1 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4 Extensions with more than 16 double VFP registers.
5 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6 to_clear_mask and all code related to it. Replace the remaining
7 entry by a sbitmap and adapt code accordingly.
9 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11 * brig-builtins.def: Change pure attributes to const.
13 2017-09-28 Joseph Myers <joseph@codesourcery.com>
15 * config.gcc (default_gnu_indirect_function): Default to yes for
16 sparc*-*-linux* with glibc.
18 2017-09-28 Joseph Myers <joseph@codesourcery.com>
20 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
21 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
22 when creating .init_array and .fini_array sections with priority
25 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
28 * config/aarch64/aarch64.c
29 (aarch64_builtin_support_vector_misalignment): Always return false
30 when misalignment is unknown.
32 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
35 this function to return false if the definition used by the swap
36 instruction is artificial, or if the memory address from which the
37 constant value is loaded is not represented by a base address held
38 in a register or if the base address register is a frame or stack
39 pointer. Additionally, return false if the base address of the
40 loaded constant is a SYMBOL_REF but is not considered to be a
42 (replace_swapped_load_constant): New function.
43 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
44 loaded constant vector with a load of a swapped constant vector.
46 2017-09-27 Carl Love <cel@us.ibm.com>
48 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
49 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
50 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
53 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
55 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
56 first, always call autopref_rank_data otherwise.
58 2017-09-27 Richard Biener <rguenther@suse.de>
60 * graphite-scop-detection.c (find_scop_parameters): Move
61 loop bound handling ...
62 (gather_bbs::before_dom_children): ... here, avoiding the need
63 to build scop_info->loop_nest.
64 (record_loop_in_sese): Remove.
65 * sese.h (sese_info_t::loop_nest): Remove.
66 * sese.c (new_sese_info): Do not allocate loop_nest.
67 (free_sese_info): Do not free loop_nest.
69 2017-09-27 Jakub Jelinek <jakub@redhat.com>
72 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
73 lhs from calls if the lhs has addressable type.
75 2017-09-27 Richard Biener <rguenther@suse.de>
77 * graphite.h (scop::max_alias_set): New member.
78 * graphite-scop-detection.c: Remove references to non-existing
80 (build_alias_sets): Record the maximum alias set used for drs.
81 (build_scops): Support zero as unlimited for
82 --param graphite-max-arrays-per-scop.
83 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
85 (build_poly_sr_1): ... here. Compute alias set based on the
86 maximum alias set used for drs rather than
87 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
89 2017-09-27 Richard Biener <rguenther@suse.de>
91 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
92 --param loop-block-tile-size=0 to disable tiling.
94 2017-09-27 Richard Biener <rguenther@suse.de>
96 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
97 (graphite-max-nb-scop-params): Document special value zero.
98 * domwalk.h (dom_walker::STOP): New symbolical constant.
99 (dom_walker::dom_walker): Add optional parameter for bb to
101 (dom_walker::~dom_walker): Declare.
102 (dom_walker::before_dom_children): Document STOP return value.
103 (dom_walker::m_user_bb_to_rpo): New member.
104 (dom_walker::m_bb_to_rpo): Likewise.
105 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
106 mapping here if not provided by the user.
107 (dom_walker::~dom_walker): Free bb to RPO mapping if not
108 provided by the user.
109 (dom_walker::STOP): Define.
110 (dom_walker::walk): Do not compute bb to RPO mapping here.
111 Support STOP return value from before_dom_children to stop
113 * graphite-optimize-isl.c (optimize_isl): If the schedule
114 is the same still generate code if -fgraphite-identity
115 or -floop-parallelize-all are given.
116 * graphite-scop-detection.c: Include cfganal.h.
117 (gather_bbs::gather_bbs): Get and pass through bb to RPO
119 (gather_bbs::before_dom_children): Return STOP for BBs
121 (build_scops): Compute bb to RPO mapping and pass it to
122 the domwalk. Treat --param graphite-max-nb-scop-params=0
123 as not limiting the number of params.
124 * graphite.c (graphite_initialize): Remove limit on the
125 number of basic-blocks in a function.
126 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
127 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
130 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
132 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
133 Adjust code to eliminate needing to do the shift right 32-bits
134 operation after XSCVDPSPN.
136 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
138 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
139 ((X / Y) != 0 -> X >= Y): Likewise.
141 2017-09-26 Carl Love <cel@us.ibm.com>
143 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
144 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
145 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
146 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
147 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
148 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
149 definitions and overloading.
150 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
151 statement for P9V_BUILTIN_XST_LEN_R.
152 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
153 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
154 define_expand and define_insn for the instructions and builtins.
155 * doc/extend.texi: Update the built-in documentation file for the new
157 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
158 define_insn for the instructions
160 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
163 * gcc/config/netbsd-protos.h: New file.
164 * gcc/config/netbsd.c: New file.
165 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
166 * gcc/config/t-netbsd: New file.
167 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
168 (tmake_file) Add t-netbsd.
169 (extra_objs) Add netbsd.o.
171 2017-09-26 Janus Weil <janus@gcc.gnu.org>
175 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
177 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
179 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
180 sign extension from a vector register to a GPR by doing a 32-bit
181 direct move and then an EXTSW.
182 (extendsi<mode>2 splitter): Likewise.
183 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
184 right or vector extract after doing XSCVDPSPN. Use
185 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
187 (movdi_from_sf_zero_ext): Likewise.
188 (reload_gpr_from_vsxsf): Likewise.
189 (p8_mfvsrd_4_disf): Delete, no longer used.
190 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
191 then needing to move the SFmode to a GPR to use the XSCVDPSP
192 instruction instead of FRSP and XSCVDPSPN.
193 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
194 it is adjacent to the other XSCVSPDP insns.
195 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
196 SFmode to be in traditional Altivec registers.
197 (vsx_xscvdpspn): Eliminate useless alternative constraint.
198 (vsx_xscvspdpn): Likewise.
199 (vsx_xscvspdpn_scalar): Likewise.
201 2017-09-26 Martin Jambor <mjambor@suse.cz>
203 * tree-sra.c (compare_access_positions): Put integral types first,
204 stabilize sorting of integral types, remove conditions putting
205 non-full-precision integers last.
206 (sort_and_splice_var_accesses): Disable scalarization if a
207 non-integert would be represented by a non-full-precision integer.
209 2017-09-26 Joseph Myers <joseph@codesourcery.com>
211 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
212 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
213 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
214 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
215 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
216 conditionals inside the function instead of around it. Call
217 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
218 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
220 2017-09-26 Richard Biener <rguenther@suse.de>
222 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
224 (scop_detection::build_scop_breadth): ... this. Removed.
225 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
226 (scop_detection::harmful_stmt_in_bb): Likewise.
227 (scop_detection::graphite_can_represent_stmt): Likewise.
228 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
229 (scop_detection::can_represent_loop): Remove recursion, fold in ...
230 (scop_detection::can_represent_loop_1): ... this. Removed.
231 (scop_detection::harmful_loop_in_region): Simplify after inlining
232 the above and remove more quadraticness.
233 (build_scops): Adjust.
234 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
237 2017-09-26 Jakub Jelinek <jakub@redhat.com>
240 * config/i386/i386.c (ix86_print_operand_address_as): Only test
241 REGNO (base) == SP_REG if base is a REG.
244 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
245 if it is different SSA_NAME.
246 (optimize_range_tests_cmp_bitwise): New function.
247 (optimize_range_tests): Call it.
249 2017-09-26 Richard Biener <rguenther@suse.de>
251 PR tree-optimization/82321
252 * graphite.c (canonicalize_loop_closed_ssa): Properly check
253 for the def being inside the loop.
255 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
257 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
259 * config/s390/s390-builtins.def: Fix constraint on op4.
261 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
263 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
264 independent expanders.
265 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
266 ("vec_ordered", "vec_unordered"): New expanders.
268 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
270 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
273 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
275 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
276 vec_unpacks_lo_v16qi.
277 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
279 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
281 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
282 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
283 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
285 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
287 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
289 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
291 ("*vec_slb<mode>"): New insn pattern.
292 ("vec_shr_<mode>"): New expander.
293 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
294 and force the shift count operand to V16QImode.
295 ("vec_srb<mode>"): Set shift count mode to V16QI.
297 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
299 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
300 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
301 ("vec_widen_smult_hi_<mode>"): New expander definitions.
303 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
306 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
308 2017-09-26 Richard Biener <rguenther@suse.de>
310 PR tree-optimization/82320
311 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
314 2017-09-25 Jeff Law <law@redhat.com>
316 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
317 prototype for new argument.
318 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
319 mostly extracted from rs6000_emit_allocate_stack.
320 (rs6000_emit_probe_stack_range_stack_clash): New function.
321 (rs6000_emit_allocate_stack): Call
322 rs6000_emit_probe_stack_range_stack_clash as needed.
323 (rs6000_emit_probe_stack_range): Add additional argument
324 to call to gen_probe_stack_range{si,di}.
325 (output_probe_stack_range): New.
326 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
327 (output_probe_stack_range_stack_clash): New.
328 (rs6000_emit_prologue): Emit notes into dump file as requested.
329 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
330 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
331 Add additional operand and pass it to output_probe_stack_range.
333 2017-09-25 Bin Cheng <bin.cheng@arm.com>
335 PR tree-optimization/82163
336 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
337 (checking_verify_loop_closed_ssa): New parameter.
338 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
339 (check_loop_closed_ssa_stmt): Delete.
340 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
341 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
342 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
345 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
347 * brig-builtins.def: Treat HSAIL barrier builtins as
348 setjmp/longjump style functions.
350 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
352 * target.def (constant_alignment): New hook.
353 * defaults.h (CONSTANT_ALIGNMENT): Delete.
354 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
355 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
356 * doc/tm.texi: Regenerate.
357 * targhooks.h (default_constant_alignment): Declare.
358 (constant_alignment_word_strings): Likewise.
359 * targhooks.c (default_constant_alignment): New function.
360 (constant_alignment_word_strings): Likewise.
361 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
362 instead of CONSTANT_ALIGNMENT.
363 * varasm.c (align_variable, get_variable_align, build_constant_desc)
364 (force_const_mem): Likewise.
365 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
366 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
367 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
368 (TARGET_CONSTANT_ALIGNMENT): Redefine.
369 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
371 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
372 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
373 constant_alignment_word_strings.
374 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
375 (CONSTANT_ALIGNMENT): Likewise.
376 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
377 (arm_constant_alignment): New function.
378 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
379 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
380 constant_alignment_word_strings.
381 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
382 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
383 constant_alignment_word_strings.
384 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
385 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
386 (cris_constant_alignment): New function.
387 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
388 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
389 (epiphany_constant_alignment): New function.
390 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
391 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
392 constant_alignment_word_strings.
393 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
394 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
395 constant_alignment_word_strings.
396 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
397 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
398 constant_alignment_word_strings.
399 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
400 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
401 * config/i386/i386.c (ix86_constant_alignment): Make static.
402 Use the same interface as the target hook.
403 (TARGET_CONSTANT_ALIGNMENT): Redefine.
404 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
405 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
406 constant_alignment_word_strings.
407 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
408 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
409 (TARGET_CONSTANT_ALIGNMENT): Redefine.
410 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
411 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
412 constant_alignment_word_strings.
413 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
414 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
415 constant_alignment_word_strings.
416 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
417 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
418 constant_alignment_word_strings.
419 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
420 * config/microblaze/microblaze.c (microblaze_constant_alignment):
422 (TARGET_CONSTANT_ALIGNMENT): Redefine.
423 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
424 * config/mips/mips.c (mips_constant_alignment): New function.
425 (TARGET_CONSTANT_ALIGNMENT): Redefine.
426 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
427 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
428 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
429 (mmix_constant_alignment): Make static. Use the same interface
431 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
432 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
433 constant_alignment_word_strings.
434 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
435 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
436 constant_alignment_word_strings.
437 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
438 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
439 constant_alignment_word_strings.
440 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
441 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
442 (rs6000_constant_alignment): New function.
443 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
444 * config/riscv/riscv.c (riscv_constant_alignment): New function.
445 (TARGET_CONSTANT_ALIGNMENT): Redefine.
446 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
447 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
448 (rs6000_constant_alignment): New function.
449 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
450 * config/s390/s390.c (s390_constant_alignment): New function.
451 (TARGET_CONSTANT_ALIGNMENT): Redefine.
452 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
453 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
454 constant_alignment_word_strings.
455 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
456 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
457 (sparc_constant_alignment): New function.
458 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
459 * config/spu/spu.c (spu_constant_alignment): New function.
460 (TARGET_CONSTANT_ALIGNMENT): Redefine.
461 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
462 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
463 constant_alignment_word_strings.
464 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
465 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
466 constant_alignment_word_strings.
467 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
468 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
469 constant_alignment_word_strings.
470 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
471 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
472 (visium_constant_alignment): New function.
473 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
474 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
475 (xtensa_constant_alignment): New function.
476 * system.h (CONSTANT_ALIGNMENT): Poison.
478 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
481 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
482 (rs6000_builtin_valid_without_lhs): New helper function.
483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
484 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
486 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
488 * target.h (vec_perm_indices): Use unsigned short rather than
490 (auto_vec_perm_indices): Likewise.
491 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
492 Use unsigned int rather than unsigned char.
493 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
495 2017-09-25 Richard Biener <rguenther@suse.de>
497 * cfgloop.h (sort_sibling_loops): Declare.
498 * cfgloop.c (sort_sibling_loops_cmp): New helper.
499 (sort_sibling_loops): New function sorting the sibling loop list
501 * graphite.c (graphite_transform_loops): Sort sibling loops.
503 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
505 * target.def (vec_perm_const_ok): Change sel parameter to
507 * optabs-query.c (can_vec_perm_p): Update accordingly.
508 * doc/tm.texi: Regenerate.
509 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
510 auto_vec_perm_indices and remove separate nelt field.
511 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
512 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
513 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
514 (aarch64_expand_vec_perm_const): Update accordingly.
515 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
517 * config/arm/arm.c (expand_vec_perm_d): Change perm to
518 auto_vec_perm_indices and remove separate nelt field.
519 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
520 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
521 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
523 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
525 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
526 sel to vec_perm_indices.
527 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
528 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
529 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
531 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
534 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
537 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
538 on the FUNCTION_DECL function context if it has a DIE that is a
541 2017-09-25 Richard Biener <rguenther@suse.de>
543 PR tree-optimization/82285
544 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
547 2017-09-25 Tom de Vries <tom@codesourcery.com>
551 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
554 2017-09-25 Richard Biener <rguenther@suse.de>
556 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
557 ISL errors other than isl_error_quota happen. Dump if the
558 schedule is the same.
559 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
560 errors instead of aborting inside ISL.
562 2017-09-25 Iain Sandoe <iain@codesourcery.com>
565 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
566 of libgcc_eh for m64.
567 * config/i386/darwin64.h: Likewise.
569 2017-09-25 Richard Biener <rguenther@suse.de>
572 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
573 attribute for incomplete types nor twice for complete ones.
575 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
578 * config/i386/i386.c (ix86_print_operand_address_as): Encode
579 %esp as %rsp to avoid 0x67 prefix if there is no index or base
582 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
585 * config/i386/i386.opt (mprefer-avx256): Use
586 ix86_target_flags variable.
587 * config/i386/i386.c (ix86_target_string): Move
588 -mprefer-avx256 to flag2_opts.
590 2017-09-22 Jakub Jelinek <jakub@redhat.com>
593 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
594 and x != -1 | y != -1 into (x & y) != -1.
596 2017-09-22 Steve Ellcey <sellcey@cavium.com>
598 * config.gcc: Add new case statement to set
599 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
600 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
601 s390x-*-linux* case statements. Added aarch64 to the list of
602 supported architectures.
604 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
606 PR tree-optimization/82289
607 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
608 STMT_VINFO_RELEVANT_P.
610 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
611 Alan Hayward <alan.hayward@arm.com>
612 David Sherwood <david.sherwood@arm.com>
614 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
615 for VR_RANGE only; don't allow VR_ANTI_RANGE.
616 (extract_range_from_binary_expr_1): Don't call
617 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
619 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
620 Alan Hayward <alan.hayward@arm.com>
621 David Sherwood <david.sherwood@arm.com>
623 * target.def (preferred_vector_alignment): New hook.
624 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
626 * doc/tm.texi: Regenerate.
627 * targhooks.h (default_preferred_vector_alignment): Declare.
628 * targhooks.c (default_preferred_vector_alignment): New function.
629 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
631 (DR_TARGET_ALIGNMENT): New macro.
632 (aligned_access_p): Update commentary.
633 (vect_known_alignment_in_bytes): New function.
634 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
636 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
637 Calculate the misalignment based on the target alignment rather than
639 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
640 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
641 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
642 the target alignment, rather than masking the element misalignment
643 with the number of elements in a vector. Also use the target
644 alignment when calculating the maximum number of peels.
645 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
646 instead of TYPE_ALIGN_UNIT.
647 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
648 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
649 (vect_create_addr_base_for_vector_ref): Update call accordingly.
650 (vect_create_data_ref_ptr): Likewise.
651 (vect_setup_realignment): Realign by ANDing with
652 -DR_TARGET_MISALIGNMENT.
653 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
654 the number of peels based on DR_TARGET_ALIGNMENT.
655 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
656 with the guaranteed alignment boundary when deciding whether
658 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
659 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
660 (ensure_base_align): Remove stmt_info parameter. Get the
661 target base alignment from DR_TARGET_ALIGNMENT.
662 (vectorizable_store): Update call accordingly. Interpret
663 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
665 (vectorizable_load): Likewise.
667 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
668 Alan Hayward <alan.hayward@arm.com>
669 David Sherwood <david.sherwood@arm.com>
671 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
672 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
673 (vect_enhance_data_refs_alignment): Likewise.
675 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
677 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
678 error. Only quit immediately if parsing is complete.
679 (BEGIN): Initialize fatal_err and parse_done.
680 (begin fpu, end fpu): Check number of arguments.
681 (begin arch, end arch): Likewise.
682 (begin cpu, end cpu): Likewise.
683 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
684 (optalias): Likewise.
686 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
688 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
689 * config/arm/arm-isa.h: Delete. Move definitions to ...
690 * arm-cpus.in: ... here. Use new feature and fgroup values.
691 * config/arm/arm.c (arm_option_override): Use lower case for feature
693 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
694 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
695 * config/arm/parsecpu.awk (END): Add new command 'isa'.
697 (print_isa_bits_for): New function.
698 (gen_isa): New function.
699 (gen_comm_data): Use print_isa_bits_for.
700 (define feature): New keyword.
701 (define fgroup): New keyword.
702 * config/arm/t-arm (TM_H): Remove.
703 (GTM_H): Add arm-isa.h.
704 (arm-isa.h): Add rule to generate file.
705 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
706 case for feature bit names.
708 2017-09-22 Richard Biener <rguenther@suse.de>
710 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
712 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
713 print of no dependency loops ...
714 * graphite.c (graphite_transform_loops): ... here.
715 (canonicalize_loop_closed_ssa_form): Work from inner to outer
717 (same_close_phi_node, remove_duplicate_close_phi,
718 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
719 (canonicalize_loop_closed_ssa): ... here and simplify.
720 * graphite-optimize-isl.c: Include tree-vectorizer.h.
721 (optimize_isl): Use dump_printf_loc to tell when we stopped
722 optimizing because of an ISL timeout.
724 2017-09-22 Richard Biener <rguenther@suse.de>
726 PR tree-optimization/82291
727 * tree-if-conv.c (predicate_mem_writes): Make sure to
728 remove writes in blocks predicated with false.
730 2017-09-22 Richard Biener <rguenther@suse.de>
732 * sese.c: Include cfganal.h.
733 (if_region_set_false_region): Remove.
734 (create_if_region_on_edge): Likewise.
735 (move_sese_in_condition): Re-implement without destroying
738 2017-09-22 Richard Biener <rguenther@suse.de>
740 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
741 Verify both BBs contain loop PHI nodes before dispatching to
743 (graphite_regenerate_ast_isl): Do not recompute dominators,
744 do not verify three times. Restructure for clarity.
745 * graphite-scop-detection.c (same_close_phi_node,
746 remove_duplicate_close_phi, make_close_phi_nodes_unique,
747 defined_in_loop_p, canonicalize_loop_closed_ssa,
748 canonicalize_loop_closed_ssa_form): Simplify, remove excess
749 checking and SSA rewrite, move to ...
750 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
751 (graphite_initialize): Do not pass in ctx, do not reset the
752 SCEV cache, compute only dominators.
753 (graphite_transform_loops): Allocate ISL ctx after
754 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
755 Maintain post-dominators only around build_scops.
756 * sese.c (if_region_set_false_region): Make static. Free
757 and recompute dominators.
758 (move_sese_in_condition): Assert we don't get called with
759 post-dominators computed.
760 * sese.h (if_region_set_false_region): Remove.
762 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
764 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
765 mode attribute for TARGET_AVX512VL.
767 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
769 * config/i386/i386.opt (mprefer-avx256): New option.
770 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
772 (ix86_preferred_simd_mode): Return 256-bit AVX modes
773 for TARGET_PREFER_AVX256.
774 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
776 2017-09-21 Jeff Law <law@redhat.com>
778 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
779 Fix dump output if the only stack space is for pushed registers.
781 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
783 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
786 2017-09-21 Martin Sebor <msebor@redhat.com>
789 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
790 code (in C++) or code that triggers warnings.
792 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
794 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
796 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
798 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
799 * sched-rgn.c: Ditto.
800 * sel-sched-ir.c: Ditto.
802 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
804 * toplev.h (set_random_seed): Adjust return type.
805 * toplev.c (init_local_tick): Move eager initialization of random_seed
806 to get_random_seed. Adjust comment.
807 (init_random_seed): Inline to get_random_seed, delete.
808 (get_random_seed): Initialize random_seed lazily.
809 (set_random_seed): Do not return previous value.
810 (print_switch_value): Do not call get_random_seed.
812 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
814 * cgraph.c (delete_function_version): New, broken out from...
815 (cgraph_node::delete_function_version): ...here. Rename to
816 cgraph_node::delete_function_version_by_decl. Update all uses.
817 (cgraph_node::remove): Call delete_function_version.
819 2017-09-21 Jakub Jelinek <jakub@redhat.com>
822 * tree-inline.c (expand_call_inline): Emit clobber stmts for
823 VAR_DECLs to which addressable non-volatile parameters are mapped
824 and for id->retvar after the return value assignment. Clear
825 id->retval and id->retbnd after inlining.
827 2017-09-21 Richard Biener <rguenther@suse.de>
829 PR tree-optimization/82276
830 PR tree-optimization/82244
831 * tree-vrp.c (build_assert_expr_for): Set
832 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
834 (remove_range_assertions): Revert earlier change.
836 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
839 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
841 2017-09-21 Richard Biener <rguenther@suse.de>
843 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
844 Restore valid IL after code generation errors.
845 * graphite.c (graphite_transform_loops): Diagnose code
846 generation issues as MSG_MISSED_OPTIMIZATION and continue
847 with processing SCOPs.
849 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
850 Alan Hayward <alan.hayward@arm.com>
851 David Sherwood <david.sherwood@arm.com>
853 * calls.c (compute_argument_addresses): Use simplify_gen_binary
854 rather than choosing between plus_constant and gen_rtx_<CODE>.
855 * expr.c (emit_push_insn): Likewise.
856 (expand_expr_real_2): Likewise.
858 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
859 Alan Hayward <alan.hayward@arm.com>
860 David Sherwood <david.sherwood@arm.com>
862 * loop-unroll.c (split_iv): Call copy_rtx on the step.
864 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
865 Alan Hayward <alan.hayward@arm.com>
866 David Sherwood <david.sherwood@arm.com>
868 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
869 calling tree_to_uhwi.
871 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
872 Alan Hayward <alan.hayward@arm.com>
873 David Sherwood <david.sherwood@arm.com>
875 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
876 INTEGER_CST rather than a negative test for ADDR_EXPR.
878 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
879 Alan Hayward <alan.hayward@arm.com>
880 David Sherwood <david.sherwood@arm.com>
882 * tree-vrp.c (extract_range_from_binary_expr_1): Check
883 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
885 2017-09-21 Richard Biener <rguenther@suse.de>
887 PR tree-optimization/71351
888 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
889 graphite_create_new_loop_guard): Remove, fold remaining parts
891 (translate_isl_ast_node_for): ... here and simplify.
893 2017-09-21 Jakub Jelinek <jakub@redhat.com>
896 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
897 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
898 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
899 alternative always use QI mode, for -Os imov (=R,R) alternative
900 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
903 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
904 Jeff Law <law@redhat.com>
906 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
907 (allocate_stack_space): New function, partially extracted from
909 (s390_emit_prologue): Track offset to most recent stack probe.
910 Code to allocate space moved into allocate_stack_space.
911 Dump actions when no stack is allocated.
912 (s390_prologue_plus_offset): New function.
913 (s390_emit_stack_probe): Likewise.
915 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
917 * common.opt (Wa, Wl, Wp, g, gz=): Add
919 (gno-column-info): Remove.
920 (gcolumn-info): Drop RejectNegative.
922 (gno-record-gcc-switches): Remove.
923 (grecord-gcc-switches): Drop RejectNegative.
924 (gno-split-dwarf): Remove.
925 (gsplit-dwarf): Drop RejectNegative.
926 (gno-strict-dwarf): Remove.
927 (gstrict-dwarf): Drop RejectNegative.
928 * config/darwin.opt (gfull, gused): Add RejectNegative.
929 * dwarf2out.c (gen_producer_string): Drop
930 gno-record-gcc-switches handler.
931 * optc-gen.awk: Add g to prefixes with negative forms.
932 * opts-common.c (remapping_prefix_p): New.
933 (find_opt): Check it.
934 (generate_canonical_option): Test g prefix.
935 (option_map): Add -gno- mapping.
936 (add_misspelling_candidates): Check remapping_prefix_p.
938 2017-09-20 Jeff Law <law@redhat.com>
940 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
941 thinko in stack clash protection support.
943 * explow.c (compute_stack_clash_protection_loop_data): Use
944 CONST_INT_P instead of explicit test. Verify object is a
945 CONST_INT_P before looking at INTVAL.
946 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
947 instead of explicit test.
949 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
952 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
953 address instead of to r1 and r11.
955 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
957 * config.gcc: Support "knm".
958 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
959 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
961 * config/i386/i386.c (m_KNM): Define.
962 (processor_target_table): Add "knm".
964 (ix86_option_override_internal): Add "knm".
965 (ix86_issue_rate): Add PROCESSOR_KNM.
966 (ix86_adjust_cost): Ditto.
967 (ia32_multipass_dfa_lookahead): Ditto.
968 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
969 (fold_builtin_cpu): Add M_INTEL_KNM.
970 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
971 (processor_type): Add PROCESSOR_KNM.
972 * config/i386/x86-tune.def: Add m_KNM.
973 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
975 2017-09-20 Richard Biener <rguenther@suse.de>
977 PR tree-optimization/80213
978 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
979 are allowed in empty BBs as well.
980 (canonicalize_loop_closed_ssa): Also look for other complex
982 (scop_detection::get_sese): Include the loop-closed PHI block
984 (scop_detection::merge_sese): Remove code adding extra blocks.
985 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
986 (build_scops): Assert the final returned scop is invalid.
988 2017-09-20 Richard Biener <rguenther@suse.de>
990 PR tree-optimization/82264
991 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
993 (vn_phi_lookup): Likewise.
994 (vn_phi_insert): Likewise.
996 2017-09-20 Jakub Jelinek <jakub@redhat.com>
998 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
999 that fits into uhwi or shwi, add DW_AT_const_value regardless
1000 of early_dwarf without going through RTL, using add_AT_unsigned
1003 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1004 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1005 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1006 (DEBUG_MACRO_SECTION): Likewise.
1007 (DEBUG_LTO_MACRO_SECTION): Likewise.
1008 (DEBUG_STR_DWO_SECTION): Likewise.
1009 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1010 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1011 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1012 (DEBUG_LTO_LINE_STR_SECTION): Define.
1013 (init_sections_and_labels): Initialize debug_line_str_section
1014 variable. Initialize debug_loc_section for -gdwarf-5 to
1015 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1017 2017-09-20 Richard Biener <rguenther@suse.de>
1019 * graphite-sese-to-poly.c (extract_affine): Properly handle
1020 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1022 2017-09-20 Richard Biener <rguenther@suse.de>
1024 PR tree-optimization/81373
1025 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1026 Force SESE live-out defs to be handled even if they are
1029 2017-09-19 Jeff Law <law@redhat.com>
1031 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1032 nothing for stack adjustments with REG_STACK_CHECK.
1033 * sched-deps.c (parse_add_or_inc): Reject insns with
1034 REG_STACK_CHECK from dependency breaking.
1035 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1036 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1037 * reg-notes.def (STACK_CHECK): New note.
1039 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1040 (ix86_expand_prologue): Dump stack clash info as needed.
1041 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1043 * function.c (dump_stack_clash_frame_info): New function.
1044 * function.h (dump_stack_clash_frame_info): Prototype.
1045 (enum stack_clash_probes): New enum.
1047 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1048 flag_stack_clash_protection.
1049 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1050 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1051 (arm_frame_pointer_required): Likewise.
1052 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1053 (ia64_expand_prologue): Likewise.
1054 * config/mips/mips.c (mips_expand_prologue): Likewise.
1055 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1056 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1057 (sparc_flat_expand_prologue): Likewise.
1058 * config/spu/spu.c (spu_expand_prologue): Likewise.
1060 * explow.c: Include "params.h".
1061 (anti_adjust_stack_and_probe_stack_clash): New function.
1062 (get_stack_check_protect): Likewise.
1063 (compute_stack_clash_protection_loop_data): Likewise.
1064 (emit_stack_clash_protection_loop_start): Likewise.
1065 (emit_stack_clash_protection_loop_end): Likewise.
1066 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1067 Use anti_adjust_stack_and_probe_stack_clash.
1068 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1069 (emit_stack_clash_protection_loop_start): Likewise.
1070 (emit_stack_clash_protection_loop_end): Likewise.
1071 * rtl.h (get_stack_check_protect): Prototype.
1072 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1073 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1074 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1076 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1078 * doc/tm.texi: Rebuilt.
1079 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1080 get_stack_check_protect.
1081 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1082 * config/arm/arm.c (arm_expand_prologue): Likewise.
1083 (arm_frame_pointer_required): Likewise.
1084 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1085 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1086 * config/mips/mips.c (mips_expand_prologue): Likewise.
1087 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1088 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1089 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1090 (sparc_flat_expand_prologue): Likewise.
1092 * common.opt (-fstack-clash-protection): New option.
1093 * flag-types.h (enum stack_check_type): Note difference between
1094 -fstack-check= and -fstack-clash-protection.
1095 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1096 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1097 * toplev.c (process_options): Issue warnings/errors for cases
1098 not handled with -fstack-clash-protection.
1099 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1100 (-fstack-check): Note additional problem with -fstack-check=generic.
1101 Note that -fstack-check is primarily for Ada and refer users
1102 to -fstack-clash-protection for stack-clash-protection.
1103 Document new params for stack clash protection.
1105 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1107 * config/i386/i386.c (ix86_split_long_move): Do not handle
1108 address used for LEA in a special way.
1110 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1112 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1114 2017-09-19 Martin Sebor <msebor@redhat.com>
1117 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1118 of incompatible types.
1120 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1122 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1123 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1124 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1125 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1127 2017-09-19 Richard Biener <rguenther@suse.de>
1129 PR tree-optimization/82244
1130 * tree-vrp.c (remove_range_assertions): Do not propagate
1131 a constant to abnormals but replace the assert with a copy.
1133 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1135 PR rtl-optimization/57878
1136 PR rtl-optimization/68988
1137 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1138 avoidance test involving non_reload_pseudos. Move frequency test
1139 below the general fragmentation avoidance test.
1141 2017-09-19 Richard Biener <rguenther@suse.de>
1143 PR tree-optimization/69728
1144 * graphite-sese-to-poly.c (schedule_error): New global.
1145 (add_loop_schedule): Handle empty domain by failing the
1147 (build_original_schedule): Handle schedule_error.
1149 2017-09-19 Richard Biener <rguenther@suse.de>
1151 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1152 Do not iterate to sibling loops but only to siblings of inner
1155 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1158 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1159 that the registers are different.
1161 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1163 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1164 to processor_model and "amdfam17h" to arch_names_table.
1165 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1167 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1170 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1172 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1173 Alan Hayward <alan.hayward@arm.com>
1174 David Sherwood <david.sherwood@arm.com>
1176 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1178 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1180 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1181 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1182 vect_schedule_slp_instance.
1183 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1184 Update call to vect_slp_analyze_node_operations. Simplify return
1186 (vect_slp_analyze_bb_1): Update call accordingly.
1187 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1188 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1189 (vect_schedule_slp): Update call accordingly.
1191 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1192 Alan Hayward <alan.hayward@arm.com>
1193 David Sherwood <david.sherwood@arm.com>
1195 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1196 with types that aren't in fact scalar.
1198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1200 * tree-vect-slp.c (vect_record_max_nunits): New function,
1202 (vect_build_slp_tree_1): ...here.
1203 (vect_build_slp_tree_2): Call it for phis too.
1205 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1207 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1208 to vect_get_vec_def_for_operand when getting the mask operand.
1210 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1211 Alan Hayward <alan.hayward@arm.com>
1212 David Sherwood <david.sherwood@arm.com>
1214 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1217 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1222 calculation for vector booleans.
1224 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1225 Alan Hayward <alan.hayward@arm.com>
1226 David Sherwood <david.sherwood@arm.com>
1228 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1230 (vect_transform_stmt): ...here.
1231 (vect_analyze_stmt): Use it instead of calling
1232 vectorizable_live_operation directly.
1234 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1236 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1237 non-SIMT targets in acc vector loops.
1239 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1241 * configure.ac: Add arc and check if assembler supports gdwarf2.
1242 * configure: Regenerate.
1244 2017-09-18 Richard Biener <rguenther@suse.de>
1246 PR tree-optimization/82220
1247 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1248 epilogue niters from the min_profitable_iters compute.
1250 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1253 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1254 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1255 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1257 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1260 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1261 switching to a new text section.
1263 2017-09-18 Richard Biener <rguenther@suse.de>
1265 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1267 (build_alias_set): Reject aliases with no access function.
1269 2017-09-18 Richard Biener <rguenther@suse.de>
1271 PR tree-optimization/79622
1272 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1274 (build_cross_bb_scalars_use): Likewise.
1276 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1278 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1280 2017-09-18 Alan Modra <amodra@gmail.com>
1283 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1284 stack_pointer_rtx for count 0. Update comments. Break up
1285 large rtl expression.
1287 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1289 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1290 Increase to 20 bytes.
1291 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1292 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1293 or avx version of the stub.
1295 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1298 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1299 compute the minimum stack alignment. Also update preferred stack
1300 boundary for leaf functions.
1302 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1304 PR tree-optimization/82228
1305 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1308 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1310 * common/config/nds32/nds32-common.c
1311 (nds32_option_optimization_table): Refine formatting.
1312 (nds32_option_optimization_table): Use -fsched-pressure and
1313 -fomit-frame-pointer for specific optimization level.
1315 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1317 * config/nds32/nds32.c: Refine formatting and comments.
1318 * config/nds32/nds32.h: Likewise.
1319 * config/nds32/nds32.md: Likewise.
1320 * config/nds32/nds32-cost.c: Likewise.
1321 * config/nds32/nds32-isr.c: Likewise.
1322 * config/nds32/nds32-md-auxiliary.c: Likewise.
1323 * config/nds32/nds32-multiple.md: Likewise.
1324 * config/nds32/nds32-predicates.c: Likewise.
1326 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1327 Jakub Jelinek <jakub@redhat.com>
1329 Add support for -std=c++2a.
1330 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1332 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1334 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1337 * doc/extend.texi (Common Function Attributes): Add
1338 references to ARM, AArch64, and S/390 specific attributes.
1339 (Function Specific Option Pragmas): Add AArch64 and S/390
1340 to list of back ends that support the target pragma.
1342 2017-09-15 Nathan Sidwell <nathan@acm.org>
1344 * doc/standards.texi: Fix C++17 description. Update URLs for
1347 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1349 * common.opt (Wcast-align=strict): New warning option.
1350 * doc/invoke.texi: Document -Wcast-align=strict.
1352 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1354 * cgraph.h (cgraph_thunk_info): Add comments.
1355 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1356 assert for VIRTUAL_* arguments stricter.
1358 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1360 PR tree-optimization/71026
1361 * match.pd: Move RDIV patterns from fold-const.c
1362 * fold-const.c (distribute_real_division): Removed.
1363 (fold_binary_loc): Remove calls to distribute_real_divison.
1365 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1367 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1368 c++1z and gnu++1z as deprecated. Change other references to
1369 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1370 Change -Wc++1z-compat to -Wc++17-compat.
1371 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1372 * dwarf2out.c (highest_c_language): Handle C++17.
1373 (gen_compile_unit_die): Likewise.
1375 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1377 PR rtl-optimization/82192
1378 * combine.c (make_extraction): Don't look through non-paradoxical
1379 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1382 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1383 Alan Hayward <alan.hayward@arm.com>
1384 David Sherwood <david.sherwood@arm.com>
1386 * target.def (function_arg_offset): New hook.
1387 * targhooks.h (default_function_arg_offset): Declare.
1388 * targhooks.c (default_function_arg_offset): New function.
1389 * function.c (locate_and_pad_parm): Use
1390 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1391 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1392 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1393 * doc/tm.texi: Regenerate.
1394 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1395 * config/spu/spu.c (spu_function_arg_offset): New function.
1396 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1397 * system.h (FUNCTION_ARG_OFFSET): Poison.
1399 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1400 Alan Hayard <alan.hayward@arm.com>
1401 David Sherwood <david.sherwood@arm.com>
1403 * target.def (truly_noop_truncation): New hook.
1404 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1405 than TRULY_NOOP_TRUNCATION.
1406 * hooks.h (hook_bool_uint_uint_true): Declare.
1407 * hooks.c (hook_bool_uint_uint_true): New function.
1408 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1409 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1410 * doc/tm.texi: Regenerate.
1411 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1412 rather than TRULY_NOOP_TRUNCATION in comments.
1413 (simplify_comparison): Likewise.
1414 (record_truncated_value): Likewise.
1415 * expmed.c (extract_bit_field_1): Likewise.
1416 (extract_split_bit_field): Likewise.
1417 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1418 instead of TRULY_NOOP_TRUNCATION.
1419 * function.c (assign_parm_setup_block): Likewise.
1420 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1421 * rtlhooks.c: Include target.h.
1422 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1423 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1424 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1425 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1426 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1427 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1428 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1429 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1430 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1431 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1432 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1433 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1434 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1435 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1436 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1437 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1438 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1439 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1440 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1441 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1442 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1443 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1444 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1445 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1446 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1447 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1448 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1449 rather than TRULY_NOOP_TRUNCATION in comments.
1450 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1451 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1452 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1453 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1454 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1455 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1456 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1457 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1458 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1459 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1460 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1461 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1462 rather than TRULY_NOOP_TRUNCATION in comments.
1463 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1464 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1465 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1466 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1467 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1468 TRULY_NOOP_TRUNCATION condition.
1469 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1470 (TRULY_NOOP_TRUNCATION): Delete.
1471 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1472 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1473 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1474 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1475 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1476 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1477 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1478 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1479 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1480 rather than TRULY_NOOP_TRUNCATION in comments.
1481 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1482 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1483 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1484 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1485 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1486 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1488 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1491 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1492 (*cmp_ior): Likewise.
1493 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1494 (*ior_scc_scc_cmp): Likewise.
1495 (*and_scc_scc): Likewise.
1496 (*and_scc_scc_cmp): Likewise.
1498 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1499 Alan Hayard <alan.hayward@arm.com>
1500 David Sherwood <david.sherwood@arm.com>
1502 * target.def (can_change_mode_class): New hook.
1503 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1504 (hard_regno_nregs): Likewise.
1505 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1506 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1507 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1508 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1509 (LOAD_EXTEND_OP): Update accordingly.
1510 * doc/tm.texi: Regenerate.
1511 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1512 CANNOT_CHANGE_MODE_CLASS.
1513 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1514 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1515 * combine.c (simplify_set): Update accordingly.
1516 * emit-rtl.c (validate_subreg): Likewise.
1517 * recog.c (general_operand): Likewise.
1518 * regcprop.c (mode_change_ok): Likewise.
1519 * reload1.c (choose_reload_regs): Likewise.
1520 (inherit_piecemeal_p): Likewise.
1521 * rtlanal.c (simplify_subreg_regno): Likewise.
1522 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1523 instead of CANNOT_CHANGE_MODE_CLASS.
1524 (reload_cse_simplify_operands): Likewise.
1525 * reload.c (push_reload): Use targetm.can_change_mode_class
1526 instead of CANNOT_CHANGE_MODE_CLASS.
1527 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1528 REG_CANNOT_CHANGE_MODE_P.
1529 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1530 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1531 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1532 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1533 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1534 (arm_can_change_mode_class): New function.
1535 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1536 than CANNOT_CHANGE_MODE_CLASS in comments.
1537 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1538 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1539 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1540 (ix86_can_change_mode_class): ...this new function, inverting the
1541 sense of the return value.
1542 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1543 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1544 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1545 (ia64_can_change_mode_class): New function.
1546 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1547 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1548 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1549 (m32c_can_change_mode_class): ...this new function, inverting the
1550 sense of the return value.
1551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1552 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1553 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1554 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1555 (mips_can_change_mode_class): ...this new function, inverting the
1556 sense of the return value.
1557 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1558 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1559 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1560 (msp430_can_change_mode_class): New function.
1561 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1562 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1563 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1564 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1565 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1566 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1567 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1568 (pa_cannot_change_mode_class): Replace with...
1569 (pa_can_change_mode_class): ...this new function, inverting the
1570 sense of the return value.
1571 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1572 than CANNOT_CHANGE_MODE_CLASS in comments.
1573 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1574 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1575 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1576 (pdp11_cannot_change_mode_class): Replace with...
1577 (pdp11_can_change_mode_class): ...this new function, inverting the
1578 sense of the return value.
1579 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1580 * config/powerpcspe/powerpcspe-protos.h
1581 (rs6000_cannot_change_mode_class_ptr): Delete.
1582 * config/powerpcspe/powerpcspe.c
1583 (rs6000_cannot_change_mode_class_ptr): Delete.
1584 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1585 (rs6000_option_override_internal): Assign to
1586 targetm.can_change_mode_class instead of
1587 rs6000_cannot_change_mode_class_ptr.
1588 (rs6000_cannot_change_mode_class): Replace with...
1589 (rs6000_can_change_mode_class): ...this new function, inverting the
1590 sense of the return value.
1591 (rs6000_debug_cannot_change_mode_class): Replace with...
1592 (rs6000_debug_can_change_mode_class): ...this new function.
1593 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1594 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1595 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1596 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1597 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1599 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1600 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1601 (rs6000_option_override_internal): Assign to
1602 targetm.can_change_mode_class instead of
1603 rs6000_cannot_change_mode_class_ptr.
1604 (rs6000_cannot_change_mode_class): Replace with...
1605 (rs6000_can_change_mode_class): ...this new function, inverting the
1606 sense of the return value.
1607 (rs6000_debug_cannot_change_mode_class): Replace with...
1608 (rs6000_debug_can_change_mode_class): ...this new function.
1609 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1610 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1611 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1612 (s390_can_change_mode_class): ...this new function, inverting the
1613 sense of the return value.
1614 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1615 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1616 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1617 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1618 (sh_cannot_change_mode_class): Replace with...
1619 (sh_can_change_mode_class): ...this new function, inverting the
1620 sense of the return value.
1621 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1622 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1623 (sparc_can_change_mode_class): New function.
1624 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1625 * config/spu/spu.c (spu_can_change_mode_class): New function.
1626 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1627 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1628 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1629 (visium_can_change_mode_class): New function.
1630 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1632 2017-09-15 Richard Biener <rguenther@suse.de>
1634 PR tree-optimization/82217
1635 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1636 but not undefined case.
1638 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1641 * postreload.c (reload_cse_simplify_operands): Skip
1642 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1644 2017-09-15 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/68823
1647 * graphite-scop-detection.c (build_alias_set): If we have a
1648 possible dependence check whether we can handle them by just
1649 looking at the DRs DR_ACCESS_FNs.
1650 (build_scops): If build_alias_set fails, fail the SCOP.
1652 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1654 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1655 to support float128 built-in functions that require the ISA 3.0
1657 (BU_FLOAT128_3_HW): Likewise.
1658 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1660 (FMAF128): Likewise.
1662 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1663 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1664 floating point instructions.
1665 (rs6000_invalid_builtin): Likewise.
1666 (rs6000_builtin_mask_names): Likewise.
1667 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1668 (RS6000_BTM_FLOAT128_HW): Likewise.
1669 (RS6000_BTM_COMMON): Likewise.
1670 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1672 * doc/extend.texi (RS/6000 built-in functions): Document the
1673 IEEE 128-bit floating point square root and fused multiply-add
1676 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1678 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1679 reg (r2) isn't in the set of registers defined in the prologue.
1681 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1682 Alan Hayward <alan.hayward@arm.com>
1683 David Sherwood <david.sherwood@arm.com>
1685 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1686 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1687 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1688 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1689 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1691 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1692 max_vectorization_factor.
1693 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1695 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1700 (vect_worthwhile_without_simd_p): Declare.
1701 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1702 (vectorizable_reduction): Use it.
1703 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1704 (vectorizable_operation): Likewise.
1706 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1707 Alan Hayward <alan.hayward@arm.com>
1708 David Sherwood <david.sherwood@arm.com>
1710 * tree-vectorizer.h (vect_get_num_copies): New function.
1711 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1712 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1713 (vectorizable_induction): Likewise.
1714 (vectorizable_live_operation): Likewise.
1715 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1716 (vectorizable_bswap): Likewise.
1717 (vectorizable_call): Likewise.
1718 (vectorizable_conversion): Likewise.
1719 (vectorizable_assignment): Likewise.
1720 (vectorizable_shift): Likewise.
1721 (vectorizable_operation): Likewise.
1722 (vectorizable_store): Likewise.
1723 (vectorizable_load): Likewise.
1724 (vectorizable_condition): Likewise.
1725 (vectorizable_comparison): Likewise.
1726 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1728 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1729 Alan Hayward <alan.hayward@arm.com>
1730 David Sherwood <david.sherwood@arm.com>
1732 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1733 of vect_init_vector.
1735 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1736 Alan Hayward <alan.hayward@arm.com>
1737 David Sherwood <david.sherwood@arm.com>
1739 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1740 an inline wrapper that provides a location.
1741 (gimple_build_vector): Likewise.
1742 * gimple-fold.c (gimple_build_vector_from_val): New function.
1743 (gimple_build_vector): Likewise.
1744 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1745 functions to build the initial value. Always return a gimple value.
1746 (get_initial_defs_for_reduction): Likewise. Only compute
1748 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1749 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1750 (vectorizable_induction): Use gimple_build_vector rather than
1753 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1754 Alan Hayward <alan.hayward@arm.com>
1755 David Sherwood <david.sherwood@arm.com>
1757 * target.h (vec_perm_indices): New typedef.
1758 (auto_vec_perm_indices): Likewise.
1759 * optabs-query.h: Include target.h
1760 (can_vec_perm_p): Take a vec_perm_indices *.
1761 * optabs-query.c (can_vec_perm_p): Likewise.
1762 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1763 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1764 * tree-vect-generic.c (lower_vec_perm): Likewise.
1765 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1766 (vect_grouped_load_supported): Likewise.
1767 (vect_shift_permute_load_chain): Likewise.
1768 (vect_permute_store_chain): Use auto_vec_perm_indices.
1769 (vect_permute_load_chain): Likewise.
1770 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
1771 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
1772 Update uses of can_vec_perm_p.
1773 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
1774 mode with a number of elements. Take a vec_perm_indices *.
1775 (vect_create_epilog_for_reduction): Update accordingly.
1776 Use auto_vec_perm_indices.
1777 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
1778 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
1779 (vect_transform_slp_perm_load): Likewise.
1780 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
1781 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
1782 (vect_gen_perm_mask_checked): Likewise.
1783 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
1784 (vect_gen_perm_mask_checked): Likewise.
1785 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
1786 (vectorizable_store): Likewise.
1787 (vectorizable_load): Likewise.
1788 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
1789 (vectorizable_bswap): Likewise.
1791 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1792 Alan Hayward <alan.hayward@arm.com>
1793 David Sherwood <david.sherwood@arm.com>
1795 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1796 * tree.c (build_vector): Likewise.
1797 (build_vector_from_ctor): Update accordingly.
1798 (build_vector_from_val): Likewise.
1799 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1800 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1801 * tree-vect-generic.c (add_rshift): Likewise.
1802 (expand_vector_divmod): Likewise.
1803 (optimize_vector_constructor): Likewise.
1804 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1805 (vect_transform_slp_perm_load): Likewise.
1806 (vect_schedule_slp_instance): Likewise.
1807 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1808 (vectorizable_call): Likewise.
1809 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1810 * expmed.c (make_tree): Likewise.
1811 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1812 a vector passed to build_vector.
1813 (fold_convert_const): Likewise.
1814 (exact_inverse): Likewise.
1815 (fold_ternary_loc): Likewise.
1816 (fold_relational_const): Likewise.
1817 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1818 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1819 (const_unop): Likewise. Store the reduction accumulator in a
1820 variable rather than an array.
1821 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1822 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1823 the new vector, rather than constructing it after the input arrays.
1824 (native_interpret_vector): Use auto_vec<tree> when building
1825 a vector passed to build_vector. Add elements in order.
1826 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1827 auto_vec<tree> when building a vector passed to build_vector.
1828 (vect_create_epilog_for_reduction): Likewise.
1829 (vectorizable_induction): Likewise.
1830 (get_initial_def_for_reduction): Likewise. Fix indentation of
1832 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1834 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1835 passed to build_vector.
1837 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1838 Alan Hayward <alan.hayward@arm.com>
1839 David Sherwood <david.sherwood@arm.com>
1841 * tree-core.h (tree_base::u): Add an "nelts" field.
1842 (tree_vector): Use VECTOR_CST_NELTS as the length.
1843 * tree.c (tree_size): Likewise.
1844 (make_vector): Initialize VECTOR_CST_NELTS.
1845 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1846 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1847 TYPE_VECTOR_SUBPARTS.
1848 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1849 as the number of units, setting it from VECTOR_CST_NELTS.
1850 (const_vector_from_tree): Likewise.
1851 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1852 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1853 (fold_negate_expr_1): Likewise.
1854 (fold_convert_const): Likewise.
1855 (const_binop): Likewise. Differentiate the number of output and
1857 (const_unop): Likewise.
1858 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1859 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1860 in cases that did the opposite.
1862 2017-09-14 Richard Biener <rguenther@suse.de>
1864 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1867 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1869 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1871 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1874 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1875 if and where to split a bb, except for splitting before debug insn
1876 sequences followed by non-label real insn. Delete debug insns
1877 in between basic blocks.
1879 * combine.c (make_compound_operation_int): Formatting fixes.
1881 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1882 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1883 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1884 * config/sol2.h (LINK_EH_SPEC): Likewise.
1885 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1886 * config/s390/linux.h (LINK_SPEC): Likewise.
1887 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1888 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1889 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1890 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1891 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1892 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1893 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1894 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1896 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1898 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1900 (ENDFILE_LINUX_SPEC): Likewise.
1901 (LINK_EH_SPEC): Likewise.
1902 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1903 (LINK_OS_LINUX_SPEC32): Likewise.
1904 (LINK_OS_LINUX_SPEC64): Likewise.
1905 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1906 (LINK_OS_LINUX_SPEC): Likewise.
1908 2017-09-13 Martin Liska <mliska@suse.cz>
1911 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1912 CASE_HIGH is NULL_TREE.
1914 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1915 Alan Hayward <alan.hayward@arm.com>
1916 David Sherwood <david.sherwood@arm.com>
1918 * target.def (secondary_memory_needed): New hook.
1919 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1920 instead of SECONDARY_MEMORY_NEEDED.
1921 (secondary_memory_needed_mode): Likewise.
1922 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1923 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1924 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1925 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1926 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1927 * doc/tm.texi: Regenerate.
1928 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1929 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1930 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1931 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1932 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1933 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1934 mode argument first and change the reg_class arguments to reg_class_t.
1935 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1936 Make static. Update the call to inline_secondary_memory_needed.
1937 (ix86_register_move_cost): Update the call to
1938 inline_secondary_memory_needed.
1939 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1940 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1942 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1943 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1945 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1946 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1947 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1948 and match hook interface. Add comment from mips.h.
1949 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1950 * config/mmix/mmix.md (truncdfsf2): Refer to
1951 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1953 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1954 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1955 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1956 (pa_secondary_memory_needed): New function.
1957 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1958 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1959 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1960 (pdp11_secondary_memory_needed): Make static and match hook interface.
1961 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1962 * config/powerpcspe/powerpcspe-protos.h
1963 (rs6000_secondary_memory_needed_ptr): Delete.
1964 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1966 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1967 (rs6000_option_override_internal): Assign to
1968 targetm.secondary_memory_needed rather than
1969 rs6000_secondary_memory_needed_ptr.
1970 (rs6000_secondary_memory_needed): Match hook interface.
1971 (rs6000_debug_secondary_memory_needed): Likewise.
1972 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1973 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1974 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1975 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1976 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1977 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1979 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1980 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1981 (rs6000_option_override_internal): Assign to
1982 targetm.secondary_memory_needed rather than
1983 rs6000_secondary_memory_needed_ptr.
1984 (rs6000_secondary_memory_needed): Match hook interface.
1985 (rs6000_debug_secondary_memory_needed): Likewise.
1986 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1987 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1988 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1989 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1990 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1991 (sparc_secondary_memory_needed): New function.
1992 * lra-constraints.c (check_and_process_move): Refer to
1993 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1995 (curr_insn_transform): Likewise.
1996 (process_alt_operands): Use targetm.secondary_memory_needed
1997 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1998 (check_secondary_memory_needed_p): Likewise.
1999 (choose_split_class): Likewise.
2000 * reload.c: Unconditionally include code that was previously
2001 conditional on SECONDARY_MEMORY_NEEDED.
2002 (push_secondary_reload): Use targetm.secondary_memory_needed
2003 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2004 (push_reload): Likewise.
2005 * reload1.c: Unconditionally include code that was previously
2006 conditional on SECONDARY_MEMORY_NEEDED.
2007 (choose_reload_regs): Use targetm.secondary_memory_needed
2008 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2009 (gen_reload): Likewise.
2010 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2012 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2013 Alan Hayward <alan.hayward@arm.com>
2014 David Sherwood <david.sherwood@arm.com>
2016 * target.def (secondary_memory_needed_mode): New hook:
2017 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2018 * targhooks.h (default_secondary_memory_needed_mode): New function.
