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25326 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25327 "RTL Tests" to menu.
25328 (GIMPLE Tests): New node.
25329 (RTL Tests): New node.
25331 2017-01-27 Richard Biener <rguenther@suse.de>
25333 PR tree-optimization/79245
25334 * tree-loop-distribution.c (distribute_loop): Apply cost
25335 modeling also to detected patterns.
25337 2017-01-27 Richard Biener <rguenther@suse.de>
25339 PR tree-optimization/71433
25340 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25341 (compare_assert_loc): New function.
25342 (process_assert_insertions): Sort and optimize assert locations
25343 to remove duplicates and push down identical assertions on
25344 edges to their destination block.
25346 2017-01-27 Richard Biener <rguenther@suse.de>
25348 PR tree-optimization/79244
25349 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25350 out SSA names even if abnormal.
25352 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25354 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25355 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25356 instead of MPFR_RNDN.
25358 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25361 * arm.c (arm_option_override): Don't call build_target_option_node
25362 until after doing all option overrides.
25363 (arm_valid_target_attribute_tree): Likewise.
25365 2017-01-27 Martin Liska <mliska@suse.cz>
25367 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25368 for {cd}tors and C++ {cd}tors.
25370 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25372 * config/s390/s390.md ("*setmem_long_and")
25373 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25375 2017-01-26 Martin Sebor <msebor@redhat.com>
25377 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25380 2017-01-26 Martin Sebor <msebor@redhat.com>
25382 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25383 HAVE_DFmode before using XFmode or DFmode.
25384 (parse_directive): Avoid using the z length modifier to avoid
25385 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25387 PR middle-end/78703
25388 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25389 to accept adjustment as an array.
25390 (get_int_range): New function.
25391 (struct directive): Make width and prec arrays.
25392 (directive::set_width, directive::set_precision): Call get_int_range.
25393 (format_integer, format_floating): Handle width and precision ranges.
25394 (format_string, parse_directive): Same.
25396 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25399 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25400 comdat_type_p set, just clone them, but keep the children in the
25404 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25405 which have direct callers with -fvar-tracking-assignments enabled
25407 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25408 inside of type units.
25410 2017-01-26 Martin Sebor <msebor@redhat.com>
25412 PR middle-end/78703
25413 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25415 (struct format_result): Replace number_chars, number_chars_min,
25416 and number_chars_max with a single member of struct result_range.
25418 (format_result::operator+=): Adjust.
25419 (struct fmtresult): Remove bounded. Handle likely and unlikely
25421 (fmtresult::adjust_for_width_or_precision): New function.
25422 (fmtresult:type_max_digits): New function.
25423 (bytes_remaining): Handle likely and unlikely counters.
25424 (min_bytes_remaining): Remove.
25425 (format_percent): Simplify.
25426 (format_integer, format_floating): Set likely and unlikely counters.
25427 (get_string_length, format_character, format_string): Same.
25428 (format_plain, should_warn_p): New function.
25429 (maybe_warn): Call should_warn_p. Update diagnostic messages
25430 and handle those for all directives, including plain strings.
25431 (format_directive): Handle likely and unlikely counters.
25432 Remove unnecessary quoting from diagnostics. Add an informational
25434 (add_bytes): Remove.
25435 (pass_sprintf_length::compute_format_length): Simplify.
25436 (try_substitute_return_value): Handle likely and unlikely counters.
25438 2017-01-26 Carl Love <cel@us.ibm.com>
25440 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25441 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25443 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25446 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25447 endianess for subregs into account.
25448 * lra-constraints.c (lra_constraints): Do risky transformations
25449 always on the first iteration.
25450 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25451 last_call_used_reg_set.
25452 (process_bb_lives): Define and use last_call_used_reg_set.
25453 * lra.c (lra): Always continue after lra_constraints on the first
25456 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25458 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25460 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25462 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25464 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25465 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25466 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25467 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25468 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25469 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25470 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25471 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25472 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25474 2017-01-26 Marek Polacek <polacek@redhat.com>
25477 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25478 for the third operand.
25480 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25482 PR middle-end/79236
25483 * omp-low.c (struct omp_context): Add simt_stmt field.
25484 (scan_omp_for): Return omp_context *.
25485 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25486 context to the _simt_ SIMD stmt.
25487 (lower_omp_for): For combined SIMD with sibling _simt_
25488 SIMD, make sure to use the same decls in _looptemp_
25489 clauses as in the sibling.
25491 2017-01-26 David Sherwood <david.sherwood@arm.com>
25493 PR middle-end/79212
25494 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25497 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25500 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25501 emit fld b; fld a; if possible.
25503 * brig-builtins.def: Update copyright years.
25504 * config/arm/arm_acle_builtins.def: Update copyright years.
25506 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25509 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25510 constraint instead of o for the stxsd instruction.
25512 2017-01-25 Carl Love <cel@us.ibm.com>
25514 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25515 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25517 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25519 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25521 2017-01-25 Richard Biener <rguenther@suse.de>
25523 PR tree-optimization/69264
25524 * target.def (vector_alignment_reachable): Improve documentation.
25525 * doc/tm.texi: Regenerate.
25526 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25528 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25529 earlier changes with respect to TYPE_USER_ALIGN.
25530 (vector_alignment_reachable_p): Likewise. Improve dumping.
25532 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25535 * config/arm/arm.md (xordi3): Force constant operand into a register
25538 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25540 * doc/invoke.texi (-fstore-merging): Correct default optimization
25541 levels at which it is enabled.
25542 (-O): Move -fstore-merging from list to...
25545 2017-01-25 Richard Biener <rguenther@suse.de>
25548 * omp-expand.c: Include debug.h.
25549 (expand_omp_taskreg): Make sure to generate early debug before
25550 outlining anything from a function.
25551 (expand_omp_target): Likewise.
25552 (grid_expand_target_grid_body): Likewise.
25554 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25557 * asan.c (get_translation_unit_decl): New function.
25558 (asan_add_global): Extract modules file name from globals
25559 TRANSLATION_UNIT_DECL name.
25561 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
25564 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
25565 for long calls with APCS frame and VFP.
25567 2017-01-24 David Malcolm <dmalcolm@redhat.com>
25569 * cfg.c (original_copy_tables_initialized_p): New function.
25570 * cfg.h (original_copy_tables_initialized_p): New decl.
25571 * cfgrtl.c (relink_block_chain): Guard the call to
25572 free_original_copy_tables with a call to
25573 original_copy_tables_initialized_p.
25574 * cgraph.h (symtab_node::native_rtl_p): New decl.
25575 * cgraphunit.c (symtab_node::native_rtl_p): New function.
25576 (symtab_node::needed_p): Don't assert for early assembly output
25577 for __RTL functions.
25578 (cgraph_node::finalize_function): Set "force_output" for __RTL
25580 (cgraph_node::analyze): Bail out early for __RTL functions.
25581 (analyze_functions): Update assertion to support __RTL functions.
25582 (cgraph_node::expand): Bail out early for __RTL functions.
25583 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25585 * function.h (struct function): Update comment for field
25587 * gimple-expr.c: Include "tree-pass.h".
25588 (gimple_has_body_p): Return false for __RTL functions.
25589 * Makefile.in (OBJS): Add run-rtl-passes.o.
25590 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25592 (gcc::pass_manager::get_clean_slate): New accessor.
25593 * passes.c: Include "insn-addr.h".
25594 (should_skip_pass_p): Add logging. Update logic for running
25595 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25596 property-provider override so it is only done for gimple passes.
25598 (skip_pass): New function.
25599 (execute_one_pass): Call skip_pass when skipping passes.
25600 * read-md.c (md_reader::read_char): Support filtering
25601 the input to a subset of line numbers.
25602 (md_reader::md_reader): Initialize fields
25603 m_first_line and m_last_line.
25604 (md_reader::read_file_fragment): New function.
25605 * read-md.h (md_reader::read_file_fragment): New decl.
25606 (md_reader::m_first_line): New field.
25607 (md_reader::m_last_line): New field.
25608 * read-rtl-function.c (function_reader::create_function): Only
25609 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25610 curr_properties. Set DECL_INITIAL to a dummy block.
25611 (read_rtl_function_body_from_file_range): New function.
25612 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25614 * run-rtl-passes.c: New file.
25615 * run-rtl-passes.h: New file.
25617 2017-01-24 Jeff Law <law@redhat.com>
25619 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25622 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25624 PR tree-optimization/79159
25625 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25626 (record_nonwrapping_iv): Improve boundary using above function if no
25627 value range information.
25629 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25630 Martin Jambor <mjambor@suse.cz>
25632 * brig-builtins.def: New file.
25633 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25634 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25635 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25636 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25637 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25638 * builtin-types.def (BT_INT8): New.
25639 (BT_INT16): Likewise.
25640 (BT_UINT8): Likewise.
25641 (BT_UINT16): Likewise.
25642 (BT_FN_ULONG): Likewise.
25643 (BT_FN_UINT_INT): Likewise.
25644 (BT_FN_UINT_ULONG): Likewise.
25645 (BT_FN_UINT_LONG): Likewise.
25646 (BT_FN_UINT_PTR): Likewise.
25647 (BT_FN_ULONG_PTR): Likewise.
25648 (BT_FN_INT8_FLOAT): Likewise.
25649 (BT_FN_INT16_FLOAT): Likewise.
25650 (BT_FN_UINT32_FLOAT): Likewise.
25651 (BT_FN_UINT16_FLOAT): Likewise.
25652 (BT_FN_UINT8_FLOAT): Likewise.
25653 (BT_FN_UINT64_FLOAT): Likewise.
25654 (BT_FN_UINT16_UINT32): Likewise.
25655 (BT_FN_UINT32_UINT16): Likewise.
25656 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25657 (BT_FN_INT_PTR_INT): Likewise.
25658 (BT_FN_UINT_PTR_UINT): Likewise.
25659 (BT_FN_LONG_PTR_LONG): Likewise.
25660 (BT_FN_ULONG_PTR_ULONG): Likewise.
25661 (BT_FN_VOID_UINT64_UINT64): Likewise.
25662 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25663 (BT_FN_INT8_INT8_INT8): Likewise.
25664 (BT_FN_INT16_INT16_INT16): Likewise.
25665 (BT_FN_INT_INT_INT): Likewise.
25666 (BT_FN_UINT_FLOAT_UINT): Likewise.
25667 (BT_FN_FLOAT_UINT_UINT): Likewise.
25668 (BT_FN_ULONG_UINT_UINT): Likewise.
25669 (BT_FN_ULONG_UINT_PTR): Likewise.
25670 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25671 (BT_FN_UINT_UINT_UINT): Likewise.
25672 (BT_FN_VOID_UINT_PTR): Likewise.
25673 (BT_FN_UINT_UINT_PTR: Likewise.
25674 (BT_FN_UINT32_UINT64_PTR): Likewise.
25675 (BT_FN_INT_INT_UINT_UINT): Likewise.
25676 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25677 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25678 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25679 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25680 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25681 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25682 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25683 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25684 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25685 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25686 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25687 * doc/frontends.texi: List BRIG FE.
25688 * doc/install.texi (Testing): Add BRIG tesring requirements.
25689 * doc/invoke.texi (Overall Options): Mention BRIG.
25690 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25692 2017-01-24 Richard Biener <rguenther@suse.de>
25694 PR translation/79208
25695 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25697 2017-01-24 Martin Jambor <mjambor@suse.cz>
25700 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25701 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25702 and known_contexts.
25704 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25706 PR middle-end/79123
25707 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25708 casts from signed to unsigned really don't have a range.
25710 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25712 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25713 GMP_RNDx for compatiblity.
25715 2017-01-24 Martin Liska <mliska@suse.cz>
25718 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25719 that would prevent us to call alloca with -1 as argument.
25721 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25723 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25724 Avoid -Wformat-security warning.
25726 2017-01-23 Andrew Pinski <apinski@cavium.com>
25728 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25731 2017-01-23 Martin Sebor <msebor@redhat.com>
25733 PR middle-end/78703
25734 * gimple-ssa-sprintf.c (warn_level): New global.
25735 (format_integer): Use it here and throughout the rest of the file.
25736 Use the same switch to compute sign as base.
25737 (maybe_warn): New function.
25738 (format_directive): Factor out warnings into maybe_warn.
25739 Add debugging output. Use warn_level.
25740 (add_bytes): Use warn_level.
25741 (pass_sprintf_length::compute_format_length): Add debugging output.
25742 (try_substitute_return_value): Same.
25743 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25745 PR middle-end/78703
25746 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25747 (struct fmtresult, format_integer, format_floating): Adjust.
25748 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25749 (get_string_length, format_string,format_directive): Same.
25750 (pass_sprintf_length::compute_format_length): Same.
25751 (try_substitute_return_value): Simplify slightly.
25753 PR middle-end/78703
25754 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25755 (fmtresult::operator+=): Outlined.
25756 (struct fmtresult): Add ctors.
25757 (struct conversion_spec): Rename...
25758 (struct directive): ...to this. Add and remove data members.
25759 (directive::set_width, directive::set_precision): New functions.
25760 (format_percent): Use fmtresult ctor.
25761 (get_width_and_precision): Remove.
25762 (format_integer): Make naming changes. Avoid computing width and
25764 (format_floating): Same. Adjust indentation.
25765 (format_character, format_none): New functions.
25766 (format_string): Moved character handling to format_character.
25767 (format_directive): Remove arguments, change return type.
25768 (parse_directive): New function.
25769 (pass_sprintf_length::compute_format_length): Move directive
25770 parsing to parse_directive.
25772 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25774 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25775 (assign_assembler_name_if_needed): ... this.
25776 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25777 (assign_assembler_name_if_needed): ... this.
25778 (free_lang_data_in_cgraph): Adjust callers.
25779 * cgraphunit.c (cgraph_node::analyze): Likewise.
25780 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25783 2017-01-23 Richard Biener <rguenther@suse.de>
25785 PR tree-optimization/79088
25786 PR tree-optimization/79188
25787 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25788 resetting loop bounds after last path deletion. Reset loop
25789 bounds of the target loop, make code match the comments.
25790 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25791 Make sure loops need no fixups.
25793 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25795 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25796 exponent support with double type for first argument.
25797 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25798 type returned by __builtin_vec_extract_sig,
25799 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25800 functions from "vector int" to "vector unsigned int" or from
25801 "vector long long int" to "vector unsigned long long int".
25802 Changed type returned by __builtin_vec_extract_exp,
25803 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25804 functions from "vector int" to "vector unsigned int" or from
25805 "vector long long int" to "vector unsigned long long int".
25806 Changed return type of __builtin_vec_test_data_class,
25807 __builtin_vec_test_data_class_sp, and
25808 __builtin_vec_test_data_class_dp from "vector int" to
25809 "vector bool int" or from "vector long long int" to "vector bool
25810 long long int" and changed second argument type from "unsigned
25811 int" to "int". Added new overloaded function forms "vector float
25812 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25813 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25814 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25815 double, vector unsigned long long int)" and "vector double
25816 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25817 long int)". Changed return type of
25818 __builtin_scalar_test_data_class and
25819 __builtin_scalar_test_data_class_sp and
25820 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25821 int" and changed second argument from "unsigned int" to "int".
25822 Changed type returned by __builtin_scalar_test_neg,
25823 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25824 from "int" to "bool int". Added new overloaded function form
25825 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25826 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25827 exponent double-precision with floating point first argument.
25828 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25829 documentation of scalar_test_data_class, scalar_test_neg,
25830 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25831 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25832 vec_test_data_class built-in functions to reflect refinements in
25833 their type signatures.
25835 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25837 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25839 (aarch64_elf_asm_destructor): Likewise.
25841 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25843 PR rtl-optimization/78634
25844 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25845 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25846 * ifcvt.c (noce_try_cmove): Add missing cost check.
25848 PR rtl-optimization/71724
25849 * combine.c (if_then_else_cond): Look for situations where it is
25850 beneficial to undo the work of one of the recursive calls.
25852 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25854 PR tree-optimization/70754
25855 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25856 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25857 combined stmt before it if not NULL.
25858 (combine_chains): Process refs reversely and compute dominance point
25861 2017-01-23 Martin Liska <mliska@suse.cz>
25863 PR tree-optimization/79196
25864 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25865 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25867 (strlen_optimize_stmt): Call the renamed function.
25869 2017-01-23 Michael Matz <matz@suse.de>
25871 PR tree-optimization/78384
25872 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25874 2017-01-23 Richard Biener <rguenther@suse.de>
25876 PR tree-optimization/79186
25877 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25878 both incoming edges before moving an assert.
25880 2017-01-23 Martin Jambor <mjambor@suse.cz>
25882 * ipa-prop.c (load_from_param_1): Removed.
25883 (load_from_unmodified_param): Bits from load_from_param_1 put back
25885 (load_from_param): Removed.
25886 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25887 with stmt. Reverted back to use of load_from_unmodified_param.
25889 2017-01-23 Martin Jambor <mjambor@suse.cz>
25892 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25893 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25894 field a pointer to garbage collected vector, mark lattices and
25895 ipcp_orig_node with GTY((skip)).
25896 (ipa_get_param_count): Adjust to descriptors being a pointer.
25897 (ipa_get_param): Likewise.
25898 (ipa_get_type): Likewise.
25899 (ipa_get_param_move_cost): Likewise.
25900 (ipa_set_param_used): Likewise.
25901 (ipa_get_controlled_uses): Likewise.
25902 (ipa_set_controlled_uses): Likewise.
25903 (ipa_is_param_used): Likewise.
25904 (ipa_node_params_t): Move into garbage collector. New methods insert
25906 (ipa_node_params_sum): Annotate wth GTY(()).
25907 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25909 (ipa_load_from_parm_agg): Adjust declaration.
25910 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25911 * ipa-profile.c (ipa_profile): Likewise.
25912 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25913 (ipa_populate_param_decls): Make descriptors parameter garbage
25915 (ipa_dump_param): Adjust to descriptors being a pointer.
25916 (ipa_alloc_node_params): Likewise.
25917 (ipa_initialize_node_params): Likewise.
25918 (load_from_param_1): Make descriptors parameter garbage collected.
25919 (load_from_unmodified_param): Likewise.
25920 (load_from_param): Likewise.
25921 (ipa_load_from_parm_agg): Likewise.
25922 (ipa_node_params::~ipa_node_params): Removed.
25923 (ipa_free_all_node_params): Remove call to delete operator.
25924 (ipa_node_params_t::insert): New.
25925 (ipa_node_params_t::remove): Likewise.
25926 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25927 copy known_csts and known_contexts vectors.
25928 (ipa_read_node_info): Adjust to descriptors being a pointer.
25929 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25931 (ipcp_transform_function): Make descriptors variable garbage
25934 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25936 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25937 * config/i386/avx512dqintrin.h: Ditto.
25938 * config/i386/avx512fintrin.h: Ditto.
25939 * config/i386/i386.c: Handle new builtins.
25940 * config/i386/i386-builtin.def: Add new builtins.
25941 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25942 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25944 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25945 Martin Liska <mliska@suse.cz>
25947 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25948 * asan.c (asan_expand_poison_ifn): Support stores and use
25949 appropriate ASAN report function.
25950 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25951 * internal-fn.def (ASAN_POISON_USE): Declare.
25952 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25953 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25954 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25955 ASAN_POISON calls w/o LHS.
25956 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25957 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25958 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25959 * gimplify.c (asan_poison_variables): Add attribute
25960 use_after_scope_memory to variables that really needs to live
25962 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25963 having the attribute.
25965 2017-01-23 Martin Liska <mliska@suse.cz>
25967 * asan.c (create_asan_shadow_var): New function.
25968 (asan_expand_poison_ifn): Likewise.
25969 * asan.h (asan_expand_poison_ifn): New declaration.
25970 * internal-fn.c (expand_ASAN_POISON): Likewise.
25971 * internal-fn.def (ASAN_POISON): New builtin.
25972 * sanopt.c (pass_sanopt::execute): Expand
25973 asan_expand_poison_ifn.
25974 * tree-inline.c (copy_decl_for_dup_finish): Make function
25976 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25977 * tree-ssa.c (is_asan_mark_p): New function.
25978 (execute_update_addresses_taken): Rewrite local variables
25979 (identified just by use-after-scope as addressable) into SSA.