2019 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2020 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2021 * doc/tm.texi: Regenerate.
2022 * lra-constraints.c (check_and_process_move): Use
2023 targetm.secondary_memory_needed_mode instead of
2024 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2025 (curr_insn_transform): Likewise.
2026 * reload.c (get_secondary_mem): Likewise.
2027 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2028 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2030 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2031 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2032 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2033 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2034 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2036 * config/powerpcspe/powerpcspe-protos.h
2037 (rs6000_secondary_memory_needed_mode): Delete.
2038 * config/powerpcspe/powerpcspe.c
2039 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2040 (rs6000_secondary_memory_needed_mode): Make static.
2041 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2042 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2044 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2046 (rs6000_secondary_memory_needed_mode): Make static.
2047 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2048 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2049 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2050 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2051 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2053 (sparc_secondary_memory_needed_mode): New function.
2054 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2056 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2058 * config/aarch64/constraints.md (Umq): New constraint.
2059 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2061 (mov<mode>): Update condition.
2063 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2065 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2066 when bitposition is the same.
2068 2017-09-13 Richard Biener <rguenther@suse.de>
2070 * dwarf2out.c (output_die_symbol): Remove.
2071 (output_die): Do not output a DIEs symbol.
2073 2017-09-13 Richard Biener <rguenther@suse.de>
2076 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2077 default-def to avoid breaking iterator update with the weird
2078 interaction with cgraph_update_edges_for_call_stmt_node.
2080 2017-09-13 Richard Biener <rguenther@suse.de>
2082 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2083 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2084 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2085 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2087 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2089 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2090 Disable pc relative literal load irrespective of
2091 TARGET_FIX_ERR_A53_84341 for default.
2093 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2095 * config/sparc/sparc.c (output_return): Output the source location of
2096 the insn in the delay slot, if any.
2097 (output_sibcall): Likewise.
2099 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2102 * common.opt (-static-pie): New alias.
2103 (shared): Negate static-pie.
2104 (-no-pie): Update help text.
2106 (static-pie): New option.
2107 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2108 -static-pie support.
2109 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2110 (LINK_EH_SPEC): Likewise.
2111 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2112 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2113 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2114 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2115 (init_gcc_specs): Likewise.
2116 (init_spec): Likewise.
2117 (display_help): Update help message for -pie.
2118 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2121 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2123 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2124 (movdi_aarch64): Likewise.
2125 (movti_aarch64): Likewise.
2127 2017-09-12 Simon Wright <simon@pushface.org>
2130 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2131 calculation of the minor version, always output as 0.
2133 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2136 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2137 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2138 on it early, rather than manual conversion late. For
2139 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2140 instead of performing manual conversion.
2142 2017-09-12 Carl Love <cel@us.ibm.com>
2144 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2145 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2147 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2148 VMULOSW): Add definitions.
2149 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2150 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2151 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2152 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2153 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2155 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2157 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2159 (movti_aarch64): Likewise.
2160 (movdf_aarch64): Likewise.
2161 (movtf_aarch64): Likewise.
2162 (load_pairdi): Likewise.
2163 (store_pairdi): Likewise.
2164 (load_pairdf): Likewise.
2165 (store_pairdf): Likewise.
2166 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2167 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2168 (ldr_got_small_<mode>): Likewise.
2169 (ldr_got_small_28k_<mode>): Likewise.
2170 (ldr_got_tiny): Likewise.
2171 * config/aarch64/iterators.md (ldst_sz): New.
2172 (ldpstp_sz): Likewise.
2173 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2175 (thunderx_load): Split load_8 to load_16.
2176 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2178 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2179 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2180 (xgene1_store_pair): Split store_8 to store_16.
2181 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2182 (falkor_st_0_st_sd): Split store_8 to store_16.
2184 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2186 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2187 and store1/2/3/4 to store_4/8/12/16.
2188 * config/aarch64/aarch64.md: Update for rename.
2189 * config/arm/arm.md: Likewise.: Likewise.
2190 * config/arm/arm.c: Likewise.
2191 * config/arm/thumb1.md: Likewise.
2192 * config/arm/thumb2.md: Likewise.
2193 * config/arm/vfp.md: Likewise.
2194 * config/arm/arm-generic.md: Likewise.
2195 * config/arm/arm1020e.md: Likewise.
2196 * config/arm/arm1026ejs.md: Likewise.
2197 * config/arm/arm1136jfs.md: Likewise.
2198 * config/arm/arm926ejs.md: Likewise.
2199 * config/arm/cortex-a15.md: Likewise.
2200 * config/arm/cortex-a17.md: Likewise.
2201 * config/arm/cortex-a5.md: Likewise.
2202 * config/arm/cortex-a53.md: Likewise.
2203 * config/arm/cortex-a57.md: Likewise.
2204 * config/arm/cortex-a7.md: Likewise.
2205 * config/arm/cortex-a8.md: Likewise.
2206 * config/arm/cortex-a9.md: Likewise.
2207 * config/arm/cortex-m4.md: Likewise.
2208 * config/arm/cortex-m7.md: Likewise.
2209 * config/arm/cortex-r4.md: Likewise.
2210 * config/arm/exynos-m1.md: Likewise.
2211 * config/arm/fa526.md: Likewise.
2212 * config/arm/fa606te.md: Likewise.
2213 * config/arm/fa626te.md: Likewise.
2214 * config/arm/fa726te.md: Likewise.
2215 * config/arm/fmp626.md: Likewise.
2216 * config/arm/iwmmxt.md: Likewise.
2217 * config/arm/ldmstm.md: Likewise.
2218 * config/arm/marvell-pj4.md: Likewise.
2219 * config/arm/xgene1.md: Likewise.
2220 * config/aarch64/thunderx.md: Likewise.
2221 * config/aarch64/thunderx2t99.md: Likewise.
2222 * config/aarch64/falkor.md: Likewise.
2224 2017-09-12 Martin Liska <mliska@suse.cz>
2226 * attribs.c (private_lookup_attribute): New function.
2227 * attribs.h (private_lookup_attribute): Declared here.
2228 (lookup_attribute): Called from this place.
2230 2017-09-12 Richard Biener <rguenther@suse.de>
2232 PR tree-optimization/82157
2233 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2234 stmts with side-effects.
2236 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2237 Alan Hayward <alan.hayward@arm.com>
2238 David Sherwood <david.sherwood@arm.com>
2240 * target.def (hard_regno_nregs): New hook.
2241 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2242 * targhooks.h (default_hard_regno_nregs): Declare.
2243 * targhooks.c (default_hard_regno_nregs): New function.
2244 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2245 (TARGET_HARD_REGNO_NREGS): ...this hook.
2246 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2247 (CLASS_MAX_NREGS): Likewise.
2248 * doc/tm.texi: Regenerate.
2249 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2250 instead of HARD_REGNO_NREGS.
2251 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2252 HARD_REGNO_NREGS in the comment.
2253 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2254 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2255 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2256 Return an unsigned int.
2257 (TARGET_HARD_REGNO_NREGS): Redefine.
2258 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2259 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2260 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2261 (arc_hard_regno_nregs): New function.
2262 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2263 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2264 (arm_hard_regno_nregs): New function.
2265 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2266 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2267 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2268 (TARGET_HARD_REGNO_NREGS): Redefine.
2269 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2270 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2271 (HARD_REGNO_NREGS): Delete.
2272 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2273 (cr16_hard_regno_nregs): New function.
2274 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2275 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2276 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2277 (cris_hard_regno_nregs): New function.
2278 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2279 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2280 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2281 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2282 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2283 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2284 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2285 (frv_hard_regno_nregs): Make static. Take and return an
2287 (frv_class_max_nregs): Remove outdated copy of documentation.
2288 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2289 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2290 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2291 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2292 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2293 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2294 (TARGET_HARD_REGNO_NREGS): Redefine.
2295 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2296 (CLASS_MAX_NREGS): Update comment.
2297 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2298 (ia64_hard_regno_nregs): New function.
2299 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2300 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2301 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2302 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2303 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2305 (m32c_hard_regno_nregs): Likewise. Make static.
2306 (TARGET_HARD_REGNO_NREGS): Redefine.
2307 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2308 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2309 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2310 (m68k_hard_regno_nregs): New function.
2311 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2312 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2313 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2314 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2315 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2316 Take and return an unsigned int.
2317 (TARGET_HARD_REGNO_NREGS): Redefine.
2318 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2319 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2320 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2321 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2322 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2323 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2324 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2325 (msp430_hard_regno_nregs): Make static. Take and return an
2327 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2328 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2329 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2330 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2331 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2332 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2333 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2334 (TARGET_HARD_REGNO_NREGS): Redefine.
2335 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2336 (PA_HARD_REGNO_NREGS): ...this.
2337 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2338 (PA_HARD_REGNO_NREGS): ...this.
2339 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2340 (pa_hard_regno_nregs): New function.
2341 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2342 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2343 (pdp11_hard_regno_nregs): New function.
2344 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2345 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2346 (rs6000_hard_regno_nregs_hook): New function.
2347 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2348 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2349 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2350 Take and return an unsigned int. Move earlier in file.
2351 (TARGET_HARD_REGNO_NREGS): Redefine.
2352 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2353 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2354 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2355 (rl78_hard_regno_nregs): Make static. Take and return an
2357 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2358 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2359 (rs6000_hard_regno_nregs_hook): New function.
2360 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2361 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2362 (TARGET_HARD_REGNO_NREGS): Redefine.
2363 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2364 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2365 instead of HARD_REGNO_NREGS.
2366 (s390_hard_regno_nregs): New function.
2367 (s390_hard_regno_mode_ok): Add comment from s390.h.
2368 (TARGET_HARD_REGNO_NREGS): Redefine.
2369 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2370 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2371 (sh_hard_regno_nregs): New function.
2372 (sh_pass_in_reg_p): Use it.
2373 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2374 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2375 (sparc_hard_regno_nregs): New function.
2376 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2377 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2378 (spu_function_arg_advance): Use it, supplying a valid register number.
2379 (TARGET_HARD_REGNO_NREGS): Redefine.
2380 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2381 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2382 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2383 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2384 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2385 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2386 (CLASS_MAX_NREGS): Remove copy of old documentation.
2387 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2388 (visium_hard_regno_nregs): New function.
2389 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2390 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2391 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2392 xtensa_hard_regno_nregs): New function.
2393 * system.h (HARD_REGNO_NREGS): Poison.
2395 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2397 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2398 hard_regno_nregs instead of HARD_REGNO_NREGS.
2399 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2400 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2401 (c6x_expand_epilogue): Likewise.
2402 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2403 (frv_read_iacc_argument): Likewise.
2404 * config/sh/sh.c: Include regs.h.
2405 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2406 (regs_used): Likewise.
2407 (output_stack_adjust): Likewise.
2408 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2409 * expmed.c: Include regs.h.
2410 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2411 * ree.c: Include regs.h.
2412 (combine_reaching_defs): Use hard_regno_nregs instead of
2414 (add_removable_extension): Likewise.
2416 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2418 * regs.h (hard_regno_nregs): Turn into a function.
2419 (end_hard_regno): Update accordingly.
2420 * caller-save.c (setup_save_areas): Likewise.
2421 (save_call_clobbered_regs): Likewise.
2422 (replace_reg_with_saved_mem): Likewise.
2423 (insert_restore): Likewise.
2424 (insert_save): Likewise.
2425 * combine.c (can_change_dest_mode): Likewise.
2426 (move_deaths): Likewise.
2427 (distribute_notes): Likewise.
2428 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2429 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2430 (rs6000_split_multireg_move): Likewise.
2431 (rs6000_register_move_cost): Likewise.
2432 (rs6000_memory_move_cost): Likewise.
2433 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2434 (rs6000_split_multireg_move): Likewise.
2435 (rs6000_register_move_cost): Likewise.
2436 (rs6000_memory_move_cost): Likewise.
2437 * cselib.c (cselib_reset_table): Likewise.
2438 (cselib_lookup_1): Likewise.
2439 * emit-rtl.c (set_mode_and_regno): Likewise.
2440 * function.c (aggregate_value_p): Likewise.
2441 * ira-color.c (setup_profitable_hard_regs): Likewise.
2442 (check_hard_reg_p): Likewise.
2443 (calculate_saved_nregs): Likewise.
2444 (assign_hard_reg): Likewise.
2445 (improve_allocation): Likewise.
2446 (calculate_spill_cost): Likewise.
2447 * ira-emit.c (modify_move_list): Likewise.
2448 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2449 (ira_hard_reg_in_set_p): Likewise.
2450 * ira.c (setup_reg_mode_hard_regset): Likewise.
2451 (clarify_prohibited_class_mode_regs): Likewise.
2452 (check_allocation): Likewise.
2453 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2454 (lra_setup_reg_renumber): Likewise.
2455 (setup_try_hard_regno_pseudos): Likewise.
2456 (spill_for): Likewise.
2457 (assign_hard_regno): Likewise.
2458 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2459 * lra-constraints.c (in_class_p): Likewise.
2460 (lra_constraint_offset): Likewise.
2461 (simplify_operand_subreg): Likewise.
2462 (lra_constraints): Likewise.
2463 (split_reg): Likewise.
2464 (split_if_necessary): Likewise.
2465 (invariant_p): Likewise.
2466 (inherit_in_ebb): Likewise.
2467 * lra-lives.c (process_bb_lives): Likewise.
2468 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2469 (get_hard_regs): Likewise.
2470 (do_remat): Likewise.
2471 * lra-spills.c (assign_spill_hard_regs): Likewise.
2472 * mode-switching.c (create_pre_exit): Likewise.
2473 * postreload.c (reload_combine_recognize_pattern): Likewise.
2474 * recog.c (peep2_find_free_register): Likewise.
2475 * regcprop.c (kill_value_regno): Likewise.
2476 (set_value_regno): Likewise.
2477 (copy_value): Likewise.
2478 (maybe_mode_change): Likewise.
2479 (find_oldest_value_reg): Likewise.
2480 (copyprop_hardreg_forward_1): Likewise.
2481 * regrename.c (check_new_reg_p): Likewise.
2482 (regrename_do_replace): Likewise.
2483 * reload.c (push_reload): Likewise.
2484 (combine_reloads): Likewise.
2485 (find_dummy_reload): Likewise.
2486 (operands_match_p): Likewise.
2487 (find_reloads): Likewise.
2488 (find_equiv_reg): Likewise.
2489 (reload_adjust_reg_for_mode): Likewise.
2490 * reload1.c (count_pseudo): Likewise.
2491 (count_spilled_pseudo): Likewise.
2492 (find_reg): Likewise.
2493 (clear_reload_reg_in_use): Likewise.
2494 (free_for_value_p): Likewise.
2495 (allocate_reload_reg): Likewise.
2496 (choose_reload_regs): Likewise.
2497 (reload_adjust_reg_for_temp): Likewise.
2498 (emit_reload_insns): Likewise.
2499 (delete_output_reload): Likewise.
2500 * rtlanal.c (subreg_get_info): Likewise.
2501 * sched-deps.c (sched_analyze_reg): Likewise.
2502 * sel-sched.c (init_regs_for_mode): Likewise.
2503 (mark_unavailable_hard_regs): Likewise.
2504 (choose_best_reg_1): Likewise.
2505 (verify_target_availability): Likewise.
2506 * valtrack.c (dead_debug_insert_temp): Likewise.
2507 * var-tracking.c (track_loc_p): Likewise.
2508 (emit_note_insn_var_location): Likewise.
2509 * varasm.c (make_decl_rtl): Likewise.
2510 * reginfo.c (choose_hard_reg_mode): Likewise.
2511 (init_reg_modes_target): Refer directly to
2512 this_target_regs->x_hard_regno_nregs.
2514 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2516 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2517 instead of hard_regno_nregs.
2519 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2521 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2522 end_hard_regno instead of hard_regno_nregs.
2523 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2524 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2525 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2526 * ira-color.c (improve_allocation): Likewise.
2527 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2528 * lra-lives.c (mark_regno_live): Likewise.
2529 (mark_regno_dead): Likewise.
2530 * lra-remat.c (operand_to_remat): Likewise.
2531 * lra.c (collect_non_operand_hard_regs): Likewise.
2532 * postreload.c (reload_combine_note_store): Likewise.
2533 (move2add_valid_value_p): Likewise.
2534 * reload.c (regno_clobbered_p): Likewise.
2536 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2538 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2540 * config/v850/v850.c (v850_reorg): Likewise.
2541 * reload.c (refers_to_regno_for_reload_p): Likewise.
2542 (find_equiv_reg): Likewise.
2543 * reload1.c (reload_reg_reaches_end_p): Likewise.
2545 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2547 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2549 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2550 * config/arm/arm.c (output_move_neon): Likewise.
2551 (arm_attr_length_move_neon): Likewise.
2552 (neon_split_vcombine): Likewise.
2553 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2554 (c6x_mark_reg_written): Likewise.
2555 (c6x_dwarf_register_span): Likewise.
2556 * config/i386/i386.c (ix86_save_reg): Likewise.
2557 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2558 (rws_access_reg): Likewise.
2559 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2560 * mode-switching.c (create_pre_exit): Likewise.
2561 * ree.c (combine_reaching_defs): Likewise.
2562 (add_removable_extension): Likewise.
2563 * regcprop.c (find_oldest_value_reg): Likewise.
2564 (copyprop_hardreg_forward_1): Likewise.
2565 * reload.c (reload_inner_reg_of_subreg): Likewise.
2566 (push_reload): Likewise.
2567 (combine_reloads): Likewise.
2568 (find_dummy_reload): Likewise.
2569 (reload_adjust_reg_for_mode): Likewise.
2570 * reload1.c (find_reload_regs): Likewise.
2571 (forget_old_reloads_1): Likewise.
2572 (reload_reg_free_for_value_p): Likewise.
2573 (reload_adjust_reg_for_temp): Likewise.
2574 (emit_reload_insns): Likewise.
2575 (delete_output_reload): Likewise.
2576 * sel-sched.c (choose_best_reg_1): Likewise.
2577 (choose_best_pseudo_reg): Likewise.
2579 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2580 Alan Hayward <alan.hayward@arm.com>
2581 David Sherwood <david.sherwood@arm.com>
2583 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2584 * target.def (slow_unaligned_access): New hook.
2585 * targhooks.h (default_slow_unaligned_access): Declare.
2586 * targhooks.c (default_slow_unaligned_access): New function.
2587 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2588 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2589 * doc/tm.texi: Regenerate.
2590 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2591 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2592 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2594 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2595 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2597 (rs6000_slow_unaligned_access): New function.
2598 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2599 (expand_block_compare): Likewise.
2600 (expand_strn_compare): Likewise.
2601 (rs6000_rtx_costs): Likewise.
2602 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2603 (riscv_slow_unaligned_access): Likewise.
2604 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2605 (riscv_slow_unaligned_access_p): ...this and make static.
2606 (riscv_option_override): Update accordingly.
2607 (riscv_slow_unaligned_access): New function.
2608 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2609 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2610 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2611 (rs6000_slow_unaligned_access): New function.
2612 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2613 (rs6000_rtx_costs): Likewise.
2614 * config/rs6000/rs6000-string.c (expand_block_compare)
2615 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2616 of SLOW_UNALIGNED_ACCESS.
2617 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2618 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2619 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2620 of SLOW_UNALIGNED_ACCESS.
2621 * expmed.c (simple_mem_bitfield_p): Likewise.
2622 * expr.c (alignment_for_piecewise_move): Likewise.
2623 (emit_group_load_1): Likewise.
2624 (emit_group_store): Likewise.
2625 (copy_blkmode_from_reg): Likewise.
2626 (emit_push_insn): Likewise.
2627 (expand_assignment): Likewise.
2628 (store_field): Likewise.
2629 (expand_expr_real_1): Likewise.
2630 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2631 * lra-constraints.c (simplify_operand_subreg): Likewise.
2632 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2633 * gimple-ssa-store-merging.c: Likewise in block comment at start
2635 * tree-ssa-strlen.c: Include target.h.
2636 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2637 of SLOW_UNALIGNED_ACCESS.
2638 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2640 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2642 PR rtl-optimization/82185
2643 * expmed.c (emit_store_flag_int): Only test tem if it has been
2646 2017-09-12 Richard Biener <rguenther@suse.de>
2649 * match.pd ((FTYPE) N CMP CST): Fix typo.
2651 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2653 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2655 (mips_near_type_p): Add 'short_call' attribute as a synonym
2657 * doc/extend.texi (short_call): Document new function attribute.
2659 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2662 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2663 assertion check that in the condition.
2664 (get_atomic_generic_size): Likewise. Before testing if parameter
2665 has pointer type, if it has array type, call for C++
2666 default_conversion to perform array-to-pointer conversion.
2668 2017-09-12 Richard Biener <rguenther@suse.de>
2670 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2671 for operations we cannot scalarize.
2673 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2675 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2676 vectors heap vectors. Clean up comments.
2677 Make visited_bbs a reference.
2678 (profitable_jump_thread_path): Make GC
2679 vectors heap vectors. Clean up comments.
2681 (convert_and_register_jump_thread_path): Make GC vectors heap
2683 (check_subpath_and_update_thread_path): Same. Clean up comments.
2684 Make visited_bbs a reference.
2685 (handle_phi): Abstract common code to to
2686 register_jump_thread_path_if_profitable.
2687 Rename VAR_BB to DEF_BB.
2689 Make GC vectors heap vectors.
2690 Make visited_bbs a reference.
2691 (handle_assignment): Same.
2692 (register_jump_thread_path_if_profitable): New.
2693 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2695 Make GC vectors heap vectors. Clean up comments.
2696 Make visited_bbs a reference.
2697 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2698 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2701 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2704 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2705 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2707 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2709 Revert r251800 and r251799.
2711 2017-09-11 Martin Jambor <mjambor@suse.cz>
2714 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2715 arguments in advance.
2716 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2717 use it to find predecessor edges.
2718 (naive_outof_ssa): Collect vector of predecessors.
2720 2017-09-08 Jason Merrill <jason@redhat.com>
2722 PR c++/70029 - ICE with ref-qualifier and -flto
2723 * langhooks.h (struct lang_hooks_for_types): Add
2724 copy_lang_qualifiers.
2725 * attribs.c (build_type_attribute_qual_variant): Use it.
2726 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2728 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2729 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2731 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2734 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2735 (*mulsi3_sp64): New instruction.
2736 (mulsi3): New expander.
2738 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2740 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2742 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2744 * sancov.c: Include memmodel.h.
2746 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2749 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2752 2017-09-07 Carl Love <cel@us.ibm.com>
2754 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2755 the sldi instruction.
2757 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2759 * sancov.c: Include tm_p.h.
2761 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2764 * output.h (switch_to_other_text_partition): New declaration.
2765 * varasm.c (switch_to_other_text_partition): New function.
2766 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
2767 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
2768 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
2769 to the other text partition before emitting LCL label and switch back
2770 after emitting the word after it.
2772 2017-09-07 Richard Biener <rguenther@suse.de>
2774 * passes.def (pass_split_crit_edges): Remove instance before PRE.
2775 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
2776 critical edges here, after loop init.
2777 (pass_data_pre): Remove PROP_no_crit_edges flags.
2778 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
2779 for valueization of call args to avoid leaking VN_TOP.
2780 (visit_use): Assert we do not visit default defs.
2781 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
2782 Use error_mark_node to more easily detect leaking VN_TOP.
2783 All default-defs are varying, not VN_TOP. Mark them visited.
2784 (run_scc_vn): Make code match comment.
2786 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2788 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
2789 OPTION_MASK_FLOAT128_KEYWORD.
2790 (POWERPC_MASKS): Likewise.
2791 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2792 support for the -mfloat128-type option, and make -mfloat128
2793 default on PowerPC Linux systems. Define or undefine
2794 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2795 Define __float128 to be __ieee128 if IEEE 128-bit support is
2796 enabled, or undefine it.
2797 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2798 Delete defining __FLOAT128_TYPE__.
2799 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2800 -mfloat128-type option and make -mfloat128 default on PowerPC
2802 (TARGET_FLOAT128_TYPE): Likewise.
2803 (-mfloat128-type): Likewise.
2804 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2805 Delete the -mfloat128-type option and make -mfloat128 default on
2806 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2807 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2808 128-bit floating point is enabled. Change tests from using
2809 -mfloat128-type to -mfloat128.
2810 (rs6000_mangle_type): Use the correct mangling for the __float128
2811 type even if normal long double is restricted to 64-bits.
2812 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2814 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2815 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2816 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2818 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2819 documentation for -mfloat128.
2821 2017-09-06 Olivier Hainque <hainque@adacore.com>
2823 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2825 2017-09-06 Wish Wu <wishwu007@gmail.com>
2826 Jakub Jelinek <jakub@redhat.com>
2828 * asan.c (initialize_sanitizer_builtins): Add
2829 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2830 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2831 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2832 BT_FN_VOID_UINT64_PTR variables.
2833 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2834 (BT_FN_VOID_UINT16_UINT16): Likewise.
2835 (BT_FN_VOID_UINT32_UINT32): Likewise.
2836 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2837 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2838 (BT_FN_VOID_UINT64_PTR): Likewise.
2839 * common.opt (flag_sanitize_coverage): New variable.
2840 (fsanitize-coverage=trace-pc): Remove.
2841 (fsanitize-coverage=): Add.
2842 * flag-types.h (enum sanitize_coverage_code): New enum.
2843 * fold-const.c (fold_range_test): Disable non-short-circuit
2844 optimization if flag_sanitize_coverage.
2845 (fold_truth_andor): Likewise.
2846 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2847 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2848 (coverage_sanitizer_opts): New array.
2849 (get_closest_sanitizer_option): Add OPTS argument, handle also
2850 OPT_fsanitize_coverage_.
2851 (parse_sanitizer_options): Adjusted to also handle
2852 OPT_fsanitize_coverage_.
2853 (common_handle_option): Add OPT_fsanitize_coverage_.
2854 * sancov.c (instrument_comparison, instrument_switch): New function.
2855 (sancov_pass): Add trace-cmp support.
2856 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2857 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2858 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2859 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2860 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2861 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2862 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2863 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2864 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2865 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2867 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2869 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2870 error. Only quit immediately if parsing is complete.
2871 (BEGIN): Initialize fatal_err and parse_done.
2872 (begin fpu, end fpu): Check number of arguments.
2873 (begin arch, end arch): Likewise.
2874 (begin cpu, end cpu): Likewise.
2875 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2876 (optalias): Likewise.
2878 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2880 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2881 * config/arm/arm-isa.h: Delete. Move definitions to ...
2882 * arm-cpus.in: ... here. Use new feature and fgroup values.
2883 * config/arm/arm.c (arm_option_override): Use lower case for feature
2885 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2886 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2887 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2889 (print_isa_bits_for): New function.
2890 (gen_isa): New function.
2891 (gen_comm_data): Use print_isa_bits_for.
2892 (define feature): New keyword.
2893 (define fgroup): New keyword.
2894 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2895 (arm-isa.h): Add rule to generate file.
2896 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2897 case for feature bit names.
2899 2017-09-06 Richard Biener <rguenther@suse.de>
2901 * tree-ssa-pre.c (NECESSARY): Remove.
2902 (create_expression_by_pieces): Do not touch pass-local flags.
2903 (insert_into_preds_of_block): Likewise.
2904 (do_pre_regular_insertion): Likewise.
2905 (eliminate_insert): Likewise.
2906 (eliminate_dom_walker::before_dom_children): Likewise.
2907 (fini_eliminate): Do not look at inserted_exprs.
2908 (mark_operand_necessary): Remove.
2909 (remove_dead_inserted_code): Replace with simple work-list
2910 algorithm based on inserted_exprs and SSA uses.
2911 (pass_pre::execute): Re-order fini_eliminate and
2912 remove_dead_inserted_code.
2914 2017-09-06 Olivier Hainque <hainque@adacore.com>
2916 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2917 for VxWorks 7. Adjust surrounding comments.
2919 2017-09-06 Richard Biener <rguenther@suse.de>
2921 * gimple-ssa-strength-reduction.c
2922 (find_candidates_dom_walker::before_dom_children): Also allow
2925 2017-09-06 Richard Biener <rguenther@suse.de>
2927 PR tree-optimization/82108
2928 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2929 for gap in the non-permutation SLP case.
2931 2017-09-06 Martin Jambor <mjambor@suse.cz>
2933 PR tree-optimization/82078
2934 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2935 add_access_to_work_queue...
2936 (build_accesses_from_assign): ...here.
2937 (propagate_all_subaccesses): Make sure racc is the group
2938 representative, if there is one.
2940 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2943 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2946 2017-09-06 Richard Biener <rguenther@suse.de>
2948 * gimple-ssa-strength-reduction.c
2949 (find_candidates_dom_walker::before_doom_children): Use a
2950 type and not a mode check.
2952 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2955 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2957 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2959 2017-09-05 Jeff Law <law@redhat.com>
2961 PR tree-optimization/64910
2962 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2963 cases where we have 3 or more operands.
2965 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2968 * omp-low.c (lower_omp_for): Recompute tree invariant if
2969 gimple_omp_for_initial/final is ADDR_EXPR.
2972 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2973 into gimple val before gimplification fo the COND_EXPR.
2975 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2977 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2978 REGION_COPY argument.
2979 (thread_through_all_blocks): Remove unused argument to
2980 duplicate_thread_path.
2982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2983 Alan Hayward <alan.hayward@arm.com>
2984 David Sherwood <david.sherwood@arm.com>
2986 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2987 Take a scalar_mode rather than a machine_mode.
2988 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2989 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2990 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2991 (aarch64_gen_adjusted_ldpstp): Likewise.
2992 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2994 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2995 Alan Hayward <alan.hayward@arm.com>
2996 David Sherwood <david.sherwood@arm.com>
2998 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2999 Take a scalar_int_mode instead of a machine_mode.
3000 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3001 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3002 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3003 (aarch64_simd_attr_length_rglist): Delete.
3004 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3005 a scalar_int_mode instead of a machine_mode.
3006 (aarch64_add_offset): Likewise.
3007 (aarch64_internal_mov_immediate): Likewise
3008 (aarch64_add_constant_internal): Likewise.
3009 (aarch64_add_constant): Likewise.
3010 (aarch64_movw_imm): Likewise.
3011 (aarch64_rtx_arith_op_extract_p): Likewise.
3012 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3013 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3014 Remove assert that the mode isn't a vector.
3015 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3016 (aarch64_expand_mov_immediate): Update calls after above changes.
3017 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3018 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3019 (aarch64_move_imm): Likewise.
3020 (aarch64_can_const_movi_rtx_p): Likewise.
3021 (aarch64_strip_extend): Likewise.
3022 (aarch64_extr_rtx_p): Likewise.
3023 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3024 a CONST_INT when the mode parameter is VOIDmode.
3025 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3027 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3029 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3030 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3031 (bitwise_type_for_mode): Update accordingly.
3033 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3035 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3036 * stor-layout.c (mode_for_size_tree): Likewise.
3037 (mode_for_array): Update accordingly.
3038 (layout_decl): Likewise.
3039 (compute_record_mode): Likewise. Only set the mode once.
3041 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3043 * target.def (get_mask_mode): Change return type to opt_mode.
3045 * doc/tm.texi: Regenerate.
3046 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3047 * targhooks.c (default_get_mask_mode): Likewise.
3048 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3049 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3050 targetm.get_mask_mode.
3051 * tree.c (build_truth_vector_type): Likewise.
3053 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3055 * machmode.h (mode_for_vector): Return an opt_mode.
3056 * stor-layout.c (mode_for_vector): Likewise.
3057 (mode_for_int_vector): Update accordingly.
3058 (layout_type): Likewise.
3059 * config/i386/i386.c (emit_memmov): Likewise.
3060 (ix86_expand_set_or_movmem): Likewise.
3061 (ix86_expand_vector_init): Likewise.
3062 (ix86_get_mask_mode): Likewise.
3063 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3065 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3066 * expmed.c (extract_bit_field_1): Likewise.
3067 * expr.c (expand_expr_real_2): Likewise.
3068 * optabs-query.c (can_vec_perm_p): Likewise.
3069 (can_vec_mask_load_store_p): Likewise.
3070 * optabs.c (expand_vec_perm): Likewise.
3071 * targhooks.c (default_get_mask_mode): Likewise.
3072 * tree-vect-stmts.c (vectorizable_store): Likewise.
3073 (vectorizable_load): Likewise.
3074 (get_vectype_for_scalar_type_and_size): Likewise.
3076 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3078 * machmode.h (mode_for_int_vector): New function.
3079 * stor-layout.c (mode_for_int_vector): Likewise.
3080 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3081 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3082 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3083 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3084 (s390_expand_vcond): Likewise.
3086 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3088 * machmode.h (opt_machine_mode): New type.
3089 (opt_mode<T>): Allow construction from anything that can be
3091 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3092 (mode_for_size): Return an opt_machine_mode.
3093 * stor-layout.c (mode_for_size): Likewise.
3094 (mode_for_size_tree): Update call accordingly.
3095 (bitwise_mode_for_mode): Likewise.
3096 (make_fract_type): Likewise.
3097 (make_accum_type): Likewise.
3098 * caller-save.c (replace_reg_with_saved_mem): Update call
3100 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3101 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3102 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3103 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3104 * expmed.c (extract_bit_field_1): Likewise.
3105 * reload.c (get_secondary_mem): Likewise.
3106 * varasm.c (assemble_integer): Likewise.
3107 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3110 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3112 * machmode.h (decimal_float_mode_for_size): New function.
3113 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3114 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3115 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3116 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3117 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3118 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3122 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3123 (get_builtin_sync_mode): Likewise.
3124 (expand_ifn_atomic_compare_exchange): Likewise.
3125 (expand_builtin_atomic_clear): Likewise.
3126 (expand_builtin_atomic_test_and_set): Likewise.
3127 (fold_builtin_atomic_always_lock_free): Likewise.
3128 * calls.c (compute_argument_addresses): Likewise.
3129 (emit_library_call_value_1): Likewise.
3130 (store_one_arg): Likewise.
3131 * combine.c (combine_instructions): Likewise.
3132 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3133 * config/arm/arm.c (arm_function_value): Likewise.
3134 (aapcs_allocate_return_reg): Likewise.
3135 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3136 * config/i386/i386.c (construct_container): Likewise.
3137 (ix86_gimplify_va_arg): Likewise.
3138 (ix86_expand_sse_cmp): Likewise.
3139 (emit_memmov): Likewise.
3140 (emit_memset): Likewise.
3141 (expand_small_movmem_or_setmem): Likewise.
3142 (ix86_expand_pextr): Likewise.
3143 (ix86_expand_pinsr): Likewise.
3144 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3145 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3147 * config/mips/mips.c (mips_function_value_1) Likewise.
3148 (mips_block_move_straight): Likewise.
3149 (mips_expand_ins_as_unaligned_store): Likewise.
3150 * config/powerpcspe/powerpcspe.c
3151 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3152 (rs6000_darwin64_record_arg_flush): Likewise.
3153 * config/rs6000/rs6000.c
3154 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3155 (rs6000_darwin64_record_arg_flush): Likewise.
3156 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3157 (sparc_function_value_1): Likewise.
3158 * config/spu/spu.c (adjust_operand): Likewise.
3159 (spu_emit_branch_or_set): Likewise.
3160 (arith_immediate_p): Likewise.
3161 * emit-rtl.c (gen_lowpart_common): Likewise.
3162 * expr.c (expand_expr_real_1): Likewise.
3163 * function.c (assign_parm_setup_block): Likewise.
3164 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3165 * reload1.c (alter_reg): Likewise.
3166 * stor-layout.c (mode_for_vector): Likewise.
3167 (layout_type): Likewise.
3169 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3171 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3172 (spu_convert_move): Likewise.
3173 * lower-subreg.c (resolve_simple_move): Likewise.
3175 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3178 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3179 define_insn to a define_expand.
3180 (altivec_vsum2sws_direct): New define_insn.
3181 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3183 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3185 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3188 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3193 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3194 realign stack if stack alignment needed is less than incoming
3197 2017-09-05 Marek Polacek <polacek@redhat.com>
3200 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3203 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3205 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3207 2017-09-05 Richard Biener <rguenther@suse.de>
3209 PR tree-optimization/82084
3210 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3212 2017-09-05 Richard Biener <rguenther@suse.de>
3214 PR tree-optimization/82102
3215 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3217 2017-09-05 Martin Liska <mliska@suse.cz>
3219 PR tree-optimization/82032
3220 * tree-cfg.c (generate_range_test): New function.
3221 * tree-cfg.h (generate_range_test): Declared here.
3222 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3223 (cleanup_control_expr_graph): Use it.
3224 * tree-switch-conversion.c (try_switch_expansion): Remove
3226 (emit_case_nodes): Use generate_range_test.
3228 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3231 * config/i386/i386.md (*<btsc><mode>_mask): Add
3232 TARGET_USE_BT to insn constraint.
3233 (*btr<mode>_mask): Ditto.
3235 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3237 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3238 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3240 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3243 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3244 TARGET_NEON and TARGET_IWMMXT.
3245 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3246 TARGET_NEON and TARGET_IWMMXT.
3247 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3249 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3251 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3252 (ix86_rewrite_tls_address): Ditto.
3253 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3254 (ix86_rewrite_tls_address_1): Ditto.
3255 (ix86_rewrite_tls_address): Ditto.
3256 * config/i386/predicates.md (tls_address_pattern): New predicate.
3257 * config/i386/i386.md (TLS address splitter): New splitter.
3259 2017-09-04 Richard Biener <rguenther@suse.de>
3261 PR tree-optimization/82084
3262 * fold-const.h (can_native_encode_string_p): Declare.
3263 * fold-const.c (can_native_encode_string_p): Factor out from ...
3264 (native_encode_string): ... here.
3265 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3266 vectorizing stores from constants we later cannot handle.
3268 2017-09-04 Marek Polacek <polacek@redhat.com>
3271 * doc/invoke.texi: Update -Wtautological-compare documentation.
3273 2017-09-04 Jeff Law <law@redhat.com>
3275 PR tree-optimization/64910
3276 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3277 swap the first and last operand if the last is a constant.
3279 2017-09-04 Marek Polacek <polacek@redhat.com>
3282 * convert.c (do_narrow): When sanitizing signed integer overflows,
3283 bail out for signed types.
3284 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3286 2017-09-04 Richard Biener <rguenther@suse.de>
3288 PR tree-optimization/82060
3289 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3290 Move devirtualization after stmt folding and before EH/AB/noreturn
3291 cleanup to get the stmt refs canonicalized. Use a bool instead
3292 of gimple_modified_p since that doesn't work for NOPs. Schedule
3293 NOPs generated by folding for removal.
3295 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3296 Alan Hayward <alan.hayward@arm.com>
3297 David Sherwood <david.sherwood@arm.com>
3299 * coretypes.h (pad_direction): New enum.
3300 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3301 (FUNCTION_ARG_PADDING): Likewise.
3302 * target.def (function_arg_padding): New hook.
3303 * targhooks.h (default_function_arg_padding): Declare.
3304 * targhooks.c (default_function_arg_padding): New function.
3305 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3306 (TARGET_FUNCTION_ARG_PADDING): ...this.
3307 * doc/tm.texi: Regenerate.
3308 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3309 instead of direction.
3310 (compute_argument_addresses): Likewise.
3311 (load_register_parameters): Likewise.
3312 (emit_library_call_value_1): Likewise.
3313 (store_one_arg): Use targetm.calls.function_arg_padding instead
3314 of FUNCTION_ARG_PADDING.
3315 (must_pass_in_stack_var_size_or_pad): Likewise.
3316 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3317 (emit_group_store): Likewise.
3318 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3319 instead of FUNCTION_ARG_PADDING.
3320 (emit_push_insn): Likewise, and propagate enum change throughout
3322 * function.h (direction): Delete.
3323 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3325 * function.c (assign_parm_find_stack_rtl): Likewise.
3326 (assign_parm_setup_block_p): Likewise.
3327 (assign_parm_setup_block): Likewise.
3328 (gimplify_parameters): Likewise.
3329 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3330 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3332 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3333 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3334 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3335 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3336 (aarch64_function_arg_padding): ...this new function.
3337 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3338 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3339 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3340 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3341 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3342 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3343 (arm_pad_arg_upward): Replace with...
3344 (arm_function_arg_padding): ...this new function.
3345 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3347 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3348 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3349 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3350 (ia64_hpux_function_arg_padding): Replace with...
3351 (ia64_function_arg_padding): ...this new function. Use pad_direction
3352 instead of direction. Check for TARGET_HPUX.
3353 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3354 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3355 (iq2000_function_arg_padding): New function.
3356 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3357 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3358 (mips_function_arg_padding): ...this new function.
3359 (mips_pad_reg_upward): Update accordingly.
3360 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3361 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3362 targetm.calls.function_arg_padding.
3363 (FUNCTION_ARG_PADDING): Delete.
3364 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3365 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3366 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3367 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3368 (nios2_block_reg_padding): Return pad_direction instead of direction.
3369 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3370 instead of direction.
3371 (nios2_function_arg_padding): Likewise. Make static.
3372 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3373 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3374 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3375 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3376 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3377 (pa_function_arg_padding): Make static. Return pad_direction instead
3379 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3380 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3381 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3382 instead of direction. Use targetm.calls.function_arg_padding.
3383 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3384 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3385 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3386 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3387 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3389 (function_arg_padding): Rename to...
3390 (rs6000_function_arg_padding): ...this. Make static. Return
3391 pad_direction instead of direction.
3392 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3393 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3394 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3395 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3396 instead of direction. Use targetm.calls.function_arg_padding.
3397 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3398 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3399 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3400 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3401 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3402 (function_arg_padding): Rename to...
3403 (rs6000_function_arg_padding): ...this. Make static. Return
3404 pad_direction instead of direction.
3405 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3406 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3407 * config/s390/s390.c (s390_function_arg_padding): New function.
3408 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3409 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3410 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3411 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3412 (function_arg_padding): Rename to...
3413 (sparc_function_arg_padding): ...this. Make static. Return
3414 pad_direction instead of direction.
3415 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3416 * config/spu/spu.c (spu_function_arg_padding): New function.
3417 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3418 * system.h (FUNCTION_ARG_PADDING): Poison.
3420 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3421 Alan Hayward <alan.hayward@arm.com>
3422 David Sherwood <david.sherwood@arm.com>
3424 * target.def (modes_tieable_p): New hook.
3425 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3426 (TARGET_MODES_TIEABLE_P): ...this.
3427 * doc/tm.texi.in: Regenerate.
3428 * hooks.h (hook_bool_mode_mode_true): Declare.
3429 * hooks.c (hook_bool_mode_mode_true): New function.
3430 * combine.c (subst): Use targetm.modes_tieable_p instead of
3432 * dse.c (find_shift_sequence): Likewise.
3433 * expmed.c (extract_low_bits): Likewise.
3434 * lower-subreg.c: Include target.h.
3435 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3437 * rtlanal.c (rtx_cost): Likewise.
3438 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3439 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3440 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3441 (TARGET_MODES_TIEABLE_P): Redefine.
3442 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3443 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3444 (TARGET_MODES_TIEABLE_P): Redefine.
3445 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3446 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3447 (arc_modes_tieable_p): New function.
3448 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3449 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3450 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3451 (arm_modes_tieable_p): Make static.
3452 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3453 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3454 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3455 (TARGET_MODES_TIEABLE_P): Redefine.
3456 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3457 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3458 (TARGET_MODES_TIEABLE_P): Redefine.
3459 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3460 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3461 (cr16_modes_tieable_p): New function.
3462 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3463 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3464 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3465 (TRULY_NOOP_TRUNCATION): Update comment.
3466 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3467 (TRULY_NOOP_TRUNCATION): Update comment.
3468 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3469 (frv_modes_tieable_p): New function.
3470 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3471 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3472 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3473 (TARGET_MODES_TIEABLE_P): Redefine.
3474 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3475 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3476 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3477 (TARGET_MODES_TIEABLE_P): Redefine.
3478 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3479 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3480 (ia64_modes_tieable_p): New function.
3481 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3482 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3483 (iq2000_modes_tieable_p): New function.
3484 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3485 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3486 (lm32_modes_tieable_p): New function.
3487 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3488 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3489 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3490 (TARGET_MODES_TIEABLE_P): Redefine.
3491 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3492 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3493 (m32r_modes_tieable_p): New function.
3494 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3495 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3496 (m68k_modes_tieable_p): New function.
3497 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3498 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3499 (mcore_modes_tieable_p): New function.
3500 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3501 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3503 (TARGET_MODES_TIEABLE_P): Redefine.
3504 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3505 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3506 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3507 (TARGET_MODES_TIEABLE_P): Redefine.
3508 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3509 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3510 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3511 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3512 (mn10300_modes_tieable_p): ...this and make static.
3513 (TARGET_MODES_TIEABLE_P): Redefine.
3514 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3515 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3516 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3517 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3518 (msp430_modes_tieable_p): Make static.
3519 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3520 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3521 (TARGET_MODES_TIEABLE_P): Redefine.
3522 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3523 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3524 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3525 (TARGET_MODES_TIEABLE_P): Redefine.
3526 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3527 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3528 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3529 (TARGET_MODES_TIEABLE_P): Redefine.
3530 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3531 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3532 (pdp11_modes_tieable_p): New function.
3533 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3534 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3535 (rs6000_modes_tieable_p): New function.
3536 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3537 * config/powerpcspe/powerpcspe.md: Update comment.
3538 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3539 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3540 (TARGET_MODES_TIEABLE_P): Redefine.
3541 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3542 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3543 (rl78_modes_tieable_p): New function.
3544 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3545 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3546 (rs6000_modes_tieable_p): New function.
3547 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3548 * config/rs6000/rs6000.md: Update comment.
3549 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3550 * config/rx/rx.c (rx_modes_tieable_p): New function.
3551 (TARGET_MODES_TIEABLE_P): Redefine.
3552 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3553 * config/s390/s390.c (s390_modes_tieable_p): New function.
3554 (TARGET_MODES_TIEABLE_P): Redefine.
3555 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3556 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3557 (sh_modes_tieable_p): New function.
3558 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3559 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3560 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3561 (sparc_modes_tieable_p): Make static.
3562 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3563 * config/spu/spu.c (spu_modes_tieable_p): New function.
3564 (TARGET_MODES_TIEABLE_P): Redefine.
3565 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3566 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3567 (TARGET_MODES_TIEABLE_P): Redefine.
3568 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3569 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3570 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3571 * config/v850/v850.c (v850_modes_tieable_p): New function.
3572 (TARGET_MODES_TIEABLE_P): Redefine.
3573 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3574 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3575 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3576 (visium_modes_tieable_p): New function.
3577 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3578 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3579 (xtensa_modes_tieable_p): New function.
3580 * system.h (MODES_TIEABLE_P): Poison.
3582 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3583 Alan Hayward <alan.hayward@arm.com>
3584 David Sherwood <david.sherwood@arm.com>
3586 * target.def (hard_regno_mode_ok): New hook.
3587 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3588 (TARGET_HARD_REGNO_MODE_OK): ...this.
3589 * doc/tm.texi.in: Regenerate.
3590 * hooks.h (hook_bool_uint_mode_true): Declare.
3591 * hooks.c (hook_bool_uint_mode_true): New function.
3592 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3594 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3595 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3596 instead of HARD_REGNO_MODE_OK.
3597 * caller-save.c: Include target.h.
3598 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3600 * combine.c (can_combine_p): Likewise.
3601 (combinable_i3pat): Likewise.
3602 (can_change_dest_mode): Likewise.
3603 * expr.c (init_expr_target): Likewise.
3604 (convert_move): Likewise.
3605 (convert_modes): Likewise.
3606 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3607 (setup_prohibited_mode_move_regs): Likewise.
3608 * ira.h (target_ira): Likewise.
3609 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3610 * lra-constraints.c (process_alt_operands): Likewise.
3611 (split_reg): Likewise.
3612 * recog.c (peep2_find_free_register): Likewise.
3613 * ree.c (combine_reaching_defs): Likewise.
3614 * regcprop.c (maybe_mode_change): Likewise.
3615 * reginfo.c (init_reg_sets_1): Likewise.
3616 (choose_hard_reg_mode): Likewise.
3617 (simplifiable_subregs): Likewise.
3618 * regrename.c (check_new_reg_p): Likewise.
3619 * reload.c (find_valid_class): Likewise.
3620 (find_valid_class_1): Likewise.
3621 (reload_inner_reg_of_subreg): Likewise.
3622 (push_reload): Likewise.
3623 (combine_reloads): Likewise.
3624 (find_dummy_reload): Likewise.
3625 (find_reloads): Likewise.
3626 * reload1.c (find_reg): Likewise.
3627 (set_reload_reg): Likewise.
3628 (allocate_reload_reg): Likewise.
3629 (choose_reload_regs): Likewise.
3630 (reload_adjust_reg_for_temp): Likewise.
3631 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3632 (simplify_subreg_regno): Likewise.
3633 * sel-sched.c (init_regs_for_mode): Likewise.
3634 * varasm.c (make_decl_rtl): Likewise.
3635 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3636 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3638 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3639 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3641 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3642 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3643 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3644 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3645 (arc_mode_class): Delete.
3646 (HARD_REGNO_MODE_OK): Delete.
3647 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3648 (arc_hard_regno_mode_ok): Rename old array to...
3649 (arc_hard_regno_mode_ok_modes): ...this.
3650 (arc_conditional_register_usage): Update accordingly.
3651 (arc_mode_class): Make static.
3652 (arc_hard_regno_mode_ok): New function.
3653 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3654 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3655 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3656 (arm_hard_regno_mode_ok): Make static.
3657 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3659 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3660 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3661 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3664 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3665 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3666 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3667 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3669 * config/bfin/predicates.md (valid_reg_operand): Use
3670 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3671 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3672 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3673 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3674 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3675 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3676 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3677 (cr16_hard_regno_mode_ok): Make static and return a bool.
3678 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3679 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3680 (cris_hard_regno_mode_ok): New function.
3681 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3682 (epiphany_mode_class): Delete.
3683 (HARD_REGNO_MODE_OK): Delete.
3684 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3685 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3686 (hard_regno_mode_ok): Rename to...
3687 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3688 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3689 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3691 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3692 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3693 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3694 (frv_hard_regno_mode_ok): Make static and return a bool.
3695 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3697 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3698 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3699 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3700 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3702 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3703 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3704 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3705 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3707 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3708 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3709 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3710 (ia64_hard_regno_mode_ok): New function.
3711 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3712 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3713 (iq2000_hard_regno_mode_ok): New function.
3714 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3715 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3716 (lm32_hard_regno_mode_ok): New function.
3717 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3718 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3719 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3720 instead of HARD_REGNO_MODE_OK.
3721 (m32c_hard_regno_ok): Rename to...
3722 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3723 (m32c_cannot_change_mode_class): Update accordingly.
3724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3725 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3726 (m32r_mode_class): Delete.
3727 (HARD_REGNO_MODE_OK): Delete.
3728 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3729 (m32r_hard_regno_mode_ok): Rename to...
3730 (m32r_hard_regno_modes): ...this.
3731 (m32r_mode_class): Make static.
3732 (m32r_hard_regno_mode_ok): New function.
3733 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3734 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3735 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3736 (m68k_hard_regno_mode_ok): Make static.
3737 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3738 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3739 (mcore_hard_regno_mode_ok): New function.
3740 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3741 (HARD_REGNO_MODE_OK): Delete.
3742 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3744 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3745 (microblaze_hard_regno_mode_ok): New function.
3746 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3747 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3748 (mips_hard_regno_mode_ok): Delete.
3749 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3750 (mips_hard_regno_mode_ok_p): ...this and make static.
3751 (mips_hard_regno_mode_ok_p): Rename to...
3752 (mips_hard_regno_mode_ok_uncached): ...this.
3753 (mips_hard_regno_mode_ok): New function.
3754 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3755 of HARD_REGNO_MODE_OK.
3756 (mips_option_override): Update after above name changes.
3757 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3758 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3759 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3760 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3761 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3763 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3764 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3765 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3766 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3767 (msp430_hard_regno_mode_ok): Make static and return a bool.
3768 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
3769 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
3770 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
3772 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3773 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
3774 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
3775 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
3776 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
3777 (PA_HARD_REGNO_MODE_OK): ...this
3778 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
3779 (PA_HARD_REGNO_MODE_OK): ...this.
3780 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3781 (pa_hard_regno_mode_ok): New function.
3782 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
3783 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3784 (pdp11_hard_regno_mode_ok): New function.
3785 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
3786 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
3788 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
3790 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3791 (rs6000_hard_regno_mode_ok): Rename to...
3792 (rs6000_hard_regno_mode_ok_uncached): ...this.
3793 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3794 (rs6000_hard_regno_mode_ok): New function.
3795 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3796 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3797 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3798 (riscv_hard_regno_mode_ok): ...this and make static.
3799 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3800 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3801 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3802 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3803 (rl78_hard_regno_mode_ok): Make static and return bool.
3804 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3805 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3807 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3809 (rs6000_hard_regno_mode_ok): Rename to...
3810 (rs6000_hard_regno_mode_ok_uncached): ...this.
3811 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3812 (rs6000_hard_regno_mode_ok): New function.
3813 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3814 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3816 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3817 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3818 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3820 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3821 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3822 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3823 (sh_hard_regno_mode_ok): Make static.
3824 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3825 instead of HARD_REGNO_MODE_OK.
3826 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3827 (sparc_mode_class): Delete.
3828 (HARD_REGNO_MODE_OK): Delete.
3829 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3830 (hard_regno_mode_classes): Make static.
3831 (sparc_mode_class): Likewise.
3832 (sparc_hard_regno_mode_ok): New function.
3833 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3834 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3835 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3837 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3838 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3839 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3840 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3841 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3842 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3843 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3844 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3845 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3846 (visium_hard_regno_mode_ok): New function.
3847 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3848 instead of HARD_REGNO_MODE_OK.
3849 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3850 (HARD_REGNO_MODE_OK): Delete.
3851 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3852 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3853 (xtensa_option_override): Update accordingly.
3854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3855 (xtensa_hard_regno_mode_ok): New function.
3856 * system.h (HARD_REGNO_MODE_OK): Poison.
3858 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3859 Alan Hayward <alan.hayward@arm.com>
3860 David Sherwood <david.sherwood@arm.com>
3862 * target.def (hard_regno_call_part_clobbered): New hook.
3863 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3864 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3865 * doc/tm.texi: Regenerate.
3866 * hooks.h (hook_bool_uint_mode_false): Declare.
3867 * hooks.c (hook_bool_uint_mode_false): New function.
3868 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3869 * cselib.c (cselib_process_insn): Use
3870 targetm.hard_regno_call_part_clobbered instead of
3871 HARD_REGNO_CALL_PART_CLOBBERED.
3872 * ira-conflicts.c (ira_build_conflicts): Likewise.
3873 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3874 * lra-constraints.c (need_for_call_save_p): Likewise.