25981 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25983 * doc/install.texi (Specific): opensource.apple.com uses https
25984 now. Remove trailing slash.
25986 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25988 * README.Portability: Remove note on an Irix compatibility issue.
25990 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25992 * gcov.c (INCLUDE_ALGORITHM): Define.
25993 (INCLUDE_VECTOR): Define.
25994 No longer include <vector> and <algorithm> directly.
25996 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25998 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26000 * doc/invoke.texi (Code Gen Options): Ditto.
26002 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26005 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26007 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26009 rtl-optimization/79125
26010 * cprop.c (local_cprop_pass): Handle cases where we make an
26011 unconditional trap.
26013 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26017 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26018 read from, for big endian.
26020 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26022 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26023 register pauth builtins for LP64 only.
26025 2017-01-20 Marek Polacek <polacek@redhat.com>
26028 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26031 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26033 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26035 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26036 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26037 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26039 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26042 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26043 in big-endian mode when they are not a single duplicated value.
26045 2017-01-20 Richard Biener <rguenther@suse.de>
26047 * BASE-VER: Bump to 7.0.1.
26049 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26051 * omp-low.c (omplow_simd_context): New struct. Use it...
26052 (lower_rec_simd_input_clauses): ...here and...
26053 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26054 references to idx, lane, max_vf, is_simt.
26056 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26058 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26061 2017-01-20 Martin Jambor <mjambor@suse.cz>
26063 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26064 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26066 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26067 (GTFILES): Rename hsa.c to hsa-common.c.
26068 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26069 * hsa-dump.c: Likewise.
26070 * hsa-gen.c: Likewise.
26071 * hsa-regalloc.c: Likewise.
26072 * ipa-hsa.c: Likewise.
26073 * omp-expand.c: Likewise.
26074 * omp-low.c: Likewise.
26075 * toplev.c: Likewise.
26077 2017-01-20 Marek Polacek <polacek@redhat.com>
26080 * doc/invoke.texi: Document -Wduplicated-branches.
26081 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26082 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26083 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26084 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26085 return 0 only when not OEP_LEXICOGRAPHIC.
26086 (fold_build_cleanup_point_expr): Use the expression
26087 location when building CLEANUP_POINT_EXPR.
26088 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26089 * tree.c (add_expr): Handle error_mark_node.
26091 2017-01-20 Martin Liska <mliska@suse.cz>
26094 * tree-profile.c (init_ic_make_global_vars): Do not call
26096 (gimple_init_gcov_profiler): Likewise.
26098 2017-01-20 Martin Liska <mliska@suse.cz>
26101 * cgraph.h (maybe_create_reference): Remove argument and
26103 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26105 * ipa-cp.c (create_specialized_node): Likewise.
26106 * symtab.c (symtab_node::maybe_create_reference): Handle
26107 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26109 2017-01-20 Martin Liska <mliska@suse.cz>
26111 * read-rtl-function.c (function_reader::create_function): Use
26112 build_decl instread of build_decl_stat.
26114 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26116 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26117 * config/i386/avx512dqintrin.h: Ditto.
26118 * config/i386/avx512fintrin.h: Ditto.
26119 * config/i386/i386-builtin-types.def: Add new types.
26120 * config/i386/i386.c: Handle new types.
26121 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26122 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26123 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26124 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26125 (__builtin_ia32_kshiftridi): New.
26126 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26128 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26132 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26133 define to rs6000_init_stack_protect_guard.
26134 (rs6000_init_stack_protect_guard): New function.
26136 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26137 Yunqiang Su <yunqiang.su@imgtec.com>
26139 * config.gcc (supported_defaults): Add madd4.
26140 (with_madd4): Add validation.
26141 (all_defaults): Add madd4.
26142 * config/mips/mips.opt (mmadd4): New option.
26143 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26145 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26147 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26148 (ISA_HAS_FUSED_MADD4): Likewise.
26149 * doc/invoke.texi (-mmadd4): Document the new option.
26150 * doc/install.texi (--with-madd4): Document the new option.
26152 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26154 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26155 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26156 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26157 (aarch64_init_pauth_hint_builtins): New.
26158 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26159 (aarch64_expand_builtin): Expand new builtins.
26161 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26163 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26164 * combine-stack-adj.c (no_unhandled_cfa): Handle
26165 REG_CFA_TOGGLE_RA_MANGLE.
26166 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26167 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26168 info for return address signing.
26169 (aarch64_expand_epilogue): Likewise.
26171 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26173 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26174 * config/aarch64/aarch64-protos.h
26175 (aarch64_return_address_signing_enabled): New declaration.
26176 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26178 (aarch64_expand_prologue): Sign return address before it's pushed onto
26180 (aarch64_expand_epilogue): Authenticate return address fetched from
26182 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26183 (aarch64_attributes): New function attributes for "sign-return-address".
26184 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26185 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26186 ("*do_return"): Generate combined instructions according to key index.
26187 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26188 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26190 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26191 * config/aarch64/aarch64.opt (msign-return-address=): New.
26192 * doc/extend.texi (AArch64 Function Attributes): Documents
26193 "sign-return-address=".
26194 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26196 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26198 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26199 overall option summary.
26201 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26203 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26204 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26205 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26206 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26208 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26210 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26211 -mpower9-minmax by default for -mcpu=power9.
26212 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26213 128-bit floating point.
26215 2017-01-20 Alan Modra <amodra@gmail.com>
26217 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26218 optimizing for size.
26220 2017-01-20 Alan Modra <amodra@gmail.com>
26223 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26224 for strcmp and strncmp from corresponding builtin decl.
26226 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26228 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26229 instead of i386/rtems-64.h.
26230 * config/i386/rtems-64.h: Remove.
26232 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26236 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26238 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26240 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26242 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26243 Change int to HOST_WIDE_INT.
26244 * config/aarch64/aarch64-protos.h
26245 (aarch64_simd_gen_const_vector_dup): Likewise.
26246 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26248 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26250 * langhooks-def.h (lhd_type_for_size): New decl.
26251 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26252 * langhooks.c (lhd_type_for_size): New function, taken from
26255 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26257 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26258 define_bypass for CR latency.
26259 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26260 (power9-alu2): Add define_bypass for CR latency.
26262 (power9-mul): Update insn latency.
26263 (power9-mul-compare): Update insn latency, bypass latency and remove
26266 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26268 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26270 * config/aarch64/aarch64.md
26271 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26272 aarch64_nopcrelative_literal_loads.
26273 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26275 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26277 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26278 TARGET_LOONGSON_3A.
26279 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26281 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26284 * config.gcc (supported_defaults): Add lxc1-sxc1.
26285 (with_lxc1_sxc1): Add validation.
26286 (all_defaults): Add lxc1-sxc1.
26287 * config/mips/mips.opt (mlxc1-sxc1): New option.
26288 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26290 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26291 __mips_no_lxc1_sxc1.
26292 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26293 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26294 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26296 2017-01-19 Richard Biener <rguenther@suse.de>
26298 PR tree-optimization/72488
26299 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26300 sure to restore SSA info.
26301 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26303 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26305 PR rtl-optimization/79121
26306 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26307 of the inner type when shifting an extended value.
26309 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26312 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26313 interposable aliases.
26315 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26318 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26319 Use the evmergelohi instruction.
26320 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26321 (mov_sitf_e500_subreg8_2_be): Likewise.
26322 (mov_sitf_e500_subreg12_2_le): Likewise.
26323 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26324 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26325 (mov_sitf_e500_subreg8_2_le): Likewise.
26326 (mov_sitf_e500_subreg12_2_be): Likewise.
26328 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26330 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26331 attribute from vecsimple to vecperm.
26332 (altivec_vbpermq2): Likewise.
26334 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26337 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26339 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26340 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26341 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26342 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26343 case where N arg is SIZE_MAX.
26344 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26345 (cmpstrsi): Add pattern.
26347 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26350 __builtin_vec_revb builtins.
26351 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26352 built-in functions to support generation of the ISA 3.0 XXBR<x>
26353 vector byte reverse instructions.
26354 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26355 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26356 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26357 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26358 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26359 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26360 (P9V_BUILTIN_VEC_REVB): Likewise.
26361 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26362 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26363 (p9_xxbrq_v16qi): Likewise.
26364 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26365 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26366 (p9_xxbrh_v8hi): Likewise.
26367 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26368 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26369 vec_revb built-in functions.
26371 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26373 PR rtl-optimization/78952
26374 * config/i386/i386.md (any_extract): New code iterator.
26375 (*insvqi_2): Use any_extract for source operand.
26376 (*insvqi_3): Use any_shiftrt for source operand.
26378 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26380 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26382 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26384 2017-01-18 Matthias Klose <doko@ubuntu.com>
26386 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26388 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26390 * config/rs6000/altivec.h (vec_bperm): Change #define.
26391 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26392 (altivec_vbpermq2): New define_insn.
26393 (altivec_vbpermd): Likewise.
26394 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26395 function interface.
26396 (VBPERMD): Likewise.
26397 (VBPERM): New polymorphic function interface.
26398 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26399 Add entries for P9V_BUILTIN_VEC_VBPERM.
26400 * doc/extend.texi: Add interfaces for vec_bperm.
26402 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26404 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26405 first letter of error messages.
26406 (s390_resolve_overloaded_builtin): Likewise.
26407 * config/s390/s390.c (s390_expand_builtin): Likewise.
26408 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26409 (s390_valid_target_attribute_inner_p): Likewise.
26410 * config/s390/s390.md ("tabort"): Likewise.
26412 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26414 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26415 (ISA_AVOID_DIV_HILO): New macro.
26416 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26417 (ISA_HAS_DDIV): Likewise.
26419 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26421 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26423 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26425 * doc/invoke.texi (fabi-version): Spelling fix.
26427 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26430 * doc/invoke.texi (fabi-version): Mention mangling fix for
26433 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26436 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26438 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26441 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26444 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26446 (stack_protect_test): Ditto.
26447 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26448 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26450 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26451 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26452 -mstack-protector-guard-offset=.
26453 (RS/6000 and PowerPC Options): Ditto.
26455 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26457 * config/i386/i386.h (MASK_CLASS_P): New define.