3875 * lra-lives.c: Include target.h.
3876 (check_pseudos_live_through_calls): Use
3877 targetm.hard_regno_call_part_clobbered instead of
3878 HARD_REGNO_CALL_PART_CLOBBERED.
3879 * regcprop.c: Include target.h.
3880 (copyprop_hardreg_forward_1): Use
3881 targetm.hard_regno_call_part_clobbered instead of
3882 HARD_REGNO_CALL_PART_CLOBBERED.
3883 * reginfo.c (choose_hard_reg_mode): Likewise.
3884 * regrename.c (check_new_reg_p): Likewise.
3885 * reload.c (find_equiv_reg): Likewise.
3886 * reload1.c (emit_reload_insns): Likewise.
3887 * sched-deps.c (deps_analyze_insn): Likewise.
3888 * sel-sched.c (init_regs_for_mode): Likewise.
3889 (mark_unavailable_hard_regs): Likewise.
3890 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3891 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3892 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3894 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3895 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3896 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3898 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3900 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3901 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3902 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3904 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3905 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3906 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3908 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3909 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3911 * config/powerpcspe/powerpcspe.c
3912 (rs6000_hard_regno_call_part_clobbered): New function.
3913 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3914 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3915 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3917 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3918 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3919 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3921 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3922 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3923 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3925 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3926 Alan Hayward <alan.hayward@arm.com>
3927 David Sherwood <david.sherwood@arm.com>
3929 * rtl.h (subreg_memory_offset): Declare.
3930 * emit-rtl.c (subreg_memory_offset): New function.
3931 * expmed.c (store_bit_field_1): Use it.
3932 * expr.c (undefined_operand_subword_p): Likewise.
3933 * simplify-rtx.c (simplify_subreg): Likewise.
3935 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3937 PR rtl-optimization/57448
3940 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3941 using atomic_load pattern.
3942 (expand_atomic_store): Likewise.
3944 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3947 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3948 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3949 handling. Use replace_call_with_value with NULL instead of
3950 gsi_replace, unlink_stmt_vdef and release_defs.
3952 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3955 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3957 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3960 * rtl.h (emit_library_call_value_1): Declare.
3961 (emit_library_call): Replace declaration with a series of overloads.
3962 Remove the parameter count argument.
3963 (emit_library_call_value): Likewise.
3964 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3965 with an "rtx_mode_t *".
3966 (emit_library_call_value): Delete.
3967 (emit_library_call): Likewise.
3968 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3969 (asan_emit_allocas_unpoison): Likewise.
3970 * builtins.c (expand_builtin_powi): Likewise.
3971 (expand_asan_emit_allocas_unpoison): Likewise.
3972 * cfgexpand.c (expand_main_function): Likewise.
3973 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3974 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3975 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3976 * config/arm/arm.c (arm_trampoline_init): Likewise.
3977 (arm_call_tls_get_addr): Likewise.
3978 (arm_expand_divmod_libfunc): Likewise.
3979 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3980 (smulsi3_highpart): Likewise.
3981 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3982 (c6x_expand_compare): Likewise.
3983 (c6x_expand_movmem): Likewise.
3984 * config/frv/frv.c (frv_trampoline_init): Likewise.
3985 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3986 (ix86_expand_divmod_libfunc): Likewise.
3987 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3988 (ia64_expand_compare): Likewise.
3989 (ia64_profile_hook): Likewise.
3990 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3991 (nonlocal_goto): Likewise.
3992 (restore_stack_nonlocal): Likewise.
3993 * config/m32r/m32r.c (block_move_call): Likewise.
3994 (m32r_trampoline_init): Likewise.
3995 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3996 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3997 (m68k_call_m68k_read_tp): Likewise.
3998 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3999 (microblaze_expand_divide): Likewise.
4000 * config/mips/mips.h (mips_args): Likewise.
4001 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4002 (MIPS_ICACHE_SYNC): Likewise.
4003 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4004 (nios2_trampoline_init): Likewise.
4005 * config/pa/pa.c (hppa_tls_call): Likewise.
4006 (pa_trampoline_init): Likewise.
4007 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4008 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4009 (expand_strn_compare): Likewise.
4010 (rs6000_generate_compare): Likewise.
4011 (rs6000_expand_float128_convert): Likewise.
4012 (output_profile_hook): Likewise.
4013 (rs6000_trampoline_init): Likewise.
4014 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4015 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4016 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4017 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4018 (rs6000_generate_compare): Likewise.
4019 (rs6000_expand_float128_convert): Likewise.
4020 (output_profile_hook): Likewise.
4021 (rs6000_trampoline_init): Likewise.
4022 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4023 * config/sh/sh.c (sh_trampoline_init): Likewise.
4024 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4025 (sparc_emit_float_lib_cmp): Likewise.
4026 (sparc32_initialize_trampoline): Likewise.
4027 (sparc64_initialize_trampoline): Likewise.
4028 (sparc_profile_hook): Likewise.
4029 * config/spu/spu.c (ea_load_store): Likewise.
4030 * config/spu/spu.md (floatunssidf2): Likewise.
4031 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4032 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4033 * config/visium/visium.c (expand_block_move_4): Likewise.
4034 (expand_block_move_2): Likewise.
4035 (expand_block_move_1): Likewise.
4036 (expand_block_set_4): Likewise.
4037 (expand_block_set_2): Likewise.
4038 (expand_block_set_1): Likewise.
4039 (visium_trampoline_init): Likewise.
4040 (visium_profile_hook): Likewise.
4041 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4042 (xtensa_setup_frame_addresses): Likewise.
4043 (xtensa_trampoline_init): Likewise.
4044 * except.c (sjlj_emit_function_enter): Likewise.
4045 (sjlj_emit_function_exit): Likewise.
4046 * explow.c (allocate_dynamic_stack_space): Likewise.
4047 (probe_stack_range): Likewise.
4048 * expr.c (convert_mode_scalar): Likewise.
4049 * optabs.c (expand_binop): Likewise.
4050 (expand_twoval_binop_libfunc): Likewise.
4051 (expand_unop): Likewise.
4052 (prepare_cmp_insn): Likewise.
4053 (prepare_float_lib_cmp): Likewise.
4054 (expand_float): Likewise.
4055 (expand_fix): Likewise.
4056 (expand_fixed_convert): Likewise.
4057 (maybe_emit_sync_lock_test_and_set): Likewise.
4058 (expand_atomic_compare_and_swap): Likewise.
4059 (expand_mem_thread_fence): Likewise.
4060 (expand_atomic_fetch_op): Likewise.
4062 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4064 * doc/generic.texi (OpenACC): Adjust URL.
4065 * doc/invoke.texi (C Dialect Options): Ditto.
4067 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4069 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4070 predicate for operand 1. Add (m,<S>) constraint.
4071 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4072 Prevent memory operand 1 with register operand 2.
4074 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4076 PR rtl-optimization/82024
4077 * combine.c (try_combine): If the combination result is a PARALLEL,
4078 and we only need to retain the SET in there that would be placed
4079 at I2, check that we can place that at I3 instead, before doing so.
4081 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4084 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4086 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4087 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4090 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4091 Jeff Law <law@redhat.com>
4093 * varasm.c (bss_initializer_p): Do not put constants into .bss
4094 (categorize_decl_for_section): Handle bss_initializer_p returning
4095 false when DECL_INITIAL is NULL.
4097 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4100 * config/s390/s390.c (s390_can_inline_p): New function.
4102 2017-09-01 Jeff Law <law@redhat.com>
4104 PR tree-optimization/82052
4105 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4106 Always initialize the returned slot after a hash table miss
4107 when INSERT is true.
4109 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4111 * config/s390/s390.md (mem_signal_fence): Remove.
4112 * doc/md.texi (mem_signal_fence): Remove.
4113 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4115 * target-insns.def (mem_signal_fence): Remove.
4117 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4120 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4123 * asan.c (create_odr_indicator): Strip name encoding from assembler
4124 name before appending it after __odr_asan_.
4126 2017-09-01 Martin Liska <mliska@suse.cz>
4128 PR tree-optimization/82059
4129 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4130 frequency only when an edge is redirected.
4132 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4134 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4135 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4136 (arc_conditional_register_usage): Remove ARC600 lp_count
4138 (arc_file_start): Emit Tag_ARC_CPU_variation.
4139 (arc_can_use_doloop_p): New conditions to use ZOLs.
4140 (hwloop_fail): New function.
4141 (hwloop_optimize): Likewise.
4142 (hwloop_pattern_reg): Likewise.
4143 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4144 (arc_reorg_loops): New function, calls reorg_loops.
4145 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4146 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4148 (arc_loop_hazard): Remove function, functionality moved into
4150 (arc_hazard): Remove arc_loop_hazard call.
4151 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4152 into hwloop_optimize.
4153 (arc_label_align): Remove ZOL handling.
4154 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4155 * config/arc/arc.md (doloop_begin): Remove pattern.
4156 (doloop_begin_i): Likewise.
4157 (doloop_end_i): Likewise.
4158 (doloop_fallback): Likewise.
4159 (doloop_fallback_m): Likewise.
4160 (doloop_end): Reimplement expand.
4161 (arc_lp): New pattern for LP instruction.
4162 (loop_end): New pattern.
4163 (loop_fail): Likewise.
4164 (decrement_and_branch_until_zero): Likewise.
4165 * config/arc/arc.opt (mlpc-width): New option.
4166 * doc/invoke.texi (mlpc-width): Document option.
4168 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4170 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4171 (arc_ccfsm_advance): Fix checking for delay slots.
4172 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4174 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4176 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4178 (movhi_insn): Update store instruction constraint which are saving
4179 6-bit short immediates.
4180 (movsi_insn): Consider also short scaled load operations.
4181 (zero_extendhisi2_i): Use Usd constraint instead of T.
4182 (extendhisi2_i): Add q constraint.
4183 (arc_clzsi2): Add type and length attributes.
4184 (arc_ctzsi2): Likewise.
4185 * config/arc/constraints.md (Usc): Update constraint, the
4186 assembler can parse two relocations for a single instruction.
4188 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4190 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4191 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4193 2017-08-31 Olivier Hainque <hainque@adacore.com>
4195 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4196 match as powerpc-wrs-vxworks*.
4198 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4200 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4201 register constraint for by-element operand.
4202 (aarch64_mls_elt_merge<mode>): Likewise.
4204 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4206 * config/arc/arc.c (arc_can_follow_jump): Check for short
4209 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4211 * config.gcc: Use g.opt for arc.
4212 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4213 functionality moved to ...
4214 (legitimate_scaled_address_p): New function, ...here.
4215 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4216 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4217 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4219 (arc_override_options): Handle G option.
4220 (arc_output_pic_addr_const): Correct function definition.
4221 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4222 (arc_decl_anon_ns_mem_p): Delete.
4223 (arc_in_small_data_p): Overhaul this function to take into
4224 consideration the value given via G option.
4225 (arc_rewrite_small_data_1): Renamed and corrected old
4226 arc_rewrite_small_data function.
4227 (arc_rewrite_small_data): New function.
4228 (small_data_pattern): Don't use pic_offset_table_rtx.
4229 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4230 * config/arc/simdext.md (movmisalignv2hi): Use
4231 prepare_move_operands function.
4233 (movmisalign*): Likewise.
4234 * doc/invoke.texi (ARC options): Document -G option.
4236 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4238 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4240 * config/arc/arc.c (arc_print_operand): Output scalled address for
4241 sdata whenever is possible.
4242 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4243 load/stores are available.
4244 (compact_sda_memory_operand): Check for the alignment required by
4245 code density instructions.
4246 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4248 * config/arc/constraints.md (Usd): Update constraint.
4249 (Us0): New constraint.
4250 (Usc): Update constraint.
4252 2017-08-31 Richard Biener <rguenther@suse.de>
4255 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4258 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4260 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4261 Resize type_signature.
4263 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4264 Alan Hayward <alan.hayward@arm.com>
4265 David Sherwood <david.sherwood@arm.com>
4267 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4268 subregs whose inner modes can be stored in GPRs.
4269 (aarch64_classify_index): Likewise.
4271 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4272 Alan Hayward <alan.hayward@arm.com>
4273 David Sherwood <david.sherwood@arm.com>
4275 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4276 (V_INT_EQUIV): ...this.
4277 (v_cmp_result): Rename to...
4278 (v_int_equiv): ...this.
4279 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4280 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4281 (copysign<mode>3): Likewise.
4282 (aarch64_simd_bsl<mode>_internal): Likewise.
4283 (aarch64_simd_bsl<mode>): Likewise.
4284 (vec_cmp<mode><mode>): Likewise.
4285 (vcond<mode><mode>): Likewise.
4286 (vcond<v_cmp_mixed><mode>): Likewise.
4287 (vcondu<mode><v_cmp_mixed>): Likewise.
4288 (aarch64_cm<optab><mode>): Likewise.
4289 (aarch64_cmtst<mode>): Likewise.
4290 (aarch64_fac<optab><mode>): Likewise.
4291 (vec_perm_const<mode>): Likewise.
4292 (vcond_mask_<mode><v_cmp_result>): Rename to...
4293 (vcond_mask_<mode><v_int_equiv>): ...this.
4294 (vec_cmp<mode><v_cmp_result>): Rename to...
4295 (vec_cmp<mode><v_int_equiv>): ...this.
4297 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4298 Alan Hayward <alan.hayward@arm.com>
4299 David Sherwood <david.sherwood@arm.com>
4301 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4303 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4304 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4305 (UNSPEC_LD4_DREG): New unspecs.
4306 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4307 (aarch64_ld2<mode>_dreg_be): Replace with...
4308 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4310 (aarch64_ld3<mode>_dreg_le)
4311 (aarch64_ld3<mode>_dreg_be): Replace with...
4312 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4314 (aarch64_ld4<mode>_dreg_le)
4315 (aarch64_ld4<mode>_dreg_be): Replace with...
4316 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4319 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4321 PR tree-optimization/81987
4322 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4323 insert an initializer in a location not dominated by the stride
4326 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4328 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4329 on the entire header of the finally block in the fallthru case.
4331 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4333 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4335 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4337 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4338 rs6000_emit_move_from_cr and call renamed function.
4339 (rs6000_emit_prologue): Call renamed functions.
4340 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4341 movesi_from_cr, remove volatile CRs.
4343 2017-08-30 Jon Beniston <jon@beniston.com>
4344 Richard Biener <rguenther@suse.de>
4346 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4347 of VECTOR_MODE_P check.
4348 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4349 element vector types.
4351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4353 * df.h (df_read_modify_subreg_p): Remove in favor of...
4354 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4355 const_rtx instead of an rtx.
4356 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4357 * df-problems.c (df_word_lr_mark_ref): Likewise.
4358 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4359 (mark_pseudo_reg_dead): Likewise.
4360 (mark_ref_dead): Likewise.
4361 * reginfo.c (init_subregs_of_mode): Likewise.
4362 * sched-deps.c (sched_analyze_1): Likewise.
4363 * df-scan.c (df_def_record_1): Likewise.
4364 (df_uses_record): Likewise.
4365 (df_read_modify_subreg_p): Remove in favor of...
4366 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4367 const_rtx instead of an rtx.
4369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4370 Alan Hayward <alan.hayward@arm.com>
4371 David Sherwood <david.sherwood@arm.com>
4373 * rtl.h (partial_subreg_p): New function.
4374 * caller-save.c (save_call_clobbered_regs): Use it.
4375 * calls.c (expand_call): Likewise.
4376 * combine.c (combinable_i3pat): Likewise.
4377 (simplify_set): Likewise.
4378 (make_extraction): Likewise.
4379 (make_compound_operation_int): Likewise.
4380 (gen_lowpart_or_truncate): Likewise.
4381 (force_to_mode): Likewise.
4382 (make_field_assignment): Likewise.
4383 (reg_truncated_to_mode): Likewise.
4384 (record_truncated_value): Likewise.
4385 (move_deaths): Likewise.
4386 * cse.c (record_jump_cond): Likewise.
4387 (cse_insn): Likewise.
4388 * cselib.c (cselib_lookup_1): Likewise.
4389 * expmed.c (extract_bit_field_using_extv): Likewise.
4390 * function.c (assign_parm_setup_reg): Likewise.
4391 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4392 * ira-build.c (create_insn_allocnos): Likewise.
4393 * lra-coalesce.c (merge_pseudos): Likewise.
4394 * lra-constraints.c (match_reload): Likewise.
4395 (simplify_operand_subreg): Likewise.
4396 (curr_insn_transform): Likewise.
4397 * lra-lives.c (process_bb_lives): Likewise.
4398 * lra.c (new_insn_reg): Likewise.
4399 (lra_substitute_pseudo): Likewise.
4400 * regcprop.c (mode_change_ok): Likewise.
4401 (maybe_mode_change): Likewise.
4402 (copyprop_hardreg_forward_1): Likewise.
4403 * reload.c (push_reload): Likewise.
4404 (find_reloads): Likewise.
4405 (find_reloads_subreg_address): Likewise.
4406 * reload1.c (alter_reg): Likewise.
4407 (eliminate_regs_1): Likewise.
4408 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4410 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4412 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4413 back to if statements, including unpack.
4415 2017-08-30 Martin Liska <mliska@suse.cz>
4418 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4420 (func_checker::compare_asm_inputs_outputs): ... this function.
4421 (func_checker::compare_gimple_asm): Use the function to compare
4422 also ASM constrains.
4423 * ipa-icf-gimple.h: Rename the function.
4425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4426 Alan Hayward <alan.hayward@arm.com>
4427 David Sherwood <david.sherwood@arm.com>
4429 * coretypes.h (complex_mode): New type.
4430 * gdbhooks.py (build_pretty_printer): Handle it.
4431 * machmode.h (complex_mode): New class.
4432 (complex_mode::includes_p): New function.
4433 (is_complex_int_mode): Likewise.
4434 (is_complex_float_mode): Likewise.
4435 * genmodes.c (get_mode_class): Handle complex mode classes.
4436 * function.c (expand_function_end): Use is_complex_int_mode.
4438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4439 Alan Hayward <alan.hayward@arm.com>
4440 David Sherwood <david.sherwood@arm.com>
4442 * coretypes.h (scalar_mode_pod): New typedef.
4443 * gdbhooks.py (build_pretty_printer): Handle it.
4444 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4445 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4446 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4447 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4451 Alan Hayward <alan.hayward@arm.com>
4452 David Sherwood <david.sherwood@arm.com>
4454 * machmode.h (mode_for_vector): Take a scalar_mode instead
4456 * stor-layout.c (mode_for_vector): Likewise.
4457 * explow.c (promote_mode): Use as_a <scalar_mode>.
4458 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4461 Alan Hayward <alan.hayward@arm.com>
4462 David Sherwood <david.sherwood@arm.com>
4464 * target.def (preferred_simd_mode): Take a scalar_mode
4465 instead of a machine_mode.
4466 * targhooks.h (default_preferred_simd_mode): Likewise.
4467 * targhooks.c (default_preferred_simd_mode): Likewise.
4468 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4469 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4470 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4471 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4472 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4473 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4474 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4475 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4477 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4478 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4479 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4480 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4481 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4482 * doc/tm.texi: Regenerate.
4483 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4487 Alan Hayward <alan.hayward@arm.com>
4488 David Sherwood <david.sherwood@arm.com>
4490 * target.def (scalar_mode_supported_p): Take a scalar_mode
4491 instead of a machine_mode.
4492 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4493 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4494 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4495 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4496 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4497 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4498 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4499 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4500 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4501 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4502 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4503 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4504 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4505 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4507 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4508 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4509 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4510 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4511 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4513 * doc/tm.texi: Regenerate.
4515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4516 Alan Hayward <alan.hayward@arm.com>
4517 David Sherwood <david.sherwood@arm.com>
4519 * coretypes.h (opt_scalar_mode): New typedef.
4520 * gdbhooks.py (build_pretty_printers): Handle it.
4521 * machmode.h (mode_iterator::get_2xwider): Add overload for
4523 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4525 * expr.c (convert_mode_scalar): Likewise.
4526 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4527 * optabs.c (expand_float): Likewise.
4528 (expand_fix): Likewise.
4529 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4532 Alan Hayward <alan.hayward@arm.com>
4533 David Sherwood <david.sherwood@arm.com>
4535 * optabs.c (expand_float): Explicitly check for scalars before
4536 using a branching expansion.
4537 (expand_fix): Likewise.
4539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4540 Alan Hayward <alan.hayward@arm.com>
4541 David Sherwood <david.sherwood@arm.com>
4543 * expr.c (convert_mode): Split scalar handling out into...
4544 (convert_mode_scalar): ...this new function. Treat the modes
4547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4548 Alan Hayward <alan.hayward@arm.com>
4549 David Sherwood <david.sherwood@arm.com>
4551 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4553 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4556 Alan Hayward <alan.hayward@arm.com>
4557 David Sherwood <david.sherwood@arm.com>
4559 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4560 rather than a machine_mode.
4561 (fixed_from_string): Likewise.
4562 (fixed_convert): Likewise.
4563 (fixed_convert_from_int): Likewise.
4564 (fixed_convert_from_real): Likewise.
4565 (real_convert_from_fixed): Likewise.
4566 * fixed-value.c (fixed_from_double_int): Likewise.
4567 (fixed_from_string): Likewise.
4568 (fixed_convert): Likewise.
4569 (fixed_convert_from_int): Likewise.
4570 (fixed_convert_from_real): Likewise.
4571 (real_convert_from_fixed): Likewise.
4572 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4575 Alan Hayward <alan.hayward@arm.com>
4576 David Sherwood <david.sherwood@arm.com>
4578 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4579 of separate mode class checks. Do not allow vector modes here.
4580 (immed_wide_int_const): Use as_a <scalar_mode>.
4581 * explow.c (trunc_int_for_mode): Likewise.
4582 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4583 (wi::shwi): Likewise.
4584 (wi::min_value): Likewise.
4585 (wi::max_value): Likewise.
4586 * dwarf2out.c (loc_descriptor): Likewise.
4587 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4591 Alan Hayward <alan.hayward@arm.com>
4592 David Sherwood <david.sherwood@arm.com>
4594 * tree.h (SCALAR_TYPE_MODE): New macro.
4595 * expr.c (expand_expr_addr_expr_1): Use it.
4596 (expand_expr_real_2): Likewise.
4597 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4598 (fold_convert_const_fixed_from_int): Likewise.
4599 (fold_convert_const_fixed_from_real): Likewise.
4600 (native_encode_fixed): Likewise
4601 (native_encode_complex): Likewise
4602 (native_encode_vector): Likewise.
4603 (native_interpret_fixed): Likewise.
4604 (native_interpret_real): Likewise.
4605 (native_interpret_complex): Likewise.
4606 (native_interpret_vector): Likewise.
4607 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4608 (simd_clone_adjust_argument_types): Likewise.
4609 (simd_clone_init_simd_arrays): Likewise.
4610 (simd_clone_adjust): Likewise.
4611 * stor-layout.c (layout_type): Likewise.
4612 * tree.c (build_minus_one_cst): Likewise.
4613 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4614 * tree-inline.c (estimate_move_cost): Likewise.
4615 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4616 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4617 (vectorizable_reduction): Likewise.
4618 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4619 (vect_recog_mixed_size_cond_pattern): Likewise.
4620 (check_bool_pattern): Likewise.
4621 (adjust_bool_pattern): Likewise.
4622 (search_type_for_mask_1): Likewise.
4623 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4624 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4625 (vectorizable_load): Likewise.
4626 (vectorizable_store): Likewise.
4627 * ubsan.c (ubsan_encode_value): Likewise.
4628 * varasm.c (output_constant): Likewise.
4630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4631 Alan Hayward <alan.hayward@arm.com>
4632 David Sherwood <david.sherwood@arm.com>
4634 * coretypes.h (scalar_mode): New class.
4635 * machmode.h (scalar_mode): Likewise.
4636 (scalar_mode::includes_p): New function.
4637 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4638 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4639 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4640 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4641 * expmed.c (store_bit_field_1): Likewise.
4642 (extract_bit_field_1): Likewise.
4643 * expr.c (write_complex_part): Likewise.
4644 (read_complex_part): Likewise.
4645 (emit_move_complex_push): Likewise.
4646 (expand_expr_real_2): Likewise.
4647 * function.c (assign_parm_setup_reg): Likewise.
4648 (assign_parms_unsplit_complex): Likewise.
4649 * optabs.c (expand_binop): Likewise.
4650 * rtlanal.c (subreg_get_info): Likewise.
4651 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4652 * varasm.c (output_constant_pool_2): Likewise.
4654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4655 Alan Hayward <alan.hayward@arm.com>
4656 David Sherwood <david.sherwood@arm.com>
4658 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4660 (expmed_mult_highpart_optab): Likewise.
4661 (expmed_mult_highpart): Likewise.
4663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4664 Alan Hayward <alan.hayward@arm.com>
4665 David Sherwood <david.sherwood@arm.com>
4667 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4668 instead of a machine_mode.
4669 (builtin_memset_read_str): Likewise.
4670 * builtins.c (c_readstr): Likewise.
4671 (builtin_memcpy_read_str): Likewise.
4672 (builtin_strncpy_read_str): Likewise.
4673 (builtin_memset_read_str): Likewise.
4674 (builtin_memset_gen_str): Likewise.
4675 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4676 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4677 instead of a machine_mode.
4678 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4680 (make_extraction): Likewise.
4681 (try_widen_shift_mode): Take and return scalar_int_modes instead
4683 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4684 a scalar_int_mode instead of a machine_mode.
4685 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4686 (avr_addr_space_pointer_mode): Likewise.
4687 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4688 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4689 (msp430_unwind_word_mode): Likewise.
4690 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4691 (spu_addr_space_pointer_mode): Likewise.
4692 (spu_addr_space_address_mode): Likewise.
4693 (spu_libgcc_cmp_return_mode): Likewise.
4694 (spu_libgcc_shift_count_mode): Likewise.
4695 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4696 (rl78_addr_space_pointer_mode): Likewise.
4697 (fl78_unwind_word_mode): Likewise.
4698 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4700 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4701 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4702 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4703 (mips_valid_pointer_mode): Likewise.
4704 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4705 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4706 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4708 (ft32_addr_space_address_mode): Likewise.
4709 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4710 scalar_int_mode instead of a machine_mode.
4711 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4713 (m32c_addr_space_address_mode): Likewise.
4714 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4715 (rs6000_eh_return_filter_mode): Likewise.
4716 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4717 (rs6000_eh_return_filter_mode): Likewise.
4718 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4719 (s390_libgcc_shift_count_mode): Likewise.
4720 (s390_unwind_word_mode): Likewise.
4721 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4723 * target.def (mode_rep_extended): Likewise.
4724 (valid_pointer_mode): Likewise.
4725 (addr_space.valid_pointer_mode): Likewise.
4726 (eh_return_filter_mode): Return a scalar_int_mode rather than
4728 (libgcc_cmp_return_mode): Likewise.
4729 (libgcc_shift_count_mode): Likewise.
4730 (unwind_word_mode): Likewise.
4731 (addr_space.pointer_mode): Likewise.
4732 (addr_space.address_mode): Likewise.
4733 * doc/tm.texi: Regenerate.
4734 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4736 (do_jump): Use scalar_int_mode for local variables.
4737 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4738 rather than a machine_mode.
4739 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4740 (scompare_loc_descriptor_wide): Likewise.
4741 (scompare_loc_descriptor_narrow): Likewise.
4742 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4744 * except.c (sjlj_emit_dispatch_table): Likewise.
4745 (expand_builtin_eh_copy_values): Likewise.
4746 * explow.c (convert_memory_address_addr_space_1): Likewise.
4747 Take a scalar_int_mode rather than a machine_mode.
4748 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4749 than a machine_mode.
4750 (memory_address_addr_space): Use scalar_int_mode for local variables.
4751 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4752 rather than a machine_mode.
4753 * expmed.c (mask_rtx): Likewise.
4754 (init_expmed_one_conv): Likewise.
4755 (expand_mult_highpart_adjust): Likewise.
4756 (extract_high_half): Likewise.
4757 (expmed_mult_highpart_optab): Likewise.
4758 (expmed_mult_highpart): Likewise.
4759 (expand_smod_pow2): Likewise.
4760 (expand_sdiv_pow2): Likewise.
4761 (emit_store_flag_int): Likewise.
4762 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4764 (extract_low_bits): Likewise.
4765 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
4767 * expr.c (pieces_addr::adjust): Likewise.
4768 (can_store_by_pieces): Likewise.
4769 (store_by_pieces): Likewise.
4770 (clear_by_pieces_1): Likewise.
4771 (expand_expr_addr_expr_1): Likewise.
4772 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
4773 (expand_expr_real_1): Likewise.
4774 (try_casesi): Likewise.
4775 * final.c (shorten_branches): Likewise.
4776 * fold-const.c (fold_convert_const_int_from_fixed): Change the
4777 type of "mode" to machine_mode.
4778 * internal-fn.c (expand_arith_overflow_result_store): Take a
4779 scalar_int_mode rather than a machine_mode.
4780 (expand_mul_overflow): Use scalar_int_mode for local variables.
4781 * loop-doloop.c (doloop_modify): Likewise.
4782 (doloop_optimize): Likewise.
4783 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
4784 than a machine_mode.
4785 (expand_doubleword_shift_condmove): Likewise.
4786 (expand_doubleword_shift): Likewise.
4787 (expand_doubleword_clz): Likewise.
4788 (expand_doubleword_popcount): Likewise.
4789 (expand_doubleword_parity): Likewise.
4790 (expand_absneg_bit): Use scalar_int_mode for local variables.
4791 (prepare_float_lib_cmp): Likewise.
4792 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4793 rather than a machine_mode.
4794 (convert_memory_address_addr_space): Likewise.
4795 (get_mode_bounds): Likewise.
4796 (get_address_mode): Return a scalar_int_mode rather than a
4798 * rtlanal.c (get_address_mode): Likewise.
4799 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4800 than a machine_mode.
4801 * targhooks.c (default_mode_rep_extended): Likewise.
4802 (default_valid_pointer_mode): Likewise.
4803 (default_addr_space_valid_pointer_mode): Likewise.
4804 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4805 than a machine_mode.
4806 (default_libgcc_cmp_return_mode): Likewise.
4807 (default_libgcc_shift_count_mode): Likewise.
4808 (default_unwind_word_mode): Likewise.
4809 (default_addr_space_pointer_mode): Likewise.
4810 (default_addr_space_address_mode): Likewise.
4811 * targhooks.h (default_eh_return_filter_mode): Likewise.
4812 (default_libgcc_cmp_return_mode): Likewise.
4813 (default_libgcc_shift_count_mode): Likewise.
4814 (default_unwind_word_mode): Likewise.
4815 (default_addr_space_pointer_mode): Likewise.
4816 (default_addr_space_address_mode): Likewise.
4817 (default_mode_rep_extended): Take a scalar_int_mode rather than
4819 (default_valid_pointer_mode): Likewise.
4820 (default_addr_space_valid_pointer_mode): Likewise.
4821 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4823 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4824 rather than a machine_mode.
4825 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4826 for local variables.
4827 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4828 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4829 than a machine_mode.
4831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4832 Alan Hayward <alan.hayward@arm.com>
4833 David Sherwood <david.sherwood@arm.com>
4835 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4836 the mode argument to scalar_int_mode.
4837 (do_jump_by_parts_zero_rtx): Likewise.
4838 (do_jump_by_parts_equality_rtx): Likewise.
4839 (do_jump_by_parts_greater): Take a mode argument.
4840 (do_jump_by_parts_equality): Likewise.
4841 (do_jump_1): Update calls accordingly.
4843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4844 Alan Hayward <alan.hayward@arm.com>
4845 David Sherwood <david.sherwood@arm.com>
4847 * is-a.h (safe_dyn_cast): New function.
4848 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4849 (jump_table_for_label): Likewise.
4850 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4851 instead of an rtx_insn *.
4852 (shorten_branches): Use dyn_cast instead of LABEL_P and
4853 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4854 rtx_jump_table_data::get_data_mode.
4855 (final_scan_insn): Likewise.
4857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4858 Alan Hayward <alan.hayward@arm.com>
4859 David Sherwood <david.sherwood@arm.com>
4861 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4862 trying to combine a full-register integer set with a subreg
4865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4866 Alan Hayward <alan.hayward@arm.com>
4867 David Sherwood <david.sherwood@arm.com>
4869 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4870 that is always either address_mode or pointer_mode.
4872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4873 Alan Hayward <alan.hayward@arm.com>
4874 David Sherwood <david.sherwood@arm.com>
4876 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4877 when the two are known to be equal.
4879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4880 Alan Hayward <alan.hayward@arm.com>
4881 David Sherwood <david.sherwood@arm.com>
4883 * simplify-rtx.c (simplify_const_unary_operation): Use
4884 is_a <scalar_int_mode> instead of checking for a nonzero
4885 precision. Forcibly convert op_mode to a scalar_int_mode
4886 in that case. More clearly differentiate the operand and
4887 result modes and use the former when deciding what the value
4888 of a count-bits operation should be. Use is_int_mode instead
4889 of checking for a MODE_INT. Remove redundant check for whether
4890 this mode has a zero precision.
4892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4893 Alan Hayward <alan.hayward@arm.com>
4894 David Sherwood <david.sherwood@arm.com>
4896 * optabs.c (widen_leading): Change the type of the mode argument
4897 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4898 (widen_bswap): Likewise.
4899 (expand_parity): Likewise.
4900 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4901 (expand_ffs): Likewise.
4902 (epand_unop): Check for scalar integer modes before calling the
4905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4906 Alan Hayward <alan.hayward@arm.com>
4907 David Sherwood <david.sherwood@arm.com>
4909 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4911 (expand_expr_real_1): Update call accordingly.
4913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4914 Alan Hayward <alan.hayward@arm.com>
4915 David Sherwood <david.sherwood@arm.com>
4917 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4918 value_mode arguments. Use scalar_int_mode internally.
4919 (store_bit_field_1): Rename the new integer mode from imode
4920 to op0_mode and use it instead of GET_MODE (op0). Update calls
4921 to store_split_bit_field, store_bit_field_using_insv and
4922 store_fixed_bit_field.
4923 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4924 Use scalar_int_mode internally. Use a bit count rather than a mode
4925 when calculating the largest bit size for get_best_mode.
4926 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4927 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4928 Remove assertion that OP0 has a scalar integer mode.
4929 (store_split_bit_field): Add op0_mode and value_mode arguments.
4930 Update calls to extract_fixed_bit_field.
4931 (extract_bit_field_using_extv): Add an op0_mode argument.
4932 Use scalar_int_mode internally.
4933 (extract_bit_field_1): Rename the new integer mode from imode to
4934 op0_mode and use it instead of GET_MODE (op0). Update calls to
4935 extract_split_bit_field, extract_bit_field_using_extv and
4936 extract_fixed_bit_field.
4937 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4938 to extract_split_bit_field and extract_fixed_bit_field_1.
4939 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4940 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4942 (extract_split_bit_field): Add an op0_mode argument. Update call
4943 to extract_fixed_bit_field.
4945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4946 Alan Hayward <alan.hayward@arm.com>
4947 David Sherwood <david.sherwood@arm.com>
4949 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4950 * explow.c (hard_function_value): Likewise.
4951 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4952 convert_to_mode call outside the loop.
4953 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4954 for the mode iterator. Require the mode specified by max_pieces
4956 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4958 (copy_blkmode_to_reg): Likewise.
4959 (set_storage_via_setmem): Likewise.
4960 * optabs.c (prepare_cmp_insn): Likewise.
4961 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4962 * stor-layout.c (finish_bitfield_representative): Likewise.
4964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4965 Alan Hayward <alan.hayward@arm.com>
4966 David Sherwood <david.sherwood@arm.com>
4968 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4969 * expr.c (convert_move): Use them.
4970 (convert_modes): Likewise.
4971 (store_expr_with_bounds): Likewise.
4973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4974 Alan Hayward <alan.hayward@arm.com>
4975 David Sherwood <david.sherwood@arm.com>
4977 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4978 parameter for the mode of "x". Remove the "known_x", "known_mode"
4979 and "known_ret" arguments. Change the type of the mode argument
4981 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4982 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4983 (reg_num_sign_bit_copies_for_combine): Likewise.
4984 * rtlanal.c (nonzero_bits1): Likewise.
4985 (num_sign_bit_copies1): Likewise.
4986 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4987 (reg_num_sign_bit_copies_general): Likewise.
4988 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4989 (reg_nonzero_bits_general): Likewise.
4991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4992 Alan Hayward <alan.hayward@arm.com>
4993 David Sherwood <david.sherwood@arm.com>
4995 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4996 than in subroutines. Return 1 for non-integer modes.
4997 (cached_num_sign_bit_copies): Change the type of the mode parameter
4999 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5000 classes. Handle CONST_INT_P first and then check whether X also
5001 has a scalar integer mode. Check the same thing for inner registers
5002 of a SUBREG and for values that are being extended or truncated.
5004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5005 Alan Hayward <alan.hayward@arm.com>
5006 David Sherwood <david.sherwood@arm.com>
5008 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5009 in subroutines. Return the mode mask for non-integer modes.
5010 (cached_nonzero_bits): Change the type of the mode parameter
5012 (nonzero_bits1): Likewise. Remove early exit for other mode
5013 classes. Handle CONST_INT_P first and then check whether X
5014 also has a scalar integer mode.
5016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5017 Alan Hayward <alan.hayward@arm.com>
5018 David Sherwood <david.sherwood@arm.com>
5020 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5021 Return a scalar_int_mode and assert that the size is greater than
5023 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5024 (op_by_pieces_d::op_by_pieces_d): Likewise.
5025 (op_by_pieces_d::run): Likewise.
5026 (can_store_by_pieces): Likewise.
5028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5029 Alan Hayward <alan.hayward@arm.com>
5030 David Sherwood <david.sherwood@arm.com>
5032 * combine.c (extract_left_shift): Add a mode argument and update
5034 (make_compound_operation_int): Change the type of the mode parameter
5035 to scalar_int_mode and update the call to extract_left_shift.
5037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5038 Alan Hayward <alan.hayward@arm.com>
5039 David Sherwood <david.sherwood@arm.com>
5041 * combine.c (simplify_and_const_int): Change the type of the mode
5042 parameter to scalar_int_mode.
5043 (simplify_and_const_int_1): Likewise. Update recursive call.
5045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5046 Alan Hayward <alan.hayward@arm.com>
5047 David Sherwood <david.sherwood@arm.com>
5049 * combine.c (simplify_compare_const): Check that the mode is a
5050 scalar_int_mode (rather than VOIDmode) before testing its
5052 (simplify_comparison): Move COMPARISON_P handling out of the
5053 loop and restrict the latter part of the loop to scalar_int_modes.
5054 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5055 and when considering SUBREG_REGs. Use is_int_mode instead of
5056 checking GET_MODE_CLASS against MODE_INT.
5058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5059 Alan Hayward <alan.hayward@arm.com>
5060 David Sherwood <david.sherwood@arm.com>
5062 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5063 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5064 shift_unit_mode and for modes involved in scalar shifts.
5066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5067 Alan Hayward <alan.hayward@arm.com>
5068 David Sherwood <david.sherwood@arm.com>
5070 * combine.c (force_int_to_mode): New function, split out from...
5071 (force_to_mode): ...here. Keep xmode up-to-date and use it
5072 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
5079 opt_scalar_int_mode and update comment.
5080 (extraction_insn::field_mode): Change type to scalar_int_mode.
5081 (extraction_insn::pos_mode): Likewise.
5082 * combine.c (make_extraction): Update accordingly.
5083 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5084 (get_optab_extraction_insn): Likewise.
5085 * recog.c (simplify_while_replacing): Likewise.
5086 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5087 parameter to opt_scalar_int_mode.
5089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5090 Alan Hayward <alan.hayward@arm.com>
5091 David Sherwood <david.sherwood@arm.com>
5093 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5094 to a scalar_int_mode instead of a machine_mode.
5095 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5096 (get_best_mode): Return a boolean and use a pointer argument to store
5097 the selected mode. Replace the limit mode parameter with a bit limit.
5098 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5099 for the values returned by bit_field_mode_iterator::next_mode.
5100 (store_bit_field): Update call to get_best_mode.
5101 (store_fixed_bit_field): Likewise.
5102 (extract_fixed_bit_field): Likewise.
5103 * expr.c (optimize_bitfield_assignment_op): Likewise.
5104 * fold-const.c (optimize_bit_field_compare): Likewise.
5105 (fold_truth_andor_1): Likewise.
5106 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5107 Update for new type of m_mode.
5108 (get_best_mode): As above.
5110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5111 Alan Hayward <alan.hayward@arm.com>
5112 David Sherwood <david.sherwood@arm.com>
5114 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5115 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5116 (store_bit_field): Check is_a <scalar_int_mode> before calling
5117 strict_volatile_bitfield_p.
5118 (extract_bit_field): 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 * target.def (cstore_mode): Return a scalar_int_mode.
5125 * doc/tm.texi: Regenerate.
5126 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5127 * targhooks.h (default_cstore_mode): Likewise.
5128 * targhooks.c (default_cstore_mode): Likewise, using a forced
5130 * expmed.c (emit_cstore): Expect the target of the cstore to be
5133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5134 Alan Hayward <alan.hayward@arm.com>
5135 David Sherwood <david.sherwood@arm.com>
5137 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5139 (niter_desc): Likewise mode.
5140 (iv_analyze): Add a mode parameter.
5142 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5143 and change its type to scalar_int_mode.
5144 * loop-iv.c: Update commentary at head of file.
5145 (iv_constant): Pass the mode paraeter before the rtx it describes
5146 and change its type to scalar_int_mode. Remove VOIDmode handling.
5147 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5148 (iv_extend): Likewise.
5149 (shorten_into_mode): Likewise.
5150 (iv_add): Use scalar_int_mode.
5151 (iv_mult): Likewise.
5152 (iv_shift): Likewise.
5153 (canonicalize_iv_subregs): Likewise.
5154 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5155 it describes and change its mode to scalar_int_mode. Also change
5156 the type of the returned inner_mode to scalar_int_mode.
5157 (get_biv_step): Likewise, turning outer_mode from a pointer
5158 into a direct parameter. Update call to get_biv_step_1.
5159 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5160 iv_constant and get_biv_step.
5161 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5162 and change its type to scalar_int_mode. Don't initialise iv->mode
5163 to VOIDmode and remove later checks for its still being VOIDmode.
5164 Update calls to iv_analyze_op and iv_analyze_expr. Check
5165 is_a <scalar_int_mode> when changing the mode under consideration.
5166 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5167 Update call to iv_analyze_expr.
5168 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5169 inner register is not also a scalar_int_mode. Update call to
5171 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5172 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5173 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5174 separate mode class checks. Update calls to iv_analyze. Remove
5175 fix-up of VOIDmodes after iv_analyze_biv.
5176 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5177 don't have a scalar_int_mode. Update call to biv_p.
5179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5180 Alan Hayward <alan.hayward@arm.com>
5181 David Sherwood <david.sherwood@arm.com>
5183 * cfgexpand.c (convert_debug_memory_address): Use
5184 as_a <scalar_int_mode>.
5185 * combine.c (expand_compound_operation): Likewise.
5186 (make_extraction): Likewise.
5187 (change_zero_ext): Likewise.
5188 (simplify_comparison): Likewise.
5189 * cse.c (cse_insn): Likewise.
5190 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5191 (mem_loc_descriptor): Likewise.
5192 (loc_descriptor): Likewise.
5193 * expmed.c (init_expmed_one_mode): Likewise.
5194 (synth_mult): Likewise.
5195 (emit_store_flag_1): Likewise.
5196 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5197 of a comparison with size.
5198 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5199 (reduce_to_bit_field_precision): Likewise.
5200 * function.c (expand_function_end): Likewise.
5201 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5202 * loop-doloop.c (doloop_modify): Likewise.
5203 * optabs.c (expand_binop): Likewise.
5204 (expand_unop): Likewise.
5205 (expand_copysign_absneg): Likewise.
5206 (prepare_cmp_insn): Likewise.
5207 (maybe_legitimize_operand): Likewise.
5208 * recog.c (const_scalar_int_operand): Likewise.
5209 * rtlanal.c (get_address_mode): Likewise.
5210 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5211 (simplify_cond_clz_ctz): Likewise.
5212 * tree-nested.c (get_nl_goto_field): Likewise.
5213 * tree.c (build_vector_type_for_mode): Likewise.
5214 * var-tracking.c (use_narrower_mode): Likewise.
5216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5217 Alan Hayward <alan.hayward@arm.com>
5218 David Sherwood <david.sherwood@arm.com>
5220 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5221 * builtins.c (expand_builtin_signbit): Use it.
5222 * cfgexpand.c (expand_debug_expr): Likewise.
5223 * dojump.c (do_jump): Likewise.
5224 (do_compare_and_jump): Likewise.
5225 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5226 * expmed.c (make_tree): Likewise.
5227 * expr.c (expand_expr_real_2): Likewise.
5228 (expand_expr_real_1): Likewise.
5229 (try_casesi): Likewise.
5230 * fold-const-call.c (fold_const_call_ss): Likewise.
5231 * fold-const.c (unextend): Likewise.
5232 (extract_muldiv_1): Likewise.
5233 (fold_single_bit_test): Likewise.
5234 (native_encode_int): Likewise.
5235 (native_encode_string): Likewise.
5236 (native_interpret_int): Likewise.
5237 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5238 * internal-fn.c (expand_addsub_overflow): Likewise.
5239 (expand_neg_overflow): Likewise.
5240 (expand_mul_overflow): Likewise.
5241 (expand_arith_overflow): Likewise.
5242 * match.pd: Likewise.
5243 * stor-layout.c (layout_type): Likewise.
5244 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5245 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5246 * tree-ssanames.c (get_range_info): Likewise.
5247 * tree-switch-conversion.c (array_value_type) Likewise.
5248 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5249 (vect_recog_divmod_pattern): Likewise.
5250 (vect_recog_mixed_size_cond_pattern): Likewise.
5251 * tree-vrp.c (extract_range_basic): Likewise.
5252 (simplify_float_conversion_using_ranges): Likewise.
5253 * tree.c (int_fits_type_p): Likewise.
5254 * ubsan.c (instrument_bool_enum_load): Likewise.
5255 * varasm.c (mergeable_string_section): Likewise.
5256 (narrowing_initializer_constant_valid_p): Likewise.
5257 (output_constant): 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 (NARROWEST_INT_MODE): New macro.
5264 * expr.c (alignment_for_piecewise_move): Use it instead of
5265 GET_CLASS_NARROWEST_MODE (MODE_INT).
5266 (push_block): Likewise.
5267 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5269 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5272 Alan Hayward <alan.hayward@arm.com>
5273 David Sherwood <david.sherwood@arm.com>
5275 * postreload.c (move2add_valid_value_p): Change the type of the
5276 mode parameter to scalar_int_mode.
5277 (move2add_use_add2_insn): Add a mode parameter and use it instead
5279 (move2add_use_add3_insn): Likewise.
5280 (reload_cse_move2add): Update accordingly.
5282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5283 Alan Hayward <alan.hayward@arm.com>
5284 David Sherwood <david.sherwood@arm.com>
5286 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5288 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5289 * optabs.c (expand_unop): Likewise.
5291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5296 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5297 (popcount_loc_descriptor): Likewise.
5298 (bswap_loc_descriptor): Likewise.
5299 (rotate_loc_descriptor): Likewise.
5300 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5301 calling the functions above.
5303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5304 Alan Hayward <alan.hayward@arm.com>
5305 David Sherwood <david.sherwood@arm.com>
5307 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5309 (try_combine): Likewise.
5310 (simplify_if_then_else): Likewise.
5311 * cse.c (cse_insn): Likewise.
5312 * dwarf2out.c (mem_loc_descriptor): Likewise.
5313 * emit-rtl.c (gen_lowpart_common): Likewise.
5314 * simplify-rtx.c (simplify_truncation): Likewise.
5315 (simplify_binary_operation_1): Likewise.
5316 (simplify_const_relational_operation): Likewise.
5317 (simplify_ternary_operation): Likewise.
5318 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5321 Alan Hayward <alan.hayward@arm.com>
5322 David Sherwood <david.sherwood@arm.com>
5324 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5325 * reload.c (push_reload): Likewise.
5326 (find_reloads): Likewise.
5328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5329 Alan Hayward <alan.hayward@arm.com>
5330 David Sherwood <david.sherwood@arm.com>
5332 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5333 (make_compound_operation_int): Likewise.
5334 (change_zero_ext): Likewise.
5335 * expr.c (convert_move): Likewise.
5336 (convert_modes): Likewise.
5337 * fwprop.c (forward_propagate_subreg): Likewise.
5338 * loop-iv.c (get_biv_step_1): Likewise.
5339 * optabs.c (widen_operand): Likewise.
5340 * postreload.c (move2add_valid_value_p): Likewise.
5341 * recog.c (simplify_while_replacing): Likewise.
5342 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5343 (simplify_binary_operation_1): Likewise. Remove redundant
5344 mode equality check.
5346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5347 Alan Hayward <alan.hayward@arm.com>
5348 David Sherwood <david.sherwood@arm.com>
5350 * combine.c (combine_simplify_rtx): Add checks for
5351 is_a <scalar_int_mode>.
5352 (simplify_if_then_else): Likewise.
5353 (make_field_assignment): Likewise.
5354 (simplify_comparison): Likewise.
5355 * ifcvt.c (noce_try_bitop): Likewise.
5356 * loop-invariant.c (canonicalize_address_mult): Likewise.
5357 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5360 Alan Hayward <alan.hayward@arm.com>
5361 David Sherwood <david.sherwood@arm.com>
5363 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5364 is_a <scalar_int_mode> instead of != BLKmode.
5366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5367 Alan Hayward <alan.hayward@arm.com>
5368 David Sherwood <david.sherwood@arm.com>
5370 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5371 instead of != VOIDmode.
5372 * combine.c (if_then_else_cond): Likewise.
5373 (change_zero_ext): Likewise.
5374 * dwarf2out.c (mem_loc_descriptor): Likewise.
5375 (loc_descriptor): Likewise.
5376 * rtlanal.c (canonicalize_condition): Likewise.
5377 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5380 Alan Hayward <alan.hayward@arm.com>
5381 David Sherwood <david.sherwood@arm.com>
5383 * simplify-rtx.c (simplify_binary_operation_1): Use
5384 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5387 Alan Hayward <alan.hayward@arm.com>
5388 David Sherwood <david.sherwood@arm.com>
5390 * wide-int.h (int_traits<unsigned char>) New class.
5391 (int_traits<unsigned short>) Likewise.
5392 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5393 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5395 * combine.c (set_nonzero_bits_and_sign_copies): Use
5396 is_a <scalar_int_mode>.
5397 (find_split_point): Likewise.
5398 (combine_simplify_rtx): Likewise.
5399 (simplify_logical): Likewise.
5400 (expand_compound_operation): Likewise.
5401 (expand_field_assignment): Likewise.
5402 (make_compound_operation): Likewise.
5403 (extended_count): Likewise.
5404 (change_zero_ext): Likewise.
5405 (simplify_comparison): Likewise.
5406 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5407 (ucompare_loc_descriptor): Likewise.
5408 (minmax_loc_descriptor): Likewise.
5409 (mem_loc_descriptor): Likewise.
5410 (loc_descriptor): Likewise.
5411 * expmed.c (init_expmed_one_mode): Likewise.
5412 * lra-constraints.c (lra_constraint_offset): Likewise.
5413 * optabs.c (prepare_libcall_arg): Likewise.
5414 * postreload.c (move2add_note_store): Likewise.
5415 * reload.c (operands_match_p): Likewise.
5416 * rtl.h (load_extend_op): Likewise.
5417 * rtlhooks.c (gen_lowpart_general): Likewise.
5418 * simplify-rtx.c (simplify_truncation): Likewise.
5419 (simplify_unary_operation_1): Likewise.
5420 (simplify_binary_operation_1): Likewise.
5421 (simplify_const_binary_operation): Likewise.
5422 (simplify_const_relational_operation): Likewise.
5423 (simplify_subreg): Likewise.
5424 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5425 * var-tracking.c (adjust_mems): Likewise.
5426 (prepare_call_arguments): Likewise.
5428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5429 Alan Hayward <alan.hayward@arm.com>
5430 David Sherwood <david.sherwood@arm.com>
5432 * machmode.h (is_int_mode): New fuction.
5433 * combine.c (find_split_point): Use it.
5434 (combine_simplify_rtx): Likewise.
5435 (simplify_if_then_else): Likewise.
5436 (simplify_set): Likewise.
5437 (simplify_shift_const_1): Likewise.
5438 (simplify_comparison): Likewise.
5439 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5440 * cse.c (notreg_cost): Likewise.
5441 (cse_insn): Likewise.
5442 * cselib.c (cselib_lookup_1): Likewise.
5443 * dojump.c (do_jump_1): Likewise.
5444 (do_compare_rtx_and_jump): Likewise.
5445 * dse.c (get_call_args): Likewise.
5446 * dwarf2out.c (rtl_for_decl_init): Likewise.
5447 (native_encode_initializer): Likewise.
5448 * expmed.c (emit_store_flag_1): Likewise.
5449 (emit_store_flag): Likewise.
5450 * expr.c (convert_modes): Likewise.
5451 (store_field): Likewise.
5452 (expand_expr_real_1): Likewise.
5453 * fold-const.c (fold_read_from_constant_string): Likewise.
5454 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5455 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5456 * optabs.c (expand_binop): Likewise.
5457 (expand_unop): Likewise.
5458 (expand_abs_nojump): Likewise.
5459 (expand_one_cmpl_abs_nojump): Likewise.
5460 * simplify-rtx.c (mode_signbit_p): Likewise.
5461 (val_signbit_p): Likewise.
5462 (val_signbit_known_set_p): Likewise.
5463 (val_signbit_known_clear_p): Likewise.
5464 (simplify_relational_operation_1): Likewise.
5465 * tree.c (vector_type_mode): Likewise.
5467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5468 Alan Hayward <alan.hayward@arm.com>
5469 David Sherwood <david.sherwood@arm.com>
5471 * machmode.h (smallest_mode_for_size): Fix formatting.
5472 (smallest_int_mode_for_size): New function.
5473 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5474 instead of smallest_mode_for_size.
5475 * combine.c (make_extraction): Likewise.
5476 * config/arc/arc.c (arc_expand_movmem): Likewise.
5477 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5478 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5479 * config/s390/s390.c (s390_expand_insv): Likewise.
5480 * config/sparc/sparc.c (assign_int_registers): Likewise.
5481 * config/spu/spu.c (spu_function_value): Likewise.
5482 (spu_function_arg): Likewise.
5483 * coverage.c (get_gcov_type): Likewise.
5484 (get_gcov_unsigned_t): Likewise.
5485 * dse.c (find_shift_sequence): Likewise.
5486 * expmed.c (store_bit_field_1): Likewise.
5487 * expr.c (convert_move): Likewise.
5488 (store_field): Likewise.
5489 * internal-fn.c (expand_arith_overflow): Likewise.
5490 * optabs-query.c (get_best_extraction_insn): Likewise.
5491 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5492 * stor-layout.c (layout_type): Likewise.
5493 (initialize_sizetypes): Likewise.
5494 * targhooks.c (default_get_mask_mode): Likewise.
5495 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5498 Alan Hayward <alan.hayward@arm.com>
5499 David Sherwood <david.sherwood@arm.com>
5501 * machmode.h (opt_mode::else_blk): New function.
5502 (int_mode_for_mode): Declare.
5503 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5504 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5506 * cfgexpand.c (expand_debug_expr): Likewise.
5507 * combine.c (gen_lowpart_or_truncate): Likewise.
5508 (gen_lowpart_for_combine): Likewise.
5509 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5510 * config/avr/avr.c (avr_to_int_mode): Likewise.
5511 (avr_out_plus_1): Likewise.
5512 (avr_out_plus): Likewise.
5513 (avr_out_round): Likewise.
5514 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5515 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5516 (s390_expand_vcond): Likewise.
5517 * config/spu/spu.c (spu_split_immediate): Likewise.
5518 (spu_expand_mov): Likewise.
5519 * dse.c (get_stored_val): Likewise.
5520 * expmed.c (store_bit_field_1): Likewise.
5521 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5523 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5524 (extract_low_bits): Likewise.
5525 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5526 handling rather than repeating the check.
5527 (emit_group_store): Likewise.
5528 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5529 * optabs.c (expand_absneg_bit): Likewise.
5530 (expand_copysign_absneg): Likewise.
5531 (expand_copysign_bit): Likewise.
5532 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5533 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5534 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5535 * var-tracking.c (prepare_call_arguments): Likewise.
5536 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5537 int_mode_for_mode instead of mode_for_size.
5538 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5541 Alan Hayward <alan.hayward@arm.com>
5542 David Sherwood <david.sherwood@arm.com>
5544 * machmode.h (int_mode_for_size): New function.
5545 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5546 instead of mode_for_size.
5547 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5549 * combine.c (expand_field_assignment): Use int_mode_for_size
5550 instead of mode_for_size.
5551 (make_extraction): Likewise.
5552 (simplify_shift_const_1): Likewise.
5553 (simplify_comparison): Likewise.
5554 * dojump.c (do_jump): Likewise.
5555 * dwarf2out.c (mem_loc_descriptor): Likewise.
5556 * emit-rtl.c (init_derived_machine_modes): Likewise.
5557 * expmed.c (flip_storage_order): Likewise.
5558 (convert_extracted_bit_field): Likewise.
5559 * expr.c (copy_blkmode_from_reg): Likewise.
5560 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5561 * internal-fn.c (expand_mul_overflow): Likewise.
5562 * lower-subreg.c (simple_move): Likewise.
5563 * optabs-libfuncs.c (init_optabs): Likewise.
5564 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5565 * tree.c (vector_type_mode): Likewise.
5566 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5567 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5568 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5569 * tree-vect-stmts.c (vectorizable_load): Likewise.
5570 (vectorizable_store): Likewise.
5572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5573 Alan Hayward <alan.hayward@arm.com>
5574 David Sherwood <david.sherwood@arm.com>
5576 * coretypes.h (pod_mode): New type.
5577 (scalar_int_mode_pod): New typedef.
5578 * machmode.h (pod_mode): New class.
5579 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5580 * genmodes.c (emit_mode_int_n): Update accordingly.
5581 * lower-subreg.h (target_lower_subreg): Change type to
5582 scalar_int_mode_pod.
5583 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5584 scalar_int_mode_pod.
5586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5587 Alan Hayward <alan.hayward@arm.com>
5588 David Sherwood <david.sherwood@arm.com>
5590 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5591 machine_mode to scalar_int_mode.
5592 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5593 (rs6000_option_override_internal): Remove cast to int.
5594 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5595 machine_mode to scalar_int_mode.
5596 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5597 (rs6000_option_override_internal): Remove cast to int.
5598 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5599 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5601 * config/s390/s390.c (s390_expand_builtin): Likewise.
5602 * coretypes.h (scalar_int_mode): New type.
5603 (opt_scalar_int_mode): New typedef.
5604 * machmode.h (scalar_int_mode): New class.
5605 (scalar_int_mode::includes_p): New function.
5606 (byte_mode): Change type to scalar_int_mode.
5607 (word_mode): Likewise.
5608 (ptr_mode): Likewise.
5609 * emit-rtl.c (byte_mode): Likewise.
5610 (word_mode): Likewise.
5611 (ptr_mode): Likewise.
5612 (init_derived_machine_modes): Update accordingly.
5613 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5614 and MODE_PARTIAL_INT.
5615 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5616 opt_scalar_int_mode.
5618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5619 Alan Hayward <alan.hayward@arm.com>
5620 David Sherwood <david.sherwood@arm.com>
5622 * target.def (libgcc_floating_mode_supported_p): Take a
5624 * doc/tm.texi: Regenerate.
5625 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5627 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5628 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5632 Alan Hayward <alan.hayward@arm.com>
5633 David Sherwood <david.sherwood@arm.com>
5635 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5636 * doc/tm.texi: Regenerate.
5637 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5638 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5639 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5640 * targhooks.h (default_floatn_mode): Likewise.
5641 * targhooks.c (default_floatn_mode): Likewise.
5642 * tree.c (build_common_tree_nodes): Update accordingly.
5644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5645 Alan Hayward <alan.hayward@arm.com>
5646 David Sherwood <david.sherwood@arm.com>
5648 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5649 (mode_iterator::iterate_p): Likewise.
5650 (mode_iterator::get_wider): Likewise.
5651 * expr.c (init_expr_target): Use opt_scalar_float_mode.
5653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5654 Alan Hayward <alan.hayward@arm.com>
5655 David Sherwood <david.sherwood@arm.com>
5657 * coretypes.h (opt_scalar_float_mode): New typedef.
5658 * machmode.h (float_mode_for_size): New function.
5659 * emit-rtl.c (double_mode): Delete.
5660 (init_emit_once): Use float_mode_for_size.
5661 * stor-layout.c (layout_type): Likewise.
5662 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5665 Alan Hayward <alan.hayward@arm.com>
5666 David Sherwood <david.sherwood@arm.com>
5668 * output.h (assemble_real): Take a scalar_float_mode.
5669 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5670 * config/arm/arm.md (consttable_4): Likewise.
5671 (consttable_8): Likewise.
5672 (consttable_16): Likewise.
5673 * config/mips/mips.md (consttable_float): Likewise.
5674 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5675 * varasm.c (assemble_real): Take a scalar_float_mode.
5676 (output_constant_pool_2): Update accordingly.
5677 (output_constant): Likewise.
5679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5680 Alan Hayward <alan.hayward@arm.com>
5681 David Sherwood <david.sherwood@arm.com>
5683 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5684 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5685 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5686 (native_encode_real): Likewise.
5687 (native_interpret_real): Likewise.
5688 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5689 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5692 Alan Hayward <alan.hayward@arm.com>
5693 David Sherwood <david.sherwood@arm.com>
5695 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5696 <scalar_float_mode>. Simplify.
5697 (gen_extend_conv_libfunc): Likewise.
5699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5700 Alan Hayward <alan.hayward@arm.com>
5701 David Sherwood <david.sherwood@arm.com>
5703 * coretypes.h (scalar_float_mode): New type.
5704 * machmode.h (mode_traits::from_int): Use machine_mode if
5705 USE_ENUM_MODES is defined.
5706 (is_a): New function.
5708 (dyn_cast): Likewise.
5709 (scalar_float_mode): New class.
5710 (scalar_float_mode::includes_p): New function.
5711 (is_float_mode): Likewise.
5712 * gdbhooks.py (MachineModePrinter): New class.
5713 (build_pretty_printer): Use it for scalar_float_mode.
5714 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5715 (format_helper::format_helper): Turn into a template.
5716 * genmodes.c (get_mode_class): New function.
5717 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5718 or machine_mode if none.
5719 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5720 as_a <scalar_float_mode>.
5721 * dwarf2out.c (mem_loc_descriptor): Likewise.
5722 (insert_float): Likewise.
5723 (add_const_value_attribute): Likewise.
5724 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5725 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5726 (expand_unop): Update accordingly.
5727 (expand_abs_nojump): Likewise.
5728 (expand_copysign_absneg): Take a scalar_float_mode.
5729 (expand_copysign_bit): Likewise.
5730 (expand_copysign): Update accordingly.
5732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5733 Alan Hayward <alan.hayward@arm.com>
5734 David Sherwood <david.sherwood@arm.com>
5736 * coretypes.h (opt_mode): New class.
5737 * machmode.h (opt_mode): Likewise.
5738 (opt_mode::else_void): New function.
5739 (opt_mode::require): Likewise.
5740 (opt_mode::exists): Likewise.
5741 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5742 (GET_MODE_2XWIDER_MODE): Likewise.
5743 (mode_iterator::get_wider): Update accordingly.
5744 (mode_iterator::get_2xwider): Likewise.
5745 (mode_iterator::get_known_wider): Likewise, turning into a template.
5746 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5747 forcing a wider mode to exist.
5748 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5749 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5750 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5751 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5752 * lower-subreg.c (init_lower_subreg): Likewise.
5753 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5754 on the final iteration.
5755 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5756 a wider mode exists before asking for a move pattern.
5757 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5758 forcing a wider mode to exist.
5759 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5760 returning false if no such mode exists.
5761 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5762 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5763 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5764 Avoid checking for a MODE_INT if we already know the mode is not a
5766 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
5767 forcing a wider mode to exist.
5768 (expmed_mult_highpart_optab): Likewise.
5769 (expmed_mult_highpart): Likewise.
5770 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
5772 * lto-streamer-in.c (lto_input_mode_table): Likewise.
5773 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
5774 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5775 * internal-fn.c (expand_mul_overflow): Update use of
5776 GET_MODE_2XWIDER_MODE.
5777 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5778 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
5779 GET_MODE_WIDER_MODE.
5780 (convert_plusminus_to_widen): Likewise.
5781 * tree-switch-conversion.c (array_value_type): Likewise.
5782 * var-tracking.c (emit_note_insn_var_location): Likewise.
5783 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5784 Return false inside rather than outside the loop if no wider mode
5786 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
5787 and GET_MODE_2XWIDER_MODE
5788 (can_compare_p): Use else_void.
5789 * gdbhooks.py (OptMachineModePrinter): New class.
5790 (build_pretty_printer): Use it for opt_mode.
5792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5793 Alan Hayward <alan.hayward@arm.com>
5794 David Sherwood <david.sherwood@arm.com>
5796 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5797 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5800 Alan Hayward <alan.hayward@arm.com>
5801 David Sherwood <david.sherwood@arm.com>
5803 * machmode.h (mode_traits): New structure.
5804 (get_narrowest_mode): New function.
5805 (mode_iterator::start): Likewise.
5806 (mode_iterator::iterate_p): Likewise.
5807 (mode_iterator::get_wider): Likewise.
5808 (mode_iterator::get_known_wider): Likewise.
5809 (mode_iterator::get_2xwider): Likewise.
5810 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5811 (FOR_EACH_MODE): Likewise.
5812 (FOR_EACH_MODE_FROM): Likewise.
5813 (FOR_EACH_MODE_UNTIL): Likewise.
5814 (FOR_EACH_WIDER_MODE): Likewise.
5815 (FOR_EACH_2XWIDER_MODE): Likewise.
5816 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5817 * combine.c (simplify_comparison): Likewise
5818 * config/i386/i386.c (type_natural_mode): Likewise.
5819 * cse.c (cse_insn): Likewise.
5820 * dse.c (find_shift_sequence): Likewise.
5821 * emit-rtl.c (init_derived_machine_modes): Likewise.
5822 (init_emit_once): Likewise.
5823 * explow.c (hard_function_value): Likewise.
5824 * expmed.c (extract_fixed_bit_field_1): Likewise.
5825 (extract_bit_field_1): Likewise.
5826 (expand_divmod): Likewise.
5827 (emit_store_flag_1): Likewise.
5828 * expr.c (init_expr_target): Likewise.
5829 (convert_move): Likewise.
5830 (alignment_for_piecewise_move): Likewise.
5831 (widest_int_mode_for_size): Likewise.
5832 (emit_block_move_via_movmem): Likewise.
5833 (copy_blkmode_to_reg): Likewise.
5834 (set_storage_via_setmem): Likewise.
5835 (compress_float_constant): Likewise.
5836 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5837 * optabs-query.c (get_best_extraction_insn): Likewise.
5838 * optabs.c (expand_binop): Likewise.
5839 (expand_twoval_unop): Likewise.
5840 (expand_twoval_binop): Likewise.
5841 (widen_leading): Likewise.
5842 (widen_bswap): Likewise.
5843 (expand_parity): Likewise.
5844 (expand_unop): Likewise.
5845 (prepare_cmp_insn): Likewise.
5846 (prepare_float_lib_cmp): Likewise.
5847 (expand_float): Likewise.
5848 (expand_fix): Likewise.
5849 (expand_sfix_optab): Likewise.
5850 * postreload.c (move2add_use_add2_insn): Likewise.
5851 * reg-stack.c (reg_to_stack): Likewise.
5852 * reginfo.c (choose_hard_reg_mode): Likewise.
5853 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5854 * stor-layout.c (mode_for_size): Likewise.
5855 (smallest_mode_for_size): Likewise.
5856 (mode_for_vector): Likewise.
5857 (finish_bitfield_representative): Likewise.
5858 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5859 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5860 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5861 * var-tracking.c (prepare_call_arguments): Likewise.
5863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5864 Alan Hayward <alan.hayward@arm.com>
5865 David Sherwood <david.sherwood@arm.com>
5867 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5868 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5869 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5870 * machmode.h (mode_size): Move earlier in file.
5871 (mode_precision): Likewise.
5872 (mode_inner): Likewise.
5873 (mode_nunits): Likewise.
5874 (mode_unit_size): Likewise.
5875 (unit_unit_precision): Likewise.
5876 (mode_wider): Likewise.
5877 (mode_2xwider): Likewise.
5878 (machine_mode): New class.
5879 (mode_to_bytes): New function.
5880 (mode_to_bits): Likewise.
5881 (mode_to_precision): Likewise.
5882 (mode_to_inner): Likewise.
5883 (mode_to_unit_size): Likewise.
5884 (mode_to_unit_precision): Likewise.
5885 (mode_to_nunits): Likewise.
5886 (GET_MODE_SIZE): Use mode_to_bytes.
5887 (GET_MODE_BITSIZE): Use mode_to_bits.
5888 (GET_MODE_PRECISION): Use mode_to_precision.
5889 (GET_MODE_INNER): Use mode_to_inner.
5890 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5891 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5892 (GET_MODE_NUNITS): Use mode_to_nunits.
5893 * system.h (ALWAYS_INLINE): New macro.
5894 * config/powerpcspe/powerpcspe-c.c
5895 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5896 int for arg1_mode and arg2_mode.
5898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5899 Alan Hayward <alan.hayward@arm.com>
5900 David Sherwood <david.sherwood@arm.com>
5902 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5903 Prefix mode names with E_ in case statements.
5904 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5905 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5906 (aarch64_split_simd_move): Likewise.
5907 (aarch64_gen_storewb_pair): Likewise.
5908 (aarch64_gen_loadwb_pair): Likewise.
5909 (aarch64_gen_store_pair): Likewise.
5910 (aarch64_gen_load_pair): Likewise.
5911 (aarch64_get_condition_code_1): Likewise.
5912 (aarch64_constant_pool_reload_icode): Likewise.
5913 (get_rsqrte_type): Likewise.
5914 (get_rsqrts_type): Likewise.
5915 (get_recpe_type): Likewise.
5916 (get_recps_type): Likewise.
5917 (aarch64_gimplify_va_arg_expr): Likewise.
5918 (aarch64_simd_container_mode): Likewise.
5919 (aarch64_emit_load_exclusive): Likewise.
5920 (aarch64_emit_store_exclusive): Likewise.
5921 (aarch64_expand_compare_and_swap): Likewise.
5922 (aarch64_gen_atomic_cas): Likewise.
5923 (aarch64_emit_bic): Likewise.
5924 (aarch64_emit_atomic_swap): Likewise.
5925 (aarch64_emit_atomic_load_op): Likewise.
5926 (aarch64_evpc_trn): Likewise.
5927 (aarch64_evpc_uzp): Likewise.
5928 (aarch64_evpc_zip): Likewise.
5929 (aarch64_evpc_ext): Likewise.
5930 (aarch64_evpc_rev): Likewise.
5931 (aarch64_evpc_dup): Likewise.
5932 (aarch64_gen_ccmp_first): Likewise.
5933 (aarch64_gen_ccmp_next): Likewise.
5934 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5935 (alpha_emit_xfloating_libcall): Likewise.
5936 (emit_insxl): Likewise.
5937 (alpha_arg_type): Likewise.
5938 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5939 (arc_preferred_simd_mode): Likewise.
5940 (arc_secondary_reload): Likewise.
5941 (get_arc_condition_code): Likewise.
5942 (arc_print_operand): Likewise.
5943 (arc_legitimate_constant_p): Likewise.
5944 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5945 * config/arc/arc.md (casesi_load): Likewise.
5946 (casesi_compact_jump): Likewise.
5947 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5948 (cc_use_register): Likewise.
5949 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5950 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5951 (arm_init_iwmmxt_builtins): Likewise.
5952 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5953 (neon_expand_vector_init): Likewise.
5954 (arm_attr_length_move_neon): Likewise.
5955 (maybe_get_arm_condition_code): Likewise.
5956 (arm_emit_vector_const): Likewise.
5957 (arm_preferred_simd_mode): Likewise.
5958 (arm_output_iwmmxt_tinsr): Likewise.
5959 (thumb1_output_casesi): Likewise.
5960 (thumb2_output_casesi): Likewise.
5961 (arm_emit_load_exclusive): Likewise.
5962 (arm_emit_store_exclusive): Likewise.
5963 (arm_expand_compare_and_swap): Likewise.
5964 (arm_evpc_neon_vuzp): Likewise.
5965 (arm_evpc_neon_vzip): Likewise.
5966 (arm_evpc_neon_vrev): Likewise.
5967 (arm_evpc_neon_vtrn): Likewise.
5968 (arm_evpc_neon_vext): Likewise.
5969 (arm_validize_comparison): Likewise.
5970 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5971 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5972 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5973 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5974 (c6x_preferred_simd_mode): Likewise.
5975 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5976 (epiphany_rtx_costs): Likewise.
5977 * config/epiphany/predicates.md (proper_comparison_operator):
5979 * config/frv/frv.c (condexec_memory_operand): Likewise.
5980 (frv_emit_move): Likewise.
5981 (output_move_single): Likewise.
5982 (output_condmove_single): Likewise.
5983 (frv_hard_regno_mode_ok): Likewise.
5984 (frv_matching_accg_mode): Likewise.
5985 * config/h8300/h8300.c (split_adds_subs): Likewise.
5986 (h8300_rtx_costs): Likewise.
5987 (h8300_print_operand): Likewise.
5988 (compute_mov_length): Likewise.
5989 (output_logical_op): Likewise.
5990 (compute_logical_op_length): Likewise.
5991 (compute_logical_op_cc): Likewise.
5992 (h8300_shift_needs_scratch_p): Likewise.
5993 (output_a_shift): Likewise.
5994 (compute_a_shift_length): Likewise.
5995 (compute_a_shift_cc): Likewise.
5996 (expand_a_rotate): Likewise.
5997 (output_a_rotate): Likewise.
5998 * config/i386/i386.c (classify_argument): Likewise.
5999 (function_arg_advance_32): Likewise.
6000 (function_arg_32): Likewise.
6001 (function_arg_64): Likewise.
6002 (function_value_64): Likewise.
6003 (ix86_gimplify_va_arg): Likewise.
6004 (ix86_legitimate_constant_p): Likewise.
6005 (put_condition_code): Likewise.
6006 (split_double_mode): Likewise.
6007 (ix86_avx256_split_vector_move_misalign): Likewise.
6008 (ix86_expand_vector_logical_operator): Likewise.
6009 (ix86_split_idivmod): Likewise.
6010 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6011 (ix86_build_const_vector): Likewise.
6012 (ix86_build_signbit_mask): Likewise.
6013 (ix86_match_ccmode): Likewise.
6014 (ix86_cc_modes_compatible): Likewise.
6015 (ix86_expand_branch): Likewise.
6016 (ix86_expand_sse_cmp): Likewise.
6017 (ix86_expand_sse_movcc): Likewise.
6018 (ix86_expand_int_sse_cmp): Likewise.
6019 (ix86_expand_vec_perm_vpermi2): Likewise.
6020 (ix86_expand_vec_perm): Likewise.
6021 (ix86_expand_sse_unpack): Likewise.
6022 (ix86_expand_int_addcc): Likewise.
6023 (ix86_split_to_parts): Likewise.
6024 (ix86_vectorize_builtin_gather): Likewise.
6025 (ix86_vectorize_builtin_scatter): Likewise.
6026 (avx_vpermilp_parallel): Likewise.
6027 (inline_memory_move_cost): Likewise.
6028 (ix86_tieable_integer_mode_p): Likewise.
6029 (x86_maybe_negate_const_int): Likewise.
6030 (ix86_expand_vector_init_duplicate): Likewise.
6031 (ix86_expand_vector_init_one_nonzero): Likewise.
6032 (ix86_expand_vector_init_one_var): Likewise.
6033 (ix86_expand_vector_init_concat): Likewise.
6034 (ix86_expand_vector_init_interleave): Likewise.
6035 (ix86_expand_vector_init_general): Likewise.
6036 (ix86_expand_vector_set): Likewise.
6037 (ix86_expand_vector_extract): Likewise.
6038 (emit_reduc_half): Likewise.
6039 (ix86_emit_i387_round): Likewise.
6040 (ix86_mangle_type): Likewise.
6041 (ix86_expand_round_sse4): Likewise.
6042 (expand_vec_perm_blend): Likewise.
6043 (canonicalize_vector_int_perm): Likewise.
6044 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6045 (expand_vec_perm_1): Likewise.
6046 (expand_vec_perm_interleave3): Likewise.
6047 (expand_vec_perm_even_odd_pack): Likewise.
6048 (expand_vec_perm_even_odd_1): Likewise.
6049 (expand_vec_perm_broadcast_1): Likewise.
6050 (ix86_vectorize_vec_perm_const_ok): Likewise.
6051 (ix86_expand_vecop_qihi): Likewise.
6052 (ix86_expand_mul_widen_hilo): Likewise.
6053 (ix86_expand_sse2_abs): Likewise.
6054 (ix86_expand_pextr): Likewise.
6055 (ix86_expand_pinsr): Likewise.
6056 (ix86_preferred_simd_mode): Likewise.
6057 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6058 * config/i386/sse.md (*andnot<mode>3): Likewise.
6059 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6060 (*<code><mode>3): Likewise.
6061 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6062 (ia64_expand_atomic_op): Likewise.
6063 (ia64_arg_type): Likewise.
6064 (ia64_mode_to_int): Likewise.
6065 (ia64_scalar_mode_supported_p): Likewise.
6066 (ia64_vector_mode_supported_p): Likewise.
6067 (expand_vec_perm_broadcast): Likewise.
6068 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6069 (iq2000_function_arg_advance): Likewise.
6070 (iq2000_function_arg): Likewise.
6071 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6072 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6073 (m68k_libcall_value): Likewise.
6074 (m68k_function_value): Likewise.
6075 (sched_attr_op_type): Likewise.
6076 * config/mcore/mcore.c (mcore_output_move): Likewise.
6077 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6079 (microblaze_function_arg): Likewise.
6080 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6081 (mips_print_operand): Likewise.
6082 (mips_mode_ok_for_mov_fmt_p): Likewise.
6083 (mips_vector_mode_supported_p): Likewise.
6084 (mips_preferred_simd_mode): Likewise.
6085 (mips_expand_vpc_loongson_even_odd): Likewise.
6086 (mips_expand_vec_unpack): Likewise.
6087 (mips_expand_vi_broadcast): Likewise.
6088 (mips_expand_vector_init): Likewise.
6089 (mips_expand_vec_reduc): Likewise.
6090 (mips_expand_msa_cmp): Likewise.
6091 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6092 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6093 (cc_flags_for_mode): Likewise.
6094 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6095 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6096 (nds32_output_casesi_pc_relative): Likewise.
6097 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6098 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6099 (nvptx_gen_unpack): Likewise.
6100 (nvptx_gen_pack): Likewise.
6101 (nvptx_gen_shuffle): Likewise.
6102 (nvptx_gen_wcast): Likewise.
6103 (nvptx_preferred_simd_mode): Likewise.
6104 * config/pa/pa.c (pa_secondary_reload): Likewise.
6105 * config/pa/predicates.md (base14_operand): Likewise.
6106 * config/powerpcspe/powerpcspe-c.c
6107 (altivec_resolve_overloaded_builtin): Likewise.
6108 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6110 (rs6000_preferred_simd_mode): Likewise.
6111 (output_vec_const_move): Likewise.
6112 (rs6000_expand_vector_extract): Likewise.
6113 (rs6000_split_vec_extract_var): Likewise.
6114 (reg_offset_addressing_ok_p): Likewise.
6115 (rs6000_legitimate_offset_address_p): Likewise.
6116 (rs6000_legitimize_address): Likewise.
6117 (rs6000_emit_set_const): Likewise.
6118 (rs6000_const_vec): Likewise.
6119 (rs6000_emit_move): Likewise.
6120 (spe_build_register_parallel): Likewise.
6121 (rs6000_darwin64_record_arg_recurse): Likewise.
6122 (swap_selector_for_mode): Likewise.
6123 (spe_init_builtins): Likewise.
6124 (paired_init_builtins): Likewise.
6125 (altivec_init_builtins): Likewise.
6126 (do_load_for_compare): Likewise.
6127 (rs6000_generate_compare): Likewise.
6128 (rs6000_expand_float128_convert): Likewise.
6129 (emit_load_locked): Likewise.
6130 (emit_store_conditional): Likewise.
6131 (rs6000_output_function_epilogue): Likewise.
6132 (rs6000_handle_altivec_attribute): Likewise.
6133 (rs6000_function_value): Likewise.
6134 (emit_fusion_gpr_load): Likewise.
6135 (emit_fusion_p9_load): Likewise.
6136 (emit_fusion_p9_store): Likewise.
6137 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6138 (fusion_gpr_mem_load): Likewise.
6139 (fusion_addis_mem_combo_load): Likewise.
6140 (fusion_addis_mem_combo_store): Likewise.
6141 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6142 (fusion_gpr_mem_load): Likewise.
6143 (fusion_addis_mem_combo_load): Likewise.
6144 (fusion_addis_mem_combo_store): Likewise.
6145 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6147 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6148 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6149 (rs6000_preferred_simd_mode): Likewise.
6150 (output_vec_const_move): Likewise.
6151 (rs6000_expand_vector_extract): Likewise.
6152 (rs6000_split_vec_extract_var): Likewise.
6153 (reg_offset_addressing_ok_p): Likewise.
6154 (rs6000_legitimate_offset_address_p): Likewise.
6155 (rs6000_legitimize_address): Likewise.
6156 (rs6000_emit_set_const): Likewise.
6157 (rs6000_const_vec): Likewise.
6158 (rs6000_emit_move): Likewise.
6159 (rs6000_darwin64_record_arg_recurse): Likewise.
6160 (swap_selector_for_mode): Likewise.
6161 (paired_init_builtins): Likewise.
6162 (altivec_init_builtins): Likewise.
6163 (rs6000_expand_float128_convert): Likewise.
6164 (emit_load_locked): Likewise.
6165 (emit_store_conditional): Likewise.
6166 (rs6000_output_function_epilogue): Likewise.
6167 (rs6000_handle_altivec_attribute): Likewise.
6168 (rs6000_function_value): Likewise.
6169 (emit_fusion_gpr_load): Likewise.
6170 (emit_fusion_p9_load): Likewise.
6171 (emit_fusion_p9_store): Likewise.
6172 * config/rx/rx.c (rx_gen_move_template): Likewise.
6173 (flags_from_mode): Likewise.
6174 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6175 (s390_slb_comparison): Likewise.
6176 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6177 (s390_vector_mode_supported_p): Likewise.
6178 (s390_cc_modes_compatible): Likewise.
6179 (s390_match_ccmode_set): Likewise.
6180 (s390_canonicalize_comparison): Likewise.
6181 (s390_emit_compare_and_swap): Likewise.
6182 (s390_branch_condition_mask): Likewise.
6183 (s390_rtx_costs): Likewise.
6184 (s390_secondary_reload): Likewise.
6185 (__SECONDARY_RELOAD_CASE): Likewise.
6186 (s390_expand_cs): Likewise.
6187 (s390_preferred_simd_mode): Likewise.
6188 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6189 * config/sh/sh.c (sh_print_operand): Likewise.
6190 (dump_table): Likewise.
6191 (sh_secondary_reload): Likewise.
6192 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6193 * config/sh/sh.md (casesi_worker_1): Likewise.
6194 (casesi_worker_2): Likewise.
6195 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6196 (fcc_comparison_operator): Likewise.
6197 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6198 (emit_soft_tfmode_cvt): Likewise.
6199 (sparc_preferred_simd_mode): Likewise.
6200 (output_cbranch): Likewise.
6201 (sparc_print_operand): Likewise.
6202 (sparc_expand_vec_perm_bmask): Likewise.
6203 (vector_init_bshuffle): Likewise.
6204 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6205 (spu_vector_mode_supported_p): Likewise.
6206 (spu_expand_insv): Likewise.
6207 (spu_emit_branch_or_set): Likewise.
6208 (spu_handle_vector_attribute): Likewise.
6209 (spu_builtin_splats): Likewise.
6210 (spu_builtin_extract): Likewise.
6211 (spu_builtin_promote): Likewise.
6212 (spu_expand_sign_extend): Likewise.
6213 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6214 (tilegx_simd_int): Likewise.
6215 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6216 (tilepro_simd_int): Likewise.
6217 * config/v850/v850.c (const_double_split): Likewise.
6218 (v850_print_operand): Likewise.
6219 (ep_memory_offset): Likewise.
6220 * config/vax/vax.c (vax_rtx_costs): Likewise.
6221 (vax_output_int_move): Likewise.
6222 (vax_output_int_add): Likewise.
6223 (vax_output_int_subtract): Likewise.
6224 * config/visium/predicates.md (visium_branch_operator): Likewise.
6225 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6226 (visium_print_operand_address): Likewise.
6227 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6228 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6229 (xtensa_expand_conditional_branch): Likewise.
6230 (xtensa_copy_incoming_a7): Likewise.
6231 (xtensa_output_literal): Likewise.
6232 * dfp.c (decimal_real_maxval): Likewise.
6233 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6236 Alan Hayward <alan.hayward@arm.com>
6237 David Sherwood <david.sherwood@arm.com>
6239 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6240 (mode_nunits_inline): Likewise.
6241 (mode_inner_inline): Likewise.
6242 (mode_unit_size_inline): Likewise.
6243 (mode_unit_precision_inline): Likewise.
6244 (emit_insn_modes_h): Likewise. Also emit a #define of the
6246 (emit_mode_wider): Add an E_ prefix to mode names.
6247 (emit_mode_complex): Likewise.
6248 (emit_mode_inner): Likewise.
6249 (emit_mode_adjustments): Likewise.
6250 (emit_mode_int_n): Likewise.
6251 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6252 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6253 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6254 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6255 (CRC32_BUILTIN, ENTRY): Likewise.
6256 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6257 (aarch64_pop_regs): Likewise.
6258 (aarch64_process_components): Likewise.
6259 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6260 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6261 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6262 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6263 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6264 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6266 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6267 (pcmpistr): Likewise.
6268 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6269 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6270 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6272 * config/rl78/rl78.c (mduc_regs): Likewise.
6273 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6274 (htm_expand_builtin): Likewise.
6275 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6276 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6277 * config/xtensa/xtensa.c (print_operand): Likewise.
6278 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6279 (NUM_MODE_VECTOR_INT): Likewise.
6280 * genoutput.c (null_operand): Likewise.
6281 (output_operand_data): Likewise.
6282 * genrecog.c (print_parameter_value): Likewise.
6283 * lra.c (debug_operand_data): Likewise.
6285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6287 * dwarf2out.c (scompare_loc_descriptor_wide)
6288 (scompare_loc_descriptor_narrow): New functions, split out from...
6289 (scompare_loc_descriptor): ...here.
6290 * expmed.c (emit_store_flag_int): New function, split out from...
6291 (emit_store_flag): ...here.
6293 2017-08-30 Richard Biener <rguenther@suse.de>
6295 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6296 (dwarf2out_early_finish): Move setting of AT_pubnames from
6297 early debug output to early finish.
6299 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6301 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6302 and -mdata-region to the assembler.
6304 2017-08-30 Richard Biener <rguenther@suse.de>
6306 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6308 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6309 (dwarf2out_early_global_decl): Only generate a DIE for the
6310 abstract origin if it doesn't already exist or is a declaration DIE.
6311 (resolve_addr): Do not add the linkage name twice when
6312 generating a stub DIE for the DW_TAG_GNU_call_site target.
6314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6316 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6317 Use machine_mode rather than int for arg1_mode.
6319 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6322 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6323 that the second argument of the built-in functions to unpack
6324 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6325 switch statement instead a lot of if statements.
6326 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6327 Allow 64-bit values to be in Altivec registers as well as
6328 traditional floating point registers.
6329 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6331 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6333 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6334 MAX_REGS_PER_ADDRESS == 1.
6336 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6338 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6339 * config/i386/i386.c (ix86_option_override_internal): Simplify
6340 setting of opts->x_flag_entry.
6342 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6343 Jakub Jelinek <jakub@redhat.com>
6344 Richard Biener <rguenther@suse.de>
6346 PR tree-optimization/81503
6347 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6348 folded constant fits in the target type; reorder tests for clarity.
6350 2017-08-29 Martin Liska <mliska@suse.cz>
6352 * passes.def: Include pass_lower_switch.
6353 * stmt.c (dump_case_nodes): Remove and move to
6354 tree-switch-conversion.
6355 (case_values_threshold): Likewise.
6356 (expand_switch_as_decision_tree_p): Likewise.
6357 (emit_case_decision_tree): Likewise.
6358 (expand_case): Likewise.
6359 (balance_case_nodes): Likewise.
6360 (node_has_low_bound): Likewise.
6361 (node_has_high_bound): Likewise.
6362 (node_is_bounded): Likewise.
6363 (emit_case_nodes): Likewise.
6364 (struct simple_case_node): New struct.
6365 (add_case_node): Remove.
6366 (emit_case_dispatch_table): Use vector instead of case_list.
6367 (reset_out_edges_aux): Remove.
6368 (compute_cases_per_edge): Likewise.
6369 (expand_case): Build list of simple_case_node.
6370 (expand_sjlj_dispatch_table): Use it.
6371 * tree-switch-conversion.c (struct case_node): Moved from
6372 stmt.c and adjusted.
6373 (emit_case_nodes): Likewise.
6374 (node_has_low_bound): Likewise.
6375 (node_has_high_bound): Likewise.
6376 (node_is_bounded): Likewise.
6377 (case_values_threshold): Likewise.
6378 (reset_out_edges_aux): Likewise.
6379 (compute_cases_per_edge): Likewise.
6380 (add_case_node): Likewise.
6381 (dump_case_nodes): Likewise.
6382 (balance_case_nodes): Likewise.
6383 (expand_switch_as_decision_tree_p): Likewise.
6384 (emit_jump): Likewise.
6385 (emit_case_decision_tree): Likewise.
6386 (try_switch_expansion): Likewise.
6387 (do_jump_if_equal): Likewise.
6388 (emit_cmp_and_jump_insns): Likewise.
6389 (fix_phi_operands_for_edge): New function.
6390 (record_phi_operand_mapping): Likewise.
6391 (class pass_lower_switch): New pass.
6392 (pass_lower_switch::execute): New function.
6393 (make_pass_lower_switch): Likewise.
6394 (conditional_probability):
6395 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6396 * tree-pass.h: Add make_pass_lower_switch.
6398 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6401 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6404 2017-08-29 Richard Biener <rguenther@suse.de>
6406 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6407 we do not add a DW_AT_inline attribute twice.
6408 (gen_subprogram_die): Remove code setting DW_AT_inline on
6409 DECL_ABSTRACT_P nodes.
6411 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6413 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6414 calls to internal functions.
6415 (gimplify_modify_expr): Likewise.
6416 * tree-call-cdce.c (use_internal_fn): Likewise.
6417 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6418 (convert_to_divmod): Set the nothrow flag.
6419 * tree-if-conv.c (predicate_mem_writes): Likewise.
6420 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6421 (vectorizable_call): Likewise.
6422 (vectorizable_store): Likewise.
6423 (vectorizable_load): Likewise.
6424 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6425 (vect_recog_mask_conversion_pattern): Likewise.
6427 2017-08-29 Martin Liska <mliska@suse.cz>
6430 * gcc.c (driver_handle_option): Add new argument.
6431 * opts-common.c (handle_option): Pass
6432 target_option_override_hook.
6433 * opts-global.c (lang_handle_option): Add new option.
6434 (set_default_handlers): Add new argument.
6435 (decode_options): Likewise.
6436 * opts.c (target_handle_option): Likewise.
6437 (common_handle_option): Call target_option_override_hook.
6438 * opts.h (struct cl_option_handler_func): Add hook for
6439 target option override.
6440 (struct cl_option_handlers): Likewise.
6441 (set_default_handlers): Add new argument.
6442 (decode_options): Likewise.
6443 (common_handle_option): Likewise.
6444 (target_handle_option): Likewise.
6445 * toplev.c (toplev::main): Pass targetm.target_option.override
6448 2017-08-29 Richard Biener <rguenther@suse.de>
6449 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6451 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6452 life to the active subtree.
6454 2017-08-28 Jeff Law <law@redhat.com>
6456 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6457 derive_equivalences.
6458 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6460 (edge_info::derive_equivalences): New private member function
6462 * tree-ssa-dom.c (class edge_info): Changed from a struct
6463 to a class. Add ctor/dtor, methods and data members.
6464 (edge_info::edge_info): Renamed from allocate_edge_info.
6465 Initialize additional members.
6466 (edge_info::~edge_info): New.
6467 (free_dom_edge_info): Delete the edge info.
6468 (record_edge_info): Use new class & associated member functions.
6469 Tighten forms for testing for edge equivalences.
6470 (record_temporary_equivalences): Iterate over the simple
6471 equivalences rather than assuming there's only one per edge.
6472 (cprop_into_successor_phis): Iterate over the simple
6473 equivalences rather than assuming there's only one per edge.
6474 (optimize_stmt): Use operand_equal_p rather than pointer
6475 equality for mini-DSE code.
6477 2017-08-28 Nathan Sidwell <nathan@acm.org>
6479 * gcc.c (execute): Fold SIGPIPE handling into switch
6480 statement. Adjust internal error message.
6482 2017-08-28 Richard Biener <rguenther@suse.de>
6485 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6486 Do nothing for removed DIEs.
6488 2017-08-28 Richard Biener <rguenther@suse.de>
6490 PR tree-optimization/81977
6491 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6494 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6497 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6498 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6499 using targetm.gen_mem_thread_fence.
6501 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6504 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6505 predicate to register_operand. Reorder operands.
6506 (*btr<mode>): Ditto.
6507 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6508 (*btr<mode>_mask): Ditto.
6510 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6512 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6513 * config/rs6000/xmmintrin.h: New file.
6514 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6516 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6519 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6520 parameter and_insn and return it.
6521 (recombine_lvx_pattern): Insert a copy to ensure availability of
6522 the base register of the copied masking operation at the point of
6523 the instruction replacement.
6524 (recombine_stvx_pattern): Likewise.
6526 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6528 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6529 undocumented switches.
6530 (-mpower9-dform-vector): Likewise.
6531 (-mpower9-dform): Likewise.
6532 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6533 comments to delete references to -mpower9-dform* switches.
6534 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6535 Delete reference to -mpower9-dform* switches, test for
6536 -mpower9-vector instead.
6537 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6538 (OTHER_P9_VECTOR_MASKS): Likewise.
6539 (POWERPC_MASKS): Likewise.
6540 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6541 tests against -mpower9-dform* to -mpower9-vector. Delete code
6542 that checked for -mpower9-dform* consistancy with other options.
6543 Add test for -mpower9-misc to enable other power9 switches.
6544 (rs6000_init_hard_regno_mode_ok): Likewise.
6545 (rs6000_option_override_internal): Likewise.
6546 (rs6000_emit_prologue): Likewise.
6547 (rs6000_emit_epilogue): Likewise.
6548 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6549 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6550 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6552 (emit_fusion_p9_store): Likewise.
6553 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6554 resetting these macros if the assembler does not support ISA 3.0
6556 (TARGET_P9_DFORM_VECTOR): Likewise.
6557 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6558 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6560 2017-08-25 Alan Modra <amodra@gmail.com>
6562 PR rtl-optimization/81747
6563 * cse.c (cse_extended_basic_block): Don't attempt to record
6564 equivalences for degenerate conditional branches that branch
6565 to their fall-through.
6567 2017-08-24 Martin Sebor <msebor@redhat.com>
6570 * gimple-fold.c (size_must_be_zero_p): New function.
6571 (gimple_fold_builtin_memory_op): Call it.
6573 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6575 * config/rs6000/mm_malloc.h: New file.
6577 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6579 PR tree-optimization/81913
6580 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6581 analysis when either IVs in condition can wrap.
6583 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6585 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6586 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6588 2017-08-24 Richard Biener <rguenther@suse.de>
6591 * targhooks.c (default_target_can_inline_p): Properly
6592 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6593 is present and always compare.
6594 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6595 infer -mfpmath=sse from TARGET_SSE_P.
6596 (ix86_can_inline_p): Properly use target_option_default_node when
6597 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6599 2017-08-24 Richard Biener <rguenther@suse.de>
6602 * dwarf2out.c (output_die): Handle flag_generate_offload like
6604 (output_comp_unit): Likewise.
6605 (gen_array_type_die): Likewise.
6606 (dwarf2out_early_finish): Likewise.
6607 (note_variable_value_in_expr): Likewise.
6608 (dwarf2out_finish): Likewise. Adjust assert.
6609 * cgraphunit.c (symbol_table::compile): Move setting of
6610 flag_generate_offload earlier ...
6611 (symbol_table::finalize_compilation_unit): ... here, before
6612 early debug finalization.
6614 2017-08-24 Richard Biener <rguenther@suse.de>
6616 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6617 and ipa-fnsummary.h.
6618 (ix86_can_inline_p): When ix86_fpmath flags do not match
6619 check whether the callee uses FP math at all.
6621 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6624 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6625 instead of TYPE_PRECISION.
6627 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6628 Alan Hayward <alan.hayward@arm.com>
6629 David Sherwood <david.sherwood@arm.com>
6631 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6633 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6635 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6637 (ix86_frame::stack_realign_allocate): New field.
6638 (struct machine_frame_state): Modify comments.
6639 (machine_frame_state::sp_realigned_fp_end): New field.
6640 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6642 (sp_valid_at): Add assertion to assure no attempt to access invalid
6643 offset of a realigned stack.
6644 (fp_valid_at): Likewise.
6645 (choose_baseaddr): Modify comments.
6646 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6647 ix86_expand_prologue.
6648 (ix86_expand_prologue): Modify stack realignment and allocation.
6649 (ix86_expand_epilogue): Modify comments.
6650 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6651 avx2_runtime, avx512f, and avx512f_runtime.
6653 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6655 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6656 (mstackrealign): Do not init to -1.
6657 * config/i386/i386.c (ix86_option_override_internal):
6658 Check opts_set, not opts when setting default value of
6659 opts->x_ix86_force_align_arg_pointer.
6661 2017-08-23 Richard Biener <rguenther@suse.de>
6663 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6664 lang_hooks.decl_printable_name.
6665 * print-rtl-function.c (print_rtx_function): Likewise.
6666 * tree-pretty-print.c (dump_function_header): Likewise.
6668 2017-08-23 Richard Biener <rguenther@suse.de>
6671 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6672 -g0 at compile-time.
6674 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6677 * doc/sourcebuild.texi (Other hardware attributes):
6680 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6683 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6684 Fix single-use check.
6686 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6688 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6690 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6692 * doc/install.texi: Modify to add more details on running selected
6695 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6697 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6698 is combined with -mabi=ms.
6699 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6702 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6704 PR tree-optimization/81488
6705 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6706 and cached_basis fields.
6707 (MAX_SPREAD): New constant.
6708 (alloc_cand_and_find_basis): Initialize new fields.
6709 (clear_visited): New function.
6710 (create_phi_basis_1): Rename from create_phi_basis, set visited
6711 and cached_basis fields.
6712 (create_phi_basis): New wrapper function.
6713 (phi_add_costs_1): Rename from phi_add_costs, add spread
6714 parameter, set visited field, short-circuit when limits reached.
6715 (phi_add_costs): New wrapper function.
6716 (record_phi_increments_1): Rename from record_phi_increments, set
6718 (record_phi_increments): New wrapper function.
6719 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6720 (phi_incr_cost): New wrapper function.
6721 (all_phi_incrs_profitable_1): Rename from
6722 all_phi_incrs_profitable, set visited field.
6723 (all_phi_incrs_profitable): New wrapper function.
6725 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6726 Alan Hayward <alan.hayward@arm.com>
6727 David Sherwood <david.sherwood@arm.com>
6729 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6730 that takes the outer and inner modes.
6731 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6732 comparison as the canonical test for a paradoxical subreg.
6733 * combine.c (simplify_set): Use paradoxical_subreg_p.
6734 (make_extraction): Likewise.
6735 (force_to_mode): Likewise.
6736 (rtx_equal_for_field_assignment_p): Likewise.
6737 (gen_lowpart_for_combine): Likewise.
6738 (simplify_comparison): Likewise.
6739 * cse.c (equiv_constant): Likewise.
6740 * expmed.c (store_bit_field_1): Likewise.
6741 * final.c (alter_subreg): Likewise.
6742 * fwprop.c (propagate_rtx): Likewise.
6743 (forward_propagate_subreg): Likewise.
6744 * ira-conflicts.c (ira_build_conflicts): Likewise.
6745 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6746 * lra-constraints.c (curr_insn_transform): Likewise.
6747 (split_reg): Likewise.
6748 * lra-eliminations.c (move_plus_up): Likewise.
6749 (lra_eliminate_regs_1): Likewise.
6750 * recog.c (general_operand): Likewise.
6751 * ree.c (combine_reaching_defs): Likewise.
6752 * reload.c (push_reload): Likewise.
6753 (find_reloads): Likewise.
6754 * reload1.c (elimination_effects): Likewise.
6755 (compute_reload_subreg_offset): Likewise.
6756 (choose_reload_regs): Likewise.
6757 * rtlanal.c (subreg_lsb_1): Likewise.
6758 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6759 (simplify_subreg): Likewise.
6760 * var-tracking.c (track_loc_p): Likewise.
6761 * emit-rtl.c (byte_lowpart_offset): Likewise.
6762 (paradoxical_subreg_p): Delete out-of-line definition.
6764 2017-08-22 Jeff Law <law@redhat.com>
6766 PR tree-optimization/81741
6767 PR tree-optimization/71947
6768 * tree-ssa-dom.c: Include tree-inline.h.
6769 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
6770 equivalences if one is more expensive to compute than the other.
6771 * tree-ssa-scopedtables.h (class const_or_copies): Make
6772 record_const_or_copy_raw method private.
6773 (class avail_exprs_stack): New method simplify_binary_operation.
6774 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
6775 avail_exprs_stack::simplify_binary_operation as needed.
6776 (avail_exprs_stack::simplify_binary_operation): New function.
6778 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
6780 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
6781 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
6782 (DOT_SYMBOLS): Likewise.
6783 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
6784 (RELOCATABLE_NEEDS_FIXUP): Likewise.
6785 (RS6000_ABI_NAME): Likewise.
6786 (TARGET_CMODEL): Likewise.
6787 (TOC_SECTION_ASM_OP): Likewise.
6788 (SET_CMODEL): New macro.
6789 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6791 2017-08-22 Richard Biener <rguenther@suse.de>
6793 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6794 to allow for free-lang-data replacements similar to verify_type_variant.
6796 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6797 Alan Hayward <alan.hayward@arm.com>
6798 David Sherwood <david.sherwood@arm.com>
6800 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6801 VOIDmode for the LABEL_REF.
6803 2017-08-22 Richard Biener <rguenther@suse.de>
6805 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6807 2017-08-22 Richard Biener <rguenther@suse.de>
6809 * common.opt (feliminate-dwarf2-dups): Ignore.
6810 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6811 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6812 same_die_p_wrap, compute_section_prefix,
6813 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6814 (comdat_symbol_id, comdat_symbol_number): Likewise.
6815 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6817 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6818 (output_die): Mark unreachable path unreachable.
6819 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6820 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6821 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6822 (dwarf2out_early_finish): Likewise.
6824 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6826 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6828 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6831 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6832 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6833 (avr_attribute_table) <io, io_low, address>: Initialize
6834 .decl_required with true.
6836 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6838 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6839 undocumented debugging options.
6840 (-mvsx-scalar-double): Likewise.
6841 (-mallow-df-permute): Likewise.
6842 (-mvectorize-builtins): Likewise.
6843 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6844 (rs6000_builtin_vectorized_function): Likewise.
6845 (rs6000_builtin_md_vectorized_function): Likewise.
6846 (rs6000_opt_vars): Likewise.
6848 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6851 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6852 (*btrq_imm): Rename from *btrq.
6853 (*btcq_imm): Rename from *btcq.
6854 (btsc): New code attribute.
6855 (*<btsc><mode>): New insn pattern.
6856 (*btr<mode>): Ditto.
6857 (*<btsc><mode>_mask): New insn_and_split pattern.
6858 (*btr<mode>_mask): Ditto.
6860 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6861 Alan Hayward <alan.hayward@arm.com>
6862 David Sherwood <david.sherwood@arm.com>
6864 * function.c (pad_below): Simplify padding calculation.
6866 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6867 Alan Hayward <alan.hayward@arm.com>
6868 David Sherwood <david.sherwood@arm.com>
6870 * target.def (function_prologue): Remove frame size argument.
6871 (function_epilogue): Likewise.
6872 * doc/tm.texi: Regenerate.
6873 * final.c (final_start_function): Update call to function_prologue.
6874 (final_end_function): Update call to function_epilogue.
6875 (default_function_pro_epilogue): Remove frame size argument.
6876 * output.h (default_function_pro_epilogue): Likewise.
6877 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6878 (arm_output_function_prologue): Likewise.
6879 * config/frv/frv.c (frv_function_prologue): Likewise.
6880 (frv_function_epilogue): Likewise.
6881 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6882 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6883 (ia64_output_function_epilogue): Likewise.
6884 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6885 (m32r_output_function_epilogue): Likewise.
6886 * config/microblaze/microblaze.c (microblaze_function_prologue)
6887 (microblaze_function_epilogue): Likewise.
6888 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6889 (mips_output_function_epilogue): Likewise.
6890 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6891 (mmix_target_asm_function_epilogue): Likewise.
6892 * config/msp430/msp430.c (msp430_start_function): Likewise.
6893 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6894 (nds32_asm_function_epilogue): Likewise.
6895 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6896 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6897 (pa_output_function_epilogue): Likewise.
6898 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6899 (rs6000_output_function_epilogue): Likewise.
6900 * config/rl78/rl78.c (rl78_start_function): Likewise.
6901 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6902 (rs6000_output_function_epilogue): Likewise.
6903 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6904 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6905 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6906 (sparc_asm_function_epilogue): Likewise.
6908 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6910 * tree.h (type_has_mode_precision_p): New function.
6911 * convert.c (convert_to_integer_1): Use it.
6912 * expr.c (expand_expr_real_2): Likewise.
6913 (expand_expr_real_1): Likewise.
6914 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6915 * match.pd: Likewise.
6916 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6917 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6918 * tree-tailcall.c (process_assignment): Likewise.
6919 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6920 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6921 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6922 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6923 (vectorizable_assignment): Likewise.
6924 (vectorizable_shift): Likewise.
6925 (vectorizable_operation): Likewise.
6926 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6928 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6930 * match.pd: Add pow (C, x) simplification.
6932 2017-08-21 Richard Biener <rguenther@suse.de>
6934 PR tree-optimization/81900
6935 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6936 for blocks with abnormal predecessors.
6937 (compute_antic): Do not set visited flag prematurely.
6939 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6942 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6944 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6946 * stor-layout.h (vector_type_mode): Move to...
6947 * tree.h (vector_type_mode): ...here.
6948 * stor-layout.c (vector_type_mode): Move to...
6949 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6951 2017-08-21 Richard Biener <rguenther@suse.de>
6953 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6954 register_external_die hooks.
6955 (debug_false_tree_charstarstar_uhwistar): Declare.
6956 (debug_nothing_tree_charstar_uhwi): Likewise.
6957 * debug.c (do_nothing_debug_hooks): Adjust.
6958 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6959 (debug_nothing_tree_charstar_uhwi): Likewise.
6960 * dbxout.c (dbx_debug_hooks): Adjust.
6961 (xcoff_debug_hooks): Likewise.
6962 * sdbout.c (sdb_debug_hooks): Likewise.
6963 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6964 * dwarf2out.c (macinfo_label_base): New global.
6965 (dwarf2out_register_external_die): New function for the
6966 register_external_die hook.
6967 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6968 (dwarf2_debug_hooks): Use them.
6969 (dwarf2_lineno_debug_hooks): Adjust.
6970 (struct die_struct): Add with_offset flag.
6971 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6972 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6973 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6974 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6975 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6976 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6977 defining section names for the early LTO debug variants.
6978 (reset_indirect_string): New helper.
6979 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6980 (print_dw_val): Add support for offsetted symbol references.
6981 (get_ultimate_context): Split out from is_cxx.
6982 (is_cxx): Use get_ultimate_context.
6983 (is_fortran): Add decl overload.
6984 (compute_comp_unit_symbol): Split out worker from
6985 compute_section_prefix.
6986 (compute_section_prefix): Call compute_comp_unit_symbol and
6987 set comdat_type_p here.
6988 (output_die): Skip DIE symbol output for the LTO added one.
6989 Handle DIE symbol references with offset.
6990 (output_comp_unit): Guard section name mangling properly.
6991 For LTO debug sections emit a symbol at the section beginning
6992 which we use to refer to its DIEs.
6993 (add_abstract_origin_attribute): For DIEs registered via
6994 dwarf2out_register_external_die directly refer to the early
6995 DIE rather than indirectly through the shadow one we created.
6996 Remove obsolete call to dwarf2out_abstract_function for
6997 non-function/block origins.
6998 (gen_array_type_die): When generating early LTO debug do
6999 not emit DW_AT_string_length.