26458 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26459 there are no registers from different register sets also when
26460 mask registers are used. Update function comment.
26461 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26462 to (*k/*r) and (*k/*km) alternatives.
26464 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26466 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26467 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26468 (EH_RETURN_HANDLER_RTX): New define.
26469 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26470 Force frame pointer in EH return functions.
26471 (aarch64_expand_epilogue): Add barrier for eh_return.
26472 (aarch64_final_eh_return_addr): Remove.
26473 (aarch64_eh_return_handler_rtx): New function.
26474 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26476 (aarch64_eh_return_handler_rtx): New prototype.
26478 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26480 * config/rs6000/altivec.h (vec_rlmi): New #define.
26481 (vec_vrlnm): Likewise.
26482 (vec_rlnm): Likewise.
26483 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26484 (UNSPEC_VRLNM): Likewise.
26485 (VIlong): New mode iterator.
26486 (altivec_vrl<VI_char>mi): New define_insn.
26487 (altivec_vrl<VI_char>nm): Likewise.
26488 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26490 (VRLDNM): Likewise.
26491 (RLNM): New polymorphic function entry.
26492 (VRLWMI): New monomorphic function entry.
26493 (VRLDMI): Likewise.
26494 (RLMI): New polymorphic function entry.
26495 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26496 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26497 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26500 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26503 * dwarf2out.c (field_byte_offset): Restore the
26504 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26505 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26506 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26509 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26512 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26514 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26517 * dwarf2out.c (add_data_member_location_attribute): For constant
26518 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26519 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26520 DW_AT_byte_size attributes.
26522 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26524 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26525 after forcing to constant memory when the code model is medium.
26527 2017-01-17 Julia Koval <julia.koval@intel.com>
26530 * config/i386/avx512fintrin.h
26531 (_mm512_i32gather_ps): Change __addr type to void const*.
26532 (_mm512_mask_i32gather_ps): Ditto.
26533 (_mm512_i32gather_pd): Ditto.
26534 (_mm512_mask_i32gather_pd): Ditto.
26535 (_mm512_i64gather_ps): Ditto.
26536 (_mm512_mask_i64gather_ps): Ditto.
26537 (_mm512_i64gather_pd): Ditto.
26538 (_mm512_mask_i64gather_pd): Ditto.
26539 (_mm512_i32gather_epi32): Ditto.
26540 (_mm512_mask_i32gather_epi32): Ditto.
26541 (_mm512_i32gather_epi64): Ditto.
26542 (_mm512_mask_i32gather_epi64): Ditto.
26543 (_mm512_i64gather_epi32): Ditto.
26544 (_mm512_mask_i64gather_epi32): Ditto.
26545 (_mm512_i64gather_epi64): Ditto.
26546 (_mm512_mask_i64gather_epi64): Ditto.
26547 (_mm512_i32scatter_ps): Change __addr type to void*.
26548 (_mm512_mask_i32scatter_ps): Ditto.
26549 (_mm512_i32scatter_pd): Ditto.
26550 (_mm512_mask_i32scatter_pd): Ditto.
26551 (_mm512_i64scatter_ps): Ditto.
26552 (_mm512_mask_i64scatter_ps): Ditto.
26553 (_mm512_i64scatter_pd): Ditto.
26554 (_mm512_mask_i64scatter_pd): Ditto.
26555 (_mm512_i32scatter_epi32): Ditto.
26556 (_mm512_mask_i32scatter_epi32): Ditto.
26557 (_mm512_i32scatter_epi64): Ditto.
26558 (_mm512_mask_i32scatter_epi64): Ditto.
26559 (_mm512_i64scatter_epi32): Ditto.
26560 (_mm512_mask_i64scatter_epi32): Ditto.
26561 (_mm512_i64scatter_epi64): Ditto.
26562 (_mm512_mask_i64scatter_epi64): Ditto.
26563 * config/i386/avx512pfintrin.h
26564 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
26565 (_mm512_mask_prefetch_i32gather_ps): Ditto.
26566 (_mm512_mask_prefetch_i64gather_pd): Ditto.
26567 (_mm512_mask_prefetch_i64gather_ps): Ditto.
26568 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
26569 (_mm512_prefetch_i32scatter_ps): Ditto.
26570 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
26571 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
26572 (_mm512_prefetch_i64scatter_pd): Ditto.
26573 (_mm512_prefetch_i64scatter_ps): Ditto.
26574 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
26575 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
26576 * config/i386/avx512vlintrin.h
26577 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
26578 (_mm_mmask_i32gather_ps): Ditto.
26579 (_mm256_mmask_i32gather_pd): Ditto.
26580 (_mm_mmask_i32gather_pd): Ditto.
26581 (_mm256_mmask_i64gather_ps): Ditto.
26582 (_mm_mmask_i64gather_ps): Ditto.
26583 (_mm256_mmask_i64gather_pd): Ditto.
26584 (_mm_mmask_i64gather_pd): Ditto.
26585 (_mm256_mmask_i32gather_epi32): Ditto.
26586 (_mm_mmask_i32gather_epi32): Ditto.
26587 (_mm256_mmask_i32gather_epi64): Ditto.
26588 (_mm_mmask_i32gather_epi64): Ditto.
26589 (_mm256_mmask_i64gather_epi32): Ditto.
26590 (_mm_mmask_i64gather_epi32): Ditto.
26591 (_mm256_mmask_i64gather_epi64): Ditto.
26592 (_mm_mmask_i64gather_epi64): Ditto.
26593 (_mm256_i32scatter_ps): Change __addr type to void*.
26594 (_mm256_mask_i32scatter_ps): Ditto.
26595 (_mm_i32scatter_ps): Ditto.
26596 (_mm_mask_i32scatter_ps): Ditto.
26597 (_mm256_i32scatter_pd): Ditto.
26598 (_mm256_mask_i32scatter_pd): Ditto.
26599 (_mm_i32scatter_pd): Ditto.
26600 (_mm_mask_i32scatter_pd): Ditto.
26601 (_mm256_i64scatter_ps): Ditto.
26602 (_mm256_mask_i64scatter_ps): Ditto.
26603 (_mm_i64scatter_ps): Ditto.
26604 (_mm_mask_i64scatter_ps): Ditto.
26605 (_mm256_i64scatter_pd): Ditto.
26606 (_mm256_mask_i64scatter_pd): Ditto.
26607 (_mm_i64scatter_pd): Ditto.
26608 (_mm_mask_i64scatter_pd): Ditto.
26609 (_mm256_i32scatter_epi32): Ditto.
26610 (_mm256_mask_i32scatter_epi32): Ditto.
26611 (_mm_i32scatter_epi32): Ditto.
26612 (_mm_mask_i32scatter_epi32): Ditto.
26613 (_mm256_i32scatter_epi64): Ditto.
26614 (_mm256_mask_i32scatter_epi64): Ditto.
26615 (_mm_i32scatter_epi64): Ditto.
26616 (_mm_mask_i32scatter_epi64): Ditto.
26617 (_mm256_i64scatter_epi32): Ditto.
26618 (_mm256_mask_i64scatter_epi32): Ditto.
26619 (_mm_i64scatter_epi32): Ditto.
26620 (_mm_mask_i64scatter_epi32): Ditto.
26621 (_mm256_i64scatter_epi64): Ditto.
26622 (_mm256_mask_i64scatter_epi64): Ditto.
26623 (_mm_i64scatter_epi64): Ditto.
26624 (_mm_mask_i64scatter_epi64): Ditto.
26625 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26626 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26627 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26628 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26629 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26630 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26631 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26632 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26633 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26634 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26635 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26636 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26637 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26638 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26639 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26640 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26641 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26642 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26643 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26644 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26645 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26646 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26647 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26648 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26649 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26650 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26651 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26652 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26653 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26654 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26655 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26656 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26657 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26658 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26659 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26660 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26661 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26662 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26663 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26664 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26665 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26666 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26667 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26668 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26669 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26670 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26671 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26672 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26673 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26674 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26675 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26676 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26677 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26678 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26679 definitions accordingly.
26681 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26682 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26685 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26688 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26691 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26692 hard regs for inner regno.
26694 2017-01-17 Martin Liska <mliska@suse.cz>
26697 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26698 assumption and add comment.
26700 2017-01-17 Nathan Sidwell <nathan@acm.org>
26702 * ipa-visibility.c (localize_node): New function, broken out of ...
26703 (function_and_variable_visibility): ... here. Call it.
26705 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26707 PR middle-end/77445
26708 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26709 correctly set frequency of oudgoing edge.
26710 (duplicate_thread_path): Fix profile updating.
26712 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26715 * configure.ac: Add GCC_BASE_VER.
26716 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26717 version from BASE-VER file.
26718 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26719 (gcc.o): Depend on $(BASEVER).
26720 * common.opt (dumpfullversion): New option.
26721 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26722 * doc/invoke.texi: Document -dumpfullversion.
26723 * doc/install.texi: Document --with-gcc-major-version-only.
26724 * configure: Regenerated.
26726 2017-01-17 Richard Biener <rguenther@suse.de>
26728 PR tree-optimization/71433
26729 * tree-vrp.c (register_new_assert_for): Merge same asserts
26730 on all incoming edges.
26731 (process_assert_insertions_for): Handle insertions at the
26734 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26736 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26737 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26739 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26742 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26745 2017-01-17 Alan Modra <amodra@gmail.com>
26748 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26749 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26750 symbolic stack limit when pic.
26752 2017-01-16 Martin Sebor <msebor@redhat.com>
26754 PR tree-optimization/78608
26755 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26757 2017-01-16 Jeff Law <law@redhat.com>
26760 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26761 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26762 for several include directories that may be relative to sysroot.
26763 * config/i386/x-mingw32 (gplus_includedir): Define.
26764 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26765 (native_system_includedir): Likewise.
26766 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26767 override if TARGET_SYSTEM_ROOT is defined.
26768 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26770 PR tree-optimization/79090
26771 PR tree-optimization/33562
26772 PR tree-optimization/61912
26773 PR tree-optimization/77485
26774 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26775 and computed trims into the dump file.
26777 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26779 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26781 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26784 * gimplify.c (gimplify_init_constructor): If want_value and
26785 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26789 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26790 sequence. Formatting fixes.