7000 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7001 late when in LTO. As suggested place a gcc_unreachable for
7002 the DECL_ABSTRACT_P case.
7003 (gen_subprogram_die): Avoid another specification DIE
7004 for early built declarations/definitions for the late LTO case.
7005 (gen_variable_die): Add type references for late duplicated VLA dies
7007 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7008 we have the abstract instance already.
7009 (process_scope_var): Adjust decl DIE contexts in LTO which
7010 first puts them in limbo.
7011 (gen_decl_die): Do not generate type DIEs late apart from
7012 types for VLAs or for decls we do not yet have a DIE. Do not
7013 call dwarf2out_abstract_function late.
7014 (dwarf2out_early_global_decl): Make sure to create DIEs
7015 for abstract instances of a decl first.
7016 (dwarf2out_late_global_decl): Adjust comment.
7017 (output_macinfo_op): With multiple macro sections use
7018 macinfo_label_base to distinguish labels.
7019 (output_macinfo): Likewise. Update macinfo_label_base.
7020 Pass in the line info label.
7021 (note_variable_value_in_expr): When generating LTO resolve
7022 all variable values here by generating DIEs as needed.
7023 (init_sections_and_labels): Add early LTO debug flag parameter
7024 and generate different sections and names if set. Add generation
7025 counter for the labels so we can have multiple of them.
7026 (reset_dies): Helper to allow DIEs to be output multiple times.
7027 (dwarf2out_finish): When outputting DIEs to the fat part of an
7028 LTO object first reset DIEs.
7029 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7030 (modified_type_die): Check for decl_ultimate_origin being self
7032 (gen_type_die_with_usage): Likewise.
7033 (gen_typedef_die): Allow decl_ultimate_origin being self.
7034 (set_decl_abstract_flags): Remove.
7035 (set_block_abstract_flags): Likewise.
7036 (dwarf2out_abstract_function): Treat the early generated DIEs
7037 as the abstract copy and only add DW_AT_inline and
7038 DW_AT_artificial here and call set_decl_origin_self.
7039 If the DIE has an abstract origin don't do anything.
7040 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7041 if we have none yet (Go fails to build one, PR78628).
7042 (variably_modified_type_p): Prevent endless recursion for Ada
7043 cyclic pointer types.
7044 * lto-streamer-in.c: Include debug.h.
7045 (dref_queue): New global.
7046 (lto_read_tree_1): Stream in DIE references.
7047 (lto_input_tree): Register DIE references.
7048 (input_function): Stream DECL_DEBUG_ARGS.
7049 * lto-streamer-out.c: Include debug.h.
7050 (lto_write_tree_1): Output DIE references.
7051 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7052 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7053 (output_function): Stream DECL_DEBUG_ARGS.
7054 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7055 Stream DECL_ABSTRACT_ORIGIN.
7056 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7057 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7058 DECL_CONTEXT for file-scope decls.
7059 * lto-streamer.h (struct dref_entry): Declare.
7060 (dref_queue): Likewise.
7061 * cfgexpand.c (pass_expand::execute): Do not call the
7062 outlining_inline_function hook here.
7063 * lto-wrapper.c (debug_obj): New global.
7064 (tool_cleanup): Unlink it if required.
7065 (debug_objcopy): New function.
7066 (run_gcc): Handle early debug sections in the IL files by
7067 extracting them to separate files, partially linkin them and
7068 feeding the result back as result to the linker.
7069 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7070 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7071 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7072 sections into a separate segment.
7073 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7075 (darwin_asm_dwarf_section): Likewise.
7076 (darwin_asm_output_dwarf_offset): Likewise.
7077 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7079 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7080 Alan Hayward <alan.hayward@arm.com>
7081 David Sherwood <david.sherwood@arm.com>
7083 * read-md.h (md_reader::record_potential_iterator_use): Replace
7084 pointer argument with an rtx and an index.
7085 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7086 (apply_mode_iterator): Likewise.
7087 (apply_code_iterator): Likewise.
7088 (apply_int_iterator): Likewise.
7089 (apply_subst_iterator): Likewise.
7090 (record_iterator_use): Likewise.
7091 (record_attribute_use): Likewise.
7092 (md_reader::record_potential_iterator_use): Likewise. Update calls
7093 to record_iterator_use and apply_iterator.
7094 (iterator_use): Replace ptr with x and index.
7095 (attribute_use): Likewise.
7096 (apply_attribute_uses): Update calls to apply_iterator.
7097 (apply_iterators): Likewise. Update initialization of iterator_use.
7098 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7099 and record_potential_iterator_use.
7100 (rtx_reader::read_rtx_operand): Likewise.
7102 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7103 Alan Hayward <alan.hayward@arm.com>
7104 David Sherwood <david.sherwood@arm.com>
7106 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7109 2017-08-21 Richard Biener <rguenther@suse.de>
7112 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7113 at struct end conservatively when comparing common bases.
7115 2017-08-21 Richard Biener <rguenther@suse.de>
7117 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7118 (mem_ref_in_stmt): Remove.
7119 (determine_max_movement): Use ref index to get at the reference.
7120 (invariantness_dom_walker::before_dom_children): Deal with
7121 lim data already initialized.
7122 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7124 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7126 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7127 (TARGET_ISA_ROUND): Ditto.
7128 (TARGET_ROUND): Ditto.
7129 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7130 * config/i386/i386.md: Ditto.
7131 * config/i386/sse.md: Ditto.
7132 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7133 with OPTION_MASK_ISA_SSE4_1.
7135 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7138 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7139 __builtin_ia32_lzcnt_u16.
7141 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7144 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7145 (rs6000_set_current_function): Rewrite function to use it.
7147 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7150 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7151 and TYPE_WARN_IF_NOT_ALIGN.
7152 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7153 (handle_warn_if_not_align): New.
7154 (place_union_field): Call handle_warn_if_not_align.
7155 (place_field): Call handle_warn_if_not_align.
7156 Copy TYPE_WARN_IF_NOT_ALIGN.
7157 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7158 (layout_type): Likewise.
7159 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7161 (tree_decl_common): Add warn_if_not_align.
7162 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7163 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7164 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7165 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7166 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7167 * doc/extend.texi: Document warn_if_not_aligned attribute.
7168 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7170 2017-08-17 Martin Liska <mliska@suse.cz>
7173 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7174 (get_data_dependence): Use it as pointer type.
7175 (distribute_loop): Likewise.
7177 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7179 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7180 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7181 (p8_vmrgow_<mode>_direct): New define_insn.
7182 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7183 handle endianness for vmrgew and vmrgow permute patterns.
7185 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7187 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7188 * config/rs6000/rs6000-cpus.def: Remove comment.
7189 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7190 (POWERPC_MASKS): Likewise.
7191 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7192 use of TARGET_VSX_TIMODE.
7193 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7194 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7195 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7196 (rs6000_option_override_internal): Remove dead code.
7197 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7198 (rs6000_legitimize_reload_address): Likewise.
7199 (rs6000_legitimate_address_p): Likewise.
7200 (rs6000_opt_masks): Delete "vsx-timode".
7201 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7202 from function comment.
7203 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7204 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7205 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7207 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7208 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7209 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7212 (define_peephole2): Likewise.
7214 2017-08-17 Martin Sebor <msebor@redhat.com>
7217 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7218 past the end of an array.
7219 (test_pp_format): Add test cases.
7221 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7223 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7224 missing ECF_NOTHROW flags.
7226 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7229 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7230 operands residing in integer registers.
7231 (*vsx_le_perm_load_<mode>): Likewise.
7232 (*vsx_le_perm_store_<mode>): Likewise.
7233 (define_peephole2): Add peepholes to optimize the above.
7235 2017-08-17 Marek Polacek <polacek@redhat.com>
7238 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7239 to mimic what shorten_compare did. Change the return type to bool.
7240 (fold_cond_expr_with_comparison): Update call to
7241 operand_equal_for_comparison_p.
7242 (fold_ternary_loc): Likewise.
7244 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7246 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7248 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7250 2017-08-17 Richard Biener <rguenther@suse.de>
7252 * tree-ssa-structalias.c (solve_graph): When propagating
7253 to successors update the graphs succ edges and avoid duplicate work.
7255 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7258 * config/i386/i386.c (ix86_option_override_internal): Save target
7259 specific options after ix86_stack_protector_guard_reg was changed.
7261 2017-08-17 Richard Biener <rguenther@suse.de>
7263 PR tree-optimization/81827
7264 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7265 (new_var_info): Initialize it conservatively.
7266 (get_call_vi): Mark register vars.
7267 (new_scalar_tmp_constraint_exp): Likewise.
7268 (handle_rhs_call): Likewise.
7269 (handle_const_call): Likewise.
7270 (create_function_info_for): Likewise.
7271 (solve_constraints): Sort varinfos to separate register from
7272 non-register vars to pack points-to solution bitmaps during
7275 2017-08-17 Marek Polacek <polacek@redhat.com>
7277 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7279 2017-08-17 Richard Biener <rguenther@suse.de>
7281 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7282 to true when overflow is undefined and we saturated the result.
7284 2017-08-17 Alan Modra <amodra@gmail.com>
7287 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7288 Don't use store multiple if only one reg needs saving.
7289 (interesting_frame_related_regno): New function.
7290 (rs6000_frame_related): Don't emit frame info for regs that
7292 (rs6000_emit_epilogue): Likewise.
7294 2017-08-16 Nathan Sidwell <nathan@acm.org>
7296 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7297 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7298 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7299 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7300 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7301 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7302 (verify_type): Adjust for TYPE_BINFO move.
7303 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7304 process TYPE_BINFO directly.
7305 (hash_tree): Likewise.
7306 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7308 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7311 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7313 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7315 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7318 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7319 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7320 Add ix86_binary_operator_ok to insn constraint.
7321 (*iordi_1_bts): Ditto.
7322 (*xordi_1_btc): Ditto.
7323 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7324 Update corresponding peephole2 pattern.
7328 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7330 PR tree-optimization/81832
7331 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7332 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7334 2017-08-16 Marek Polacek <polacek@redhat.com>
7337 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7338 regarding size_zero_node.
7340 2017-08-16 Martin Liska <mliska@suse.cz>
7343 * config.gcc: Respect previously set extra_objs in case
7346 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7348 PR tree-optimization/81835
7349 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7350 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7351 not depend on the phi.
7353 2017-08-16 Alan Modra <amodra@gmail.com>
7355 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7358 2017-08-16 Alan Modra <amodra@gmail.com>
7360 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7361 (save_reg_p): ..into this. Update all callers.
7362 (first_reg_to_save): Simplify.
7364 2017-08-16 Alan Modra <amodra@gmail.com>
7366 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7369 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7373 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7374 constant count if that count is less than 32.
7376 2017-08-15 Nathan Sidwell <nathan@acm.org>
7378 * gcc.c (execute): Emit friendlier message if inferior is killed
7379 by an external cause.
7381 2017-08-15 Richard Biener <rguenther@suse.de>
7383 PR tree-optimization/81790
7384 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7385 CONSTRUCTORs from simplifying and VN.
7387 2017-08-14 Martin Sebor <msebor@redhat.com>
7389 * builtin-attrs.def: Add comments.
7391 2017-08-14 Martin Sebor <msebor@redhat.com>
7394 * doc/extend.texi (attribute nonstring): Document new attribute.
7396 2017-08-14 Martin Sebor <msebor@redhat.com>
7399 * tree-diagnostic.c (default_tree_printer): Handle %G.
7400 * gimple-pretty-print.h (percent_G_format): Declare new function.
7401 * gimple-pretty-print.c (percent_G_format): Define.
7402 * tree-pretty-print.c (percent_K_format): Add argument.
7404 2017-08-14 Martin Sebor <msebor@redhat.com>
7406 PR translation/79998
7407 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7408 Remove a stray space.
7410 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7413 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7414 (*iordi_1_bts): Ditto.
7415 (*xordi_1_btc): Ditto.
7417 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7420 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7421 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7423 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7424 quoted strings, and make more translator-friendly.
7425 (darwin_rs6000_override_options): Likewise.
7426 (rs6000_option_override_internal): Likewise.
7427 (rs6000_return_in_memory): Fix overlong line.
7428 (init_cmulative_args): Use quoted strings, and make more
7429 translator-friendly.
7430 (rs6000_pass_by_reference): Fix overlong line.
7431 (def_builtin): Use quoted strings.
7432 (altivec_expand_predicate_builtin): Use quoted strings, and make
7433 more translator-friendly.
7434 (htm_expand_builtin): Use quoted strings.
7435 (cpu_expand_builtin): Use quoted strings, and make more
7436 translator-friendly.
7437 (altivec_expand_builtin): Likewise.
7438 (paired_expand_predicate_builtin): Likewise.
7439 (rs6000_invalid_builtin): Likewise.
7440 (builtin_function_type): Use quoted strings.
7441 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7442 more translator-friendly.
7443 (rs6000_trampoline_init): Likewise.
7444 (rs6000_handle_altivec_attribute): Likewise.
7445 (rs6000_inner_target_options): Use quoted strings.
7446 (rs6000_disable_incompatible_switches): Likewise.
7447 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7448 strings, and make more translator-friendly.
7449 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7451 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7453 PR tree-optimization/81799
7454 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7455 cond_expr to simple gimple operand.
7457 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7460 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7462 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7466 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7467 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7468 TARGET_GASISR_PROLOGUES.
7469 * config/avr/avr.c (avr_option_override): Same.
7470 (avr_pass_pre_proep::execute): Same.
7472 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7475 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7476 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7478 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7480 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7481 define_insn_and_split. Split to a memory load from 0 in
7482 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7483 using PTR mode iterator.
7484 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7485 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7486 (*add_tp_<mode>): Redefine as define_insn_and_split.
7487 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7488 address space. Merge with *add_tp_x32 using PTR mode iterator.
7489 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7490 Split to an add with a memory load from 0 in
7491 DEFAULT_TLS_SEG_REG address space.
7493 2017-08-12 Andrew Pinski <apinski@cavium.com>
7495 * config/aarch64/aarch64-option-extensions.def (rdma):
7496 Fix feature string to what Linux prints out in /proc/cpuinfo.
7498 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7501 * dwarf2out.c (modified_type_die): For C typedef types that have
7502 an ultimate origin, process the ultimate origin instead of the
7504 (gen_typedef_die): Assert that input DECLs have no ultimate
7506 (gen_type_die_with_usage): For typedef variants that have an
7507 ultimate origin, just call gen_decl_die on the original DECL.
7508 (process_scope_var): Avoid creating DIEs for local typedefs and
7509 concrete static variables.
7511 2017-08-12 Alan Modra <amodra@gmail.com>
7515 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7516 match gnu-user.h startfile.
7517 (ENDFILE_LINUX_SPEC): Similarly.
7519 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7522 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7524 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7526 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7527 * config/aarch64/aarch64.md (mov<mode>): Change.
7528 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7529 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7530 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7532 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7534 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7535 for storage order barriers.
7537 2017-08-11 Martin Liska <mliska@suse.cz>
7540 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7541 variables of void type.
7543 2017-08-11 Martin Liska <mliska@suse.cz>
7545 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7546 TARGET_SUPPORTS_ALIASES.
7547 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7548 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7549 (optimize_weakref): Likewise.
7550 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7551 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7552 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7554 2017-08-11 Martin Liska <mliska@suse.cz>
7557 * config/i386/i386.c (make_resolver_func): Do complete
7558 refactoring of the function.
7560 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7563 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7564 * config/i386/i386.c (ix86_stack_protect_guard): Use
7565 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7566 * doc/invoke.texi (x86 Options): Document
7567 -mstack-protector-guard-symbol= option.
7569 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7571 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7572 * config/i386/i386.c (ix86_split_stack_guard): New function.
7573 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7574 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7575 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7576 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7577 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7578 (split_stack_space_check): Call ix86_split_stack_guard.
7580 2017-08-10 Martin Sebor <msebor@redhat.com>
7582 * print-tree.c (print_node): Print location using the established
7584 Replace spaces with colons.
7585 (debug_raw, debug): Ditto.
7587 2017-08-10 Martin Sebor <msebor@redhat.com>
7590 * pretty-print.c (pp_format): Correct the handling of %s precision.
7592 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7595 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7597 (ix86_finalize_stack_frame_flags): This. Also clear
7598 frame_pointer_needed if -fno-omit-frame-pointer is used without
7600 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7601 with ix86_finalize_stack_frame_flags.
7602 (ix86_expand_epilogue): Likewise.
7603 (ix86_expand_split_stack_prologue): Likewise.
7604 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7606 2017-08-10 Martin Liska <mliska@suse.cz>
7609 * c-attribs.c (handle_target_attribute):
7610 Report warning for an empty string argument of target attribute.
7612 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7615 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7617 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7618 or DECL_NONLOCAL labels.
7619 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7620 or DECL_NONLOCAL labels here.
7622 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7625 to indicate when early gimple folding has been disabled.
7626 (rs6000_gimple_fold_builtin): Add debug content.
7627 (rs6000_invalid_builtin): Fix whitespace.
7628 (rs6000_expand_builtin): Fix whitespace.
7629 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7631 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7634 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7635 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7637 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7639 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7640 (qdf24xx): Likewise.
7641 * config/aarch64/aarch64.md: Include falkor.md.
7642 * config/aarch64/falkor.md: New.
7644 2017-08-09 Marek Polacek <polacek@redhat.com>
7647 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7648 * diagnostic.c (emit_diagnostic): Add a comment.
7649 (emit_diagnostic_valist): New function.
7651 2017-08-09 Marek Polacek <polacek@redhat.com>
7654 * input.c (make_location): New overload.
7655 * input.h (make_location): Declare.
7657 2017-08-08 Alan Modra <amodra@gmail.com>
7658 H.J. Lu <hongjiu.lu@intel.com>
7661 * gcc.c (NO_PIE_SPEC): Delete.
7662 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7664 (LINK_PIE_SPEC): ..to here.
7665 (LINK_COMMAND_SPEC): Support -no-pie.
7666 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7667 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7669 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7670 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7671 (ENDFILE_CRTEND_SPEC): Similarly.
7673 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7676 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7677 (mstack-protector-guard-offset=): Ditto.
7678 * config/i386/i386.c (ix86_option_override): Handle
7679 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7681 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7682 ix86_stack_protect_guard_offset variables.
7683 (TARGET_STACK_PROTECT_GUARD): Always define.
7684 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7685 and -mstack-protector-guard-offset= options.
7687 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7689 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7690 boundary case for the last candidate.
7692 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7694 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7695 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7697 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7700 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7701 * config/aarch64/aarch64-builtins.c
7702 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7703 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7704 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7706 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7707 Andrew Pinski <pinskia@gmail.com>
7710 * internal-fn.def (XORSIGN): New.
7711 * optabs.def (xorsign_optab): New.
7712 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7713 (convert_expand_mult_copysign): New.
7714 (pass_optimize_widening_mul::execute): Call
7715 convert_expand_mult_copysign.
7717 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7719 PR tree-optimization/81354
7720 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7721 Insert on edges rather than explicitly creating landing pads.
7722 (analyze_candidates_and_replace): Commit edge inserts.
7724 2017-08-08 Richard Biener <rguenther@suse.de>
7727 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7728 (expand_simple_operations): Also look through ADDR_EXPRs with
7729 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7731 2017-08-08 Richard Biener <rguenther@suse.de>
7733 PR tree-optimization/81723
7734 * tree-vect-slp.c (struct bst_traits): New hash traits.
7735 (bst_fail): New global.
7736 (vect_build_slp_tree_2): New worker, split out from ...
7737 (vect_build_slp_tree): ... this now wrapping it with using
7738 bst_fail set to cache SLP tree build fails. Properly handle
7740 (vect_analyze_slp_instance): Allocate and free bst_fail.
7742 2017-08-08 Martin Liska <mliska@suse.cz>
7745 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7746 LABEL_DECLs that can be from a different function.
7748 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7750 PR tree-optimization/81744
7751 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7752 loop's number of iterations.
7754 2017-08-08 Martin Liska <mliska@suse.cz>
7756 * asan.c: Include header files.
7757 * attribs.c (build_decl_attribute_variant): New function moved
7759 (build_type_attribute_qual_variant): Likewise.
7760 (cmp_attrib_identifiers): Likewise.
7761 (simple_cst_list_equal): Likewise.
7762 (omp_declare_simd_clauses_equal): Likewise.
7763 (attribute_value_equal): Likewise.
7764 (comp_type_attributes): Likewise.
7765 (build_type_attribute_variant): Likewise.
7766 (lookup_ident_attribute): Likewise.
7767 (remove_attribute): Likewise.
7768 (merge_attributes): Likewise.
7769 (merge_type_attributes): Likewise.
7770 (merge_decl_attributes): Likewise.
7771 (merge_dllimport_decl_attributes): Likewise.
7772 (handle_dll_attribute): Likewise.
7773 (attribute_list_equal): Likewise.
7774 (attribute_list_contained): Likewise.
7775 * attribs.h (lookup_attribute): New function moved from tree.[ch].
7776 (lookup_attribute_by_prefix): Likewise.
7777 * bb-reorder.c: Include header files.
7778 * builtins.c: Likewise.
7779 * calls.c: Likewise.
7780 * cfgexpand.c: Likewise.
7781 * cgraph.c: Likewise.
7782 * cgraphunit.c: Likewise.
7783 * convert.c: Likewise.
7784 * dwarf2out.c: Likewise.
7785 * final.c: Likewise.
7786 * fold-const.c: Likewise.
7787 * function.c: Likewise.
7788 * gimple-expr.c: Likewise.
7789 * gimple-fold.c: Likewise.
7790 * gimple-pretty-print.c: Likewise.
7791 * gimple.c: Likewise.
7792 * gimplify.c: Likewise.
7793 * hsa-common.c: Likewise.
7794 * hsa-gen.c: Likewise.
7795 * internal-fn.c: Likewise.
7796 * ipa-chkp.c: Likewise.
7797 * ipa-cp.c: Likewise.
7798 * ipa-devirt.c: Likewise.
7799 * ipa-fnsummary.c: Likewise.
7800 * ipa-inline.c: Likewise.
7801 * ipa-visibility.c: Likewise.
7803 * lto-cgraph.c: Likewise.
7804 * omp-expand.c: Likewise.
7805 * omp-general.c: Likewise.
7806 * omp-low.c: Likewise.
7807 * omp-offload.c: Likewise.
7808 * omp-simd-clone.c: Likewise.
7809 * opts-global.c: Likewise.
7810 * passes.c: Likewise.
7811 * predict.c: Likewise.
7812 * sancov.c: Likewise.
7813 * sanopt.c: Likewise.
7814 * symtab.c: Likewise.
7815 * toplev.c: Likewise.
7816 * trans-mem.c: Likewise.
7817 * tree-chkp.c: Likewise.
7818 * tree-eh.c: Likewise.
7819 * tree-into-ssa.c: Likewise.
7820 * tree-object-size.c: Likewise.
7821 * tree-parloops.c: Likewise.
7822 * tree-profile.c: Likewise.
7823 * tree-ssa-ccp.c: Likewise.
7824 * tree-ssa-live.c: Likewise.
7825 * tree-ssa-loop.c: Likewise.
7826 * tree-ssa-sccvn.c: Likewise.
7827 * tree-ssa-structalias.c: Likewise.
7828 * tree-ssa.c: Likewise.
7829 * tree-streamer-in.c: Likewise.
7830 * tree-vectorizer.c: Likewise.
7831 * tree-vrp.c: Likewise.
7833 * ubsan.c: Likewise.
7834 * varasm.c: Likewise.
7835 * varpool.c: Likewise.
7836 * tree.c: Remove functions moved to attribs.[ch].
7838 * config/aarch64/aarch64.c: Add attrs.h header file.
7839 * config/alpha/alpha.c: Likewise.
7840 * config/arc/arc.c: Likewise.
7841 * config/arm/arm.c: Likewise.
7842 * config/avr/avr.c: Likewise.
7843 * config/bfin/bfin.c: Likewise.
7844 * config/c6x/c6x.c: Likewise.
7845 * config/cr16/cr16.c: Likewise.
7846 * config/cris/cris.c: Likewise.
7847 * config/darwin.c: Likewise.
7848 * config/epiphany/epiphany.c: Likewise.
7849 * config/fr30/fr30.c: Likewise.
7850 * config/frv/frv.c: Likewise.
7851 * config/ft32/ft32.c: Likewise.
7852 * config/h8300/h8300.c: Likewise.
7853 * config/i386/winnt.c: Likewise.
7854 * config/ia64/ia64.c: Likewise.
7855 * config/iq2000/iq2000.c: Likewise.
7856 * config/lm32/lm32.c: Likewise.
7857 * config/m32c/m32c.c: Likewise.
7858 * config/m32r/m32r.c: Likewise.
7859 * config/m68k/m68k.c: Likewise.
7860 * config/mcore/mcore.c: Likewise.
7861 * config/microblaze/microblaze.c: Likewise.
7862 * config/mips/mips.c: Likewise.
7863 * config/mmix/mmix.c: Likewise.
7864 * config/mn10300/mn10300.c: Likewise.
7865 * config/moxie/moxie.c: Likewise.
7866 * config/msp430/msp430.c: Likewise.
7867 * config/nds32/nds32-isr.c: Likewise.
7868 * config/nds32/nds32.c: Likewise.
7869 * config/nios2/nios2.c: Likewise.
7870 * config/nvptx/nvptx.c: Likewise.
7871 * config/pa/pa.c: Likewise.
7872 * config/pdp11/pdp11.c: Likewise.
7873 * config/powerpcspe/powerpcspe.c: Likewise.
7874 * config/riscv/riscv.c: Likewise.
7875 * config/rl78/rl78.c: Likewise.
7876 * config/rx/rx.c: Likewise.
7877 * config/s390/s390.c: Likewise.
7878 * config/sh/sh.c: Likewise.
7879 * config/sol2.c: Likewise.
7880 * config/sparc/sparc.c: Likewise.
7881 * config/spu/spu.c: Likewise.
7882 * config/stormy16/stormy16.c: Likewise.
7883 * config/tilegx/tilegx.c: Likewise.
7884 * config/tilepro/tilepro.c: Likewise.
7885 * config/v850/v850.c: Likewise.
7886 * config/vax/vax.c: Likewise.
7887 * config/visium/visium.c: Likewise.
7888 * config/xtensa/xtensa.c: Likewise.
7890 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7893 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7894 constraints since the -mupper-regs-* switches have been
7896 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7897 into a vector from a double word element that was extracted from
7898 another vector, and eliminate extra XXPERMDI instructions.
7899 (vsx_concat_<mode>_2): Likewise.
7900 (vsx_concat_<mode>_3): Likewise.
7901 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7902 concat to allow optimizing inserts from previous extracts.
7904 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7906 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7907 memory reference to a SSP offset in TLS address space.
7908 (ix86_print_operand) <case '@'>: Remove.
7909 (ix86_print_operand_punct_valid_p): Remove '@' code.
7910 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7912 (stack_tls_protect_set_<mode>): Remove.
7913 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7914 (stack_tls_protect_test_<mode>): Remove.
7915 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7917 2017-08-07 Olivier Hainque <hainque@adacore.com>
7920 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7922 2017-08-07 Douglas Rupp <rupp@adacore.com>
7924 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7925 variable was referenced as multidir in command.
7927 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7930 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7933 2017-08-07 Martin Liska <mliska@suse.cz>
7935 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7936 * config/rl78/rl78.c: Add include of attribs.h.
7937 * config/sh/sh.c: Likewise.
7938 * config/v850/v850.c: Likewise.
7940 2017-08-07 Tom de Vries <tom@codesourcery.com>
7943 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7945 2017-08-07 Martin Liska <mliska@suse.cz>
7947 * config/mips/mips.c: Include attribs.h.
7949 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7952 * doc/invoke.texi: Document change in behvaior for -Ofast for
7955 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7957 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7959 (aarch64_pop_regs): Likewise.
7960 (aarch64_gen_load_pair): Likewise.
7961 (aarch64_save_callee_saves): Likewise.
7962 (aarch64_restore_callee_saves): Likewise.
7964 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7966 * config/i386/i386.c: Revert the last change.
7968 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7971 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7973 (ix86_finalize_stack_frame_flags): This. Also clear
7974 frame_pointer_needed if -fno-omit-frame-pointer is used without
7976 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7977 with ix86_finalize_stack_frame_flags.
7978 (ix86_expand_epilogue): Likewise.
7979 (ix86_expand_split_stack_prologue): Likewise.
7981 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7984 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7985 to P_AES for Westmere.
7987 2017-08-07 Jonathan Yong <10walls@gmail.com>
7989 * config/i386/mingw.opt (fset-stack-executable): Removed.
7990 * config/i386/cygming.opt (fset-stack-executable): Moved
7992 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7994 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7996 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7998 2017-08-07 Marek Polacek <polacek@redhat.com>
8001 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8003 2017-08-07 Martin Liska <mliska@suse.cz>
8005 * attribs.h (canonicalize_attr_name): New function.
8006 (cmp_attribs): Move from c-format.c and adjusted.
8007 (is_attribute_p): Moved from tree.h.
8008 * tree-inline.c: Add new includes.
8009 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8010 (private_is_attribute_p): Remove.
8011 (private_lookup_attribute): Likewise.
8012 (private_lookup_attribute_by_prefix): Simplify.
8013 (remove_attribute): Use is_attribute_p.
8014 * tree.h: Remove removed declarations.
8016 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8019 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8020 instead of computing it in the function. Formatting fix.
8021 (expand_case): Don't rely on default_edge being the first edge,
8022 clear it if removing it, pass default_edge to
8023 emit_case_dispatch_table.
8024 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8027 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8029 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8030 insn in the function, emit NOP after the insn.
8032 2017-08-06 Tom de Vries <tom@codesourcery.com>
8034 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8037 2017-08-06 Tom de Vries <tom@codesourcery.com>
8039 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8042 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8044 PR tree-optimization/57371
8045 * match.pd: New pattern.
8047 2017-08-04 Marek Polacek <polacek@redhat.com>
8050 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8051 perform the computation in offset_int.
8053 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8055 PR tree-optimization/81136
8056 * tree-vectorizer.h: Include tree-hash-traits.h.
8057 (vec_base_alignments): New typedef.
8058 (vec_info): Add a base_alignments field.
8059 (vect_record_base_alignments): Declare.
8060 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8062 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8063 (create_data_ref): Add an is_conditional_in_stmt argument.
8064 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8065 the is_conditional_in_stmt field.
8066 (data_ref_loc): Add an is_conditional_in_stmt field.
8067 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8068 (find_data_references_in_stmt): Update call to create_data_ref.
8069 (graphite_find_data_references_in_stmt): Likewise.
8070 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8071 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8072 (vect_record_base_alignment): New function.
8073 (vect_record_base_alignments): Likewise.
8074 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8075 for nested statements even if we fail to compute a misalignment.
8076 Use pooled base alignments for unconditional references.
8077 (vect_find_same_alignment_drs): Compare base addresses instead
8079 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8080 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8082 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8084 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8085 Add an explicit name for the enum. Use auto_vec for slp_instances
8087 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8089 (_bb_vec_info): Add a constructor and destructor.
8090 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8091 (destroy_loop_vec_info): Delete.
8092 (vect_destroy_datarefs): Likewise.
8093 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8094 (vec_info::vec_info): New function.
8095 (vec_info::~vec_info): Likewise.
8096 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8097 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8098 destroy_loop_vec_info.
8099 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8100 (_loop_vec_info::_loop_vec_info): ...this.
8101 (destroy_loop_vec_info): Replace with...
8102 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8103 the stmt_vec_infos. Leave handling of vec_info information to its
8104 destructor. Remove explicit vector releases.
8105 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8106 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8107 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8108 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8109 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8110 (destroy_bb_vec_info): Replace with...
8111 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8112 information to its destructor.
8113 (vect_slp_analyze_bb_1): Use new and delete instead of
8114 new_bb_vec_info and destroy_bb_vec_info.
8115 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8118 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8120 * tree-data-ref.h (subscript): Add access_fn field.
8121 (data_dependence_relation): Add could_be_independent_p.
8122 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8123 (same_access_functions): Move to tree-data-ref.c.
8124 * tree-data-ref.c (ref_contains_union_access_p): New function.
8125 (access_fn_component_p): Likewise.
8126 (access_fn_components_comparable_p): Likewise.
8127 (dr_analyze_indices): Add a reference to access_fn_component_p.
8128 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8130 (constant_access_functions): Likewise.
8131 (add_other_self_distances): Likewise.
8132 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8133 (initialize_data_dependence_relation): Use XCNEW and remove
8134 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8135 of access functions that have the same type. Allow the
8136 subsequence to end with different bases in some circumstances.
8137 Record the chosen access functions in SUB_ACCESS_FN.
8138 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8139 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8140 (subscript_dependence_tester_1): Likewise dra and drb.
8141 (build_classic_dist_vector): Update calls accordingly.
8142 (subscript_dependence_tester): Likewise.
8143 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8144 DDR_COULD_BE_INDEPENDENT_P.
8145 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8146 comp_alias_ddrs instead of may_alias_ddrs.
8147 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8149 (vect_analyze_data_ref_dependence): Use it if
8150 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8151 distance vectors if that fails.
8152 (dependence_distance_ge_vf): New function.
8153 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8154 LOOP_VINFO_MAY_ALIAS_DDRS.
8156 2017-08-04 Richard Biener <rguenther@suse.de>
8159 * fold-const.c (fold_binary_loc): Properly restrict
8160 minus_var0 && minus_var1 case when associating undefined overflow
8163 2017-08-04 Tom de Vries <tom@codesourcery.com>
8165 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8167 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8169 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8170 Don't start diagnostic messages with a capital letter.
8171 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8173 (rs6000_invalid_builtin): Likewise.
8174 (rs6000_trampoline_init): Likewise.
8176 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8179 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8180 after setting changeable df flags.
8182 2017-08-03 Richard Biener <rguenther@suse.de>
8184 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8185 up if the use is in USE - X.
8187 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8189 * toplev.c (dumpfile.h): New include.
8190 (internal_error_reentered): New static function. Use it...
8191 (internal_error_function): ...here to handle reentered internal_error.
8193 2017-08-03 Richard Biener <rguenther@suse.de>
8196 * fold-const.c (split_tree): Add minus_var and minus_con
8197 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8198 here but always use minus_*.
8199 (associate_trees): Assert we never associate with MINUS_EXPR
8200 and NULL first operand. Do not recurse for PLUS_EXPR operands
8201 when associating as MINUS_EXPR either.
8202 (fold_binary_loc): Track minus_var and minus_con.
8204 2017-08-03 Tom de Vries <tom@codesourcery.com>
8207 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8208 ACCEL_COMPILER, apply finish_options on
8209 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8211 2017-08-03 Tom de Vries <tom@codesourcery.com>
8214 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8215 function_entry_patch_area_size > 0.
8217 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8220 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8221 instead of 10??LU, perform unit multiplication in wide_int,
8222 don't change alloc_object_size_limit if the limit is larger
8225 PR tree-optimization/81655
8226 PR tree-optimization/81588
8227 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8228 the case when ranges[i].low and high are 1 for unsigned type with
8232 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8233 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8235 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8237 * tree-vrp.h: Add include guard.
8239 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8242 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8243 (ud2): New insn pattern.
8244 * config/i386/i386.c (ix86_expand_epilogue):
8245 For naked functions, generate ud2 instead of trap insn.
8247 2017-08-02 Marek Polacek <polacek@redhat.com>
8250 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8252 2017-08-02 Tom de Vries <tom@codesourcery.com>
8253 Cesar Philippidis <cesar@codesourcery.com>
8255 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8256 Add missing edge probabilities.
8258 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8260 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8263 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8266 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8267 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8268 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8269 prologue_seq sequences - if any.
8271 2017-08-02 Richard Biener <rguenther@suse.de>
8273 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8274 via vectors if supported, integer extracts via punning if supported
8275 or otherwise vector extracts.
8277 2017-08-02 Richard Biener <rguenther@suse.de>
8279 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8281 (bitmap_insert_into_set): ... this.
8283 2017-08-02 Richard Biener <rguenther@suse.de>
8285 PR tree-optimization/81633
8287 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8289 PR tree-optimization/71752
8290 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8292 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8294 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8295 (machine_function::call_ms2sysv_pad_out): Remove field.
8296 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8297 (ix86_compute_frame_layout): Likewise.
8299 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8302 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8303 attribute with interrupt attribute.
8305 2017-08-01 Andrew Pinski <apinski@cavium.com>
8307 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8308 BIT_INSERT_EXPR's operand 1
8309 to see if the types precision matches.
8311 2017-08-01 Martin Liska <mliska@suse.cz>
8314 * builtins.c (expand_builtin_memcpy_args): Remove.
8315 (expand_builtin_memcpy): Call newly added function
8316 expand_builtin_memory_copy_args.
8317 (expand_builtin_memcpy_with_bounds): Likewise.
8318 (expand_builtin_mempcpy): Remove last argument.
8319 (expand_builtin_mempcpy_with_bounds): Likewise.
8320 (expand_builtin_memory_copy_args): New function created from
8321 expand_builtin_mempcpy_args with small modifications.
8322 (expand_builtin_mempcpy_args): Remove.
8323 (expand_builtin_stpcpy): Remove unused argument.
8324 (expand_builtin): Likewise.
8325 (expand_builtin_with_bounds): Likewise.
8327 2017-08-01 Martin Liska <mliska@suse.cz>
8330 Make mempcpy more optimal (PR middle-end/70140).
8332 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8335 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8336 __builtin_vec_cmpne verify both arguments are compatible vectors
8337 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8338 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8339 move computation of aligned to after checking the argument types.
8343 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8344 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8347 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8348 Doug Rupp <rupp@adacore.com>
8349 Olivier Hainque <hainque@adacore.com>
8351 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8352 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8354 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8355 for TARGET_OS_CPP_BUILTIN.
8356 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8357 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8359 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8360 passing required abi options to the assembler for EABI configurations.
8361 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8362 of .text.hot and .text.unlikely sections for kernel modules when
8363 using ARM style exceptions.
8364 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8365 options. Add EXTRA_CC1_SPEC.
8366 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8368 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8370 (ARM_TARGET2_DWARF_FORMAT): Define.
8371 * config/arm/t-vxworks: Adjust multilib control to removal of the
8372 Diab command line options.
8374 2017-08-01 Martin Liska <mliska@suse.cz>
8376 PR gcov-profile/81561
8377 * gcov.c (unblock): Make unblocking safe as we need to preserve
8378 index correspondence of blocks and block_lists.
8380 2017-08-01 Richard Biener <rguenther@suse.de>
8382 PR tree-optimization/81181
8383 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8384 (compute_antic): ... end of iteration here.
8386 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8388 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8389 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8390 (ftree-slp-vectorize): Likewise.
8391 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8392 can no longer be set independent of flag_tree_loop_vectorize.
8393 * omp-general.c (emp_max_vf): Likewise.
8394 * opts.c (enable_fdo_optimizations): Remove references to
8395 flag_tree_vectorize, these are now implicit.
8396 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8397 and leave it for the options machinery.
8399 2017-08-01 Martin Liska <mliska@suse.cz>
8402 * builtins.c (expand_builtin_memcpy_args): Remove.
8403 (expand_builtin_memcpy): Call newly added function
8404 expand_builtin_memory_copy_args.
8405 (expand_builtin_memcpy_with_bounds): Likewise.
8406 (expand_builtin_mempcpy): Remove last argument.
8407 (expand_builtin_mempcpy_with_bounds): Likewise.
8408 (expand_builtin_memory_copy_args): New function created from
8409 expand_builtin_mempcpy_args with small modifications.
8410 (expand_builtin_mempcpy_args): Remove.
8411 (expand_builtin_stpcpy): Remove unused argument.
8412 (expand_builtin): Likewise.
8413 (expand_builtin_with_bounds): Likewise.
8415 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8418 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8419 print "ds:" only for immediates in generic address space.
8421 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8424 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8425 (ix86_function_ok_for_sibcall): Return false for naked functions.
8427 2017-08-01 Richard Biener <rguenther@suse.de>
8429 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8430 (compute_antic): Seed worklist with exit block predecessors.
8431 * cfganal.c (dfs_find_deadend): For a cycle return the source
8432 of the edge closing it.
8434 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8436 * config/aarch64/aarch64.c
8437 (aarch64_can_const_movi_rtx_p): Move 0 check.
8439 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8441 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8442 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8444 * match.pd: Ditto in address comparison pattern.
8446 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8448 PR tree-optimization/81627
8449 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8450 closed ssa form for store-store chain.
8452 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8454 PR tree-optimization/81620
8455 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8456 for store-store chain.
8458 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8460 PR tree-optimization/81588
8461 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8462 ranges[i].in_p, invert comparison code ccode. For >/>=,
8463 swap rhs1 and rhs2 and comparison code unconditionally,
8464 for </<= don't do that. Don't swap rhs1/rhs2 again if
8465 ranges[i].in_p, instead invert comparison code ccode if
8466 opcode or oe->rank is BIT_IOR_EXPR.
8469 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8470 a direct optab to conversion optab.
8471 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8472 with GET_MODE_INNER as last argument instead of optab_handler.
8473 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8474 vector extraction if possible and optab is available.
8475 * expr.c (store_constructor): Use convert_optab_handler instead
8476 of optab_handler. Use vector initialization from smaller
8477 vectors if possible and optab is available.
8478 * tree-vect-stmts.c (vectorizable_load): Likewise.
8479 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8481 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8482 of vec_init from half-sized vectors with the same element mode.
8483 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8484 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8485 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8486 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8487 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8488 after mode in gen_vec_extract* calls.
8489 (vec_extract<mode>): Renamed to ...
8490 (vec_extract<mode><ssescalarmodelower>): ... this.
8491 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8492 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8493 element mode after mode in gen_vec_init* calls.
8494 (VEC_INIT_HALF_MODE): New mode iterator.
8495 (vec_init<mode>): Renamed to ...
8496 (vec_init<mode><ssescalarmodelower>): ... this.
8497 (vec_init<mode><ssehalfvecmodelower>): New expander.
8498 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8499 (vec_extractv2sfsf): ... this.
8500 (vec_initv2sf): Renamed to ...
8501 (vec_initv2sfsf): ... this.
8502 (vec_extractv2si): Renamed to ...
8503 (vec_extractv2sisi): ... this.
8504 (vec_initv2si): Renamed to ...
8505 (vec_initv2sisi): ... this.
8506 (vec_extractv4hi): Renamed to ...
8507 (vec_extractv4hihi): ... this.
8508 (vec_initv4hi): Renamed to ...
8509 (vec_initv4hihi): ... this.
8510 (vec_extractv8qi): Renamed to ...
8511 (vec_extractv8qiqi): ... this.
8512 (vec_initv8qi): Renamed to ...
8513 (vec_initv8qiqi): ... this.
8514 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8515 (vec_init<mode>): Renamed to ...
8516 (vec_init<mode><VEC_base_l>): ... this.
8517 (vec_extract<mode>): Renamed to ...
8518 (vec_extract<mode><VEC_base_l>): ... this.
8519 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8520 (vec_initv2sfsf): ... this.
8521 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8522 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8523 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8524 element mode after mode in gen_vec_init* calls.
8525 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8526 (vec_init<mode><Vel>): ... this.
8527 (vec_extract<mode>): Renamed to ...
8528 (vec_extract<mode><Vel>): ... this.
8529 * config/aarch64/iterators.md (Vel): New mode attribute.
8530 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8531 Add element mode after mode in gen_vec_extract* calls.
8532 * config/s390/vector.md (non_vec_l): New mode attribute.
8533 (vec_extract<mode>): Renamed to ...
8534 (vec_extract<mode><non_vec_l>): ... this.
8535 (vec_init<mode>): Renamed to ...
8536 (vec_init<mode><non_vec_l>): ... this.
8537 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8538 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8540 * config/arm/iterators.md (V_elem_l): New mode attribute.
8541 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8542 (vec_extract<mode><V_elem_l>): ... this.
8543 (vec_extractv2di): Renamed to ...
8544 (vec_extractv2didi): ... this.
8545 (vec_init<mode>): Renamed to ...
8546 (vec_init<mode><V_elem_l>): ... this.
8547 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8548 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8549 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8550 Add element mode after gen_vec_extract* calls.
8551 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8552 (vec_init<mode><unitmode>): ... this.
8553 (vec_extract<mode>): Renamed to ...
8554 (vec_extract<mode><unitmode>): ... this.
8555 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8556 (vec_init<mode><unitmode>): ... this.
8557 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8558 (vec_initv2sfsf): ... this.
8559 (vec_extractv2sf): Renamed to ...
8560 (vec_extractv2sfsf): ... this.
8561 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8562 Add element mode after gen_vec_extract* calls.
8563 * config/mips/mips.md (unitmode): New mode iterator.
8564 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8565 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8566 * config/spu/spu.md (inner_l): New mode attribute.
8567 (vec_init<mode>): Renamed to ...
8568 (vec_init<mode><inner_l>): ... this.
8569 (vec_extract<mode>): Renamed to ...
8570 (vec_extract<mode><inner_l>): ... this.
8571 * config/sparc/sparc.md (veltmode): New mode iterator.
8572 (vec_init<VMALL:mode>): Renamed to ...
8573 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8574 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8575 (vec_initv2sisi): ... this.
8576 (vec_initv2sf): Renamed to ...
8577 (vec_initv2sfsf): ... this.
8578 (vec_extractv2sf): Renamed to ...
8579 (vec_extractv2sfsf): ... this.
8580 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8581 (vec_init<mode>): Renamed to ...
8582 (vec_init<mode><VEC_base_l>): ... this.
8583 (vec_extract<mode>): Renamed to ...
8584 (vec_extract<mode><VEC_base_l>): ... this.
8585 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8586 (vec_initv2sfsf): ... this.
8587 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8588 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8589 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8590 gen_vec_init* calls.
8592 2017-08-01 Richard Biener <rguenther@suse.de>
8594 PR tree-optimization/81297
8595 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8596 TREE_OVERFLOW from INTEGER_CSTs.
8598 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8600 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8602 2017-07-31 Carl Love <cel@us.ibm.com>
8604 * config/rs6000/rs6000-c: Add support for built-in functions
8605 vector signed char vec_xl_be (signed long long, signed char *);
8606 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8607 vector signed int vec_xl_be (signed long long, signed int *);
8608 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8609 vector signed long long vec_xl_be (signed long long, signed long long *);
8610 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8611 vector signed short vec_xl_be (signed long long, signed short *);
8612 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8613 vector double vec_xl_be (signed long long, double *);
8614 vector float vec_xl_be (signed long long, float *);
8615 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8616 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8617 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8619 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8620 (altivec_expand_builtin): Add switch statement to call
8621 altivec_expand_xl_be for each builtin.
8622 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8623 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8624 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8625 __builtin_vsx_le_be_v16qi.
8626 * doc/extend.texi: Update the built-in documentation file for the
8627 new built-in functions.
8629 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8632 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8634 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8636 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8638 * config.gcc: Add z14.
8639 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8640 CPU model numbers for z13s and z14.
8641 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8643 * config/s390/s390-opts.h (enum processor_type): Rename
8644 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8645 * config/s390/s390.c (processor_table): Add field for CPU name to
8646 be passed to Binutils.
8647 (s390_asm_output_machine_for_arch): Use the new field in
8648 processor_table for Binutils.
8649 (s390_expand_builtin): Replace arch12 with z14.
8650 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8651 (s390_get_sched_attrmask): Likewise.
8652 (s390_get_unit_mask): Likewise.
8653 * config/s390/s390.opt: Add z14 to processor_type enum.
8655 2017-07-31 Martin Jambor <mjambor@suse.cz>
8658 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8659 regardless of optimization level.
8661 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8662 Martin Liska <mliska@suse.cz>
8664 * predict.def: Remove old comment and adjust probability.
8665 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8668 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8671 * config/i386/i386.c (ix86_function_naked): New function.
8672 (ix86_can_use_return_insn_p): Return false for naked functions.
8673 (ix86_expand_prologue): Skip prologue for naked functions.
8674 (ix86_expand_epilogue): Skip epilogue for naked functions
8675 and emit trap instruction.
8676 (ix86_warn_func_return): New function.
8677 (ix86_attribute_table): Add "naked" attribute specification.
8678 (TARGET_WARN_FUNC_RETURN): Define.
8679 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8681 2017-07-31 Martin Liska <mliska@suse.cz>
8683 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8684 (dump_gimple_bb_header): Always dump BB info.
8685 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8687 2017-07-31 Martin Liska <mliska@suse.cz>
8690 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8691 also with current_function_decl non-null equality.
8693 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8696 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8697 change type to the element type, instead add eltype variable and
8698 use it where we are interested in the element type.
8700 PR tree-optimization/81603
8701 * ipa-polymorphic-call.c
8702 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8703 offset arithmetic in offset_int, bail out if the resulting bit offset
8704 doesn't fit into shwi.
8706 2017-07-31 Martin Liska <mliska@suse.cz>
8708 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8709 (gimplify_save_expr): Fix comment.
8711 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8714 * config/i386/i386.c (ix86_function_arg): Update arguments for
8716 (ix86_compute_frame_layout): Set the initial stack offset to
8717 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8718 INCOMING_FRAME_SP_OFFSET.
8719 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8720 stack before exception handler returns.
8721 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8722 the 'ERROR_CODE' for exception handler.
8724 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8726 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8727 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8728 (ASM_OUTPUT_REG_POP): Ditto.
8729 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8730 instead of asm_fprintf to output pure string.
8732 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8734 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8735 to imported_module_or_decl hook.
8736 (debug_nothing_tree_tree_tree_bool): Remove.
8737 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8738 * debug.c (do_nothing_debug_hooks): Use
8739 debug_nothing_tree_tree_tree_bool_bool instead of
8740 debug_nothing_tree_tree_tree_bool.
8741 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8742 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8743 * sdbout.c (sdb_debug_hooks): Likewise.
8744 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8745 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8747 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8748 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8749 attribute, don't add anything.
8751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8753 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8754 (fold_build2_stat_loc): Likewise.
8755 (fold_build3_stat_loc): Likewise.
8756 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8757 (fold_build1_loc): Remove macro.
8758 (fold_build2_loc): Likewise.
8759 (fold_build3_loc): Likewise.
8761 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8763 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8764 (gimple_build_debug_bind_source_stat): Likewise.
8765 * gimple.h (gimple_build_debug_bind): Remove macro.
8766 (gimple_build_debug_bind_source): Likewise.
8768 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8770 * bitmap.c (bitmap_alloc): Adjust.
8771 (bitmap_gc_alloc): Likewise.
8772 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
8774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8776 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
8777 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
8778 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
8779 (bitmap_gc_alloc_stat): Likewise.
8780 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
8782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8784 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
8785 * rtl.h (shallow_copy_rtx): Remove macro.
8787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8789 * emit-rtl.c (gen_raw_REG): Adjust.
8790 * gengenrtl.c (gendef): Likewise.
8791 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8792 * rtl.h (rtx_alloc): Remove macro.
8794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8796 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8797 (build_tree_list_stat): Likewise.
8798 * tree.h (build_tree_list): Remove macro.
8799 (build_tree_list_vec): Likewise.
8801 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8803 * tree.c (make_vector_stat): Remove _stat from name.
8804 (build_vector_stat): Likewise.
8805 * tree.h (make_vector_stat): Remove macro.
8806 (build_vector_stat): Likewise.
8808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8810 * tree.h (build_var_debug_value): Remove prototype.
8812 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8814 * tree.c (tree_cons_stat): Remove _stat from name.
8815 * tree.h (tree_cons): Remove macro.
8817 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8819 * tree.c (build_vl_exp_stat): Remove _stat from name.
8820 * tree.h (build_vl_exp): Remove macro.
8822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8824 * tree.c (build_decl_stat): Remove _stat from name.
8825 * tree.h (build_decl): Remove macro.
8827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8829 * gimple.c (gimple_build_with_ops_stat): Adjust.
8830 (gimple_alloc_stat): Remove _stat from name.
8831 * gimple.h (gimple_alloc): Remove macro.
8833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8835 * tree.c (make_tree_vec_stat): Remove _stat from name.
8836 (grow_tree_vec_stat): Likewise.
8837 * tree.h (make_tree_vec_stat): Adjust prototype.
8838 (grow_tree_vec_stat): Likewise.
8839 (make_tree_vec): Remove macro.
8840 (grow_tree_vec): Likewise.
8842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8844 * fold-const.c (fold_build1_stat_loc): Adjust.
8845 (fold_build2_stat_loc): Likewise.
8846 (fold_build3_stat_loc): Likewise.
8847 * tree.c (build0_stat): Remove _stat from name.
8848 (build1_stat): Likewise.
8849 (build2_stat): Likewise.
8850 (build3_stat): Likewise.
8851 (build4_stat): Likewise.
8852 (build5_stat): Likewise.
8853 * tree.h (build1_loc): Remove macro, and rename _stat function
8855 (build2_loc): Likewise.
8856 (build3_loc): Likewise.
8857 (build4_loc): Likewise.
8858 (build5_loc): Likewise.
8860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8862 * tree.c (make_int_cst_stat): Remove _stat from name.
8863 * tree.h (make_int_cst_stat): Adjust prototype.
8864 (make_int_cst): Remove macro.
8866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8868 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8869 * tree.h (make_tree_binfo_stat): Adjust prototype.
8870 (make_tree_binfo): Remove.
8872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8874 * tree.c (copy_node_stat): Rename to copy_node.
8875 (build_distinct_type_copy): Adjust.
8876 * tree.h (copy_node_stat): Adjust prototype.
8877 (copy_node): Remove macro.
8879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8881 * tree.c (make_node_stat): rename to make_node.
8882 (build_tree_list_stat): Adjust.
8883 (build0_stat): Likewise.
8884 (build2_stat): Likewise.
8885 (build3_stat): Likewise.
8886 (build4_stat): Likewise.
8887 (build5_stat): Likewise.
8888 (build_decl_stat): Likewise.
8889 * tree.h (make_node_stat): Adjust prototype.
8890 (make_node): remove macro.
8892 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8894 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8895 (PPC_FEATURE2_SCV): Likewise.
8896 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8898 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8900 * config/aarch64/aarch64.c
8901 (aarch64_internal_mov_immediate): Add new special pattern.
8902 * config/aarch64/aarch64.md (*movdi_aarch64):
8903 Add reg/32bit const mov case.
8905 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8906 Richard Sandiford <richard.sandiford@linaro.org>
8908 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8909 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8910 Add integer and movi cases.
8911 (movi-split-hf-df-sf split, fp16): New.
8912 (enabled): Added TARGET_FP_F16INST.
8913 * config/aarch64/iterators.md (GPF_HF): New.
8914 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8916 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8918 * config/aarch64/aarch64.c
8919 (aarch64_simd_container_mode): Add prototype.
8920 (aarch64_expand_mov_immediate): Add HI support.
8921 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8922 (aarch64_can_const_movi_rtx_p): New.
8923 (aarch64_preferred_reload_class):
8924 Remove restrictions of using FP registers for certain SIMD operations.
8925 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8926 (aarch64_valid_floating_const): Add integer move validation.
8927 (aarch64_simd_imm_scalar_p): Remove.
8928 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8929 (aarch64_legitimate_constant_p): Expand list of supported cases.
8930 * config/aarch64/aarch64-protos.h
8931 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8932 (aarch64_reinterpret_float_as_int): New.
8933 (aarch64_simd_imm_scalar_p): Remove.
8934 * config/aarch64/constraints.md (Uvi): New.
8935 (Dd): Split into Ds and new Dd.
8936 * config/aarch64/aarch64.md (*movsi_aarch64):
8938 (*movdi_aarch64): Add SIMD mov case.
8940 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8942 * tree-predcom.c: (struct chain): Handle store-store chain in which
8943 stores for elimination only store loop invariant values.
8944 (execute_pred_commoning_chain): Ditto.
8945 (prepare_initializers_chain_store_elim): Ditto.
8946 (prepare_finalizers): Ditto.
8947 (is_inv_store_elimination_chain): New function.
8948 (initialize_root_vars_store_elim_1): New function.
8950 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8952 * tree-predcom.c: Revise general description of the pass.
8953 (enum chain_type): New enum type for store elimination.
8954 (struct chain): New field supporting store elimination.
8955 (struct component): Ditto.
8956 (dump_chain): Dump store-stores chain.
8957 (release_chain): Release resources.
8958 (split_data_refs_to_components): Compute and create component
8959 contains only stores for elimination.
8960 (get_chain_last_ref_at): New function.
8961 (make_invariant_chain): Initialization.
8962 (make_rooted_chain): Specify chain type in parameter and record it.
8963 (add_looparound_copies): Skip for store-stores chain.
8964 (determine_roots_comp): Compute type of chain and pass it to
8966 (initialize_root_vars_store_elim_2): New function.
8967 (finalize_eliminated_stores): New function.
8968 (remove_stmt): Handle store for elimination.
8969 (execute_pred_commoning_chain): Execute predictive commoning on
8971 (determine_unroll_factor): Skip unroll for store-stores chain.
8972 (prepare_initializers_chain_store_elim): New function.
8973 (prepare_initializers_chain): Hanlde store-store chain.
8974 (prepare_finalizers_chain, prepare_finalizers): New function.
8975 (tree_predictive_commoning_loop): Return integer value indicating
8976 if loop is unrolled or lcssa form is corrupted.
8977 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8979 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8981 * tree-predcom.c (initialize_root): Delete.
8982 (execute_pred_commoning_chain): Initialize root vars and replace
8983 reference of non-combined chain directly, rather than call above
8986 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8988 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8989 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8991 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8993 * tree-predcom.c (struct chain): New field init_seq.
8994 (release_chain): Release init_seq.
8995 (prepare_initializers_chain): Record intialization stmts in above
8997 (insert_init_seqs): New function.
8998 (tree_predictive_commoning_loop): Call insert_init_seqs.
9000 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9002 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9004 2017-07-28 Richard Biener <rguenther@suse.de>
9006 * match.pd: Remove superfluous :c.
9007 * genmatch.c (simplify::id): Add member.
9008 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9010 (current_id): New global.
9011 (dt_node::parent): Move from ...
9012 (dt_operand::parent): ... here. Add for_id member.
9013 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9014 (decision_tree::find_node): Relax order requirement when
9015 merging DT_TRUE nodes to ones inbetween the current simplify
9016 and the one we try to merge with. Add diagnostic whenever
9017 we need to enforce pattern order by not merging.
9018 (decision_tree::insert): Set current_id.
9019 (decision_tree::print_node): Dump parent node and for_id.
9020 (parser::last_id): Add member.
9021 (parser::push_simplify): Assign unique id.
9022 (parser::parser): Initialize last_id.
9024 2017-07-28 Martin Liska <mliska@suse.cz>
9027 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9028 gimple_build_debug_bind.
9030 2017-07-28 Richard Biener <rguenther@suse.de>
9032 PR tree-optimization/81502
9033 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9035 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9037 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9038 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9039 for BIT_FIELD_REF args.
9040 * fold-const.c (make_bit_field_ref): Likewise.
9041 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9043 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9046 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9047 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9048 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9049 Or it into SANITIZER_UNDEFINED.
9050 * ubsan.c: Include gimple-fold.h and varasm.h.
9051 (ubsan_expand_ptr_ifn): New function.
9052 (instrument_pointer_overflow): New function.
9053 (maybe_instrument_pointer_overflow): New function.
9054 (instrument_object_size): Formatting fix.
9055 (pass_ubsan::execute): Call instrument_pointer_overflow
9056 and maybe_instrument_pointer_overflow.
9057 * internal-fn.c (expand_UBSAN_PTR): New function.
9058 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9059 * sanitizer.def (__ubsan_handle_pointer_overflow,
9060 __ubsan_handle_pointer_overflow_abort): New builtins.
9061 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9062 * internal-fn.def (UBSAN_PTR): New internal function.
9063 * opts.c (sanitizer_opts): Add pointer-overflow.
9064 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9065 * fold-const.c (build_range_check): Compute pointer range check in
9066 integral type if pointer arithmetics would be needed. Formatting
9069 2017-07-28 Martin Liska <mliska@suse.cz>
9072 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9073 parameters that are of a variable-length.
9075 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9077 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9079 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9080 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9081 (CRT_CALL_STATIC_FUNCTION): Likewise.
9082 (ASM_DEFAULT_SPEC): New define.
9083 (ASM_SPEC32): Likewise.
9084 (ASM_SPEC64): Likewise.
9085 (ASM_SPEC_COMMON): Likewise.
9086 (ASM_SPEC): Likewise.
9087 (INVALID_64BIT): Likewise.
9088 (LINK_OS_DEFAULT_SPEC): Likewise.
9089 (LINK_OS_SPEC32): Likewise.
9090 (LINK_OS_SPEC64): Likewise.
9091 (POWERPC_LINUX): Likewise.
9092 (PTRDIFF_TYPE): Likewise.
9093 (RESTORE_FP_PREFIX): Likewise.
9094 (RESTORE_FP_SUFFIX): Likewise.
9095 (SAVE_FP_PREFIX): Likewise.
9096 (SAVE_FP_SUFFIX): Likewise.
9097 (SIZE_TYPE): Likewise.
9098 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9099 (TARGET_64BIT): Likewise.
9100 (TARGET_64BIT): Likewise.
9101 (TARGET_AIX): Likewise.
9102 (WCHAR_TYPE_SIZE): Likewise.
9103 (WCHAR_TYPE): Undefine.
9104 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9105 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9106 (CPP_OS_RTEMS_SPEC): Delete.
9107 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9108 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9110 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9112 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9114 PR tree-optimization/81578
9115 * tree-parloops.c (build_new_reduction): Bail out if
9116 reduction_code isn't one of the standard OpenMP reductions.
9117 Move the details printing after that decision.
9119 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9121 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9122 related to reload_in_progress.
9123 (splat_input_operand): Likewise.
9124 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9126 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9128 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9129 (TARGET_INSTANTIATE_DECLS): Likewise.
9130 (legitimate_indexed_address_p): Delete reload_in_progress code.
9131 (rs6000_debug_legitimate_address_p): Likewise.
9132 (rs6000_eliminate_indexed_memrefs): Likewise.
9133 (rs6000_emit_le_vsx_store): Likewise.
9134 (rs6000_emit_move_si_sf_subreg): Likewise.
9135 (rs6000_emit_move): Likewise.
9136 (register_to_reg_type): Likewise.
9137 (rs6000_pre_atomic_barrier): Likewise.
9138 (rs6000_machopic_legitimize_pic_address): Likewise.
9139 (rs6000_allocate_stack_temp): Likewise.
9140 (rs6000_address_for_fpconvert): Likewise.
9141 (rs6000_address_for_altivec): Likewise.
9142 (rs6000_secondary_memory_needed_rtx): Delete function.
9143 (rs6000_check_sdmode): Likewise.
9144 (rs6000_alloc_sdmode_stack_slot): Likewise.
9145 (rs6000_instantiate_decls): Likewise.
9146 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9147 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9148 Delete reload_in_progress.
9149 (*vec_reload_and_plus_<mptrsize>): Likewise.
9150 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9151 (vsx_div_v2di): Likewise.
9152 (vsx_udiv_v2di): Likewise.
9154 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9156 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9157 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9158 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9159 (rs6000_debug_reg_global): Delete print of LRA status.
9160 (rs6000_option_override_internal): Delete dead LRA related code.
9161 (rs6000_lra_p): Delete function.
9162 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9164 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9166 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9167 * config/riscv/rtems.h: New file.
9169 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9170 Sudakshina Das <sudi.das@arm.com>
9172 * config/aarch64/aarch64.md
9173 (define_split for and<mode>3nr_compare): Move
9174 non aarch64_logical_operand to a register.
9175 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9176 register immediate operand to a register.
9177 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9179 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9182 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9184 2017-07-27 Richard Biener <rguenther@suse.de>
9186 PR tree-optimization/81573
9187 PR tree-optimization/81494
9188 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9189 multi defuse cycle case.
9191 2017-07-27 Richard Biener <rguenther@suse.de>
9193 PR tree-optimization/81571
9194 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9197 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9199 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9200 earlier and only if MASK_FPU is set. Adjust formatting.
9202 2017-07-27 Martin Liska <mliska@suse.cz>
9204 * opt-functions.awk: Add validation of value of Init.
9205 * optc-gen.awk: Pass new argument.
9207 2017-07-27 Martin Liska <mliska@suse.cz>
9209 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9210 Fix wrong condition.
9212 2017-07-27 Martin Liska <mliska@suse.cz>
9214 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9215 BBs and edges seen by autoFDO.
9217 2017-07-27 Richard Biener <rguenther@suse.de>
9219 PR tree-optimization/81502
9220 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9221 with incompatible but same sized type.
9222 (execute_update_addresses_taken): Likewise.
9224 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9226 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9227 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9229 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9232 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9233 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9234 Change s_operand to memory_operand.
9236 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9238 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9239 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9240 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9241 Emit instructions rather than returning an expression. Handle TFmode
9242 and KFmode by casting to TImode.
9243 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9244 (rs6000_emit_le_vsx_store): Likewise.
9245 * config/rs6000/vsx.md (VSX_TI): New iterator.
9246 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9247 (*vsx_le_undo_permute_<mode>): Likewise.
9248 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9249 emit the split sequence.
9250 (*vsx_le_perm_store_<mode>): Likewise.
9252 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9254 PR tree-optimization/81555
9255 PR tree-optimization/81556
9256 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9257 if true, force CHANGED for the recursive invocation.
9258 (reassociate_bb): Remember original length of ops array, pass
9259 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9261 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9262 attributes for noipa attribute. For naked attribute use
9263 lookup_attribute first before lookup_attribute_spec.
9264 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9266 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9267 for functions with noipa attribute.
9268 (cgraph_externally_visible_p): Return true for functions with noipa
9270 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9271 for functions with noipa attribute.
9272 * doc/extend.texi: Document noipa function attribute.
9273 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9274 also for functions with noipa attribute.
9275 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9277 2017-07-26 Andrew Pinski <apinski@cavium.com>
9279 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9280 vec_unalign_load_cost and vec_unalign_store_cost.
9282 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9284 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9285 -mvsx-small-integer option.
9286 (ISA_3_0_MASKS_IEEE): Likewise.
9287 (OTHER_VSX_VECTOR_MASKS): Likewise.
9288 (POWERPC_MASKS): Likewise.
9289 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9290 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9291 code, only testing for DImode being allowed in non-VSX floating
9293 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9294 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9296 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9297 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9298 TARGET_P8_VECTOR test.
9299 (rs6000_secondary_reload_simple_move): Likewise.
9300 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9301 since TARGET_P9_VECTOR was already tested.
9302 (rs6000_opt_masks): Remove -mvsx-small-integer.
9303 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9304 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9306 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9307 test for TARGET_VEXTRACTUB was used, and that uses
9309 (p9 extract splitter): Likewise.
9310 (vsx_extract_<mode>_di_p9): Likewise.
9311 (vsx_extract_<mode>_store_p9): Likewise.
9312 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9313 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9314 the elimination of TARGET_VSX_SMALL_INTEGER.
9315 (vsx_extract_<mode>_p8): Likewise.
9316 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9317 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9318 (vsx_set_<mode>_p9): Likewise.
9319 (vsx_set_v4sf_p9): Likewise.
9320 (vsx_set_v4sf_p9_zero): Likewise.
9321 (vsx_insert_extract_v4sf_p9): Likewise.
9322 (vsx_insert_extract_v4sf_p9_2): Likewise.
9323 * config/rs6000/rs6000.md (sign extend splitter): Change
9324 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9325 (floatsi<mode>2_lfiwax_mem): Likewise.
9326 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9327 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9328 since a test for TARGET_P9_VECTOR was used.
9329 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9330 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9331 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9332 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9333 TARGET_P8_VECTOR test.
9334 (fix_trunc<mode>si2_stfiwx): Likewise.
9335 (fix_trunc<mode>si2_internal): Likewise.
9336 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9337 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9339 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9340 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9341 TARGET_P8_VECTOR test.
9342 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9343 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9344 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9346 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9347 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9348 since a test for TARGET_P9_VECTOR was used.
9349 (splitter for loading small constants): Likewise.
9351 2017-07-26 Andrew Pinski <apinski@cavium.com>
9353 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9356 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9359 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9360 (fp_valid_at): Likewise.
9362 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9364 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9365 (qdf24xx_addrcost_table): Likewise.
9366 (cortexa57_tunings): Update to use generic_branch_cost.
9367 (cortexa72_tunings): Likewise.
9368 (cortexa73_tunings): Likewise.
9369 (qdf24xx_tunings): Likewise.
9371 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9373 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9374 (thunderx2t99_branch_cost): Likewise.
9375 (cortexa35_tunings): Update to use generic_branch_cost.
9376 (cortexa53_tunings): Likewise.
9377 (cortexa57_tunings): Likewise.
9378 (cortexa72_tunings): Likewise.
9379 (cortexa73_tunings): Likewise.
9380 (thunderx2t99_tunings): Likewise.
9382 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9384 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9385 (sparc_option_override): Honour MASK_FSMULD.
9386 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9387 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9388 * config/sparc/sparc.opt (mfsmuld): New option.
9389 * doc/invoke.texi (mfsmuld): Document option.
9391 2017-07-26 Marek Polacek <polacek@redhat.com>
9394 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9396 2017-07-26 Richard Biener <rguenther@suse.de>
9398 * gimple-match-head.c (do_valueize): Return OP if valueize
9400 (get_def): New helper to get at the def stmt of a SSA name
9402 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9403 do_valueize to get at the def stmt.
9404 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9406 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9409 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9411 2017-07-26 Martin Liska <mliska@suse.cz>
9414 * function.c (expand_function_start): Make expansion of
9415 nonlocal_goto_save_area after parm_birth_insn.
9417 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9419 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9420 from all CPU target flags enable members.
9422 2017-07-26 Richard Biener <rguenther@suse.de>
9424 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9425 (decision_tree::gen): Make 'type' const.
9426 (write_predicate): Likewise.
9428 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9430 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9431 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9432 (rs6000_option_override_internal): Likewise.
9433 (rs6000_expand_vector_set): Likewise.
9434 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9435 (TARGET_UPPER_REGS_SF): Likewise.
9436 (TARGET_UPPER_REGS_DI): Likewise.
9437 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9438 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9439 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9440 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9441 (Splitters for DI constants in Altivec registers): Likewise.
9442 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9443 (vsx_set_v4sf_p9): Likewise.
9444 (vsx_set_v4sf_p9_zero): Likewise.
9445 (vsx_insert_extract_v4sf_p9): Likewise.
9446 (vsx_insert_extract_v4sf_p9_2): Likewise.
9448 2017-07-25 Carl Love <cel@us.ibm.com>
9450 * doc/extend.texi: Update the built-in documentation file for the
9451 existing built-in functions
9452 vector signed char vec_cnttz (vector signed char);
9453 vector unsigned char vec_cnttz (vector unsigned char);
9454 vector signed short vec_cnttz (vector signed short);
9455 vector unsigned short vec_cnttz (vector unsigned short);
9456 vector signed int vec_cnttz (vector signed int);
9457 vector unsigned int vec_cnttz (vector unsigned int);
9458 vector signed long long vec_cnttz (vector signed long long);
9459 vector unsigned long long vec_cnttz (vector unsigned long long);
9461 2017-07-25 Andrew Pinski <apinski@cavium.com>
9463 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9464 accesses where the use is for the first operand of a BIT_INSERT.
9466 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9469 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9470 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9472 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9474 * config/i386/gstabs.h: Delete.
9475 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9477 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9479 * config/i386/i386.c (ix86_decompose_address): Do not check for
9480 register RTX when looking at index_reg or base_reg.
9481 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9483 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9485 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9486 to update EH info here.
9488 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9490 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9492 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9494 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9496 2017-07-25 Torsten Duwe <duwe@suse.de>
9498 * common.opt: Introduce -fpatchable-function-entry
9499 command line option, and its variables function_entry_patch_area_size
9500 and function_entry_patch_area_start.
9501 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9502 including a two-value parser.
9503 * target.def (print_patchable_function_entry): New target hook.
9504 * targhooks.h (default_print_patchable_function_entry): New function.
9505 * targhooks.c (default_print_patchable_function_entry): Likewise.
9506 * toplev.c (process_options): Switch off IPA-RA if
9507 patchable function entries are being generated.
9508 * varasm.c (assemble_start_function): Look at the
9509 patchable-function-entry command line switch and current
9510 function attributes and maybe generate NOP instructions by
9511 calling the print_patchable_function_entry hook.
9512 * doc/extend.texi: Document patchable_function_entry attribute.
9513 * doc/invoke.texi: Document -fpatchable_function_entry
9514 command line option.
9515 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9517 * doc/tm.texi: Re-generate.
9519 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9522 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9523 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9525 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9527 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9529 2017-07-25 Richard Biener <rguenther@suse.de>
9531 PR tree-optimization/81455
9532 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9533 not walk in cycles when looking for guards.
9535 2017-07-25 Richard Biener <rguenther@suse.de>
9537 PR tree-optimization/81529
9538 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9539 when optimizing backedge uses.
9541 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9543 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9545 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9546 to dl_section_ref. On AIX, append an expression to subtract
9547 the size of the section length to dl_section_ref.
9549 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9551 * configure.ac: If any of the config.* scripts fail, exit 1.
9552 * configure: Regenerate.
9554 2017-07-25 Richard Biener <rguenther@suse.de>
9557 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9558 of immediate uses, be more verbose on errors.
9560 2017-07-25 Richard Biener <rguenther@suse.de>
9562 PR tree-optimization/81510
9563 * tree-vect-loop.c (vect_is_simple_reduction): When the
9564 reduction stmt is not inside the loop bail out.
9566 2017-07-25 Richard Biener <rguenther@suse.de>
9568 PR tree-optimization/81303
9569 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9570 profitability check against LOOP_VINFO_NITERSM1.
9572 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9574 * domwalk.c (cmp_bb_postorder): Simplify.
9575 (sort_bbs_postorder): New function. Use it...
9576 (dom_walker::walk): ...here to optimize common cases.
9578 2017-07-25 Martin Liska <mliska@suse.cz>
9581 * ipa-visibility.c (function_and_variable_visibility): Make the
9582 redirection just on target that supports aliasing.
9583 Fix GNU coding style.
9585 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9588 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9590 * config/arm/rtems.h: Likewise.
9591 * config/bfin/rtems.h: Likewise.
9592 * config/i386/rtemself.h: Likewise.
9593 * config/lm32/rtems.h: Likewise.
9594 * config/m32c/rtems.h: Likewise.
9595 * config/m68k/rtemself.h: Likewise.
9596 * config/microblaze/rtems.h: Likewise.
9597 * config/mips/rtems.h: Likewise.
9598 * config/moxie/rtems.h: Likewise.
9599 * config/nios2/rtems.h: Likewise.
9600 * config/powerpcspe/rtems.h: Likewise.
9601 * config/rs6000/rtems.h: Likewise.
9602 * config/rtems.h: Likewise.
9603 * config/sh/rtems.h: Likewise.
9604 * config/sh/rtemself.h: Likewise.
9605 * config/sparc/rtemself.h: Likewise.
9607 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9610 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9611 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9612 * tree-ssa-structalias.c (alias_get_name): Same.
9614 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9617 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9618 instructions if no du chain is found.
9620 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9622 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9624 2017-07-25 Richard Biener <rguenther@suse.de>
9627 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9630 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9632 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9634 (ISA_2_7_MASKS_SERVER): Likewise.
9635 (ISA_3_0_MASKS_IEEE): Likewise.
9636 (OTHER_P8_VECTOR_MASKS): Likewise.
9637 (OTHER_VSX_VECTOR_MASKS): Likewise.
9638 (POWERPC_MASKS): Likewise.
9639 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9640 duplicate list of options.
9641 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9642 explicit -mupper-regs options.
9643 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9644 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9645 alias for -mupper-regs-df.
9646 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9647 (rs6000_init_hard_regno_mode_ok): Likewise.
9648 (rs6000_option_override_internal): Likewise.
9649 (rs6000_opt_masks): Likewise.
9650 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9651 options in terms of whether -mvsx or -mpower8-vector was used.
9652 (TARGET_UPPER_REGS_DI): Likewise.
9653 (TARGET_UPPER_REGS_SF): Likewise.
9654 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9655 -mupper-regs-* options.
9657 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9659 * passes.c (emergency_dump_function): Print some empty lines and a
9660 header before the RTL dump.
9662 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9664 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9666 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9669 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9670 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9672 2017-07-24 Carl Love <cel@us.ibm.com>
9674 * config/rs6000/rs6000-c.c: Add support for built-in functions
9675 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9676 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9677 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9678 vec_extract_fp_from_shortl): Add defines for the two builtins.
9679 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9680 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9682 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9683 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9684 * doc/extend.texi: Update the built-in documentation file for the
9685 new built-in function.
9687 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9690 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9692 * doc/generic.texi: Likewise.
9693 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9694 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9696 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9698 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9699 (aarch64_mls_elt_merge<mode>): Likewise.
9701 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9703 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9704 having __cxa_atexit.
9706 2017-07-23 Michael Collison <michael.collison@arm.com>
9708 * config/arm/arm.c (arm_option_override): Deprecate
9709 use of -mstructure-size-boundary.
9710 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9711 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9713 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9716 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9717 Reduce cost estimate for direct moves.
9719 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9722 * config/i386/i386.c (ix86_option_override_internal): Disable
9723 BMI, BMI2 and TBM instructions for -m16.
9725 2017-07-21 Carl Love <cel@us.ibm.com>
9727 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9728 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9729 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9731 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9732 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9733 VMULOSW): New enum "unspec" values.
9734 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9735 altivec_vmulosw): New patterns.
9736 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9737 VMULOSW): Add definitions.
9739 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9741 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9742 (qdf24xx): Likewise.
9743 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9744 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9745 (AARCH64_FL_V8_1): Renumber.
9746 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9747 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9748 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9749 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9750 rdma to feature modifiers list.
9752 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9755 * ipa-visibility (function_and_variable_visibility): Convert
9756 recursive PLT call to direct call if appropriate.
9758 2017-07-21 Andrew Pinski <apinski@cavium.com>
9760 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9761 operand 1 to see if the types precision matches.
9762 * fold-const.c (operand_equal_p): Likewise.
9764 2017-07-21 Richard Biener <rguenther@suse.de>
9766 PR tree-optimization/81303
9767 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
9768 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
9769 (vect_peeling_hash_get_lowest_cost): Adjust.
9770 (vect_enhance_data_refs_alignment): Likewise. Use
9771 vect_get_peeling_costs_all_drs to compute the penalty for no
9772 peeling to match up costs.
9774 2017-07-21 Richard Biener <rguenther@suse.de>
9776 PR tree-optimization/81500
9777 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
9778 we didn't identify a reduction path.
9780 2017-07-21 Tom de Vries <tom@codesourcery.com>
9781 Cesar Philippidis <cesar@codesourcery.com>
9783 PR gcov-profile/81442
9784 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
9787 2017-07-21 Tom de Vries <tom@codesourcery.com>
9790 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9792 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9793 nvptx_override_options_after_change.
9795 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9797 * dwarf2out.c (output_file_names): Avoid double testing for
9800 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9802 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9804 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9806 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9808 (try_crossjump_to_edge): Do not punt on partitioned functions.
9810 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9812 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9813 Put all BBs reachable only via paths crossing cold region to cold
9815 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9817 2016-07-21 Richard Biener <rguenther@suse.de>
9819 PR tree-optimization/81303
9820 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9821 into account prologue and epilogue iterations when raising
9822 min_profitable_iters to sth at least covering one vector iteration.
9824 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9826 * config/arm/arm.c (arm_test_cpu_arch_dat):
9829 2017-07-20 Nathan Sidwell <nathan@acm.org>
9831 Remove TYPE_METHODS.
9832 * tree.h (TYPE_METHODS): Delete.
9833 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9834 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9835 (dbxout_type_methods): Scan TYPE_FIELDS.
9836 (dbxout_type): Don't check TYPE_METHODS here.
9837 * function.c (use_register_for_decl): Always ignore register for
9838 class types when not optimizing.
9839 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9840 * tree.c (free_lang_data_in_type): Stitch out member functions and
9841 templates from TYPE_FIELDS.
9842 (build_distinct_type_copy, verify_type_variant,
9843 verify_type): Member fns are on TYPE_FIELDS.
9844 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9845 * tree-pretty-print.c (dump_generic_node): Likewise.
9847 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9850 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9851 V2TImode and V4TImode.
9852 (ix86_expand_vector_extract): Likewise.
9853 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9854 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9855 (ssescalarmode): Handle V4TImode and V2TImode.
9856 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9857 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9858 (VEXTRACTI128_MODE): New mode iterator.
9859 (splitter for *vec_extractv?ti first element): New.
9860 (VEC_INIT_MODE): New mode iterator.
9861 (vec_init<mode>): Consolidate 3 expanders into one using
9862 VEC_INIT_MODE mode iterator.
9864 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9866 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9867 non_spilled_static_chain_regno_p.
9869 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9871 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9873 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9875 * bb-reorder.c (connect_traces): Allow copying of blocks within
9878 2017-07-20 Richard Biener <rguenther@suse.de>
9880 * gimple.h (gimple_phi_result): Add gphi * overload.
9881 (gimple_phi_result_ptr): Likewise.
9882 (gimple_phi_arg): Likewise. Adjust index assert to only
9883 allow actual argument accesses rather than all slots available
9885 (gimple_phi_arg_def): Add gphi * overload.
9886 * tree-phinodes.c (make_phi_node): Initialize only actual
9888 (resize_phi_node): Clear memory not covered by old node,
9889 do not initialize excess argument slots.
9890 (reserve_phi_args_for_new_edge): Initialize new argument slot
9893 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9895 PR tree-optimization/81388
9897 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9899 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9900 by removing computation of may_be_zero.
9902 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9903 Tom de Vries <tom@codesourcery.com>
9906 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9907 when gimple level profile disagrees with what RTL expander did.
9909 2017-07-20 Richard Biener <rguenther@suse.de>
9911 PR tree-optimization/61171
9912 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9913 (vect_analyze_stmt): Add slp instance parameter.
9914 (vectorizable_reduction): Likewise.
9915 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9916 (vect_is_simple_reduction): Deal with chains not detected
9917 as SLP reduction chain, specifically not properly associated
9918 chains containing a mix of plus/minus.
9919 (get_reduction_op): Remove.
9920 (get_initial_defs_for_reduction): Simplify, pass in whether
9921 this is a reduction chain, pass in the SLP node for the PHIs.
9922 (vect_create_epilog_for_reduction): Get the SLP instance as
9924 (vectorizable_reduction): Get the SLP instance as arg.
9925 During analysis remember the SLP node with the PHIs in the
9926 instance. Simplify getting at the vectorized reduction PHIs.
9927 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9928 through SLP instance.
9929 (vect_slp_analyze_operations): Likewise.
9930 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9931 (vect_transform_stmt): Likewise.
9933 2017-07-20 Tom de Vries <tom@codesourcery.com>
9935 PR tree-optimization/81489
9936 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9937 read of phi arg location to before loop that modifies phi.
9939 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9941 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9944 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9947 * except.c (execute): Fix ordering issue.
9949 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9951 PR rtl-optimization/81423
9952 * combine.c (make_compound_operation_int): Don't try to optimize
9953 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9955 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9957 PR rtl-optimization/81423
9958 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9959 with a constant that is -1 in the truncated to mode.
9961 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9963 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9964 (determine_unlikely_bbs): ... here.
9965 * predict.h (propagate_unlikely_bbs_forward): Declare.
9966 * cfgexpand.c (pass_expand::execute): Use it.
9967 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9969 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9970 propagate_unlikely_bbs_forward.
9972 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9975 * except.c (maybe_add_nop_after_section_switch): New function.
9978 2017-07-19 Tom de Vries <tom@codesourcery.com>
9980 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9982 2017-07-19 Tom de Vries <tom@codesourcery.com>
9984 * gimple.h (gimple_phi_arg): Make assert more strict.
9986 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9988 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9989 * config/rs6000/mmintrin.h: New file.
9990 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9992 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9994 PR tree-optimization/81346
9995 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9997 2017-07-19 Tom de Vries <tom@codesourcery.com>
9999 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10001 2017-07-19 Tom de Vries <tom@codesourcery.com>
10003 * config/nvptx/nvptx-modes.def: Add V2DImode.
10004 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10005 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10006 (nvptx_output_mov_insn): Handle lack of mov.b128.
10007 (nvptx_print_operand): Handle 'H' and 'L' codes.
10008 (nvptx_vector_mode_supported): Allow V2DImode.
10009 (nvptx_preferred_simd_mode): New function.
10010 (nvptx_data_alignment): New function.
10011 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10012 nvptx_preferred_simd_mode.
10013 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10015 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10017 2017-07-19 Tom de Vries <tom@codesourcery.com>
10019 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10020 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10021 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10022 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10023 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10024 (mov<VECIM>_insn): New define_insn.
10025 (define_expand "mov<VECIM>): New define_expand.
10027 2017-07-19 Tom de Vries <tom@codesourcery.com>
10029 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10031 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10033 PR tree-optimization/81346
10034 * fold-const.h (fold_div_compare, range_check_type): Declare.
10035 * fold-const.c (range_check_type): New function.
10036 (build_range_check): Use range_check_type.
10037 (fold_div_compare): No longer static, rewritten into
10038 a match.pd helper function.
10039 (fold_comparison): Don't call fold_div_compare here.
10040 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10041 as helper function.
10043 2017-07-19 Nathan Sidwell <nathan@acm.org>
10045 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10046 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10047 * tree.c (find_decls_types_r, verify_type): Use
10048 TYPE_{MIN,MAX}_VALUE_RAW.
10049 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10050 (hash_tree): Likewise.
10051 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10053 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10056 2017-07-18 Tom de Vries <tom@codesourcery.com>
10058 PR middle-end/81464
10059 * omp-expand.c (expand_omp_for_static_chunk): Handle
10060 equal-argument loop exit phi.
10062 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10065 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10066 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10067 operand 2 predicate.
10068 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10069 operand 2 predicate.
10070 (ror,rol -> rorx splitters): Use const_int_operand as
10071 operand 2 predicate.
10073 2017-06-18 Richard Biener <rguenther@suse.de>
10075 PR tree-optimization/81410
10076 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10077 the gap in the ! slp_perm SLP case after each group.
10079 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10081 PR middle-end/81463
10082 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10085 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10087 PR middle-end/81462
10088 * predict.c (set_even_probabilities): Cleanup; do not affect
10089 probabilities that are already known.
10090 (combine_predictions_for_bb): Call even when count is set.
10092 2017-07-18 Nathan Sidwell <nathan@acm.org>
10094 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10097 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10100 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10101 optimization for loop niter analysis.
10103 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10106 * config/avr/avr.c (avr_optimize_casesi): Don't use
10107 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10109 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10111 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10112 body_cost_vec from _vect_peel_extended_info.
10113 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10114 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10117 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10119 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10121 2017-07-18 Richard Biener <rguenther@suse.de>
10123 PR tree-optimization/80620
10124 PR tree-optimization/81403
10125 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10126 info when re-using a VN table entry.
10128 2017-07-18 Richard Biener <rguenther@suse.de>
10130 PR tree-optimization/81418
10131 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10132 vectype_in. Verify that with lane-reducing reduction operations
10133 we have a single def-use cycle.
10135 2017-07-17 Carl Love <cel@us.ibm.com>
10137 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10139 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10140 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10141 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10143 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10144 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10145 VMULOSW): New enum "unspec" values.
10146 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10147 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10148 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10149 altivec_vmulosw): New patterns.
10150 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10151 VMULOSW): Add definitions.
10152 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10154 * config/alpha/alpha.c: Include predict.h.
10156 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10158 * tree-vrp.c (compare_assert_loc): Fix comparison function
10159 to return predictable results.
10161 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10163 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10165 (subdi3): Likewise.
10166 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10167 * doc/invoke.texi (mexpand-adddi): Update text.
10169 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10171 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10172 that also clobbers the CC register. The old expand code is moved
10174 (*arc_clzsi2): ... here.
10175 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10176 the CC register. The old expand code is moved to ...
10177 (arc_ctzsi2): ... here.
10179 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10181 * config/arc/arc.opt (mindexed-loads): Use initial value
10182 TARGET_INDEXED_LOADS_DEFAULT.
10183 (mauto-modify-reg): Use initial value
10184 TARGET_AUTO_MODIFY_REG_DEFAULT.
10185 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10186 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10187 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10188 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10190 2017-07-17 Martin Liska <mliska@suse.cz>
10193 * opts.c (finish_options): Do not allow -fgnu-tm
10194 w/ -fsanitize={kernel-,}address. Say sorry.
10196 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10199 * tree-loop-distribution.c (classify_partition): Only assert on
10200 numer of iterations.
10201 (merge_dep_scc_partitions): Delete prameter. Update function call.
10202 (distribute_loop): Remove code handling loop with unknown niters.
10203 (pass_loop_distribution::execute): Skip loop with unknown niters.
10205 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10208 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10209 function sort_partitions_by_post_order.
10211 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10213 PR tree-optimization/81374
10214 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10215 the max index of basic blocks, rather than number of basic blocks.
10217 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10219 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10221 (arc_legitimate_pic_operand_p): Likewise.
10222 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10224 (arc_needs_pcl_p): Likewise.
10225 (arc_legitimate_pc_offset_p): Likewise.
10226 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10227 function is also used in constrains.md.
10228 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10229 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10230 PLUS. Only return true/false in known cases, otherwise assert.
10231 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10232 is already called in arc_legitimate_constant_p.
10233 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10235 (LEGITIMATE_PIC_OPERAND_P): Use
10236 arc_raw_symbolic_reference_mentioned_p function.
10237 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10242 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10243 Andrew Burgess <andrew.burgess@embecosm.com>
10245 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10246 (arc_return_address_register): New function.
10247 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10248 (arc_handle_fndecl_attribute): Add naked attribute.
10249 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10250 (TARGET_WARN_FUNC_RETURN): Likewise.
10251 (arc_allocate_stack_slots_for_args): New function.
10252 (arc_warn_func_return): Likewise.
10253 (machine_function): Change type fn_type.
10254 (arc_compute_function_type): Consider new naked function type,
10255 change function return type.
10256 (arc_must_save_register): Adapt to handle new
10257 arc_compute_function_type's return type.
10258 (arc_expand_prologue): Likewise.
10259 (arc_expand_epilogue): Likewise.
10260 (arc_return_address_regs): Delete.
10261 (arc_return_address_register): New function.
10262 (arc_epilogue_uses): Use above function.
10263 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10264 (arc_function_type): Change encoding, add naked type.
10265 (ARC_INTERRUPT_P): Change to handle the new encoding.
10266 (ARC_FAST_INTERRUPT_P): Likewise.
10267 (ARC_NORMAL_P): Define.
10268 (ARC_NAKED_P): Likewise.
10269 (arc_compute_function_type): Delete prototype.
10270 * config/arc/arc.md (in_ret_delay_slot): Use
10271 arc_return_address_register function.
10272 (simple_return): Likewise.
10273 (p_return_i): Likewise.
10275 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10277 PR tree-optimization/81428
10278 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10279 can't be built for those types.
10281 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10283 Remove stuff dead since r239246.
10285 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10286 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10287 (avr_inform_devices): Remove dead stuff.
10289 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10291 * config/arm/arm_neon.h: Fix softp typo.
10293 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10295 PR tree-optimization/81365
10296 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10297 aggregate moves onto bb predecessor edges, make sure there are no
10298 loads that could alias the lhs in between the start of bb and the
10301 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10304 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10305 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10306 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10308 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10310 PR tree-optimization/81396
10311 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10312 (init_symbolic_number): Initialize it to 1.
10313 (perform_symbolic_merge): Add n_ops from both operands into the new
10315 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10316 without base_addr as useless if they need more than one operation.
10317 (bswap_replace): Handle !bswap case for NULL base_addr.
10319 2017-07-17 Tom de Vries <tom@codesourcery.com>
10322 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10325 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10327 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10328 conditional builtin define __FIX_LEON3FT_B2BST.
10330 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10332 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10333 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10336 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10338 PR rtl-optimization/81424
10339 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10340 to remove potential trapping from operands if -fnon-call-exceptions.
10342 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10344 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10345 profile_proability for scalling.
10346 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10348 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10350 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10352 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10354 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10355 fixpoint arithmetics.
10357 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10359 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10360 fixpoint arithmetics.
10362 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10364 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10365 fixpoint arithmetics.
10367 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10369 * profile-count.h (profile_probability::from_reg_br_prob_note,
10370 profile_probability::to_reg_br_prob_note): New functions.
10371 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10372 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10373 * predict.c (probability_reliable_p): Update.
10374 (edge_probability_reliable_p): Update.
10375 (br_prob_note_reliable_p): Update.
10376 (invert_br_probabilities): Update.
10377 (add_reg_br_prob_note): New function.
10378 (combine_predictions_for_insn): Update.
10379 * asan.c (asan_clear_shadow): Update.
10380 * cfgbuild.c (compute_outgoing_frequencies): Update.
10381 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10382 (update_br_prob_note): Update.
10383 (rtl_verify_edges): Update.
10384 (purge_dead_edges): Update.
10385 (fixup_reorder_chain): Update.
10386 * emit-rtl.c (try_split): Update.
10387 * ifcvt.c (cond_exec_process_insns): Update.
10388 (cond_exec_process_if_block): Update.
10389 (dead_or_predicable): Update.
10390 * internal-fn.c (expand_addsub_overflow): Update.
10391 (expand_neg_overflow): Update.
10392 (expand_mul_overflow): Update.
10393 * loop-doloop.c (doloop_modify): Update.
10394 * loop-unroll.c (compare_and_jump_seq): Update.
10395 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10396 * predict.h: Update.
10397 * reorg.c (mostly_true_jump): Update.
10399 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10400 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10401 * config/arc/arc.c: (emit_unlikely_jump): Update.
10402 * config/arm/arm.c: (emit_unlikely_jump): Update.
10403 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10404 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10405 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10406 (ix86_print_operand): Update.
10407 (ix86_split_fp_branch): Update.
10408 (predict_jump): Update.
10409 * config/ia64/ia64.c (ia64_print_operand): Update.
10410 * config/mmix/mmix.c (mmix_print_operand): Update.
10411 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10412 (rs6000_expand_split_stack_prologue): Update.
10413 * config/rs6000/rs6000.c: Update.
10414 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10415 (s390_expand_vec_movstr): Update.
10416 (s390_expand_cs_tdsi): Update.
10417 (s390_expand_split_stack_prologue): Update.
10418 * config/sh/sh.c (sh_print_operand): Update.
10419 (expand_cbranchsi4): Update.
10420 (expand_cbranchdi4): Update.
10421 * config/sparc/sparc.c (output_v9branch): Update.
10422 * config/spu/spu.c (get_branch_target): Update.
10423 (ea_load_store_inline): Update.
10424 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10425 * config/tilepro/tilepro.c: Update.
10427 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10429 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10430 (gimplify_save_expr): Likewise.
10432 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10434 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10436 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10438 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10440 * timevar.def (TV_IPA_FNSUMMARY): Define.
10442 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10444 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10445 to back store errata sensitive sequence from being generated.
10446 (sqrtdf2_fix): Likewise.
10448 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10450 * tree-ssa-threadupdate.c (compute_path_counts,
10451 update_joiner_offpath_counts): Use profile_probability.
10453 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10456 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10458 * config/arm/arm-c.c (arm_cpu_builtins): Define
10459 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10461 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10463 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10464 array entries to represent __ieee128 versions of the
10465 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10466 scalar_extract_sig, and scalar_insert_exp built-in functions.
10467 (altivec_resolve_overloaded_builtin): Add special case handling
10468 for the __builtin_scalar_insert_exp function, as represented by
10469 the P9V_BUILTIN_VEC_VSIEDP constant.
10470 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10471 exponent support for __ieee128 argument.
10472 (VSESQP): Add scalar extract signature support for __ieee128
10474 (VSTDCNQP): Add scalar test negative support for __ieee128
10476 (VSIEQP): Add scalar insert exponent support for __int128 argument
10477 with __ieee128 result.
10478 (VSIEQPF): Add scalar insert exponent support for __ieee128
10479 argument with __ieee128 result.
10480 (VSTDCQP): Add scalar test data class support for __ieee128
10482 (VSTDCNQP): Add overload support for scalar test negative with
10483 __ieee128 argument.
10484 (VSTDCQP): Add overload support for scalar test data class
10485 __ieee128 argument.
10486 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10487 UNSPEC_VSX_SXSIGDP.
10488 (UNSPEC_VSX_SIEXPQP): New constant.
10489 (xsxexpqp): New insn for VSX scalar extract exponent quad
10491 (xsxsigqp): New insn for VSX scalar extract significand quad
10493 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10494 precision with floating point argument.
10495 (xststdcqp): New expand for VSX scalar test data class quad
10497 (xststdcnegqp): New expand for VSX scalar test negative quad
10499 (xststdcqp): New insn to match expansions for VSX scalar test data
10500 class quad precision and VSX scalar test negative quad precision.
10501 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10502 special case operand checking to enforce that second operand of
10503 VSX scalar test data class with quad precision argument is a 7-bit
10505 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10506 prototypes and descriptions of __ieee128 versions of
10507 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10508 scalar_test_data_class, and scalar_test_neg built-in functions.
10510 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10512 PR tree-optimization/81162
10513 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10514 replace a negate with an add.
10516 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10518 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10520 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10522 * config/arm/arm-c.c (arm_cpu_builtins): Define
10523 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10525 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10527 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10528 (armv8-r): Set ARM Cortex-R52 as default CPU.
10529 * config/arm/arm-tables.opt: Regenerate.
10530 * config/arm/arm-tune.md: Regenerate.
10531 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10533 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10534 extension for -mcpu=cortex-r52.
10536 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10538 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10539 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10540 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10541 (fp-armv8): Define it as FP_ARMv8 only.
10542 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10543 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10545 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10546 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10547 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10548 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10549 than TARGET_FPU_ARMV8.
10550 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10551 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10552 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10554 * config/arm/neon.md (neon_vrint): Likewise.
10555 (neon_vcvt): Likewise.
10556 (neon_<fmaxmin_op><mode>): Likewise.
10557 (<fmaxmin><mode>3): Likewise.
10558 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10559 * config/arm/predicates.md (arm_cond_move_operator): Check against
10560 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10562 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10564 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10565 to top of function.
10567 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10569 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10570 loop in comment with memset.
10572 2017-07-14 Martin Liska <mliska@suse.cz>
10574 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10575 * dwarf2out.c (is_java): Remove the function.
10576 (output_pubname): Remove usage of the function.
10577 (lower_bound_default): Remove usage of DW_LANG_Java.
10578 (gen_compile_unit_die): Likewise.
10579 * gcc.c: Remove compiler defaults for .java and .zip files.
10580 * gimple-expr.c (remove_suffix): Change as there's no longer
10581 extension than 4-letter one.
10582 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10583 (gimplify_save_expr): Likewise.
10584 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10585 as it's possible even for other languages than Java.
10586 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10587 * lto-opts.c (lto_write_options): Remove reference to Java.
10588 * opts.c (strip_off_ending): Update file extension handling.
10589 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10590 * tree-eh.c (lower_resx): Likewise.
10591 * tree.c (free_lang_data_in_type): Remove dead code.
10592 (find_decls_types_r): Likewise.
10593 (build_common_builtin_nodes): Remove Java from a comment.
10594 (verify_type): Remove dead code.
10595 * varasm.c (assemble_external): Remove Java from a comment.
10597 2017-07-14 Martin Liska <mliska@suse.cz>
10599 * opts.c (finish_options): Add quotes.
10600 (common_handle_option): Likewise.
10602 2017-07-14 Martin Liska <mliska@suse.cz>
10604 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10605 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10606 Remove N_SO_PASCAL.
10607 * dwarf2out.c (lower_bound_default): Do not handle
10609 (gen_compile_unit_die): Likewise.
10610 * gcc.c: Remove default extension binding for GNU Pascal.
10611 * stmt.c: Remove Pascal language from a comment.
10612 * xcoffout.c: Likewise.
10614 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10617 * diagnostic-show-locus.c (fixit_cmp): New function.
10618 (layout::layout): Sort m_fixit_hints.
10619 (column_range::column_range): Assert that the values are valid.
10620 (struct char_span): New struct.
10621 (correction::overwrite): New method.
10622 (struct source_line): New struct.
10623 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10624 calls in terms of classes source_line and char_span, and
10625 correction::overwrite.
10626 (selftest::test_overlapped_fixit_printing_2): New function.
10627 (selftest::diagnostic_show_locus_c_tests): Call it.
10629 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10632 early if there is no lhs.
10634 2017-07-13 Martin Liska <mliska@suse.cz>
10636 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10637 (gen_reference_type_die): Likewise.
10638 * stor-layout.c: Remove Pascal-related comment.
10640 2017-07-13 Martin Liska <mliska@suse.cz>
10642 * opts.c (finish_options): Add quotes to error messages.
10643 (parse_sanitizer_options): Likewise.
10645 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10647 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10649 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10651 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10653 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10655 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10657 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10659 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10662 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10663 provides the hardware capability bits, define the macro
10664 __BUILTIN_CPU_SUPPORTS__.
10665 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10666 if GLIBC does not provide the hardware capability bits. Add a
10667 gcc_unreachable call if the built-in cpu function is neither
10668 __builtin_cpu_is nor __builtin_cpu_supports.
10669 (rs6000_get_function_versions_dispatcher): Change the warning
10670 that an old GLIBC is used which does not export the capability
10671 bits to be an error.
10672 * doc/extend.texi (target_clones attribute): Document the
10673 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10674 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10675 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10676 the macros defined by GCC if the newer GLIBC is available.
10678 2017-07-12 Jeff Law <law@redhat.com>
10680 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10681 remaining includes slightly.
10682 * config/riscv/riscv-builtins.c: Include profile-count.h.
10684 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10687 * config/avr/avr.c (avr_set_current_function): In diagnostic
10688 messages: Quote keywords and (parts of) identifiers.
10689 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10692 2017-07-12 Carl Love <cel@us.ibm.com>
10694 * config/rs6000/rs6000-c.c: Add support for built-in functions
10695 vector bool char vec_revb (vector bool char);
10696 vector bool short vec_revb (vector short char);
10697 vector bool int vec_revb (vector bool int);
10698 vector bool long long vec_revb (vector bool long long);
10699 * doc/extend.texi: Update the built-in documentation file for the
10700 new built-in functions.
10702 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10704 * config/s390/s390.md: Remove movcc splitter.
10706 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10708 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10709 load/store on condition.
10711 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10714 * config/avr/avr.c (avr_encode_section_info)
10715 [progmem && !TREE_READONLY]: Error if progmem object needs
10718 2017-07-11 Michael Collison <michael.collison@arm.com>
10720 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10723 2017-07-11 Carl Love <cel@us.ibm.com>
10725 * config/rs6000/rs6000-c.c: Add support for builtins
10726 vector unsigned int vec_parity_lsbb (vector signed int);
10727 vector unsigned int vec_parity_lsbb (vector unsigned int);
10728 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10729 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10730 vector unsigned long long vec_parity_lsbb (vector signed long long);
10731 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10732 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10733 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10734 * doc/extend.texi: Update the built-in documentation file for the
10735 new built-in functions.
10737 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10739 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10740 (layout::m_primary_loc): New field.
10741 (layout::layout): Initialize new field. Move location filtering
10742 logic from here to...
10743 (layout::maybe_add_location_range): ...this new method. Add
10744 support for filtering to just the lines already specified by other
10746 (layout::will_show_line_p): New method.
10747 (gcc_rich_location::add_location_if_nearby): New method.
10748 (selftest::test_add_location_if_nearby): New test function.
10749 (selftest::diagnostic_show_locus_c_tests): Call it.
10750 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10753 2017-07-11 Tom de Vries <tom@codesourcery.com>
10755 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10756 (bb_first_real_insn): New function.
10757 (nvptx_single): Add extra initialization of broadcasted condition
10760 2017-07-11 Nathan Sidwell <nathan@acm.org>
10762 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10764 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10766 * doc/extend.texi (AVR Function Attributes): Remove weblink to
10767 Binutils doc as TEXI will mess them up.
10768 * doc/invoke.texi (AVR Options): Same here.
10770 2017-07-11 Daniel Cederman <cederman@gaisler.com>
10772 * config/sparc/sparc.opt (mfix-ut700): New option.
10773 (mfix-gr712rc): Likewise.
10774 (sparc_fix_b2bst): New variable.
10775 * doc/invoke.texi (SPARC options): Document them.
10776 (ARM options): Fix warnings.
10777 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
10778 instructions to prevent sequences that can trigger the store-store
10779 errata for certain LEON3FT processors.
10780 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
10781 (sparc_option_override): Set sparc_fix_b2bst appropriately.
10782 * config/sparc/sparc.md (fix_b2bst): New attribute.
10783 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
10785 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
10788 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
10790 (*rsqrtsf2_sse): Ditto.
10792 (div<mode>3): Macroize insn from divdf3 and divsf3
10793 using MODEF mode iterator.
10795 2017-07-10 Martin Sebor <msebor@redhat.com>
10797 PR tree-optimization/80397
10798 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10799 instead of testing for equality to INTEGER_TYPE.
10801 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10803 * config.gcc: Remove uclibc from arc target spec.
10805 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10807 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10809 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10812 * lto-wrapper.c (remove_option): New function.
10813 (merge_and_complain): Merge PIC/PIE options more realistically.
10815 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10817 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10821 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10822 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10823 * config.in: Regenerate.
10824 * configure: Regenerate.
10825 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10826 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10827 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10828 (TARGET_GASISR_PROLOGUES): ...target mask.
10829 * common/config/avr/avr-common.c
10830 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10831 Set -mgas-isr-prologues.
10832 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10833 INSERT_PASS_BEFORE for it.
10834 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10835 * config/avr/avr.c (avr_option_override)
10836 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10837 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10838 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10839 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10840 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10842 (make_avr_pass_pre_proep): New function.
10843 (emit_push_sfr) <treg>: Add argument to function and use it
10844 instead of TMP_REG.
10845 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10846 and set machine->gasisr.yes.
10847 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10848 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10849 __gcc_isr.n_pushed to .L__stack_usage.
10850 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10851 (avr_asm_final_postscan_insn): ...this new static function.
10852 * config/avr/avr.h (machine_function)
10853 <is_no_gccisr, use_L__stack_usage>: New fields.
10854 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10855 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10856 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10857 (gasisr, *gasisr): New expander and insn.
10858 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10859 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10860 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10862 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10864 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10867 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10869 Move jump-tables out of .text again.
10872 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10873 (ASM_OUTPUT_ADDR_VEC): New function.
10874 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10875 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10876 INSN_ADDRESSes as asm comment.
10877 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10878 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10879 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10880 * config/avr/avr.md (*tablejump): Adjust comment.
10881 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10882 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10884 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10885 (avr_output_addr_vec): New proto.
10886 (avr_log_t) <insn_addresses>: New field.
10888 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10891 * config/i386/i386.c (ix86_function_arg_advance): Set
10892 outgoing_args_on_stack to true if there are outgoing arguments
10894 (ix86_function_arg): Likewise.
10895 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10896 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10897 * config/i386/i386.h (machine_function): Add
10898 outgoing_args_on_stack.
10900 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10902 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10903 supporting pthreds.
10904 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10906 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10908 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10909 (REAL_H): Remove $(MACHMODE_H).
10910 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10912 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10913 $(MACHMODE_H) and double-int.h.
10914 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10916 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10919 2017-07-07 Andrew Pinski <apinski@cavium.com>
10921 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10922 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10924 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10926 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10927 Add warning if GCC was not configured to link against a GLIBC that
10928 exports the hardware capability bits.
10929 (make_resolver_func): Make resolver function private and not a
10930 COMDAT function. Create the name with clone_function_name instead
10931 of make_unique_name.
10934 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10935 correct operand in doing the split.
10937 2017-07-07 Carl Love <cel@us.ibm.com>
10939 * config/rs6000/rs6000-c: Add support for built-in function
10940 vector unsigned short vec_pack_to_short_fp32 (vector float,
10942 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10943 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10944 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10945 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10946 (convert_4f32_8i16): Add define_expand.
10947 * doc/extend.texi: Update the built-in documentation file for the
10948 new built-in function.
10950 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10952 * config/sparc/m8.md: New file.
10953 * config/sparc/sparc.md: Include m8.md.
10955 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10957 * config/sparc/sparc.opt: New option -mvis4b.
10958 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10959 (sparc_option_override): Handle VIS4B.
10960 (enum sparc_builtins): Define
10961 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10962 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10963 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10964 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10965 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10966 (check_constant_argument): New function.
10967 (sparc_vis_init_builtins): Define builtins
10968 __builtin_vis_dictunpack{8,16,32},
10969 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10970 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10971 __builtin_vis_fpcmpde{8,16,32}shl and
10972 __builtin_vis_fpcmpur{8,16,32}shl.
10973 (sparc_expand_builtin): Check that the constant operands to
10974 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10975 constant and in range.
10976 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10978 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10979 (SPARC_IMM5_P): Likewise.
10980 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10981 (enabled): Handle vis4b.
10982 (UNSPEC_DICTUNPACK): New unspec.
10983 (UNSPEC_FPCMPSHL): Likewise.
10984 (UNSPEC_FPUCMPSHL): Likewise.
10985 (UNSPEC_FPCMPDESHL): Likewise.
10986 (UNSPEC_FPCMPURSHL): Likewise.
10987 (cpu_feature): New CPU feature `vis4b'.