26791 (doloop_optimize): Formatting fixes.
26794 * gcc.c (debug_level_greater_than_spec_func): New function.
26795 (static_spec_functions): Add debug-level-gt spec function.
26796 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26798 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26799 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26800 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26801 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26802 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26803 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26805 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26807 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26808 QImode fixups to general and mask registers only.
26810 2017-01-16 Carl Love <cel@us.ibm.com>
26812 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26813 for built-in functions
26814 vector signed char vec_nabs (vector signed char)
26815 vector signed short vec_nabs (vector signed short)
26816 vector signed int vec_nabs (vector signed int)
26817 vector signed long long vec_nabs (vector signed long long)
26818 vector float vec_nabs (vector float)
26819 vector double vec_nabs (vector double)
26820 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26822 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26823 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26824 * doc/extend.texi: Update the documentation file for the new built-in
26827 2017-01-16 Martin Sebor <msebor@redhat.com>
26829 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26832 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26834 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26835 UNSPEC_VSX__XXSPLTD to require special splat handling.
26837 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26840 * system.h: Poison strndup.
26842 2017-01-16 Alan Modra <amodra@gmail.com>
26845 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26848 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26850 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26852 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26854 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26855 call recog here. Assert that INSN_CODE (insn) is non-negative.
26857 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26860 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26862 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26863 in the currently scheduled RTL fragment.
26865 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26867 PR rtl-optimization/78751
26868 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26871 2017-01-14 Jeff Law <law@redhat.com>
26873 PR tree-optimization/79090
26874 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26875 variable length stores.
26876 (compute_trims): Delete dead assignment to *trim_tail.
26877 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26880 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26882 PR rtl-optimization/78626
26883 PR rtl-optimization/78727
26884 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26885 of a block, and split such blocks after everything else is finished.
26887 2017-01-14 Alan Modra <amodra@gmail.com>
26890 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26891 target legitimate_combined_insn.
26892 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26893 (rs6000_legitimate_combined_insn): New function.
26894 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26896 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26897 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26898 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26900 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26902 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26904 2017-01-13 Jeff Law <law@redhat.com>
26906 PR tree-optimization/33562
26907 PR tree-optimization/61912
26908 PR tree-optimization/77485
26909 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26911 (delete_dead_assignment): Likewise.
26912 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26913 statement to delete_dead_call and delete_dead_assignment.
26915 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26918 * substring-locations.c (format_warning_va): Strengthen case 1 so
26919 that both endpoints of the substring must be within the format
26920 range for just the substring to be printed.
26922 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26924 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26925 * config/i386/i386.c (ix86_target_string): Add missing options
26926 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26927 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26928 flags_other and ix86_target_other to flags2_other. Display unknown
26930 (ix86_valid_target_attribute_inner_p): Add missing options and
26931 reorder options by implied ISAs, as in ix86_target_string.
26933 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26935 * hash-table.h (hash_table::too_empty_p): New function.
26936 (hash_table::expand): Use it.
26937 (hash_table::traverse): Likewise.
26938 (hash_table::empty_slot): Use sizeof (value_type) instead of
26939 sizeof (PTR) to convert bytes to elements. Shrink the table
26940 if the current size is excessive for the current number of
26943 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26945 * ira-costs.c (record_reg_classes): Break from the inner loop
26946 early once alt_fail is known to be true. Update outer loop
26947 handling accordingly.
26949 2017-01-13 Jeff Law <law@redhat.com>
26951 * tree-ssa-dse.c (decrement_count): New function.
26952 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26953 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26954 when we know the partially dead statement is a mem* function.
26956 PR tree-optimization/61912
26957 PR tree-optimization/77485
26958 * tree-ssa-dse.c: Include expr.h.
26959 (maybe_trim_constructor_store): New function.
26960 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26962 PR tree-optimization/33562
26963 PR tree-optimization/61912
26964 PR tree-optimization/77485
26965 * doc/invoke.texi: Document new dse-max-object-size param.
26966 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26967 * tree-ssa-dse.c: Include params.h.
26968 (dse_store_status): New enum.
26969 (initialize_ao_ref_for_dse): New, partially extracted from
26971 (valid_ao_ref_for_dse, normalize_ref): New.
26972 (setup_live_bytes_from_ref, compute_trims): Likewise.
26973 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26974 (maybe_trim_partially_dead_store): Likewise.
26975 (maybe_trim_complex_store): Likewise.
26976 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26977 Track what bytes live from the original store. Return tri-state
26978 for dead, partially dead or live.
26979 (dse_dom_walker): Add constructor, destructor and new private members.
26980 (delete_dead_call, delete_dead_assignment): New extracted from
26982 (dse_optimize_stmt): Make a member of dse_dom_walker.
26983 Use initialize_ao_ref_for_dse.
26985 PR tree-optimization/33562
26986 PR tree-optimization/61912
26987 PR tree-optimization/77485
26988 * sbitmap.h (bitmap_count_bits): Prototype.
26989 (bitmap_clear_range, bitmap_set_range): Likewise.
26990 * sbitmap.c (bitmap_clear_range): New function.
26991 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26993 2017-01-13 Martin Liska <mliska@suse.cz>
26996 * function.c (set_cfun): Add new argument force.
26997 * function.h (set_cfun): Likewise.
26998 * ipa-inline-transform.c (inline_call): Use the function when
26999 strict alising from is dropped for function we inline to.
27001 2017-01-13 Richard Biener <rguenther@suse.de>
27003 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27004 for dumping GIMPLE INTEGER_CSTs.
27006 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27008 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27009 to 201112L since C++17.
27011 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27014 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27015 if -fsanitize=kernel-address is present.
27017 2017-01-13 Richard Biener <rguenther@suse.de>
27019 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27020 as _Literal ( type ) number in case usual suffixes do not
27021 preserve all information.
27023 2017-01-13 Richard Biener <rguenther@suse.de>
27025 PR tree-optimization/77283
27026 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27027 and ssa-iterators.h.
27028 (is_feasible_trace): Implement a cost model based on joiner
27031 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27034 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27035 char or short to __float128/_Float128 directly.
27037 2017-01-12 Martin Sebor <msebor@redhat.com>
27039 to -Wformat-overflow.
27040 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27041 (min_bytes_remaining): Same.
27042 (get_string_length): Same.
27043 (format_string): Same.
27044 (format_directive): Same.
27046 (pass_sprintf_length::handle_gimple_call): Same.
27048 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27050 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27051 info.nowrite calls with no lhs that can't throw. Return bool
27052 whether gsi_remove has been called or not.
27053 (pass_sprintf_length::handle_gimple_call): Return bool whether
27054 try_substitute_return_value called gsi_remove. Formatting fix.
27055 (pass_sprintf_length::execute): Don't use gsi_remove if
27056 handle_gimple_call returned true.
27059 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27060 be removed due to side-effects, don't remove following barrier nor
27061 turn the successor edge into fallthru edge.
27063 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27066 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27067 element-reversing loads and stores as not swappable.
27069 2017-01-12 Nathan Sidwell <nathan@acm.org>
27070 Nicolai Stange <nicstange@gmail.com>
27072 * combine.c (try_combine): Don't ignore result of overlap checking
27073 loop. Combine overlap & asm check into single loop.
27075 2017-01-12 Richard Biener <rguenther@suse.de>
27077 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27078 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27080 2017-01-12 Richard Biener <rguenther@suse.de>
27082 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27083 and TS_TARGET_OPTION directly derive from TS_BASE.
27084 * tree-core.h (tree_optimization_option): Derive from tree_base.
27085 (tree_target_option): Likewise.
27087 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27089 * config/i386/i386.c (memory_address_length): Increase len
27090 only when rip_relative_addr_p returns false.
27092 2017-01-11 Julia Koval <julia.koval@intel.com>
27094 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27095 (OPTION_MASK_ISA_SGX_SET): New.
27096 (ix86_handle_option): Handle OPT_msgx.
27097 * config.gcc: Added sgxintrin.h.
27098 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27099 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27100 * config/i386/i386.c (ix86_target_string): Add -msgx.
27102 (ix86_option_override_internal): Handle new options.
27103 (ix86_valid_target_attribute_inner_p): Add sgx.
27104 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27105 * config/i386/i386.opt: Add msgx.
27106 * config/i386/sgxintrin.h: New file.
27107 * config/i386/x86intrin.h: Add sgxintrin.h.
27109 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27112 * fold-const.c (maybe_nonzero_address): Return 1 for function
27114 (tree_single_nonzero_warnv_p): Don't handle function local objects
27118 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27121 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27124 * opts.c: Include "spellcheck.h"
27125 (struct string_fragment): New struct.
27126 (struct edit_distance_traits<const string_fragment &>): New
27128 (get_closest_sanitizer_option): New function.
27129 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27131 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27133 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27135 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27136 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27137 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27138 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27139 for initial die_offset if dwarf_split_debug_info.
27140 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27141 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27143 (output_skeleton_debug_sections): Formatting fix. Use
27144 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27145 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27147 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27149 * config/arm/cortex-a53.md: Add bypasses for
27150 cortex_a53_r2f_cvt.
27151 (cortex_a53_r2f): Only use for transfers.
27152 (cortex_a53_f2r): Likewise.
27153 (cortex_a53_r2f_cvt): Add reservation for conversions.
27154 (cortex_a53_f2r_cvt): Likewise.
27156 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27158 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27159 to all inlined functions, change static to extern.
27161 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27164 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27166 (arm_assemble_integer): Likewise.
27168 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27170 * config.gcc: Use new awk script to check CPU, FPU and architecture
27171 parameters for --with-... options.
27172 * config/arm/parsecpu.awk: New file
27173 * config/arm/arm-cpus.in: New file.
27174 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27176 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27178 * config/arm/t-arm: Update dependency rules.
27179 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27180 of processing .def files.
27181 * config/arm/genopt.sh: Deleted.
27182 * config/arm/gentune.sh: Deleted.
27183 * config/arm/arm-cores.def: Deleted.
27184 * config/arm/arm-arches.def: Deleted.
27185 * config/arm/arm-fpus.def: Deleted.
27186 * config/arm/arm-tune.md: Regenerated.
27187 * config/arm/arm-tables.opt: Regenerated.