10988 (dictunpack{8,16,32}): New insns.
10989 (FPCSMODE): New mode iterator.
10990 (fpcscond): New code iterator.
10991 (fpcsucond): Likewise.
10992 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10993 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10994 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10995 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10996 * config/sparc/constraints.md: Define constraints `q' for unsigned
10997 2-bit integer constants and `t' for unsigned 5-bit integer
10999 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11001 (imm5_operand_dictunpack16): Likewise.
11002 (imm5_operand_dictunpack32): Likewise.
11003 (imm2_operand): Likewise.
11004 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11005 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11006 ditunpack* and fpcmp*shl builtins.
11008 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11010 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11011 * config.in: Add HAVE_AS_SPARC6 define.
11012 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11014 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11016 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11017 (CPP_CPU_SPEC): Handle m8.
11018 (ASM_CPU_SPEC): Likewise.
11019 * config/sparc/sparc-opts.h (enum processor_type): Add
11021 * config/sparc/sparc.c (m8_costs): New struct.
11022 (sparc_option_override): Handle TARGET_CPU_m8.
11023 (sparc32_initialize_trampoline): Likewise.
11024 (sparc64_initialize_trampoline): Likewise.
11025 (sparc_issue_rate): Likewise.
11026 (sparc_register_move_cost): Likewise.
11027 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11028 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11029 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11030 (CPP_CPU_SPEC): Handle M8.
11031 (ASM_CPU_SPEC): Likewise.
11032 (AS_M8_FLAG): Define.
11033 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11034 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11035 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11037 * configure: Regenerate.
11038 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11041 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11043 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11045 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11046 ("*movdi_insn_sp32"): Do not set v3pipe.
11047 ("*movsi_insn"): Likewise.
11048 ("*movdi_insn_sp64"): Likewise.
11049 ("*movsf_insn"): Likewise.
11050 ("*movdf_insn_sp32"): Likewise.
11051 ("*movdf_insn_sp64"): Likewise.
11052 ("*zero_extendsidi2_insn_sp64"): Likewise.
11053 ("*sign_extendsidi2_insn"): Likewise.
11054 ("*mov<VM32:mode>_insn"): Likewise.
11055 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11056 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11057 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11058 ("<vlop:code><VL:mode>3"): Likewise.
11059 ("*not_<vlop:code><VL:mode>3"): Likewise.
11060 ("*nand<VL:mode>_vis"): Likewise.
11061 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11062 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11063 ("one_cmpl<VL:mode>2"): Likewise.
11064 ("faligndata<VM64:mode>_vis"): Likewise.
11065 ("alignaddrsi_vis"): Likewise.
11066 ("alignaddrdi_vis"): Likweise.
11067 ("alignaddrlsi_vis"): Likewise.
11068 ("alignaddrldi_vis"): Likewise.
11069 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11070 ("bmaskdi_vis"): Likewise.
11071 ("bmasksi_vis"): Likewise.
11072 ("bshuffle<VM64:mode>_vis"): Likewise.
11073 ("cmask8<P:mode>_vis"): Likewise.
11074 ("cmask16<P:mode>_vis"): Likewise.
11075 ("cmask32<P:mode>_vis"): Likewise.
11076 ("pdistn<P:mode>_vis"): Likewise.
11077 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11079 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11081 * config/sparc/sparc.md ("subtype"): New insn attribute.
11082 ("*wrgsr_sp64"): Set insn subtype.
11083 ("*rdgsr_sp64"): Likewise.
11084 ("alignaddrsi_vis"): Likewise.
11085 ("alignaddrdi_vis"): Likewise.
11086 ("alignaddrlsi_vis"): Likewise.
11087 ("alignaddrldi_vis"): Likewise.
11088 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11089 ("fexpand_vis"): Likewise.
11090 ("fpmerge_vis"): Likewise.
11091 ("faligndata<VM64:mode>_vis"): Likewise.
11092 ("bshuffle<VM64:mode>_vis"): Likewise.
11093 ("cmask8<P:mode>_vis"): Likewise.
11094 ("cmask16<P:mode>_vis"): Likewise.
11095 ("cmask32<P:mode>_vis"): Likewise.
11096 ("fchksm16_vis"): Likewise.
11097 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11098 ("fmean16_vis"): Likewise.
11099 ("fp<plusminus_insn>64_vis"): Likewise.
11100 ("<plusminus_insn>v8qi3"): Likewise.
11101 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11102 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11103 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11104 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11105 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11106 ("*movqi_insn"): Likewise.
11107 ("*movhi_insn"): Likewise.
11108 ("*movsi_insn"): Likewise.
11109 ("movsi_pic_gotdata_op"): Likewise.
11110 ("*movdi_insn_sp32"): Likewise.
11111 ("*movdi_insn_sp64"): Likewise.
11112 ("movdi_pic_gotdata_op"): Likewise.
11113 ("*movsf_insn"): Likewise.
11114 ("*movdf_insn_sp32"): Likewise.
11115 ("*movdf_insn_sp64"): Likewise.
11116 ("*zero_extendhisi2_insn"): Likewise.
11117 ("*zero_extendqihi2_insn"): Likewise.
11118 ("*zero_extendqisi2_insn"): Likewise.
11119 ("*zero_extendqidi2_insn"): Likewise.
11120 ("*zero_extendhidi2_insn"): Likewise.
11121 ("*zero_extendsidi2_insn_sp64"): Likewise.
11122 ("ldfsr"): Likewise.
11123 ("prefetch_64"): Likewise.
11124 ("prefetch_32"): Likewise.
11125 ("tie_ld32"): Likewise.
11126 ("tie_ld64"): Likewise.
11127 ("*tldo_ldub_sp32"): Likewise.
11128 ("*tldo_ldub1_sp32"): Likewise.
11129 ("*tldo_ldub2_sp32"): Likewise.
11130 ("*tldo_ldub_sp64"): Likewise.
11131 ("*tldo_ldub1_sp64"): Likewise.
11132 ("*tldo_ldub2_sp64"): Likewise.
11133 ("*tldo_ldub3_sp64"): Likewise.
11134 ("*tldo_lduh_sp32"): Likewise.
11135 ("*tldo_lduh1_sp32"): Likewise.
11136 ("*tldo_lduh_sp64"): Likewise.
11137 ("*tldo_lduh1_sp64"): Likewise.
11138 ("*tldo_lduh2_sp64"): Likewise.
11139 ("*tldo_lduw_sp32"): Likewise.
11140 ("*tldo_lduw_sp64"): Likewise.
11141 ("*tldo_lduw1_sp64"): Likewise.
11142 ("*tldo_ldx_sp64"): Likewise.
11143 ("*mov<VM32:mode>_insn"): Likewise.
11144 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11145 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11147 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11149 * config/sparc/sparc.md ("type"): New insn type viscmp.
11150 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11152 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11153 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11154 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11155 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11157 ("n7_vis_logical_11cycle"): Likewise.
11158 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11159 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11160 * config/sparc/niagara.md ("niag_vis"): Likewise.
11161 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11162 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11164 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11166 * config/sparc/sparc.md: New instruction type `bmask'.
11167 (bmaskdi_vis): Use the `bmask' type.
11168 (bmasksi_vis): Likewise.
11169 * config/sparc/ultra3.md (us3_array): Likewise.
11170 * config/sparc/niagara7.md (n7_array): Likewise.
11171 * config/sparc/niagara4.md (n4_array): Likewise.
11172 * config/sparc/niagara2.md (niag2_vis): Likewise.
11173 (niag3_vis): Likewise.
11174 * config/sparc/niagara.md (niag_vis): Likewise.
11176 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11178 * ipa-comdats.c: Remove optimize check from gate.
11179 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11180 for functions not optimized.
11181 (ipa_fn_summary_read): Skip optimize check.
11182 (ipa_fn_summary_write): Likewise.
11183 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11185 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11187 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11188 (check_callers): Give up on caller which is not optimized.
11189 (inline_small_functions): Likewise.
11190 (ipa_inline): Do not give up when not optimizing.
11191 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11192 away unoptimizes cdtors.
11193 (whole_program_function_and_variable_visibility): Do
11194 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11195 * ipa.c (process_references): Do not check optimize.
11196 (symbol_table::remove_unreachable_nodes): Update optimize check.
11197 (set_writeonly_bit): Update optimize check.
11198 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11199 (pass_ipa_single_use::gate): Remove.
11201 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11203 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11204 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11205 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11206 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11207 permute_load, permute_store, adjust_extract, adjust_splat,
11208 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11209 replace_swap_with_copy, dump_swap_insn_table,
11210 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11211 recombine_lvx_pattern, recombine_stvx_pattern,
11212 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11213 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11214 to file rs6000-p8swap.c.
11215 * config/rs6000/rs6000-p8swap.c: New file.
11216 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11217 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11218 and rs6000*-*-* targets.
11220 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11222 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11224 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11226 * lto-wrapper.c (merge_and_complain): Do not merge
11227 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11228 fsigned_zeros, ftrapping_math, fwrapv.
11229 (append_compiler_options): Do not track these options.
11230 (append_linker_options): Likewie
11232 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11234 * cgraphunit.c (cgraph_node::finalize_function): When
11235 !flag_toplevel_reorde set no_reorder flag.
11236 (varpool_node::finalize_decl): Likewise.
11237 (symbol_table::compile): Drop no toplevel reorder path.
11239 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11241 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11242 edges; zero probability is not better than uninitialized.
11244 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11246 * asan.h (asan_sanitize_allocas_p): Declare.
11247 * asan.c (asan_sanitize_allocas_p): New function.
11248 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11249 (handle_builtin_alloca): Likewise.
11250 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11251 if !asan_sanitize_allocas_p.
11252 * params.def (asan-instrument-allocas): Add new option.
11253 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11254 * opts.c (common_handle_option): Disable allocas sanitization for
11257 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11259 * asan.c: Include gimple-fold.h.
11260 (get_last_alloca_addr): New function.
11261 (handle_builtin_stackrestore): Likewise.
11262 (handle_builtin_alloca): Likewise.
11263 (asan_emit_allocas_unpoison): Likewise.
11264 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11265 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11266 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11267 (instrument_builtin_call): Pass gimple iterator to
11268 get_mem_refs_of_builtin_call.
11269 (last_alloca_addr): New global.
11270 * asan.h (asan_emit_allocas_unpoison): Declare.
11271 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11272 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11273 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11274 if function calls alloca.
11275 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11276 * gimple-fold.h (replace_call_with_value): Declare.
11277 * internal-fn.c: Include asan.h.
11278 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11279 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11281 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11283 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11284 (C_SELFTEST_FLAGS): New.
11285 (CPP_SELFTEST_FLAGS): New.
11286 (SELFTEST_DEPS): New, from deps of s-selftest.
11287 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11288 (CPP_SELFTEST_DEPS): New.
11289 (selftest): Add dependency on s-selftest-c++.
11290 (s-selftest): Rename to...
11291 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11292 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11293 than SELFTEST_FLAGS.
11294 (selftest-gdb): Rename to...
11295 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11297 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11298 (selftest-valgrind): Rename to...
11299 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11301 (selftest-valgrind): Reintroduce as an alias for
11302 selftest-c-valgrind.
11303 (s-selftest-c++): New.
11304 (selftest-c++-gdb): New.
11305 (selftest-c++-valgrind): New.
11307 2017-07-06 Olivier Hainque <hainque@adacore.com>
11309 * gcc.c (process_command): When deciding if undefined variables
11310 should be ignored when processing specs, accept "gcc -v" as well.
11312 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11314 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11315 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11317 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11319 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11320 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11321 * config/arm/arm-tables.opt: Regenerate.
11322 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11324 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11326 2017-07-06 Carl Love <cel@us.ibm.com>
11328 * ChangeLog: Clean up from mid air collision
11330 2017-07-06 Carl Love <cel@us.ibm.com>
11332 * config/rs6000/rs6000-c.c: Add support for built-in functions
11333 vector signed int vec_subc (vector signed int, vector signed int);
11334 vector signed __int128 vec_subc (vector signed __int128,
11335 vector signed __int128);
11336 vector unsigned __int128 vec_subc (vector unsigned __int128,
11337 vector unsigned __int128);
11338 vector signed int vec_sube (vector signed int, vector signed int,
11339 vector signed int);
11340 vector unsigned int vec_sube (vector unsigned int,
11341 vector unsigned int,
11342 vector unsigned int);
11343 vector signed __int128 vec_sube (vector signed __int128,
11344 vector signed __int128,
11345 vector signed__int128);
11346 vector unsigned __int128 vec_sube (vector unsigned __int128,
11347 vector unsigned __int128,
11348 vector unsigned __int128);
11349 vector signed int vec_subec (vector signed int, vector signed int,
11350 vector signed int);
11351 vector unsigned int vec_subec (vector unsigned int,
11352 vector unsigned int,
11353 vector unsigned int);
11354 vector signed __int128 vec_subec (vector signed __int128,
11355 vector signed __int128,
11356 vector signed__int128);
11357 vector unsigned __int128 vec_subec (vector unsigned __int128,
11358 vector unsigned __int128,
11359 vector unsigned __int128);
11360 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11361 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11362 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11363 BU_ALTIVEC_OVERLOAD_X definitions.
11364 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11365 * doc/extend.texi: Update the built-in documentation file for the new
11366 built-in functions.
11368 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11371 * diagnostic-show-locus.c (layout::layout): Use start and finish
11372 spelling location for the start and finish of each range.
11373 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11374 Add unused aspect param.
11375 * input.c (expand_location_1): Add "aspect" param, and use it
11376 to access the correct part of the location.
11377 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11379 (expand_location_to_spelling_point): Likewise.
11380 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11381 param, and pass it to expand_location_1.
11383 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11385 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11386 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11387 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11388 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11389 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11390 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11391 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11392 _mm_maskz_getmant_ss): New intrinsics.
11393 (__builtin_ia32_getexpss128_mask): Changed to ...
11394 __builtin_ia32_getexpss128_round ... this.
11395 (__builtin_ia32_getexpsd128_mask): Changed to ...
11396 __builtin_ia32_getexpsd128_round ... this.
11397 * config/i386/i386-builtin-types.def
11398 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11399 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11400 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11401 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11402 __builtin_ia32_getmantss_mask_round): New builtins.
11403 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11404 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11405 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11406 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11407 * config/i386/sse.md
11408 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11409 avx512f_sgetexp<mode><mask_scalar_name>
11410 <round_saeonly_scalar_name> ... this.
11411 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11412 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11413 vgetexp<ssescalarmodesuffix>
11414 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11415 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11416 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11417 avx512f_vgetmant<mode><mask_scalar_name>
11418 <round_saeonly_scalar_name> ... this.
11419 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11420 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11421 vgetmant<ssescalarmodesuffix>
11422 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11423 %0<mask_scalar_operand4>, %1, %2
11424 <round_saeonly_scalar_mask_op4>, %3} ... this.
11425 * config/i386/subst.md (mask_scalar_operand4,
11426 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11427 round_saeonly_scalar_nimm_predicate): New subst attributes.
11429 2017-07-06 Julia Koval <julia.koval@intel.com>
11431 * config/i386/i386.c (ix86_erase_embedded_rounding):
11432 Remove code for old rounding pattern.
11434 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11436 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11438 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11440 * doc/sourcebuild.texi (Test Directives, Variants of
11441 dg-require-support): Add documentation for dg-require-stack-check.
11443 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11445 * config/i386/subst.md (mask_scalar, round_scalar,
11446 round_saeonly_scalar): New meta-templates.
11447 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11448 round_scalar_mask_operand3, round_scalar_mask_op3,
11449 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11450 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11451 round_saeonly_scalar_constraint,
11452 round_saeonly_scalar_prefix): New subst attribute.
11453 * config/i386/sse.md
11454 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11455 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11456 <round_scalar_name> ... this.
11457 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11458 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11459 <round_scalar_name> ... this.
11460 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11461 <sse>_vm<code><mode>3<mask_scalar_name>
11462 <round_saeonly_scalar_name> ... this.
11463 (v<plusminus_mnemonic><ssescalarmodesuffix>
11464 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11465 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11466 v<plusminus_mnemonic><ssescalarmodesuffix>
11467 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11468 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11469 (v<multdiv_mnemonic><ssescalarmodesuffix>
11470 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11471 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11472 v<multdiv_mnemonic><ssescalarmodesuffix>
11473 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11474 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11475 (v<maxmin_float><ssescalarmodesuffix>
11476 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11477 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11478 v<maxmin_float><ssescalarmodesuffix>
11479 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11480 %0<mask_scalar_operand3>, %1, %<iptr>2
11481 <round_saeonly_scalar_mask_op3>} ... this.
11483 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11485 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11486 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11488 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11489 Alan Hayward <alan.hayward@arm.com>
11490 David Sherwood <david.sherwood@arm.com>
11492 * combine.c (simplify_if_then_else): Remove "enum" before
11494 * compare-elim.c (can_eliminate_compare): Likewise.
11495 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11497 (aarch64_lookup_simd_builtin_type): Likewise.
11498 (aarch64_simd_builtin_type): Likewise.
11499 (aarch64_init_simd_builtin_types): Likewise.
11500 (aarch64_simd_expand_args): Likewise.
11501 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11503 (aarch64_reverse_mask): Likewise.
11504 (aarch64_simd_emit_reg_reg_move): Likewise.
11505 (aarch64_gen_adjusted_ldpstp): Likewise.
11506 (aarch64_ccmp_mode_to_code): Likewise.
11507 (aarch64_operands_ok_for_ldpstp): Likewise.
11508 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11509 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11511 (aarch64_min_divisions_for_recip_mul): Likewise.
11512 (aarch64_reassociation_width): Likewise.
11513 (aarch64_get_condition_code_1): Likewise.
11514 (aarch64_simd_emit_reg_reg_move): Likewise.
11515 (aarch64_simd_attr_length_rglist): Likewise.
11516 (aarch64_reverse_mask): Likewise.
11517 (aarch64_operands_ok_for_ldpstp): Likewise.
11518 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11519 (aarch64_gen_adjusted_ldpstp): Likewise.
11520 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11522 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11523 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11524 (arm_lookup_simd_builtin_type): Likewise.
11525 (arm_simd_builtin_type): Likewise.
11526 (arm_init_simd_builtin_types): Likewise.
11527 (arm_expand_builtin_args): Likewise.
11528 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11529 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11530 (ft32_setup_incoming_varargs): Likewise.
11531 (ft32_function_arg): Likewise.
11532 (ft32_function_arg_advance): Likewise.
11533 (ft32_pass_by_reference): Likewise.
11534 (ft32_arg_partial_bytes): Likewise.
11535 (ft32_valid_pointer_mode): Likewise.
11536 (ft32_addr_space_pointer_mode): Likewise.
11537 (ft32_addr_space_legitimate_address_p): Likewise.
11538 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11540 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11541 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11542 (iamcu_alignment): Likewise.
11543 (canonicalize_vector_int_perm): Likewise.
11544 (ix86_noce_conversion_profitable_p): Likewise.
11545 (ix86_mpx_bound_mode): Likewise.
11546 (ix86_operands_ok_for_move_multiple): Likewise.
11547 * config/microblaze/microblaze-protos.h
11548 (microblaze_expand_conditional_branch_reg): Likewise.
11549 * config/microblaze/microblaze.c
11550 (microblaze_expand_conditional_branch_reg): Likewise.
11551 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11553 (rs6000_reassociation_width): Likewise.
11554 (rs6000_invalid_binary_op): Likewise.
11555 (fusion_p9_p): Likewise.
11556 (emit_fusion_p9_load): Likewise.
11557 (emit_fusion_p9_store): Likewise.
11558 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11560 (riscv_hard_regno_mode_ok_p): Likewise.
11561 (riscv_address_insns): Likewise.
11562 (riscv_split_symbol): Likewise.
11563 (riscv_legitimize_move): Likewise.
11564 (riscv_function_value): Likewise.
11565 (riscv_hard_regno_nregs): Likewise.
11566 (riscv_expand_builtin): Likewise.
11567 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11568 (riscv_build_integer): Likewise.
11569 (riscv_split_integer): Likewise.
11570 (riscv_legitimate_constant_p): Likewise.
11571 (riscv_cannot_force_const_mem): Likewise.
11572 (riscv_regno_mode_ok_for_base_p): Likewise.
11573 (riscv_valid_base_register_p): Likewise.
11574 (riscv_valid_offset_p): Likewise.
11575 (riscv_valid_lo_sum_p): Likewise.
11576 (riscv_classify_address): Likewise.
11577 (riscv_legitimate_address_p): Likewise.
11578 (riscv_address_insns): Likewise.
11579 (riscv_load_store_insns): Likewise.
11580 (riscv_force_binary): Likewise.
11581 (riscv_split_symbol): Likewise.
11582 (riscv_force_address): Likewise.
11583 (riscv_legitimize_address): Likewise.
11584 (riscv_move_integer): Likewise.
11585 (riscv_legitimize_const_move): Likewise.
11586 (riscv_legitimize_move): Likewise.
11587 (riscv_address_cost): Likewise.
11588 (riscv_subword): Likewise.
11589 (riscv_output_move): Likewise.
11590 (riscv_canonicalize_int_order_test): Likewise.
11591 (riscv_emit_int_order_test): Likewise.
11592 (riscv_function_arg_boundary): Likewise.
11593 (riscv_pass_mode_in_fpr_p): Likewise.
11594 (riscv_pass_fpr_single): Likewise.
11595 (riscv_pass_fpr_pair): Likewise.
11596 (riscv_get_arg_info): Likewise.
11597 (riscv_function_arg): Likewise.
11598 (riscv_function_arg_advance): Likewise.
11599 (riscv_arg_partial_bytes): Likewise.
11600 (riscv_function_value): Likewise.
11601 (riscv_pass_by_reference): Likewise.
11602 (riscv_setup_incoming_varargs): Likewise.
11603 (riscv_print_operand): Likewise.
11604 (riscv_elf_select_rtx_section): Likewise.
11605 (riscv_save_restore_reg): Likewise.
11606 (riscv_for_each_saved_reg): Likewise.
11607 (riscv_register_move_cost): Likewise.
11608 (riscv_hard_regno_mode_ok_p): Likewise.
11609 (riscv_hard_regno_nregs): Likewise.
11610 (riscv_class_max_nregs): Likewise.
11611 (riscv_memory_move_cost): Likewise.
11612 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11613 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11614 (rl78_addr_space_address_mode): Likewise.
11615 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11617 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11618 (rs6000_reassociation_width): Likewise.
11619 (rs6000_invalid_binary_op): Likewise.
11620 (fusion_p9_p): Likewise.
11621 (emit_fusion_p9_load): Likewise.
11622 (emit_fusion_p9_store): Likewise.
11623 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11624 (ok_for_simple_move_operands): Likewise.
11625 (ok_for_simple_move_strict_operands): Likewise.
11626 (ok_for_simple_arith_logic_operands): Likewise.
11627 (visium_legitimize_reload_address): Likewise.
11628 (visium_select_cc_mode): Likewise.
11629 (output_cbranch): Likewise.
11630 (visium_split_double_move): Likewise.
11631 (visium_expand_copysign): Likewise.
11632 (visium_expand_int_cstore): Likewise.
11633 (visium_expand_fp_cstore): Likewise.
11634 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11635 (visium_function_arg): Likewise.
11636 (visium_function_arg_advance): Likewise.
11637 (visium_libcall_value): Likewise.
11638 (visium_setup_incoming_varargs): Likewise.
11639 (visium_legitimate_constant_p): Likewise.
11640 (visium_legitimate_address_p): Likewise.
11641 (visium_legitimize_address): Likewise.
11642 (visium_secondary_reload): Likewise.
11643 (visium_register_move_cost): Likewise.
11644 (visium_memory_move_cost): Likewise.
11645 (prepare_move_operands): Likewise.
11646 (ok_for_simple_move_operands): Likewise.
11647 (ok_for_simple_move_strict_operands): Likewise.
11648 (ok_for_simple_arith_logic_operands): Likewise.
11649 (visium_function_value_1): Likewise.
11650 (rtx_ok_for_offset_p): Likewise.
11651 (visium_legitimize_reload_address): Likewise.
11652 (visium_split_double_move): Likewise.
11653 (visium_expand_copysign): Likewise.
11654 (visium_expand_int_cstore): Likewise.
11655 (visium_expand_fp_cstore): Likewise.
11656 (visium_split_cstore): Likewise.
11657 (visium_select_cc_mode): Likewise.
11658 (visium_split_cbranch): Likewise.
11659 (output_cbranch): Likewise.
11660 (visium_print_operand_address): Likewise.
11661 * expmed.c (flip_storage_order): Likewise.
11662 * expmed.h (emit_cstore): Likewise.
11663 (flip_storage_order): Likewise.
11664 * genrecog.c (validate_pattern): Likewise.
11665 * hsa-gen.c (gen_hsa_addr): Likewise.
11666 * internal-fn.c (expand_arith_overflow): Likewise.
11667 * ira-color.c (allocno_copy_cost_saving): Likewise.
11668 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11669 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11670 (process_invariant_for_inheritance): Likewise.
11671 * lra-eliminations.c (move_plus_up): Likewise.
11672 * omp-low.c (lower_oacc_reductions): Likewise.
11673 * simplify-rtx.c (simplify_subreg): Likewise.
11674 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11675 (TARGET_CHKP_BOUND_MODE): Likewise..
11676 * targhooks.c (default_chkp_bound_mode): Likewise.
11677 (default_setup_incoming_vararg_bounds): Likewise.
11678 * targhooks.h (default_chkp_bound_mode): Likewise.
11679 (default_setup_incoming_vararg_bounds): Likewise.
11680 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11681 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11682 (have_whole_vector_shift): Likewise.
11683 * tree-vect-stmts.c (vectorizable_load): Likewise.
11684 * doc/tm.texi: Regenerate.
11686 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11688 Graceful degrade if Binutils PR21472 is not available.
11691 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11692 .rodata in flash test fails.
11693 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11694 * confgure: Regenerate.
11695 * config.in: Regenerate.
11696 * config/avr/avr.c (avr_asm_named_section)
11697 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11698 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11699 (avr_asm_init_sections): Same.
11701 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11703 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11704 (fma<VH:mode>4_intrinsic): Likewise.
11705 (*fmsub<VCVTF:mode>4): Likewise.
11706 (*fmsub<VH:mode>4_intrinsic): Likewise.
11708 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11711 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11712 Don't depend on "optimize > 0".
11713 (out_movhi_r_mr, out_movqi_mr_r): Same.
11714 (out_movhi_mr_r, out_movqi_r_mr): Same.
11715 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11716 io_address_operand on "optimize > 0".
11718 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11720 * tree-loop-distribution.c: Add general explanantion on the pass.
11721 (generate_loops_for_partition): Mark distributed loop.
11722 (pg_add_dependence_edges): New parameter. Handle alias data
11723 dependence specially and record it in the parameter if asked.
11724 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11725 (init_partition_graph_vertices, add_partition_graph_edge): New.
11726 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11727 (free_partition_graph_vdata, build_partition_graph): New.
11728 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11729 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11730 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11731 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11732 (version_for_distribution_p, finalize_partitions): New.
11733 (distribute_loop): Handle alias data dependence specially. Factor
11734 out loop fusion code as functions and call these functions.
11736 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11738 * tree-loop-distribution.c (classify_partition): New parameter and
11739 better handle reduction statement.
11740 (rdg_build_partitions): Revise comment.
11741 (distribute_loop): Compute statements in all partitions and pass it
11742 to classify_partition.
11744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11746 * tree-loop-distribution.c (enum partition_type): New.
11747 (struct partition): New field type.
11748 (partition_merge_into): Add parameter. Update partition type.
11749 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11750 (build_rdg_partition_for_vertex): Compute partition type.
11751 (rdg_build_partitions): Dump partition type.
11752 (distribute_loop): Update calls to partition_merge_into.
11754 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11756 * tree-loop-distribution.c (struct ddr_hasher): New.
11757 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11759 (classify_partition): Call get_data_dependence.
11760 (pg_add_dependence_edges): Ditto.
11761 (distribute_loop): Release data dependence hash table.
11763 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11765 * tree-loop-distribution.c (ref_base_address): Delete.
11766 (similar_memory_accesses): Rename ...
11767 (share_memory_accesses): ... to this. Check if partitions access
11768 the same memory reference.
11769 (distribute_loop): Call share_memory_accesses.
11771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11773 * tree-loop-distribution.c (struct partition): New field recording
11774 its data reference.
11775 (partition_alloc, partition_free): Init and release data refs.
11776 (partition_merge_into): Merge data refs.
11777 (build_rdg_partition_for_vertex): Collect data refs for partition.
11778 (pg_add_dependence_edges): Change parameters from vector to bitmap.
11780 (distribute_loop): Remve data refs from vertice data of partition
11783 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11785 * tree-loop-distribution.c (params.h): Include header file.
11786 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
11787 (datarefs_vec): New global var.
11788 (create_rdg_vertices): Use datarefs_vec directly.
11789 (free_rdg): Don't free data references.
11790 (build_rdg): Update use. Don't free data references.
11791 (distribute_loop): Compute global variable for data references.
11792 Bail out if there are too many data references.
11794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11796 * tree-loop-distribution.c (loop_nest): New global var.
11797 (build_rdg): Use loop directly, rather than loop nest.
11798 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11800 (distribute_loop): Compute global variable loop nest. Update use.
11802 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11804 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11805 (partition_merge_into): New parameter. Dump reason for fusion.
11806 (distribute_loop): Update use of partition_merge_into.
11808 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11810 * tree-loop-distribution.c (bb_top_order_index): New.
11811 (bb_top_order_index_size, bb_top_order_cmp): New.
11812 (stmts_from_loop): Use topological order.
11813 (pass_loop_distribution::execute): Compute and release topological
11814 order for basic blocks.
11816 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11818 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11821 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11823 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11824 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11825 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11826 * internal-fn.def (LOOP_DIST_ALIAS): New.
11827 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11828 (fold_loop_internal_call): ... this.
11829 (vect_loop_dist_alias_call): New function.
11830 (set_uid_loop_bbs): Call fold_loop_internal_call.
11831 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11834 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11837 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11838 Require dead FLAGS_REG at the beginning of a peephole.
11840 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11843 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11844 arguments in the call to __builtin_ia32_sbb_u32.
11845 (_subborrow_u64): Swap _X and _Y arguments in the call to
11846 __builtin_ia32_sbb_u64.
11848 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11851 * tree-vrp.c (compare_assert_loc): Turn into a function template
11852 with stable template parameter. Only test if a->e is NULL,
11853 !a->e == !b->e has been verified already. Use e == NULL or
11854 e != NULL instead of e or ! e tests. If stable is true, don't use
11855 iterative_hash_expr, on the other side allow a or b or both NULL
11856 and sort the NULLs last.
11857 (process_assert_insertions): Sort using compare_assert_loc<false>
11858 instead of compare_assert_loc, later sort using
11859 compare_assert_loc<true> before calling process_assert_insertions_for
11860 in a loop. Use break instead of continue once seen NULL pointer.
11862 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11864 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11865 Cortex-R7 and Cortex-R8 processors.
11867 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11869 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11870 uninitialized while src is not.
11872 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11874 * common/config/arm/arm-common.c: Adjust include path for
11876 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11877 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11878 (arm-cpu-data.h): Likewise.
11879 (arm-cpu-cdata.h): Likewise.
11880 * config/arm/arm-cpu.h: Delete.
11881 * config/arm/arm-cpu-cdata.h: Delete.
11882 * config/arm/arm-cpu-data.h: Delete.
11884 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11886 * config/arm/arm-cpus.in (cortex-a55): New.
11887 (cortex-a75): Likewise.
11888 (cortex-a75.cortex-a55): Likewise.
11889 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11891 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11892 * config/arm/arm-cpu-cdata.h: Regenerate.
11893 * config/arm/arm-cpu-data.h: Regenerate.
11894 * config/arm/arm-cpu.h: Regenerate.
11895 * config/arm/arm-tables.opt: Regenerate.
11896 * config/arm/arm-tune.md: Regenerate.
11898 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11900 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11902 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11904 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11907 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11909 PR tree-optimization/81292
11910 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11911 full_string_p, also call adjust_related_strinfos if the adjustment
11912 is simple, otherwise invalidate related strinfos.
11914 2017-07-04 Martin Liska <mliska@suse.cz>
11917 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11918 newly created variable as DECL_IGNORED_P.
11920 2017-07-04 Martin Liska <mliska@suse.cz>
11923 * ipa-inline.c (inline_small_functions):
11924 Use xstrdup_for_dump.
11926 2017-07-04 Tom de Vries <tom@codesourcery.com>
11928 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11930 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11933 * config/darwin.c (darwin_function_switched_text_sections):
11936 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11938 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11940 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11942 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11944 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11946 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11947 min_profitable_iters, and th as inclusive lower bounds.
11948 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11949 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11950 for min_profitable_iters and min_profitable_estimate.
11951 (vect_transform_loop): Treat th as an inclusive lower bound.
11952 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11954 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11957 * config/darwin.c (darwin_function_switched_text_sections):
11958 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11959 in two pieces, and suppress the use of buf.
11961 2017-07-03 Nathan Sidwell <nathan@acm.org>
11963 * hash-table.h (hash_table_mod1): Fix indentation.
11965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11967 PR middle-end/81290
11968 * predict.c (force_edge_cold): Be more careful about propagation
11970 * profile-count.h (profile_probability::guessed,
11971 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11973 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11975 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11977 * doc/invoke.texi (rcpc architecture extension): Document it.
11979 2017-07-03 Richard Biener <rguenther@suse.de>
11981 PR tree-optimization/60510
11982 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11983 the scalar reduction PHI and use it.
11984 (vectorizable_reduction): Properly guard the single_defuse_cycle
11985 path for non-SLP reduction chains where we cannot use it.
11986 Rework reduc_def/index and vector type deduction. Rework
11987 vector operand gathering during reduction op code-gen.
11988 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11989 chains dissolve the chain and leave it to non-SLP reduction
11992 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11994 * tree-data-ref.h (dr_alignment): Declare.
11995 * tree-data-ref.c (dr_alignment): New function.
11996 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11997 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11999 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12001 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12003 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12004 and base_misalignment fields.
12005 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12006 * tree-data-ref.c: Include builtins.h.
12007 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12008 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12009 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12010 * tree-vect-data-refs.c: Include tree-cfg.h.
12011 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12012 fields instead of calculating an alignment here.
12013 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12014 innermost_loop_behavior fields.
12016 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12018 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12020 (DR_STEP_ALIGNMENT): New macro.
12021 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12022 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12023 (create_data_ref): Print it.
12024 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12025 to tell whether the step preserves vector (mis)alignment.
12026 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12027 Move the check for an integer step and generalise to all INTEGER_CST.
12028 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12029 Print the outer step alignment.
12031 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12033 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12034 with offset_alignment.
12035 (DR_ALIGNED_TO): Delete.
12036 (DR_OFFSET_ALIGNMENT): New macro.
12037 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12038 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12039 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12040 (create_data_ref): Likewise.
12041 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12042 (vect_analyze_data_refs): Likewise.
12043 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12044 creating dummy innermost behavior.
12046 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12048 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12049 with a "innermost_loop_behavior *" and refeence tree.
12050 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12051 (create_data_ref): Update call accordingly.
12052 * tree-predcom.c (find_looparound_phi): Likewise.
12054 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12056 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12057 fields with dr_wrt_vec_loop.
12058 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12059 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12060 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12061 (vect_dr_behavior): New function.
12062 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12063 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12064 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12065 track whether the step preserves the misalignment.
12066 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12067 Use vect_dr_behavior.
12068 (vect_setup_realignment): Update call accordingly.
12069 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12070 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12071 call to vect_create_addr_base_for_vector_ref.
12072 (vect_create_cond_for_align_checks): Likewise.
12073 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12074 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12075 (vect_recog_mask_conversion_pattern): Likewise.
12076 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12077 (new_stmt_vec_info): Remove redundant zeroing.
12079 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12081 * common/config/arm/arm-common.c (arm_be8_option): New function.
12082 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12083 (ISA_ARMv6): Add isa_bit_be8.
12084 * config/arm/arm.h (arm_be8_option): Add prototype.
12085 (BE8_SPEC_FUNCTION): New define.
12086 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12087 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12088 (mlittle-endian): Similarly.
12089 (mbe8, mbe32): New options.
12090 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12091 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12093 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12095 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12097 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12099 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12100 (cleanup_tree_cfg_bb): Use it.
12101 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12103 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12108 * loop-doloop.c (add_test): Update profile.
12110 2017-07-03 Martin Liska <mliska@suse.cz>
12113 * sanopt.c (rewrite_usage_of_param): New function.
12114 (sanitize_rewrite_addressable_params): Likewise.
12115 (pass_sanopt::execute): Call rewrite_usage_of_param.
12117 2017-07-03 Richard Biener <rguenther@suse.de>
12119 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12120 back to using VIEW_CONVERT_EXPR.
12122 2017-07-03 Martin Liska <mliska@suse.cz>
12125 * doc/extend.texi: Document when a resolver function is
12126 generated for target_clones.
12128 2017-07-03 Martin Liska <mliska@suse.cz>
12130 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12131 and shadow memory of auto variables which are subject of
12132 use-after-scope sanitization.
12133 (asan_expand_mark_ifn): Add do set only when is_poison.
12135 2016-07-03 Richard Biener <rguenther@suse.de>
12137 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12139 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12140 (vectorizable_reduction): Perform reduction PHI creation when
12141 visiting a reduction PHI and adjust and simplify code generation
12142 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12143 (vect_transform_loop): Visit reduction PHIs.
12144 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12145 defs into the SLP tree.
12146 (vect_build_slp_tree): Reduction defs terminate the recursion.
12147 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12149 (vect_get_vec_defs_for_stmt_copy): Export.
12150 (vect_get_vec_defs): Likewise.
12151 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12153 (vect_get_vec_defs_for_stmt_copy): Declare.
12154 (vect_get_vec_defs): Likewise.
12156 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12158 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12159 parameter with a "loop" parameter and use it instead of the
12160 loop containing DR_STMT. Don't check simple_iv when doing
12161 BB analysis. Describe the two analysis modes in the comment.
12163 2017-07-03 Tom de Vries <tom@codesourcery.com>
12165 PR tree-optimization/69468
12166 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12167 (find_same_succ_bb): Handle ignore_edge_flags.
12169 2017-07-03 Tom de Vries <tom@codesourcery.com>
12171 PR tree-optimization/81192
12172 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12174 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12176 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12178 2017-07-03 Tom de Vries <tom@codesourcery.com>
12180 PR tree-optimization/81192
12181 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12182 BB_SAME_SUCC (bb) == NULL.
12184 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12186 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12189 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12191 * dumpfile.c: Include profile-count.h
12192 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12194 (insert_cond_bb): Update profile.
12195 * tree-cfg.h (insert_cond_bb): Update prototype.
12196 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12197 * tree-dump.c: Do not include tree-cfg.
12199 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12201 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12203 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12205 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12208 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12210 * tree-complex.c (expand_complex_div_wide): update profile.
12212 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12213 Alan Hayward <alan.hayward@arm.com>
12214 David Sherwood <david.sherwood@arm.com>
12216 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12217 (CORETYPES_H): New define.
12218 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12219 (insn-modes-inline.h, s-modes-inline-h): New rules.
12220 (generated_files): Add insn-modes-inline.h.
12221 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12222 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12223 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12224 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12225 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12226 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12227 (build/gencodes.o, build/genconditions.o): Likewise.
12228 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12229 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12230 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12231 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12232 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12233 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12234 * coretypes.h: Include everything up to real.h for generators.
12235 Include insn-modes.h first. Include wide-int-print.h after
12236 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12237 * machmode.h: Don't include insn-modes.h here.
12238 * function-tests.c: Remove includes of signop.h, machmode.h,
12239 double-int.h and wide-int.h.
12241 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12243 * optc-save-gen.awk: Likewise.
12244 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12245 * godump.c: Remove include of wide-int-print.h.
12246 * pretty-print.h: Likewise.
12247 * wide-int-print.cc: Likewise.
12248 * wide-int.cc: Likewise.
12249 * hash-map-tests.c: Remove include of signop.h.
12250 * hash-set-tests.c: Likewise.
12251 * rtl-tests.c: Likewise.
12252 * mkconfig.sh: Remove include of machmode.h.
12253 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12255 (emit_insn_modes_inline_h): ...this new function. Emit the code
12256 into an insn-modes-inline.h header file, adding appropriate
12257 include guards and end comments.
12258 (emit_insn_modes_c_header): Remove include of machmode.h.
12259 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12261 (main): Handle -i flag and call emit_insn_modes_inline_h when
12264 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12266 * tree-ssa-strlen.c (strinfo): Rename the length field to
12267 nonzero_chars. Add a full_string_p field.
12268 (compare_nonzero_chars, zero_length_string_p): New functions.
12269 (get_addr_stridx): Add an offset_out parameter.
12270 Use compare_nonzero_chars.
12271 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12272 (new_strinfo): Update after above changes to strinfo.
12273 (set_endptr_and_length): Set full_string_p.
12274 (get_string_length): Update after above changes to strinfo.
12275 (unshare_strinfo): Update call to new_strinfo.
12276 (maybe_invalidate): Likewise.
12277 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12278 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12279 as a uhwi instead of an shwi. Update after above changes to
12280 strinfo and new_strinfo.
12281 (zero_length_string): Assert that chainsi contains full strings.
12282 Use zero_length_string_p. Update call to new_strinfo.
12283 (adjust_related_strinfos): Update after above changes to strinfo.
12284 Copy full_string_p from origsi.
12285 (adjust_last_stmt): Use zero_length_string_p.
12286 (handle_builtin_strlen): Update after above changes to strinfo and
12287 new_strinfo. Install the lhs as the string length if the previous
12288 entry didn't describe a full string.
12289 (handle_builtin_strchr): Update after above changes to strinfo
12291 (handle_builtin_strcpy): Likewise.
12292 (handle_builtin_strcat): Likewise.
12293 (handle_builtin_malloc): Likewise.
12294 (handle_pointer_plus): Likewise.
12295 (handle_builtin_memcpy): Likewise. Track nonzero characters
12296 that aren't necessarily followed by a nul terminator.
12297 (handle_char_store): Likewise.
12299 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12301 PR tree-optimization/80769
12302 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12303 for malloc and calloc. Document the new invariant that all related
12304 strinfos have delayed lengths or none do.
12305 (verify_related_strinfos): Move earlier in file.
12306 (set_endptr_and_length): New function, split out from...
12307 (get_string_length): ...here. Also set the lengths of related
12309 (zero_length_string): Assert that chainsi has known (rather than
12311 (adjust_related_strinfos): Likewise.
12313 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12315 PR tree-optimization/81136
12316 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12317 assert that two references with the same misalignment have the same
12318 compile-time misalignment if those compile-time misalignments
12321 2017-07-01 Andi Kleen <ak@linux.intel.com>
12323 * print-tree.c (print_node): Print all attributes.
12325 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12327 * cfg.c (scale_bbs_frequencies): New function.
12328 * cfg.h (scale_bbs_frequencies): Declare it.
12329 * cfgloopanal.c (single_likely_exit): Cleanup.
12330 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12332 (scale_loop_profile): Likewise.
12333 (loop_version): Likewise.
12334 (create_empty_loop_on_edge): Update.
12335 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12336 scale_loop_frequencies, scale_loop_profile, loopify,
12337 loop_version): Update prototypes.
12338 * modulo-sched.c (sms_schedule): Update.
12339 * predict.c (unlikely_executed_edge_p): Also check probability.
12340 (probably_never_executed_edge_p): Fix typo.
12341 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12342 * tree-parloops.c (gen_parallel_loop): Update.
12343 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12344 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12345 * tree-ssa-loop-split.c (split_loop): Update.
12346 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12347 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12348 (vect_loop_versioning): Update.
12349 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12351 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12353 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12355 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12357 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12358 to keep profile consistent.
12360 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12362 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12363 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12364 * profile-count.h (max_safe_multiplier): Make unsigned.
12365 (profile_count::guessed_zero): New.
12367 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12369 * bb-reorder.c (fix_up_crossing_landing_pad,
12370 fix_crossing_conditional_branches): Use make_single_succ_edge
12371 to keep profile consistent.
12373 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12375 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12378 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12381 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12382 the right scopes, make sure cond_jump isn't preserved between multiple
12383 iterations. Search for fallthru edge whenever there are 3+ edges and
12384 use find_fallthru_edge for it.
12386 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12388 Patch by Alexander Monakov <amonakov@ispras.ru>
12389 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12390 probabilities consistently.
12392 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12394 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12395 emit_cmp_and_jump_insns.
12397 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12400 * tree-inline.c (expand_call_inline): Combine profile statuses.
12402 2017-06-30 Andrew Pinski <apinski@cavium.com>
12404 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12405 fold_stmt returned true.
12407 2017-06-30 Nathan Sidwell <nathan@acm.org>
12409 * ggc.h (empty_string): Delete.
12410 * cfgexpand.c (expand_asm_stmt): Use plain "".
12411 * optabs.c (expand_asm_memory_barrier): Likewise.
12412 * stringpool.c (empty_string): Delete.
12413 (digit_vector, digit_string): Delete.
12414 (ggc_alloc_string): Use plain "", don't optimize single digit
12415 strings. Use ggc_alloc_atomic.
12417 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12419 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12420 comparison set and one other set, use the cost of the non-comparison
12423 2017-06-30 Nathan Sidwell <nathan@acm.org>
12425 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12428 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12430 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12431 loops. Remove now unneeded calls to gimple_switch_set_label() that
12432 just set removed labels to NULL_TREE.
12434 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12436 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12437 (set_range_info): ...here. Only call set_range_info_raw if domain
12439 (set_nonzero_bits): Call set_range_info_raw.
12440 * tree-ssanames.h (set_range_info_raw): New.
12442 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12445 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12446 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12447 of nonimmediate_operand and <store_mask_constraint> instead of m
12448 for the input operand. For V8FI iterator, always split if input
12449 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12450 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12451 <store_mask_predicate> instead of register_operand and
12452 <store_mask_constraint> instead of v for the input operand. Make
12453 sure both operands aren't MEMs for if not <mask_applied>.
12455 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12457 * lto-wrapper.c (copy_file) Close both file descriptors before
12460 2017-06-30 Martin Liska <mliska@suse.cz>
12463 * multiple_target.c (create_dispatcher_calls): Make ifunc
12464 also for function that don't have calls or are not referenced.
12466 2017-06-30 Richard Biener <rguenther@suse.de>
12468 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12469 analyze the first scalar stmt. Move vector type computation
12470 for the BB case here from ...
12471 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12472 live operation processing in the SLP case properly.
12474 2017-06-30 Richard Biener <rguenther@suse.de>
12476 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12478 2017-06-30 Martin Liska <mliska@suse.cz>
12481 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12482 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12484 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12486 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12489 2017-06-30 Richard Biener <rguenther@suse.de>
12491 PR tree-optimization/81249
12492 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12493 condition reduction result to original scalar type.
12495 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12497 * profile-count.h (enum profile_quality): Fix typos and whitespace
12500 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12502 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12503 type for branch probabilities.
12505 2017-06-29 Julian Brown <julian@codesourcery.com>
12506 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12508 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12509 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12510 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12511 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12513 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12515 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12516 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12517 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12518 CC usage from generic code to here.
12519 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12520 CC usage into the target macros.
12522 2017-06-29 Maya Rashish <coypu@sdf.org>
12524 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12527 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12529 * arm/arm-builtins.c: Include profile-count.h
12530 * except.c (sjlj_emit_function_enter): Use
12531 profile_probability::unlikely.
12533 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12535 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12536 and tocrel_offset be pointer args rather than implicitly using
12538 (legitimate_constant_pool_address_p, rs6000_emit_move,
12539 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12540 tocrel_offset and use in toc_relative_expr_p call.
12541 (print_operand, print_operand_address): Use static tocrel_base_oac
12542 and tocrel_offset_oac.
12543 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12546 2017-06-29 Maya Rashish <coypu@sdf.org>
12548 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12550 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12552 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12553 objects, take into account only the alignment of 'op0' and 'mode1' if
12556 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12558 * ccmp.c (ccmp_tree_comparison_p): New function.
12559 (ccmp_candidate_p): Update to use above function.
12560 (get_compare_parts): New function.
12561 (expand_ccmp_next): Update to use new functions.
12562 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12564 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12565 take mode as argument.
12566 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12567 * expr.c (expand_expr_real_1): Pass mode as argument.
12569 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12571 * combine.c (combine_instructions): Print insns to dump_file, together
12574 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12576 * asan.c (asan_emit_stack_protection): Update.
12577 (create_cond_insert_point): Update.
12578 * auto-profile.c (afdo_propagate_circuit): Update.
12579 * basic-block.h (struct edge_def): Turn probability to
12580 profile_probability.
12581 (EDGE_FREQUENCY): Update.
12582 * bb-reorder.c (find_traces_1_round): Update.
12583 (better_edge_p): Update.
12584 (sanitize_hot_paths): Update.
12585 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12586 (make_single_succ_edge): Update.
12587 (check_bb_profile): Update.
12588 (dump_edge_info): Update.
12589 (update_bb_profile_for_threading): Update.
12590 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12591 probabilitycount to 0.
12592 * cfgbuild.c (compute_outgoing_frequencies): Update.
12593 * cfgcleanup.c (try_forward_edges): Update.
12594 (outgoing_edges_match): Update.
12595 (try_crossjump_to_edge): Update.
12596 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12597 (expand_gimple_tailcall): Update.
12598 (construct_init_block): Use make_single_succ_edge.
12599 (construct_exit_block): Use make_single_succ_edge.
12600 * cfghooks.c (verify_flow_info): Update.
12601 (redirect_edge_succ_nodup): Update.
12602 (split_edge): Update.
12603 (account_profile_record): Update.
12604 * cfgloopanal.c (single_likely_exit): Update.
12605 * cfgloopmanip.c (scale_loop_profile): Update.
12606 (set_zero_probability): Remove.
12607 (duplicate_loop_to_header_edge): Update.
12608 * cfgloopmanip.h (loop_version): Update prototype.
12609 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12610 (force_nonfallthru_and_redirect): Update.
12611 (update_br_prob_note): Update.
12612 (rtl_verify_edges): Update.
12613 (purge_dead_edges): Update.
12614 (rtl_lv_add_condition_to_bb): Update.
12615 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12616 * cgraphunit.c (init_lowered_empty_function): Update.
12617 (cgraph_node::expand_thunk): Update.
12618 * cilk-common.c: Include profile-count.h
12619 * dojump.c (inv): Remove.
12620 (jumpifnot): Update.
12621 (jumpifnot_1): Update.
12622 (do_jump_1): Update.
12624 (do_jump_by_parts_greater_rtx): Update.
12625 (do_compare_rtx_and_jump): Update.
12626 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12627 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12628 * dwarf2cfi.c: Include profile-count.h
12629 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12630 (sjlj_emit_dispatch_table): Likewise.
12631 * explow.c: Include profile-count.h
12632 * expmed.c (emit_store_flag_force): Update.
12633 (do_cmp_and_jump): Update.
12634 * expr.c (compare_by_pieces_d::generate): Update.
12635 (compare_by_pieces_d::finish_mode): Update.
12636 (emit_block_move_via_loop): Update.
12637 (store_expr_with_bounds): Update.
12638 (store_constructor): Update.
12639 (expand_expr_real_2): Update.
12640 (expand_expr_real_1): Update.
12641 * expr.h (try_casesi, try_tablejump): Update prototypes.
12642 * gimple-pretty-print.c (dump_probability): Update.
12643 (dump_profile): New.
12644 (dump_gimple_label): Update.
12645 (dump_gimple_bb_header): Update.
12646 * graph.c (draw_cfg_node_succ_edges): Update.
12647 * hsa-gen.c (convert_switch_statements): Update.
12648 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12649 (find_if_case_1): Update.
12650 (find_if_case_2): Update.
12651 * internal-fn.c (expand_arith_overflow_result_store): Update.
12652 (expand_addsub_overflow): Update.
12653 (expand_neg_overflow): Update.
12654 (expand_mul_overflow): Update.
12655 (expand_vector_ubsan_overflow): Update.
12656 * ipa-cp.c (good_cloning_opportunity_p): Update.
12657 * ipa-split.c (split_function): Use make_single_succ_edge.
12658 * ipa-utils.c (ipa_merge_profiles): Update.
12659 * loop-doloop.c (add_test): Update.
12660 (doloop_modify): Update.
12661 * loop-unroll.c (compare_and_jump_seq): Update.
12662 (unroll_loop_runtime_iterations): Update.
12663 * lra-constraints.c (lra_inheritance): Update.
12664 * lto-streamer-in.c (input_cfg): Update.
12665 * lto-streamer-out.c (output_cfg): Update.
12666 * mcf.c (adjust_cfg_counts): Update.
12667 * modulo-sched.c (sms_schedule): Update.
12668 * omp-expand.c (expand_omp_for_init_counts): Update.
12669 (extract_omp_for_update_vars): Update.
12670 (expand_omp_ordered_sink): Update.
12671 (expand_omp_for_ordered_loops): Update.
12672 (expand_omp_for_generic): Update.
12673 (expand_omp_for_static_nochunk): Update.
12674 (expand_omp_for_static_chunk): Update.
12675 (expand_cilk_for): Update.
12676 (expand_omp_simd): Update.
12677 (expand_omp_taskloop_for_outer): Update.
12678 (expand_omp_taskloop_for_inner): Update.
12679 * omp-simd-clone.c (simd_clone_adjust): Update.
12680 * optabs.c (expand_doubleword_shift): Update.
12681 (expand_abs): Update.
12682 (emit_cmp_and_jump_insn_1): Update.
12683 (expand_compare_and_swap_loop): Update.
12684 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12685 * predict.c (predictable_edge_p): Update.
12686 (edge_probability_reliable_p): Update.
12687 (set_even_probabilities): Update.
12688 (combine_predictions_for_insn): Update.
12689 (combine_predictions_for_bb): Update.
12690 (propagate_freq): Update.
12691 (estimate_bb_frequencies): Update.
12692 (force_edge_cold): Update.
12693 * profile-count.c (profile_count::dump): Add missing space into dump.
12694 (profile_count::debug): Add newline.
12695 (profile_count::differs_from_p): Explicitly convert to unsigned.
12696 (profile_count::stream_in): Update.
12697 (profile_probability::dump): New member function.
12698 (profile_probability::debug): New member function.
12699 (profile_probability::differs_from_p): New member function.
12700 (profile_probability::differs_lot_from_p): New member function.
12701 (profile_probability::stream_in): New member function.
12702 (profile_probability::stream_out): New member function.
12703 * profile-count.h (profile_count_quality): Rename to ...
12704 (profile_quality): ... this one.
12705 (profile_probability): New.
12706 (profile_count): Update.
12707 * profile.c (compute_branch_probabilities): Update.
12708 * recog.c (peep2_attempt): Update.