27188 * config/arm/arm-cpu.h: New generated file.
27189 * config/arm/arm-cpu-data.h: New generated file.
27190 * config/arm/arm-cpu-cdata.h: New generated file.
27192 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27195 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27197 (input_varpool_node): Unpack dynamically_initialized bit.
27199 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27201 PR rtl-optimization/79032
27202 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27203 the alignment of the adjusted memory reference against that of MODE,
27204 instead of the alignment of the original memory reference.
27206 2017-01-11 Martin Jambor <mjambor@suse.cz>
27208 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27210 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27211 decorated functions.
27213 2017-01-11 Richard Biener <rguenther@suse.de>
27215 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27216 set range/nonnull info for PHI results. Do not set it on
27217 stmts marked for removal.
27219 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27221 * expr.c (store_field): In the bitfield case, fetch the return value
27222 from the registers before applying a single big-endian adjustment.
27223 Always do a final load for a BLKmode value not larger than a word.
27225 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27228 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27229 that we correctly handle column numbers greater than
27230 LINE_MAP_MAX_COLUMN_NUMBER.
27232 2017-01-10 Martin Sebor <msebor@redhat.com>
27234 PR middle-end/78245
27235 * gimple-ssa-sprintf.c (get_destination_size): Call
27236 {init,fini}object_sizes.
27237 * tree-object-size.c (addr_object_size): Adjust.
27238 (pass_through_call): Adjust.
27239 (pass_object_sizes::execute): Adjust.
27240 * tree-object-size.h (fini_object_sizes): Declare.
27242 2017-01-10 Martin Sebor <msebor@redhat.com>
27244 PR tree-optimization/78775
27245 * builtins.c (get_size_range): Move...
27246 * calls.c: ...to here.
27247 (alloc_max_size): Accept zero argument.
27248 (operand_signed_p): Remove.
27249 (maybe_warn_alloc_args_overflow): Call get_size_range.
27250 * calls.h (get_size_range): Declare.
27252 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27254 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27255 from TI's devices.csv file as of September 2016.
27256 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27258 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27260 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27261 * doc/invoke.texi: Likewise.
27262 * doc/md.texi: Likewise.
27263 * doc/objc.texi: Likewise.
27265 2017-01-10 Joshua Conner <joshconner@google.com>
27267 * config/arm/fuchsia-elf.h: New file.
27268 * config/fuchsia.h: New file.
27269 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27270 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27272 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27274 2016-01-10 Richard Biener <rguenther@suse.de>
27276 PR tree-optimization/79034
27277 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27278 Propagate out degenerate PHIs in the joiner.
27280 2017-01-10 Martin Liska <mliska@suse.cz>
27282 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27283 (sort_congruence_classes_by_decl_uid): Likewise.
27284 (sort_congruence_class_groups_by_decl_uid): Likewise.
27285 (sem_item_optimizer::merge_classes): Sort class, groups in these
27286 classes and members in the groups by DECL_UID of declarations.
27287 This would make merge operations stable.
27289 2017-01-10 Martin Liska <mliska@suse.cz>
27291 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27292 usage of m_classes_vec.
27293 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27294 (sem_item_optimizer::get_group_by_hash): Likewise.
27295 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27296 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27297 (sem_item_optimizer::verify_classes): Likewise.
27298 (sem_item_optimizer::process_cong_reduction): Likewise.
27299 (sem_item_optimizer::dump_cong_classes): Likewise.
27300 (sem_item_optimizer::merge_classes): Likewise.
27301 * ipa-icf.h (congruence_class_hash): Rename from
27302 congruence_class_group_hash. Remove declaration of m_classes_vec.
27304 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27306 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27307 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27308 * config.gcc: Add avx512vpopcntdqintrin.h.
27309 * config/i386/avx512vpopcntdqintrin.h: New.
27310 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27311 * config/i386/i386-builtin-types.def: Add new types.
27312 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27313 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27314 __builtin_ia32_vpopcountq_v8di_mask): New.
27315 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27316 __AVX512VPOPCNTDQ__.
27317 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27318 (PTA_AVX512VPOPCNTDQ): Define.
27319 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27320 TARGET_AVX512VPOPCNTDQ_P): Define.
27321 * config/i386/i386.opt: Add mavx512vpopcntdq.
27322 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27323 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27325 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27327 PR middle-end/77484
27328 * predict.def (PRED_CALL): Set to 67.
27330 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27332 * expr.c (store_field): In the bitfield case, if the value comes from
27333 a function call and is of an aggregate type returned in registers, do
27334 not modify the field mode; extract the value in all cases if the mode
27335 is BLKmode and the size is not larger than a word.
27337 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27340 * config/i386/cpuid.h: Fix undefined behavior.
27342 2017-01-04 Jeff Law <law@redhat.com>
27344 PR tree-optimization/79007
27345 PR tree-optimization/67955
27346 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27347 conservative for pt.null when flag_non_call_exceptions is on.
27349 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27351 PR translation/79019
27352 PR translation/79020
27353 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27354 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27355 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27357 * config/avr/avr.opt (maccumulate-args): Likewise.
27358 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27359 * common.opt (freport-bug): Likewise.
27360 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27361 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27362 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27363 translatable string.
27364 * config/i386/i386.c (function_value_32): Likewise.
27365 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27366 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27368 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27369 * common/config/msp430/msp430-common.c (msp430_handle_option):
27371 * symtab.c (symtab_node::verify_base): Likewise.
27372 * opts.c (set_debug_level): Likewise.
27373 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27374 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27375 missing whitespace to translatable strings.
27376 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27377 * config/sh/superh.h: Likewise.
27378 * config/i386/xopintrin.h: Likewise.
27379 * config/i386/znver1.md: Likewise.
27380 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27381 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27382 * double-int.h (struct double_int): Likewise.
27383 * double-int.c (div_and_round_double): Likewise.
27384 * wide-int.cc: Likewise.
27385 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27386 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27387 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27388 (crossjumps_occurred): ... this.
27389 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27392 PR tree-optimization/78899
27393 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27394 returning bool return struct loop *, NULL for failure and the new
27396 (versionable_outer_loop_p): Don't version outer loop if it has
27397 dont_vectorized bit set.
27398 (tree_if_conversion): When versioning outer loop, ensure
27399 tree_if_conversion is performed also on the inner loop of the
27400 non-vectorizable outer loop copy.
27401 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27402 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27403 prevent vectorization of it.
27404 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27405 the outer loop vectorization of the non-scalar version is attempted
27406 before vectorization of the inner loop in scalar version. If
27407 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27408 vectorization of its inner loop.
27409 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27410 has 2 inner loops, rename also on edges from bb whose single pred
27411 is outer_loop->header. Fix typo in function comment.
27413 2017-01-09 Martin Sebor <msebor@redhat.com>
27416 * asan.c (asan_emit_stack_protection): Increase local buffer size
27417 to avoid snprintf truncation warning.
27419 2017-01-09 Andrew Pinski <apinski@cavium.com>
27421 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27422 to reference thunderx2t99 for the tuning structure
27423 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27425 (thunderx2t99_extra_costs): This.
27426 * config/aarch64/aarch64-tune.md: Regenerate.
27427 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27428 (vulcan_addrcost_table): This.
27429 (vulcan_regmove_cost): Rename to ...
27430 (thunderx2t99_regmove_cost): This.
27431 (vulcan_vector_cost): Rename to ...
27432 (thunderx2t99_vector_cost): this.
27433 (vulcan_branch_cost): Rename to ...
27434 (thunderx2t99_branch_cost): This.
27435 (vulcan_tunings): Rename to ...
27436 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27437 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27439 2017-01-09 Martin Jambor <mjambor@suse.cz>
27443 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27444 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27445 (propagate_vr_accross_jump_function): Use the above function for all
27446 value range computations for pass-through jump functions and type
27447 converasion from explicit value range values.
27448 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27449 parameters from TYPE_ARG_TYPES.
27450 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27451 (ipa_write_node_info): Stream type of the actual argument.
27452 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27454 2017-01-09 Martin Liska <mliska@suse.cz>
27457 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27458 (lookup_compiler): Do not show error message with have_E.
27460 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27462 PR tree-optimization/78938
27463 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27464 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27465 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27466 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27469 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27471 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27474 2017-01-09 Richard Biener <rguenther@suse.de>
27476 PR tree-optimization/78997
27477 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27478 name condition properly.
27480 2017-01-09 Richard Biener <rguenther@suse.de>
27483 * dwarf2out.c (is_cxx): New overload with context.
27484 (is_naming_typedef_decl): Use it.
27486 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27488 * invoke.texi (Option Summary): Correct spacing in option lists
27489 and add line breaks to fix over-long lines.
27491 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27493 PR middle-end/17660
27495 * extend.texi (Common Variable Attributes): Add xref to GCC
27496 Internals manual to explain mode attribute keywords.
27498 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27501 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27502 and Preprocessor Options.
27503 (Options for Linking): Document -pthread here....
27504 (RS/6000 and PowerPC Options): ...not here.
27505 (Solaris 2 Options): ...or here.
27506 * doc/cppopts.texi: Document -pthread.
27508 2017-01-08 Martin Sebor <msebor@redhat.com>
27510 PR middle-end/77708
27511 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27512 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27513 New member functions.
27514 (format_directive): Used them.
27516 (pass_sprintf_length::handle_gimple_call): Same.
27517 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27518 to avoid truncation for any argument.
27519 (extract_affine_mul): Same.
27520 * tree.c (get_file_function_name): Same.
27522 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27524 PR middle-end/77484
27525 * predict.def (PRED_INDIR_CALL): Set to 86.
27527 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27529 PR preprocessor/54124
27530 * doc/cppopts.texi: Reformat -d subtable to list the full name
27531 of the options. Add cross-reference to the docs for the general
27532 compiler -d options.
27533 * doc/invoke.texi (Developer Options): Add cross-reference to the
27534 preprocessor-specific -d option documentation.
27536 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27538 PR preprocessor/13498
27539 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27540 redudant material, and reflect new command-line options.
27541 (System Headers): Likewise.
27543 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27545 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27546 -isystem, and -idirafter. Copy-edit.
27547 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27548 default for -ftrack-macro-expansion. Delete obsolete and
27549 badly-formatted implementation details about -fdebug-cpp output.
27550 * doc/cppwarnopts.texi: Copy-edit.
27552 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27555 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27556 that the transition from a max line width >= 1<<10 to narrower
27557 lines works correctly.
27559 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
27561 * doc/options.texi (PerFunction): New.
27562 * opt-functions.awk (switch_flags): Map both Optimization and
27563 PerFunction to CL_OPTIMIZATION.
27564 * opth-gen.awk: Test for PerFunction flag along with
27566 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
27567 it only when the latter is present. Skip those that don't in
27568 the hash function generator.
27569 * common.opt (fvar-tracking): Mark as PerFunction instead of
27571 (fvar-tracking-assignments): Likewise.
27572 (fvar-tracking-assignments-toggle): Likewise.
27573 (fvar-tracking-uninit): Likewise.
27575 2017-01-07 Jakub Jelinek <jakub@redhat.com>
27577 PR translation/79018
27578 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
27581 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27584 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27585 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27586 TARGET_LONG_JUMP_TABLE_OFFSETS.
27587 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27588 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27589 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27590 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27591 * config/m68k/m68k.md (tablejump expander): Likewise.
27592 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27593 TARGET_LONG_JUMP_TABLE_OFFSETS.
27594 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27595 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27597 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27598 David Holsgrove <david.holsgrove@xilinx.com>
27600 * common/config/microblaze/microblaze-common.c
27601 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27602 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27604 * config/microblaze/microblaze.c (microblaze_must_save_register)
27605 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27607 (microblaze_eh_return): New function.
27608 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27609 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27610 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27611 * config/microblaze/microblaze.md (eh_return): New pattern.
27613 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27615 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27616 GCC_DIAGNOSTIC_STRINGIFY): Define.
27618 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27620 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27622 * config/arm/arm.md (<mcrr>): New.
27624 * config/arm/arm.c (arm_arch5te): New.
27625 (arm_option_override): Set arm_arch5te.
27626 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27628 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27629 (arm_mcrr_qualifiers): ... this. New.
27630 (MRRC_QUALIFIERS): Define to...
27631 (arm_mrrc_qualifiers): ... this. New.
27632 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27634 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27635 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27636 (MRRCI, mrrc, MRRC): New.
27637 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27638 VUNSPEC_MRRC2): New.
27640 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27642 * config/arm/arm.md (<mcr>): New.
27644 * config/arm/arm.c (arm_coproc_builtin_available): Add
27645 support for mcr, mrc, mcr2 and mrc2.
27646 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27647 (arm_mcr_qualifiers): ... this. New.
27648 (MRC_QUALIFIERS): Define to ...
27649 (arm_mrc_qualifiers): ... this. New.
27650 (MCR_QUALIFIERS): Define to ...
27651 (arm_mcr_qualifiers): ... this. New.
27652 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27654 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27655 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27656 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27657 VUNSPEC_MRC2): New.
27659 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27661 * config/arm/arm.md (*ldc): New.
27665 * config/arm/arm.c (arm_coproc_builtin_available): Add
27666 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27667 (arm_coproc_ldc_stc_legitimate_address): New.
27668 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27669 'qualifier_const_pointer'.
27670 (LDC_QUALIFIERS): Define to...
27671 (arm_ldc_qualifiers): ... this. New.
27672 (STC_QUALIFIERS): Define to...
27673 (arm_stc_qualifiers): ... this. New.
27674 * config/arm/arm-protos.h
27675 (arm_coproc_ldc_stc_legitimate_address): New.
27676 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27677 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27678 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27679 stc2, stcl, stc2l): New.
27680 * config/arm/constraints.md (Uz): New.
27681 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27682 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27683 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27684 VUNSPEC_STC2L): New.
27686 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27688 * config/arm/arm.md (<cdp>): New.
27689 * config/arm/arm.c (neon_const_bounds): Rename this ...
27690 (arm_const_bounds): ... this.
27691 (arm_coproc_builtin_available): New.
27692 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27693 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27694 (CDP_QUALIFIERS): Define to...
27695 (arm_cdp_qualifiers): ... this. New.
27697 (arm_expand_builtin_args): Add case for 6 arguments.
27698 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27699 (arm_const_bounds): ... this.
27700 (arm_coproc_builtin_available): New.
27701 * config/arm/arm_acle.h (__arm_cdp): New.
27703 * config/arm/arm_acle_builtins.def (cdp): New.
27705 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27706 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27707 'arm_const_bounds'.
27708 * config/arm/types.md (coproc): New.
27709 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27710 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27711 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27712 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27714 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27716 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27717 (UBINOP_QUALIFIERS): New.
27719 (acle_builtin_data): New. Change comment.
27720 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27721 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27722 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27723 arm_acle_builtins.def.
27724 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27725 (arm_init_acle_builtins): New.
27726 (CRC32_BUILTIN): Remove.
27727 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27728 crc32cb, crc32ch and crc32cw.
27729 (arm_init_crc32_builtins): Remove.
27730 (arm_init_builtins): Use arm_init_acle_builtins rather
27731 than arm_init_crc32_builtins.
27732 (arm_expand_acle_builtin): New.
27733 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27734 * config/arm/arm_acle_builtins.def: New.
27736 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27738 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27739 (arm_builtin_datum): ... this.
27740 (arm_init_neon_builtin): Rename to ...
27741 (arm_init_builtin): ... this. Add a new parameters PREFIX
27742 and USE_SIG_IN_NAME.
27743 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27744 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27745 'arm_builtin_datum'.
27746 (arm_init_vfp_builtins): Likewise.
27747 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27748 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27749 (arm_expand_neon_args): Rename to ...
27750 (arm_expand_builtin_args): ... this. Rename builtin_arg
27751 enum values and differentiate between ARG_BUILTIN_MEMORY
27752 and ARG_BUILTIN_NEON_MEMORY.
27753 (arm_expand_neon_builtin_1): Rename to ...
27754 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27755 values, arm_expand_builtin_args and add bool parameter NEON.
27756 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27757 (arm_expand_vfp_builtin): Likewise.
27758 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27760 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27762 PR middle-end/77484
27763 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27764 * predict.c (tree_estimate_probability_bb): Reverse direction of
27765 polymorphic call predictor.
27767 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27769 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27770 (determine_pass_name_match): ...this new function and...
27771 (should_skip_pass_p): ...this new function.
27773 2017-01-06 Nathan Sidwell <nathan@acm.org>
27775 * ipa-visibility.c (function_and_variable_visibility): Reformat
27776 comments and long lines. Remove extrneous if.
27777 * symtab.c (symtab_node::make_decl_local): Fix code format.
27778 (symtab_node::set_section_for_node): Fix comment typo.
27780 2017-01-06 Martin Liska <mliska@suse.cz>
27783 * lra-constraints.c: Rename invariant to lra_invariant.
27784 * predict.c (set_even_probabilities): Initialize e to NULL.
27786 2017-01-05 Martin Sebor <msebor@redhat.com>
27788 PR tree-optimization/78910
27789 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27790 (format_integer): Correct off-by-one error in the handling
27791 of precision with negative numbers in signed conversions..
27793 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27795 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27797 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27799 PR tree-optimization/71016
27800 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27801 factor_out_conditional_conversion. Formatting fix.
27802 (factor_out_conditional_conversion): Add cond_stmt argument.
27803 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27804 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27807 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27809 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27810 read-rtl-function.o, and selftest-rtl.o.
27811 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27812 (selftest::aarch64_test_loading_full_dump): New function.
27813 (selftest::aarch64_run_selftests): New function.
27814 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27815 selftest::aarch64_run_selftests.
27816 * config/i386/i386.c
27817 (selftest::ix86_test_loading_dump_fragment_1): New function.
27818 (selftest::ix86_test_loading_call_insn): New function.
27819 (selftest::ix86_test_loading_full_dump): New function.
27820 (selftest::ix86_test_loading_unspec): New function.
27821 (selftest::ix86_run_selftests): Call the new functions.
27822 * emit-rtl.c (maybe_set_max_label_num): New function.
27823 * emit-rtl.h (maybe_set_max_label_num): New decl.
27824 * function.c (instantiate_decls): Guard call to
27825 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27826 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27828 * gensupport.c (gen_reader::gen_reader): Pass "false"
27829 for new "compact" param of rtx_reader.
27830 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27831 rather than an empty string for NULL strings.
27832 * read-md.c: Potentially include config.h rather than bconfig.h.
27833 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27834 (have_error): New global, copied from errors.c.
27835 (md_reader::read_name): Rename to...
27836 (md_reader::read_name_1): ...this, adding "out_loc" param,
27837 and converting "missing name or number" to returning false, rather
27839 (md_reader::read_name): Reimplement in terms of read_name_1.
27840 (md_reader::read_name_or_nil): New function.
27841 (md_reader::read_string): Handle "(nil)" by returning NULL.
27842 (md_reader::md_reader): Add new param "compact".
27843 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27844 (md_reader::read_file): New method.
27845 * read-md.h (md_reader::md_reader): Add new param "compact".
27846 (md_reader::read_file): New method.
27847 (md_reader::is_compact): New accessor.
27848 (md_reader::read_name): Convert return type from void to file_location.
27849 (md_reader::read_name_or_nil): New decl.
27850 (md_reader::read_name_1): New decl.
27851 (md_reader::m_compact): New field.
27852 (noop_reader::noop_reader): Pass "false" for new "compact" param
27854 (rtx_reader::rtx_reader): Add new "compact" param.
27855 (rtx_reader::read_rtx_operand): Make virtual and convert return
27856 type from void to rtx.
27857 (rtx_reader::read_until): New decl.
27858 (rtx_reader::handle_any_trailing_information): New virtual function.
27859 (rtx_reader::postprocess): New virtual function.
27860 (rtx_reader::finalize_string): New virtual function.
27861 (rtx_reader::m_in_call_function_usage): New field.
27862 (rtx_reader::m_reuse_rtx_by_id): New field.
27863 * read-rtl-function.c: New file.
27864 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27865 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27866 (selftest::verify_three_block_rtl_cfg): New decl.
27867 * read-rtl-function.h: New file.
27868 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27869 For host, include function.h, memmodel.h, and emit-rtl.h.
27870 (one_time_initialization): New function.
27871 (struct compact_insn_name): New struct.
27872 (compact_insn_names): New array.
27873 (find_code): Handle insn codes in compact dumps.
27874 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27875 (bind_subst_iter_and_attr): Likewise.
27876 (add_condition_to_string): Likewise.
27877 (add_condition_to_rtx): Likewise.
27878 (apply_attribute_uses): Likewise.
27879 (add_current_iterators): Likewise.
27880 (apply_iterators): Likewise.
27881 (initialize_iterators): Guard usage of apply_subst_iterator with
27882 #ifdef GENERATOR_FILE.
27883 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27884 (md_reader::read_mapping): Likewise.
27885 (add_define_attr_for_define_subst): Likewise.
27886 (add_define_subst_attr): Likewise.
27887 (read_subst_mapping): Likewise.
27888 (check_code_iterator): Likewise.
27889 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27891 (one_time_initialization): New function.
27892 (rtx_reader::read_until): New method.
27893 (read_flags): New function.
27894 (parse_reg_note_name): New function.
27895 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27896 Handle reuse_rtx ids.
27897 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27898 Add parsing support for RTL dumps, mirroring the special-cases in
27899 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27900 values, and calling handle_any_trailing_information.
27901 (rtx_reader::read_rtx_operand): Convert return type from void
27902 to rtx, returning return_rtx. Handle case 'e'. Call
27903 finalize_string on XSTR and XTMPL fields.
27904 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27905 "(nil)" values were omitted. Call the postprocess vfunc on the
27907 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27908 class ctor. Initialize m_in_call_function_usage. Call
27909 one_time_initialization.
27910 * rtl-tests.c (selftest::test_uncond_jump): Call
27911 set_new_first_and_last_insn.
27912 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27913 * selftest-rtl.c: New file.
27914 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27915 (selftest::get_insn_by_uid): New decl.
27916 * selftest-run-tests.c (selftest::run_tests): Call
27917 read_rtl_function_c_tests.
27918 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27919 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27922 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27924 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27925 operands in a special way. Assert that pos+len <= mode precision.
27927 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27929 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27930 3 argument Alias with unlimited for the negative form.
27931 (fno-vect-cost-model): Removed.
27933 2017-01-05 Martin Liska <mliska@suse.cz>
27935 * hsa-gen.c (gen_hsa_divmod): New function.
27936 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27938 2017-01-05 Martin Liska <mliska@suse.cz>
27941 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27944 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27946 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27947 small constant length operands.
27949 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27951 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27952 between loop iterations.
27954 2017-01-05 Martin Liska <mliska@suse.cz>
27957 * gimplify.c (gimplify_decl_expr): Compare to
27958 asan_poisoned_variables instread of checking flags.
27959 (gimplify_target_expr): Likewise.
27960 (gimplify_expr): Likewise.
27961 (gimplify_function_tree): Conditionally initialize
27962 asan_poisoned_variables.
27964 2017-01-04 Jeff Law <law@redhat.com>
27966 PR tree-optimizatin/78812
27967 * rtl.h (contains_mem_rtx_p): Prototype.
27968 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27969 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27970 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27971 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27972 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27974 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27976 * input.c (assert_char_at_range): Default-initialize actual_range.
27978 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27980 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27981 to match the caller.
27983 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27985 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27986 insns after final jump in test to emit dummy move.
27988 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27990 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27991 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27993 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27995 * multiple_target.c (create_dispatcher_calls): Init e_next.
27996 * tree-ssa-loop-split.c (split_loop): Init border.
27997 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28000 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28005 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28006 (altivec_register_operand): Do not return true if the operand
28007 contains a SUBREG mixing SImode and SFmode.
28008 (vsx_register_operand): Likewise.
28009 (vsx_reg_sfsubreg_ok): New predicate.
28010 (vfloat_operand): Do not return true if the operand contains a
28011 SUBREG mixing SImode and SFmode.
28012 (vint_operand): Likewise.
28013 (vlogical_operand): Likewise.
28014 (gpc_reg_operand): Likewise.
28015 (int_reg_operand): Likewise.
28016 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28017 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28018 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28020 (rs6000_emit_move_si_sf_subreg): New helper function.
28021 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28022 fixup SUBREGs involving SImode and SFmode.
28023 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28024 numbers for the new peephole2 optimization.
28025 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28026 the GLIBC math library that do AND/IOR/XOR operations on single
28027 precision floating point.
28028 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28029 target macros to say whether we need to avoid SUBREGs mixing
28031 (TARGET_ALLOW_SF_SUBREG): Likewise.
28032 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28033 (UNSPEC_SI_FROM_SF): Likewise.
28034 (iorxor): Change spacing.
28035 (and_ior_xor): New iterator for AND, IOR, and XOR.
28036 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28037 (movdi_from_sf_zero_ext): Likewise.
28038 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28039 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28040 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28041 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28042 (fms<mode>4): Likewise.
28043 (fnma<mode>4): Likewise.
28044 (fnms<mode>4): Likewise.
28045 (nfma<mode>4): Likewise.
28046 (nfms<mode>4): Likewise.
28048 2017-01-04 Marek Polacek <polacek@redhat.com>
28051 * doc/invoke.texi: Document -Wpointer-compare.
28053 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28055 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28058 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28059 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28062 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28065 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28066 documentation of the powerpc_popcntb_ok attribute.
28067 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28068 code to issue warning messages if a requested CPU configuration is
28069 not supported by the binary (assembler and loader) toolchain.
28070 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28071 made to define a built-in function that has been disabled.
28072 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28073 made to define a built-in function that has been disabled.
28074 (altivec_init_builtins): Add comment explaining why definition
28075 of the DST built-in functions is not preceded by an assertion
28076 check. Add assertions to prevent ICE if attempts are made to
28077 define an altivec predicate or an abs* built-in function that has
28079 (htm_init_builtins): Add comment explaining why definition of the
28080 htm built-in functions is not preceded by an assertion check.
28082 2017-01-04 Jeff Law <law@redhat.com>
28084 PR tree-optimizatin/67955
28085 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28086 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28087 the points-to solution does not include pt_null. Use DECL_PT_UID
28090 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28092 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28093 Use gen_int_mode instead of gen_lopwart for const_int operands.
28095 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28097 PR tree-optimization/71563
28098 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28099 out of range value - has low bits known to be zero.
28101 2017-01-04 Alan Modra <amodra@gmail.com>
28103 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28104 * configure: Regenerate.
28105 * config.in: Regenerate.
28107 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28110 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28111 a substring of the message, but strcmp with the whole message. Ifdef
28112 ENABLE_NLS, translate the message first using dgettext.
28114 2017-01-03 Jeff Law <law@redhat.com>
28116 PR tree-optimizatin/78856
28117 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28118 (mark_threaded_blocks): Remove code to truncate thread paths that
28119 cross multiple loop headers. Instead invalidate the cached loop
28120 iteration information and handle case of a thread path walking
28121 into an irreducible region.
28123 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28126 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28127 assertions. Add support for doing the signbit if the IEEE 128-bit
28128 floating point value is in a GPR.
28129 * config/rs6000/rs6000.md (Fsignbit): Delete.
28130 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28131 Update the length attribute if the value is in a GPR.
28132 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28133 the sign or zero extension instruction, since the value is always 0/1.
28134 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28137 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28138 extracting SImode to a GPR register so that we can generate a
28139 store, limit the vector to be in a traditional Altivec register
28140 for the vextuwrx instruction.
28142 2017-01-03 Ian Lance Taylor <iant@google.com>
28144 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28146 2017-01-03 Martin Sebor <msebor@redhat.com>
28148 PR tree-optimization/78696
28149 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28150 precision. Use MPFR for %f for greater fidelity. Correct handling
28152 (pass_sprintf_length::compute_format_length): Set width and precision
28153 specified by asrerisk to void_node for vararg functions.
28154 (try_substitute_return_value): Adjust dump output.
28156 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28158 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28160 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28162 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28163 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28164 -mlra/-mno-lra was passed to the compiler.
28166 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28168 PR rtl-optimization/65618
28169 * emit-rtl.c (try_split): Move initialization of "before" and
28170 "after" to just before the call to emit_insn_after_setloc.
28172 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28174 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28176 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28178 * dwarf2out.c (gen_enumeration_type_die): When
28179 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28181 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28183 PR tree-optimization/78965
28184 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28185 Change first argument from const call_info & to call_info &. For %n
28186 set info.nowrite to false.
28188 PR middle-end/78901
28189 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28190 possibly throwing calls.
28192 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28193 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28194 and fns handling, rather than in a separate case SSA_NAME.
28196 2017-01-02 Jeff Law <law@redhat.com>
28198 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28199 fixes for first_period and second_period variables.
28201 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28204 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28205 (*insvqi_1): New insn pattern.
28206 (*insvqi_1_mem_rex64): Ditto.
28207 (*insvqi_2): Ditto.
28208 (*insvqi_3): Rename from *insvqi.
28210 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28212 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28214 * doc/cfg.texi (Edges): Remove reference to Java.
28215 (Maintaining the CFG): Ditto.
28217 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28219 PR middle-end/77674
28220 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28221 transparent aliases.
28223 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28225 PR middle-end/77484
28226 * predict.def (PRED_CALL): Update hitrate.
28227 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28228 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28229 into direct/indirect/polymorphic variants.
28231 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28233 Update copyright years.
28235 * gcc.c (process_command): Update copyright notice dates.
28236 * gcov-dump.c (print_version): Ditto.
28237 * gcov.c (print_version): Ditto.
28238 * gcov-tool.c (print_version): Ditto.
28239 * gengtype.c (create_file): Ditto.
28240 * doc/cpp.texi: Bump @copying's copyright year.
28241 * doc/cppinternals.texi: Ditto.
28242 * doc/gcc.texi: Ditto.
28243 * doc/gccint.texi: Ditto.
28244 * doc/gcov.texi: Ditto.
28245 * doc/install.texi: Ditto.
28246 * doc/invoke.texi: Ditto.
28248 Copyright (C) 2017 Free Software Foundation, Inc.
28250 Copying and distribution of this file, with or without modification,
28251 are permitted in any medium without royalty provided the copyright
28252 notice and this notice are preserved.