12709 * sched-ebb.c (schedule_ebbs): Update.
12710 * sched-rgn.c (find_single_block_region): Update.
12711 (compute_dom_prob_ps): Update.
12712 (schedule_region): Update.
12713 * sel-sched-ir.c (compute_succs_info): Update.
12714 * stmt.c (struct case_node): Update.
12715 (do_jump_if_equal): Update.
12716 (get_outgoing_edge_probs): Update.
12717 (conditional_probability): Update.
12718 (emit_case_dispatch_table): Update.
12719 (expand_case): Update.
12720 (expand_sjlj_dispatch_table): Update.
12721 (emit_case_nodes): Update.
12722 * targhooks.c: Update.
12723 * tracer.c (better_p): Update.
12724 (find_best_successor): Update.
12725 * trans-mem.c (expand_transaction): Update.
12726 * tree-call-cdce.c: Update.
12727 * tree-cfg.c (gimple_split_edge): Upate.
12728 (move_sese_region_to_fn): Upate.
12729 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12730 * tree-eh.c (lower_resx): Upate.
12731 (cleanup_empty_eh_move_lp): Upate.
12732 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12733 * tree-inline.c (copy_edges_for_bb): Update.
12734 (copy_cfg_body): Update.
12735 * tree-parloops.c (gen_parallel_loop): Update.
12736 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12737 (gimple_gen_time_profiler): Update.
12738 * tree-ssa-dce.c (remove_dead_stmt): Update.
12739 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12740 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12741 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12742 (unloop_loops): Update.
12743 (try_peel_loop): Update.
12744 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12745 * tree-ssa-loop-split.c (connect_loops): Update.
12746 (split_loop): Update.
12747 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12748 (hoist_guard): Update.
12749 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12750 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12751 (value_replacement): Update.
12752 * tree-ssa-reassoc.c (branch_fixup): Update.
12753 * tree-ssa-tail-merge.c (replace_block_by): Update.
12754 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12755 (create_edge_and_update_destination_phis): Update.
12756 (compute_path_counts): Update.
12757 (recompute_probabilities): Update.
12758 (update_joiner_offpath_counts): Update.
12759 (freqs_to_counts_path): Update.
12760 (duplicate_thread_path): Update.
12761 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12762 (struct switch_conv_info): Update.
12763 (gen_inbound_check): Update.
12764 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12765 (vect_do_peeling): Update.
12766 (vect_loop_versioning): Update.
12767 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12768 (optimize_mask_stores): Update.
12769 * ubsan.c (ubsan_expand_null_ifn): Update.
12770 * value-prof.c (gimple_divmod_fixed_value): Update.
12771 (gimple_divmod_fixed_value_transform): Update.
12772 (gimple_mod_pow2): Update.
12773 (gimple_mod_pow2_value_transform): Update.
12774 (gimple_mod_subtract): Update.
12775 (gimple_mod_subtract_transform): Update.
12776 (gimple_ic): Update.
12777 (gimple_stringop_fixed_value): Update.
12778 (gimple_stringops_transform): Update.
12779 * value-prof.h: Update.
12781 2017-06-29 Carl Love <cel@us.ibm.com>
12783 * config/rs6000/rs6000-c.c: Add support for built-in functions
12784 vector signed int vec_signed (vector float);
12785 vector signed long long vec_signed (vector double);
12786 vector signed int vec_signed2 (vector double, vector double);
12787 vector signed int vec_signede (vector double);
12788 vector signed int vec_signedo (vector double);
12789 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
12790 instruction generator.
12791 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12792 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12793 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12795 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12796 vunsignede_v2df): Add define_expands.
12797 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12798 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12799 VEC_UNSIGNEDO): Add definitions.
12800 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12801 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12802 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12803 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12804 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12805 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12806 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12807 vec_unsignede, vec_unsignedo): Add builtin defines.
12808 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12810 * doc/extend.texi: Update the built-in documentation file for the
12811 new built-in functions.
12813 2017-06-29 Richard Biener <rguenther@suse.de>
12815 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12816 reduction chains to LOOP_VINFO_REDUCTIONS.
12817 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12818 SLP reductions after processing reduction chains.
12820 2017-06-29 Nathan Sidwell <nathan@acm.org>
12822 * builtins.c (fold_builtin_FUNCTION): Use
12823 lang_hooks.decl_printable_name.
12825 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12827 PR middle-end/81194
12828 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12829 with only one label.
12830 * stmt.c (expand_case): Assert NCASES is greater than one.
12832 2017-06-29 Richard Biener <rguenther@suse.de>
12834 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12836 (group_case_labels): Likewise.
12837 (find_taken_edge): Push sanity checking on val to workers...
12838 (find_taken_edge_cond_expr): ... here
12839 (find_taken_edge_switch_expr): ... and here, handle cases
12840 with just a default label.
12841 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12842 (group_case_labels): Likewise.
12843 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12844 group_case_labels does anything cleanup the CFG again.
12846 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12848 PR tree-optimization/81196
12849 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12850 exit condition comparing two IVs.
12852 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12854 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12855 profile to the dummy entry at the end of the list of architectures.
12856 * config/arm/arm-cpu-cdata.h: Regenerated.
12858 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12859 Michael Collison <michael.collison@arm.com>
12862 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12864 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12865 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12866 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12867 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12868 of shift when the shift amount is masked with constant equal to
12869 the size of the mode.
12870 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12873 2017-06-29 Martin Liska <mliska@suse.cz>
12875 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12877 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12880 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12883 (edge_growth_cache_entry::edge_growth_cache_entry): New
12885 (reset_edge_growth_cache): Update to use constructor.
12887 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12889 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12890 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12891 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12893 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12895 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12896 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12898 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12900 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12901 (*-linux-uclibc*): Add t-uclibc tmake_file.
12902 * config/t-musl: New.
12903 * config/t-uclibc: New.
12905 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12907 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12909 (gen_comm_data): Emit architectural setting of arch_prof.
12910 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12912 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12913 (armv8-m.base, armv8-m.main): Likewise.
12914 * arm-protos.h (arm_build_target): Add profile field.
12915 (arch_option): Likewise.
12916 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12918 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12919 arm_active_target.profile.
12921 2017-06-28 Richard Biener <rguenther@suse.de>
12923 PR middle-end/81227
12924 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12925 TYPE_OVERFLOW_WRAPS.
12926 * match.pd (negate_expr_p): Likewise.
12927 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12928 fold_build2, not fold_binary.
12930 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12932 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12933 Convert memory address to Pmode.
12934 (aarch64_print_operand): Assert MEM operands are always Pmode.
12936 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12939 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12940 Remove redundant if.
12941 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12942 * config/arm/aarch-common-protos.h
12943 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12944 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12946 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12949 * multiple_target.c (create_dispatcher_calls): Set the default
12950 clone to be static, not public.
12952 2017-06-28 Richard Biener <rguenther@suse.de>
12954 * tree-vect-loop.c (vectorizable_reduction): Move special
12955 cond reduction IV var creation ...
12956 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12957 parameter. Use STMT_VINFO_VECTYPE.
12958 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12961 2017-06-28 Martin Liska <mliska@suse.cz>
12964 * ipa-visibility.c (non_local_p): Handle visibility.
12966 2017-06-28 Martin Liska <mliska@suse.cz>
12969 * common.opt: Add IntegerRange to various options.
12970 * opt-functions.awk (integer_range_info): New function.
12971 * optc-gen.awk: Add integer_range_info to cl_options struct.
12972 * opts-common.c (decode_cmdline_option): Handle
12973 CL_ERR_INT_RANGE_ARG.
12974 (cmdline_handle_error): Likewise.
12975 * opts.c (print_filtered_help): Show valid interval in
12976 when --help is provided.
12977 * opts.h (struct cl_option): Add range_min and range_max fields.
12978 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12980 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12982 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12983 (x * C EQ/NE y * C): New transformation.
12985 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12987 * genmultilib (combination_space): Accept '+' in option names.
12989 2017-06-28 Martin Liska <mliska@suse.cz>
12992 * asan.c (instrument_derefs): Bail out inner references
12993 that are hard register variables.
12995 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12998 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12999 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13001 2017-06-28 Richard Biener <rguenther@suse.de>
13003 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13004 (vect_get_slp_defs): Adjust.
13005 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13007 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13009 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13010 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13011 (vectorizable_reduction): Adjust.
13012 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13014 (vect_get_slp_defs): Likewise.
13015 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13016 (vectorizable_bswap): Adjust.
13017 (vectorizable_call): Likewise.
13018 (vectorizable_conversion): Likewise.
13019 (vectorizable_assignment): Likewise.
13020 (vectorizable_shift): Likewise.
13021 (vectorizable_operation): Likewise.
13022 (vectorizable_store): Likewise.
13023 (vectorizable_condition): Likewise.
13024 (vectorizable_comparison): Likewise.
13026 2017-06-28 Michael Collison <michael.collison@arm.com>
13029 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13031 (arm_gen_movmemqi): Removed unused variable 'i'.
13032 Convert 'for' loop into 'while' loop.
13033 (arm_expand_prologue): Remove last unnecessary set of insn.
13034 (thumb_pop): Remove unused variable 'pushed_words'.
13035 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13037 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13039 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13040 * config/s390/s390-protos.h: Add prototype of
13041 s390_rel_address_ok_p.
13042 * config/s390/s390.c (s390_got_symbol): New function.
13043 (s390_rel_address_ok_p): New function.
13044 (legitimize_pic_address): Use s390_rel_address_ok_p.
13045 (s390_load_got): Use s390_got_symbol.
13046 (s390_option_override): Issue error if
13047 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13048 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13050 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13052 2017-06-27 Andrew Pinski <apinski@cavium.com>
13054 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13055 (X * copysign (1.0, X)): New pattern.
13056 (X * copysign (1.0, -X)): New pattern.
13057 (copysign (-1.0, CST)): New pattern.
13059 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13061 * genmultilib (combination_space): Remove variable.
13062 Validate reuse rules against regular expression for any sequence
13063 of multilib options in any order.
13065 2017-06-27 Michael Collison <michael.collison@arm.com>
13067 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13068 call aarch64_split_simd_combine.
13069 * (aarch64_combine_internal<mode>): Delete pattern.
13070 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13071 Allow register and subreg operands.
13073 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13075 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13076 specific need, just fallback on defaults.
13077 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13079 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13080 Olivier Hainque <hainque@adacore.com>
13082 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13084 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13085 targets. Pick a default if no particular attempt applied.
13086 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13089 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13091 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13092 (x86_64-wrs-vxworks7): Likewise.
13094 2017-06-27 Marek Polacek <polacek@redhat.com>
13097 * ubsan.c (instrument_null): Check get_base_address's result for null.
13099 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13101 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13103 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13105 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13106 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13107 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13108 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13109 New function types.
13110 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13111 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13112 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13113 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13114 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13115 BUILT_IN_FEUPDATEENV): New builtins.
13116 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13117 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13118 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13119 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13121 (builtin_structptr_types): Adjust size.
13122 * tree.c (builtin_structptr_types): Add four entries.
13124 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13125 Olivier Hainque <hainque@adacore.com>
13127 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13128 (TLS_SYM): New local macro, forcing reference to __tls__ on
13129 link command lines for VxWorks 7 RTPs, triggering initialization
13131 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13132 OS features TLS support, true for RTPs on VxWorks 7.
13133 * config/vxworks.c (vxworks_override_options): Setup emutls
13136 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13138 * predict.c (test_prediction_value_range): Use -1U instead of -1
13139 to avoid narrowing conversion warning.
13140 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13141 to avoid narrowing conversion warning.
13142 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13144 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13146 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13148 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13149 64bit configurations.
13150 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13151 (SIZE_TYPE): Likewise.
13152 * config/vxworks.c (vxworks_emutls_var_fields): Use
13153 long_unsigned_type_node instead of unsigned_type_node as the offset
13154 field type, which is "pointer" mode in emutls.c.
13156 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13159 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13161 PR middle-end/81207
13162 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13163 gimple_vuse copying separately from gimple_vdef copying.
13165 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13167 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13170 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13171 Olivier Hainque <hainque@adacore.com>
13173 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13174 for all vxworks7 targets.
13175 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13176 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13177 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13178 variations for VX6/VX7 and 32/64bits later on in ...
13179 (VXWORKS_LIB_SPEC): Leverage new macros.
13180 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13181 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13183 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13185 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13186 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13188 2017-06-26 Carl Love <cel@us.ibm.com>
13190 * config/rs6000/rs6000-c.c: Add support for built-in functions
13191 vector bool char vec_reve (vector bool char);
13192 vector signed char vec_reve (vector signed char);
13193 vector unsigned char vec_reve (vector unsigned char);
13194 vector bool int vec_reve (vector bool int);
13195 vector signed int vec_reve (vector signed int);
13196 vector unsigned int vec_reve (vector unsigned int);
13197 vector bool long long vec_reve (vector bool long long);
13198 vector signed long long vec_reve (vector signed long long);
13199 vector unsigned long long vec_reve (vector unsigned long long);
13200 vector bool short vec_reve (vector bool short);
13201 vector signed short vec_reve (vector signed short);
13202 vector double vec_reve (vector double);
13203 vector float vec_reve (vector float);
13204 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13205 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13206 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13207 (altivec_vreve): New pattern.
13208 * config/rs6000/altivec.h (vec_reve): New define.
13209 * doc/extend.texi (vec_rev): Update the built-in documentation file
13210 for the new built-in functions.
13212 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13214 PR tree-optimization/71815
13215 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13217 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13219 (slsr_process_phi): Likewise.
13220 (replace_uncond_cands_and_profitable_phis): Don't replace a
13221 multiply candidate with a stride of 1 (copy or cast).
13222 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13224 (lowest_cost_path): Likewise.
13225 (total_savings): Likewise.
13227 2017-06-26 Richard Biener <rguenther@suse.de>
13230 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13231 Use def_builtin_pure for all gather builtins.
13233 2017-06-26 Richard Biener <rguenther@suse.de>
13235 PR tree-optimization/81203
13236 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13237 non-dominating BBs.
13239 2017-06-26 Marek Polacek <polacek@redhat.com>
13242 * doc/invoke.texi: Document -Wmultistatement-macros.
13244 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13246 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13247 arm_neon_ok_no_float_abi effective target.
13249 2017-06-26 Richard Biener <rguenther@suse.de>
13251 PR tree-optimization/80928
13252 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13253 (copy_bbs): Set BB_DUPLICATED flag early.
13254 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13256 (execute_on_shrinking_pred): Likewise.
13257 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13258 BB_DUPLICATED blocks.
13259 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13260 iterate over all PHIs considering removal of *gsi.
13262 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13264 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13267 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13269 * config/rs6000/rs6000-string.c: (expand_block_clear,
13270 do_load_for_compare, select_block_compare_mode,
13271 compute_current_alignment, expand_block_compare,
13272 expand_strncmp_align_check, expand_strn_compare,
13273 expand_block_move, rs6000_output_load_multiple)
13274 Move functions related to string/block move/compare
13275 to a separate file.
13276 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13277 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13278 for this function which is now used in two files.
13279 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13280 * config.gcc: Add rs6000-string.o to extra_objs for
13281 targets powerpc*-*-* and rs6000*-*-*.
13283 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13286 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13287 32-bit, since indexed is not valid for DImode.
13288 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13289 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13290 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13291 (define_peephole2 for Altivec d-form store): Likewise.
13294 * multiple_target.c (create_dispatcher_calls): Only create the
13295 dispatcher call if the function is the default clone of a
13296 versioned function.
13298 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13300 PR middle-end/80902
13301 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13302 a call, force the call to not be a tail call.
13304 2017-06-23 Jeff Law <law@redhat.com>
13306 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13309 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13311 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13312 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13313 handling for early expansion of vector shifts (sl,sr,sra,rl).
13314 (builtin_function_type): Add vector shift right instructions
13315 to the unsigned argument list.
13317 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13319 rtl-optimizatoin/79286
13320 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13321 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13322 trap. PIC register plus a const unspec without offset can never trap.
13324 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13326 * tree.h (builtin_structptr_type): New type.
13327 (builtin_structptr_types): Declare new array.
13328 * tree.c (builtin_structptr_types): New array.
13329 (free_lang_data, build_common_tree_nodes): Use it.
13331 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13334 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13337 2017-06-22 Matt Turner <mattst88@gmail.com>
13339 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13340 Lake models to skylake case. Assume skylake for unknown
13341 models with clflushopt.
13343 2017-06-22 Jeff Law <law@redhat.com>
13345 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13346 frame sizes that do not satisfy aarch64_uimm12_shift.
13348 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13350 * profile-count.h (apply_probability,
13351 apply_scale, probability_in): Fix checks for zero.
13353 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13355 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13356 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13358 2016-06-22 Richard Biener <rguenther@suse.de>
13360 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13361 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13362 REDUC_MAX_EXPR support.
13363 (vectorizable_reduction): Likewise.
13364 (vect_create_epilog_for_reduction): Likewise.
13366 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13368 * match.pd (A / (1 << B) -> A >> B): New.
13369 * generic-match-head.c: Include optabs-tree.h.
13370 * gimple-match-head.c: Likewise.
13371 * optabs-tree.h (target_supports_op_p): New.
13372 * optabs-tree.c (target_supports_op_p): New.
13374 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13376 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13377 $gcc_cv_ld --help output.
13378 (gcc_cv_ld_demangle): Likewise.
13379 (gcc_cv_ld_eh_frame_hdr): Likewise.
13380 (gcc_cv_ld_pie): Likewise.
13381 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13382 (gcc_cv_ld_buildid): Likewise.
13383 (gcc_cv_ld_sysroot): Likewise.
13384 (ld_bndplt_support): Likewise.
13385 (ld_pushpopstate_support): Likewise.
13386 * configure: Regenerate.
13387 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13389 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13392 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13393 operands indexes to avoid gap between operands and match_dups.
13395 2017-06-21 Andrew Pinski <apinski@cavium.com>
13397 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13398 Increment Arith_shift and Arith_shift_reg by 1.
13399 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13401 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13402 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13403 (aarch64_strip_extend): Add new argument and test for it.
13404 (aarch64_cheap_mult_shift_p): New function.
13405 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13406 add a cost if it is true.
13407 Update calls to aarch64_strip_extend.
13408 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13410 2017-06-21 Andrew Pinski <apinski@cavium.com>
13412 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13414 (thunderxt88): Likewise.
13415 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13416 (thunderx_prefetch_tune): New variable.
13417 (thunderx2t99_prefetch_tune): Update for the correct values.
13418 (thunderxt88_tunings): New variable.
13419 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13420 generic_prefetch_tune.
13421 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13423 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13425 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13426 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13427 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13428 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13429 (aarch64_atomic_cas<mode>, GPI): Likewise.
13431 2017-06-21 Martin Liska <mliska@suse.cz>
13433 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13434 statements on cold and hot labels.
13435 * predict.c (tree_estimate_probability_bb): Remove the
13436 prediction from this place.
13438 2017-06-21 Martin Liska <mliska@suse.cz>
13440 PR tree-optimization/79489
13441 * gimplify.c (maybe_add_early_return_predict_stmt): New
13443 (gimplify_return_expr): Call the function.
13444 * predict.c (tree_estimate_probability_bb): Remove handling
13446 * predict.def: Update comment about early return predictor.
13447 * gimple-predict.h (is_gimple_predict): New function.
13448 * predict.def: Change default value of early return to 66.
13449 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13451 * passes.def: Put pass_strip_predict_hints to the beginning of
13454 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13456 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13457 FUNCTION_DECL declarations.
13458 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13460 (dwaf2out_decl): Likewise.
13461 * godump.c (go_early_global_decl): Skip call to the real debug hook
13462 for FUNCTION_DECL declarations.
13463 * passes.c (rest_of_decl_compilation): Skip call to the
13464 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13465 -fdump-go-spec is passed.
13467 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13469 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13470 Reorder for compactness.
13471 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13472 (def_builtin_pure, def_builtin_pure2): New functions.
13473 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13475 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13477 * match.pd (nop_convert): New predicate.
13478 ((A +- CST1) +- CST2): Allow some NOP conversions.
13480 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13483 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13484 with ctors/dtors if GOVD_SHARED is set.
13486 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13488 * config/aarch64/aarch64.md (movti_aarch64):
13489 Emit mov rather than orr.
13490 (movtf_aarch64): Likewise.
13491 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13492 Emit mov rather than orr.
13494 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13496 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13497 Swap alternatives, make integer dup more expensive.
13499 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13501 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13502 Return true for non-tls symbols.
13504 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13506 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13507 (cortex-a75): Likewise.
13508 (cortex-a75.cortex-a55): Likewise.
13509 * config/aarch64/aarch64-tune.md: Regenerate.
13510 * doc/invoke.texi (-mtune): Document new values for -mtune.
13512 2017-06-21 Tom de Vries <tom@codesourcery.com>
13514 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13515 stack_size feature.
13516 (Effective-Target Keywords, Other attributes): Suggest using
13517 dg-add-options stack_size feature to get stack limit in stack_size
13518 effective target documentation.
13520 2017-06-21 Julian Brown <julian@codesourcery.com>
13521 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13523 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13524 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13525 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13527 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13528 attribute type list for neon_multiply.
13529 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13530 attribute type list for neon_multiply.
13531 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13532 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13533 attribute type list for neon_multiply.
13534 * config/arm/types.md (crypto_pmull): Add.
13535 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13536 attribute type list.
13538 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13540 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13543 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13545 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13546 to make sure not to dereference a NULL cost_classes_ptr pointer.
13548 2017-06-20 Carl Love <cel@us.ibm.com>
13550 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13551 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13552 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13553 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13554 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13555 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13556 VMULOSW): New enum "unspec" values.
13557 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13558 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13559 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13560 altivec_vmulosw): New patterns.
13561 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13562 VMULOSW): Add definitions.
13564 2017-06-20 Julia Koval <julia.koval@intel.com>
13566 * config/i386/i386.c: Fix rounding expand for new pattern.
13567 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13569 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13571 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13572 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13574 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13576 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13579 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13581 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13582 architecture, then by implementer ID.
13583 * config/aarch64/aarch64-tune.md: Regenerate.
13585 2017-06-20 Richard Biener <rguenther@suse.de>
13587 PR middle-end/81097
13588 * fold-const.c (split_tree): Fold to type before negating.
13590 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13592 * diagnostic-show-locus.c
13593 (selftest::test_fixit_deletion_affecting_newline): New function.
13594 (selftest::diagnostic_show_locus_c_tests): Call it.
13596 2017-06-20 Andreas Schwab <schwab@suse.de>
13599 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13602 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13604 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13605 __ARM_FEATURE_COPROC according to support.
13607 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13609 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13610 Rewritten to avoid overflow for > 32-bit pointers.
13613 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13614 by removing enum keyword.
13615 (ubsan_type_descriptor): Likewise. Formatting fix.
13618 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13619 splitter): Require TARGET_SSE2 in the condition.
13621 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13624 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13625 for doing vector set of SFmode on ISA 3.0.
13626 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13627 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13629 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13630 SFmode value into a V4SF variable that was extracted from another
13631 V4SF variable without converting the element to double precision
13632 and back to single precision vector format.
13633 (vsx_insert_extract_v4sf_p9_2): Likewise.
13635 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13637 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13638 in UWHI to avoid undefined overflow.
13641 * ubsan.h (enum ubsan_encode_value_phase): New.
13642 (ubsan_encode_value): Change second argument to
13643 enum ubsan_encode_value_phase with default value of
13644 UBSAN_ENCODE_VALUE_GENERIC.
13645 * ubsan.c (ubsan_encode_value): Change second argument to
13646 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13647 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13648 create_tmp_var_raw instead of create_tmp_var and use a
13650 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13651 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13652 ubsan_encode_value callers.
13655 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13656 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13657 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13659 2017-06-19 Richard Biener <rguenther@suse.de>
13661 PR middle-end/81118
13662 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13663 estimates if we changed anything.
13665 2017-06-19 Richard Biener <rguenther@suse.de>
13667 PR tree-optimization/80887
13668 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13669 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13670 simplified lookups, then reset mprts_hook.
13671 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13673 (try_to_simplify): Likewise.
13675 2017-06-19 Martin Liska <mliska@suse.cz>
13678 * gimplify.c (gimplify_switch_expr):
13679 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13681 2017-06-19 Martin Liska <mliska@suse.cz>
13683 * doc/install.texi: Document that PGO runs in 4 stages.
13685 2017-06-19 Martin Liska <mliska@suse.cz>
13688 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13689 to dispatcher function name.
13690 * multiple_target.c (replace_function_decl): New function.
13691 (create_dispatcher_calls): Redirect both edges and references.
13693 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13695 * profile-count.c (profile_count::dump): Dump quality.
13696 (profile_count::differs_from_p): Update for unsigned val.
13697 * profile-count.h (profile_count_quality): New enum.
13698 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13700 2017-06-19 Richard Biener <rguenther@suse.de>
13702 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13703 struct function as arg.
13704 (estimate_numbers_of_iterations): Export overload with loop arg.
13705 (free_numbers_of_iterations_estimates_loop): Use an overload of
13706 free_numbers_of_iterations_estimates instead.
13707 * tree-cfg.c (remove_bb): Adjust.
13708 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13709 * tree-parloops.c (gen_parallel_loop): Likewise.
13710 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13712 (tree_unroll_loops_completely): Likewise.
13713 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13714 Use an overload instead and export.
13715 (estimated_loop_iterations): Adjust.
13716 (max_loop_iterations): Likewise.
13717 (likely_max_loop_iterations): Likewise.
13718 (estimate_numbers_of_iterations): Take struct function as arg
13720 (loop_exits_before_overflow): Adjust.
13721 (free_numbers_of_iterations_estimates_loop): Use an overload.
13722 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13723 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13725 2017-06-19 Richard Biener <rguenther@suse.de>
13728 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13729 RANGE_EXPR conservatively.
13731 2017-06-16 Carl Love <cel@us.ibm.com>
13733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13734 definitions for vec_float, vec_float2, vec_floato,
13735 vec_floate built-ins.
13736 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13737 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13739 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13740 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13741 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13742 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13743 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13744 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13745 vec_floato): Add builtin defines.
13746 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13747 Update the built-in documentation file for the new built-in
13750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13752 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13753 (mthumb): Mark as the negative of -marm.
13755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13757 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13759 (ARM Options, -mtune): Document that this accepts the same
13760 extension options as -mcpu.
13761 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13765 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13766 permitted extensions.
13768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13770 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
13771 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
13772 (armv8-m.main): Add option +nodsp.
13773 * config/arm/arm-cpu-cdata.h: Regenerated.
13775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13777 * config/arm/t-fuchsia: New file.
13778 * config.gcc (arm*-*-fuchsia*): Use it.
13780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13782 * config/arm/t-symbian: Rewrite for new option infrastructure.
13784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13786 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
13787 (MULTILIB_REQUIRED): Likewise.
13789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13791 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13792 (MULTILIB_RESUE): Likewise.
13793 (MULTILIB_MATCHES): Likewise.
13794 (MULTLIB_REQUIRED): Likewise.
13796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13798 * config/arm/t-rtems: Rewrite for new option framework.
13800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13802 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13803 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13804 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13805 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13806 * config/arm/t-multilib: ... here.
13807 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13808 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13809 armv7-a and armv8*-a when A-profile libraries have not been built.
13810 * config/arm/t-rmprofile: Rewrite.
13812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13814 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13815 with a backslash. Remove the backslash after substituting unescaped
13817 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13821 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13822 the driver to pass through the default setting of -mfloat-abi.
13823 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13825 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13826 (all_feat_combs): New rule.
13827 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13829 * config/arm/t-aprofile: Rewrite.
13831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13833 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13834 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13835 fpu is not specified by the user/command-line.
13836 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13837 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13838 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13839 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13840 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13841 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13845 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13846 * config/arm/t-arm-elf: Rewritten.
13848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13850 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13851 have some floating-point instructions.
13852 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13853 (TARGET_MAYBE_HARD_FLOAT): New macro.
13854 * config/arm/arm-builtins.c (arm_init_builtins): Use
13855 TARGET_MAYBE_HARD_FLOAT.
13856 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13860 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13861 (configargs.h): Include it.
13862 (arm_print_hint_for_fpu_option): New function.
13863 (arm_parse_fpu_option): New function.
13864 (candidate_extension): New class.
13865 (arm_canon_for_multilib): New function.
13866 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13867 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13868 (ARCH_CANONICAL_SPECS): New macro.
13869 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13873 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13874 are set after handling multilib fragments. Set target_cpu_default2
13877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13879 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13881 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13882 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13884 (check_arch): Likewise.
13885 * config/arm/arm.c (arm_configure_build_target): Handle
13886 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13887 options in the default.
13889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13891 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13892 when an option is an alias of another.
13893 * config/arm/parsecpu.awk (optalias): New parser token.
13894 (gen_comm_data): Mark non-alias options as such. Emit entries
13895 for extension aliases.
13896 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13897 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13898 (armv6kz, armv6zk, armv6t2): Likewise.
13899 (armv7): Make vfpv3-d16 an alias.
13900 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13902 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13903 Sort in canonical order.
13904 (armv8-a): Sort in canonical order.
13905 (armv8.1-a, armv8.2-a): Likewise.
13906 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13908 (cortex-a9): Sort in canonical order.
13909 * config/arm/arm.c (selftests.h): Include it.
13910 (arm_test_cpu_arch_data): New function.
13911 (arm_run_self_tests): New function.
13912 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13913 (targetm): Move declaration to the end of the file.
13914 * arm-cpu-cdata.h: Regenerated.
13916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13918 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13919 call to target_mode_check describing the type of option passed.
13920 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13921 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13922 arm_parse_cpu_option_name to match parameters against list of
13924 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13925 arm_arch_core_flags data structure.
13926 * config/arm/arm-cpu_cdata.h: Regenerated.
13928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13930 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13932 (arm_print_hint_for_cpu_option): Likewise.
13933 (arm_print_hint_for_arch_option): Likewise.
13934 (arm_parse_cpu_option_name): Likewise.
13935 (arm_parse_arch_option_name): Likewise.
13936 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13937 of entries in the all_fpus list.
13938 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13939 (arm_parse_cpu_option_name): Declare.
13940 (arm_parse_arch_option_name): Declare.
13941 (arm_parse_option_features): Declare.
13942 (arm_intialize_isa): Declare.
13943 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13945 (gen_comm_data): ... here. Make definitions non-static.
13946 * config/arm/arm-cpu-data.h: Regenerated.
13947 * config/arm/arm-cpu-cdata.h: Regenerated.
13949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13951 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13952 (cpu_arch_extension): New structure.
13953 (cpu_arch_option, arch_option, cpu_option): New structures.
13954 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13955 architecture types.
13956 (gen_data): Generate new format data tables.
13957 * config/arm/arm.c (cpu_tune): New structure.
13958 (cpu_option, processors): Delete.
13959 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13960 (arm_print_hint_for_cpu_option): ... this and ...
13961 (arm_print_hint_for_arch_option): ... this.
13962 (arm_parse_arch_cpu_name): Delete. Replace with ...
13963 (arm_parse_cpu_option_name): ... this and ...
13964 (arm_parse_arch_option_name): ... this.
13965 (arm_unrecognized_feature): Change type of target parameter to
13967 (arm_parse_arch_cpu_features): Delete. Replace with ...
13968 (arm_parse_option_features): ... this.
13969 (arm_configure_build_target): Rework to use new configuration data
13971 (arm_print_tune_info): Rework for new configuration data tables.
13972 * config/arm/arm-cpu-data.h: Regenerated.
13973 * config/arm/arm-cpu.h: Regenerated.
13975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13977 * Makefile.in (OBJS): Move sbitmap.o from here ...
13978 (OBJS-libcommon): ... to here.
13980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13982 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13983 (ISA_ALL_CRYPTO): New macro.
13984 (ISA_ALL_SIMD): New macro
13985 (ISA_ALL_FP): New macro.
13986 * config/arm/arm.c (fpu_bitlist): Update initializer.
13987 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13989 (arm9e): Add fpu. Add option for nofp
13990 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13991 (arm926ej-s, arm1026ej-s): Likewise.
13992 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13993 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13994 neon-fp16, neon-vfpv4, nofp and nosimd.
13995 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13996 (cortex-a8): Add fpu. Add option for nofp.
13997 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13998 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13999 (cortex-r4f): Add fpu.
14000 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14001 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14003 (cortex-r8): Likewise.
14004 (cortex-m4): Add fpu. Add option for nofp.
14005 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14006 (cortex-a17.cortex-a7): Likewise.
14007 (cortex-a32): Add fpu. Add options for crypto and nofp.
14008 (cortex-a35, cortex-a53): Likewise.
14009 (cortex-a57): Add fpu. Add option for crypto.
14010 (cortex-a72, cortex-a73): Likewise.
14011 (exynos-m1): Likewise.
14012 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14013 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14014 (cortex-m33): Add fpu. Add option for nofp.
14015 * config/arm/arm-cpu-cdata.h: Regenerated
14016 * config/arm/arm-cpu-data.h: Regenerated.
14018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14020 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14021 (armv5te, armv5tej): Likewise.
14022 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14023 (armv7): Add options fp and vfpv3-d16.
14024 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14025 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14027 (armv7ve): Likewise.
14028 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14029 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14030 (armv8-a): Add nocrypto option.
14031 (armv8.1-a, armv8.2-a): Likewise.
14032 (armv8-m.main): add options fp, fp.dp and nofp.
14034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14036 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14038 (armv8-a+crc): Delete.
14039 (armv8.1-a): Add options simd, crypto and nofp.
14040 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14041 (armv8.2-a+fp16): Delete.
14042 (armv8-m.main): Add option dsp.
14043 (armv8-m.main+dsp): Delete.
14044 (cortex-a8): Add fpu. Add nofp option.
14045 (cortex-a9): Add fpu. Add nofp and nosimd options.
14046 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14047 link to main cpu and architecture data structures.
14048 (gen_comm_data): Only put isa attributes from the main architecture
14050 (option): New statement for architecture and CPU entries.
14051 * arm.c (struct cpu_option): New structure.
14052 (struct processors): Add entry for options.
14053 (arm_unrecognized_feature): New function.
14054 (arm_parse_arch_cpu_name): Ignore any characters after the first
14056 (arm_parse_arch_cpu_feature): New function.
14057 (arm_configure_build_target): Separate out any CPU and architecture
14058 features and parse separately. Don't error out if -mfpu=auto is
14059 used with only an architecture string.
14060 (arm_print_asm_arch_directives): New function.
14061 (arm_file_start): Call it.
14062 * config/arm/arm-cpu-cdata.h: Regenerated.
14063 * config/arm/arm-cpu-data.h: Likewise.
14064 * config/arm/arm-tables.opt: Likewise.
14066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14068 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14069 assembler when it is not -mfpu=auto.
14071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14073 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14074 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14075 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14076 (ASM_CPU_SPEC): Rewrite.
14077 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14078 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14079 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14080 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14081 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14082 copied string is NUL-terminated. Also strip any characters prefixed
14084 (arm_rewrite_selected_arch): New function.
14085 (arm_rewrite_march): New function.
14087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14089 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14090 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14091 (march, mcpu, mtune): Convert to string-based options.
14092 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14093 (arm_parse_arch_cpu_name): New function.
14094 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14095 identify selected architecture or CPU.
14096 (arm_option_save): New function.
14097 (TARGET_OPTION_SAVE): Redefine.
14098 (arm_option_restore): Restore string options.
14099 (arm_option_print): Print string options.
14101 2017-06-16 Martin Sebor <msebor@redhat.com>
14103 PR tree-optimization/80933
14104 PR tree-optimization/80934
14105 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14106 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14107 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14108 (gimple_fold_builtin): Call them.
14110 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14112 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14113 as unlikely; update profile.
14115 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14117 * predict.c (force_edge_cold): Handle declaring edges impossible
14120 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14122 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14124 (try_unroll_loop_completely): Fix reporting.
14126 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14128 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14130 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14133 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14134 if given a non-constant argument for an intrinsic which requires a
14137 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14139 * profile.c (compare_freqs): New function.
14140 (branch_prob): Sort edge list.
14141 (find_spanning_tree): Assume that the list is priority sorted.
14143 2017-06-16 Richard Biener <rguenther@suse.de>
14145 PR tree-optimization/81090
14146 * passes.def (pass_record_bounds): Remove.
14147 * tree-pass.h (make_pass_record_bounds): Likewise.
14148 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14149 make_pass_record_bounds): Likewise.
14150 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14151 not free niter estimates at the beginning but at the end.
14152 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14154 2017-06-16 Richard Biener <rguenther@suse.de>
14156 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14157 initializer to workaround ICE in host GCC 4.8.
14159 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14161 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14163 (clone_inlined_nodes): Update.
14165 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14167 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14168 prefetch settings, and enable prefetching by default at -O3.
14170 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14172 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14173 Set flag_prefetch_loop_arrays according to tuning data.
14175 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14177 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14178 New tune structure.
14179 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14180 [Unrelated to main purpose of the patch] Place the pointer field last
14181 to enable type checking errors when tune structure are wrongly merged.
14182 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14183 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14184 (thunderx2t99_prefetch_tune): New tune constants.
14185 (tune_params *_tunings): Update all tunings (no functional change).
14186 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14187 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14188 from tunings structures.
14190 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14193 * ubsan.c (instrument_null): Add T argument, use it instead
14194 of computing it based on IS_LHS.
14195 (instrument_object_size): Likewise.
14196 (pass_ubsan::execute): Adjust instrument_null and
14197 instrument_object_size callers to pass gimple_get_lhs or
14198 gimple_assign_rhs1 result to it. Use instrument_null instead of
14199 calling get_base_address and instrument_mem_ref. Handle
14200 aggregate call arguments for object-size sanitization.
14202 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14204 PR tree-optimization/81089
14205 * tree-vrp.c (is_masked_range_test): Validate operands of
14208 2017-06-15 Martin Sebor <msebor@redhat.com>
14211 * dumpfile.c (dump_register): Avoid calling memset to initialize
14212 a class with a default ctor.
14213 * gcc.c (struct compiler): Remove const qualification.
14214 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14215 * hash-table.h: Ditto.
14216 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14218 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14219 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14221 * params.h (struct param_info): Make struct members non-const.
14222 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14223 with default initialization.
14224 * vec.h (vec_copy_construct, vec_default_construct): New helper
14226 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14227 with vec_copy_construct.
14228 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14229 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14230 * doc/invoke.texi (-Wclass-memaccess): Document.
14232 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14234 * emit-rtl.h (is_leaf): Update comment about local
14235 register allocator.
14237 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14240 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14241 for a variable to have a section before checking if the section has a
14243 Set section to.persistent if persistent attribute is set.
14244 Warn if .persistent attribute is used on an automatic variable.
14246 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14248 PR rtl-optimization/80474
14249 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14251 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14253 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14256 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14258 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14259 (casesi): Emit bounds checking as RTL.
14260 (casesi_internal_mips16_<mode>): Remove bounds checking.
14262 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14264 * config/xtensa/xtensa.c (xtensa_option_override): Append
14265 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14266 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14267 xtensa_doloop_hooks): Define unconditionally.
14268 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14270 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14271 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14272 for it in xtensa_option_override.
14273 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14274 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14276 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14278 * doc/cppopts.texi: Document '-' special value to -MF.
14280 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14282 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14283 (cortex_a53_fconst): Likewise.
14284 (cortex_a53_fpmul): Likewise.
14285 (cortex_a53_f_load_64): Likewise.
14286 (cortex_a53_f_load_many): Likewise.
14287 (cortex_a53_advsimd_alu): Likewise.
14288 (cortex_a53_advsimd_alu_q): Likewise.
14289 (cortex_a53_advsimd_mul): Likewise.
14290 (cortex_a53_advsimd_mul_q): Likewise.
14291 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14292 Add missing fmul, r2f_cvt and fconst cases.
14294 2017-06-14 Richard Biener <rguenther@suse.de>
14296 PR middle-end/81088
14297 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14299 (fold_binary_loc): When associating do not treat pre-existing
14300 TREE_OVERFLOW on literal constants as a reason to allow
14301 TREE_OVERFLOW on associated literal constants.
14303 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14305 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14306 (MASK_FEATURES): New macro.
14307 * config/sparc/sparc.c (sparc_option_override): Remove the special
14308 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14310 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14312 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14313 a division of 0 if non-call exceptions are enabled.
14315 2017-06-14 Andrew Pinski <apinski@cavium.com>
14316 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14319 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14320 Improve vector initialization code gen for only variable case.
14322 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14324 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14326 2017-06-14 Richard Biener <rguenther@suse.de>
14328 PR tree-optimization/81083
14329 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14332 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14334 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14335 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14336 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14337 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14338 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14339 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14341 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14343 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14344 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14346 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14348 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14352 * config/rs6000/t-rtems: Don't handle SPE.
14354 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14356 * config/rs6000/t-linux: Don't handle SPE.
14358 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14360 * config/rs6000/eabispe.h: Delete file.
14362 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14364 * config/rs6000/t-spe: Delete file.
14366 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14368 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14369 (rs6000_legitimate_offset_address_p): Return false for anything in
14370 V2SImode or V2SFmode.
14372 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14374 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14375 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14376 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14378 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14379 (rs6000_legitimate_offset_address_p): Ditto.
14380 (rs6000_emit_move): Ditto.
14381 (rs6000_init_builtins): Remove V4HI_type_node.
14383 2017-06-13 Martin Liska <mliska@suse.cz>
14386 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14387 (gate_asan): Likewise.
14388 * asan.h (asan_no_sanitize_address_p): Remove the function.
14389 (sanitize_flags_p): New function.
14390 * builtins.def: Fix coding style.
14391 * common.opt: Use renamed enum value.
14392 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14393 * doc/extend.texi: Document no_sanitize attribute.
14394 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14395 to SANITIZE_UNDEFINED_NONDEFAULT.
14396 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14397 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14398 Use sanitize_flags_p.
14399 * gimplify.c (gimplify_function_tree): Likewise.
14400 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14401 * opts.c (parse_no_sanitize_attribute): New function.
14402 (common_handle_option): Use renamed enum value.
14403 * opts.h (parse_no_sanitize_attribute): Declare.
14404 * tree.c (sanitize_flags_p): New function.
14405 * tree.h: Declared here.
14406 * tsan.c: Use sanitize_flags_p.
14407 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14408 (instrument_mem_ref): Likewise.
14409 (instrument_bool_enum_load): Likewise.
14410 (do_ubsan_in_current_function): Remove the function.
14411 (pass_ubsan::execute): Use sanitize_flags_p.
14412 * ubsan.h: Remove do_ubsan_in_current_function
14413 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14414 (dump_function_to_file): Use it here.
14416 2017-06-13 Martin Jambor <mjambor@suse.cz>
14418 PR tree-optimization/80803
14419 PR tree-optimization/81063
14420 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14421 (propagate_subaccesses_across_link): Enqueue subtree whenever
14422 necessary instead of relying on the caller.
14424 2017-06-13 Martin Jambor <mjambor@suse.cz>
14426 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14427 that have a first_link.
14428 (sort_and_splice_var_accesses): Do not check first_link before
14430 (subtree_mark_written_and_enqueue): Likewise.
14431 (propagate_all_subaccesses): Likewise and do not stop at first
14432 parent with a first_link.
14434 2017-06-13 Martin Jambor <mjambor@suse.cz>
14436 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14439 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14441 * match.pd: New pattern.
14443 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14445 * tree-vrp.c (is_masked_range_test): New function.
14446 (register_edge_assert_for): Determine ranges for
14449 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14451 PR tree-optimization/67328
14452 * fold-const.c (maskable_range_p): New function.
14453 (build_range_check): Generate bittests if possible.
14455 2017-06-13 Martin Liska <mliska@suse.cz>
14457 * gimple-pretty-print.c (dump_probability): Add new argument.
14458 (dump_edge_probability): Dump both probability and count.
14459 (dump_gimple_label): Likewise.
14460 (dump_gimple_bb_header): Likewise.
14462 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14465 * config/avr/avr-devices.c: Fix indentation.
14466 * config/avr/gen-avr-mmcu-specs.c: Dito.
14468 2017-06-13 Richard Biener <rguenther@suse.de>
14470 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14471 instead get vector type from stmt_info.
14472 (vectorizable_reduction): Adjust. Remove dead code.
14474 2017-06-13 Richard Biener <rguenther@suse.de>
14476 PR middle-end/81065
14477 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14478 case of C * (x * C2 + C3).
14479 (fold_addr_of_array_ref_difference): Properly fold index difference.
14481 2017-06-12 David S. Miller <davem@davemloft.net>
14484 * config/sparc/sparc.md (return expander): Emit frame blockage if
14485 function uses alloca.
14487 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14489 * combine.c (make_field_assignment): Check len rather than the mode
14490 precision when calling force_to_mode.
14492 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14494 Support multilibs and devices that see flash in RAM address range.
14497 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14498 (avr_mcu_t) <flash_pm_offset>: New field.
14499 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14500 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14501 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14502 (AVR_TINY_PM_OFFSET): Remove macro.
14503 * config/avr/avr.opt (-mshort-calls): New option.
14504 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14505 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14506 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14507 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14508 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14509 instead of avr_arch->have_jmp_call.
14510 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14511 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14512 avr_arch->flash_pm_offset to define.
14513 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14514 new field flash_pm_offset. Add entry for avrxmega3.
14515 (avr_texinfo): Add entry for avrxmega3.
14516 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14517 attiny212, attiny214,
14518 attiny412, attiny414, attiny416, attiny417,
14519 attiny814, attiny816, attiny817,
14520 attiny1614, attiny1616, attiny1617,
14521 attiny3214, attiny3216, attiny3217.
14522 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14523 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14524 (avr_print_operand_address) [AVR_TINY]: Same.
14525 (avr_asm_init_sections) <readonly_data_section>: Only patch
14526 callback if avr_arch->flash_pm_offset = 0.
14527 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14528 for rodata if avr_arch->flash_pm_offset != 0.
14529 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14530 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14531 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14532 (m_options): Append opt_rcall.
14533 (m_dirnames): Append dir_rcall.
14534 * config/avr/t-multilib: Regenerate.
14536 * configure.ac [target=avr]: Check whether avrxmega3 default
14537 linker description file works as needed.
14538 * configure: Regenerate.
14539 * doc/avr-mmcu.texi: Regenerate.
14540 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14541 <__AVR_ARCH__>: Document avrxmega3 and 103.
14542 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14543 <__AVR_SHORT_CALLS__>: Document it.
14544 <__AVR_PM_BASE_ADDRESS__>: Document it.
14545 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14546 (AVR Variable Attributes) <progmem>: Document this is
14547 not needed for avrxmega3.
14548 (AVR Named Address Spaces) <__flash>: Dito.
14550 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14552 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14554 2017-06-12 Doug Rupp <rupp@adacore.com>
14556 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14557 Append vxworks-stdint.h to the tm_file list.
14558 * config/vxworks-stdint.h: New file.
14560 2017-06-12 Martin Liska <mliska@suse.cz>
14562 PR tree-optimization/81041
14563 * tree-profile.c (gimple_gen_ic_func_profiler):
14564 Create an extra BB in profile-generate
14565 (gimple_gen_time_profiler): Likewise.
14567 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14569 PR tree-optimization/81003
14570 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14571 (update_range_test): Use it instead of force_gimple_operand_gsi.
14573 2017-06-12 Richard Biener <rguenther@suse.de>
14575 PR tree-optimization/81053
14576 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14577 with backedge value not defined in loop. Simplify def stmt
14580 2017-06-11 Tom de Vries <tom@codesourcery.com>
14583 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14585 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14586 nvptx_cannot_force_const_mem.
14588 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14590 * opts.c (finish_options): Move test for flag_split_stack after
14591 it has been initialized.
14593 2017-06-11 Jason Merrill <jason@redhat.com>
14595 * tree.h (id_equal): New.
14596 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14597 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14598 instead of strcmp of IDENTIFIER_POINTER.
14600 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14602 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14603 (mark_all_inlined_calls_cdtor): Fix formating.
14604 (inline_transform): Rescale profile before inlining.
14606 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14608 * cgraph.h (cgraph_edge::clone): Update prototype.
14609 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14610 (cgraph_node::create_clone): Update.
14611 (cgraph_node::create_version_clone): Update.
14612 * tree-inline.c (copy_bb): Update.
14613 (expand_call_inline): Update.
14615 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14617 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14618 factored out from ...
14619 (rs6000_emit_prologue): ... here.
14621 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14623 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14624 factored out from ...
14625 (rs6000_emit_prologue): ... here.
14627 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14629 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14631 (handle_missing_profiles): Fix computation of tp_first_run.
14632 (counts_to_freqs): Do not touch freqs when count is 0.
14634 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14636 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14639 2017-06-10 Tom de Vries <tom@codesourcery.com>
14641 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14642 attributes): Document signal effective target.
14644 2017-06-10 Tom de Vries <tom@codesourcery.com>
14646 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14647 Document effective target stack_size.
14649 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14651 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14652 to the edit_context if they can be auto-applied.
14654 2017-06-9 Ian Lance Taylor <iant@golang.org>
14656 * opts.c (finish_options): If -fsplit-stack, disable implicit
14657 -forder-blocks-and-partition.
14658 * doc/invoke.texi (Optimize Options): Document that when using
14659 -fsplit-stack -forder-blocks-and-partition is not implicitly
14662 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14664 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14665 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14666 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14667 * builtins.def (abort, trap, unreachable): Declare cold.
14668 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14669 * tree-core.h (ECF_COLD): New.
14670 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14671 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14673 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14675 * predict.c (unlikely_executed_stmt_p): Cleanup.
14677 2017-06-09 Richard Biener <rguenther@suse.de>
14679 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14680 model if the ref is always written to.
14682 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14684 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14686 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14688 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14691 2017-06-09 Tom de Vries <tom@codesourcery.com>
14694 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14695 "target cannot support label values" when encountering LABEL_REF.
14697 2017-06-09 Martin Liska <mliska@suse.cz>
14699 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14700 (gimple_gen_ic_func_profiler): Emit direct comparison
14701 of __gcov_indirect_call_callee with NULL.
14702 (gimple_gen_time_profiler): Change probability from
14703 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14705 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14707 * profile.c (edge_gcov_counts): Turn to pointer.
14708 (compute_branch_probabilities, compute_branch_probabilities): Update.
14709 (branch_prob): Do not clear edge_gcov_count.
14710 * profile.h (edge_gcov_counts): Turn to pointer.
14711 (edge_gcov_count): Update.
14713 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14715 * gimple.h (gimple_check_failed): Mark cold.
14717 2017-06-09 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/66623
14720 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14721 refactor check_reduction into two parts, properly computing
14722 whether we have to check reduction validity for outer loop
14725 2017-06-09 Richard Biener <rguenther@suse.de>
14727 PR tree-optimization/79483
14728 * graphite-scop-detection.c (order): New global.
14729 (get_order): Compute bb to order mapping that satisfies code
14730 generation constraints.
14731 (cmp_pbbs): New helper.
14732 (build_scops): Start domwalk at entry block, sort generated
14735 2017-06-09 Richard Biener <rguenther@suse.de>
14737 PR middle-end/81007
14738 * ipa-polymorphic-call.c
14739 (ipa_polymorphic_call_context::restrict_to_inner_class):
14740 Skip FIELD_DECLs with error_mark_node type.
14741 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14744 2017-06-09 Martin Liska <mliska@suse.cz>
14746 * predict.c (struct branch_predictor): New struct.
14747 (test_prediction_value_range): New test.
14748 (predict_c_tests): New function.
14749 * selftest-run-tests.c (selftest::run_tests): Run the function.
14750 * selftest.h: Declare new tests.
14752 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14755 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14756 gen_add3_insn did not fail.
14757 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14758 r0, construct that number in a temporary reg and add that reg to r0.
14759 If asked to put the result in r0 as well, fail.
14761 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14763 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14764 for early expansion of vec_eqv.
14766 2017-06-08 Jakub Jelinek <jakub@redhat.com>
14768 PR middle-end/81005
14769 * ubsan.c (instrument_null): Avoid pointless code temporary.
14770 (pass_ubsan::execute): Instrument aggregate arguments of calls.
14772 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
14776 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
14779 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
14782 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14784 * predict.c (unlikely_executed_edge_p): Move ahead.
14785 (probably_never_executed_edge_p): Use it.
14787 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14789 PR middle-end/79988
14790 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14791 gimple_call_builtin_p call.
14793 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14795 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14796 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14797 rtl_check_failed_type2, rtl_check_failed_code1,
14798 rtl_check_failed_code2, rtl_check_failed_code_mode,
14799 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14800 rtvec_check_failed_bounds, rtl_check_failed_flag,
14801 _fatal_insn_not_found, _fatal_insn): Likewise.
14802 * tree.h (tree_contains_struct_check_failed,
14803 tree_check_failed, tree_not_check_failed,
14804 tree_class_check_failed, tree_range_check_failed,
14805 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14806 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14807 tree_operand_check_failed, omp_clause_check_failed,
14808 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14811 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14813 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14814 flag_branch_probabilities.
14815 * ipa-inline.c (edge_badness): Likewise.
14816 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14817 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14818 * predict.c (maybe_hot_frequency_p): Likewise.
14819 (probably_never_executed): Likewise.
14820 * sched-ebb.c (schedule_ebbs): Likewise.
14821 * sched-rgn.c (find_single_block_region): Likewise.
14822 * tracer.c (tail_duplicate): Likewise.
14824 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14826 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14827 longer requires x_flag_profile_use.
14829 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14831 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14832 instead of flag_reorder_blocks_and_partition.
14833 * dbxout.c (dbxout_function_end): Likewise.
14834 * dwarf2out.c (gen_subprogram_die): Likewise.
14835 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14836 * hw-doloop.c (reorg_loops): Likewise.
14837 * varasm.c (assemble_start_function,
14838 assemble_end_function): Likewise.
14839 (decide_function_section): Do not check for
14840 flag_reorder_blocks_and_partition.
14842 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14844 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14846 (chkp_get_hard_register_fake_addr_expr): Ditto.
14847 (chkp_build_addr_expr): Add check for hard reg case.
14848 (chkp_parse_array_and_component_ref): Ditto.
14849 (chkp_find_bounds_1): Ditto.
14850 (chkp_process_stmt): Don't generate bounds store for
14853 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14855 * predict.c (maybe_hot_bb_p): Do not check profile status.
14856 (maybe_hot_edge_p): Likewise.
14857 (probably_never_executed): Check for zero counts even if profile
14859 (unlikely_executed_edge_p): New function.
14860 (unlikely_executed_stmt_p): New function.
14861 (unlikely_executed_bb_p): New function.
14862 (set_even_probabilities): Use unlikely predicates.
14863 (combine_predictions_for_bb): Likewise.
14864 (predict_paths_for_bb): Likewise.
14865 (predict_paths_leading_to_edge): Likewise.
14866 (determine_unlikely_bbs): New function.
14867 (estimate_bb_frequencies): Use it.
14868 (compute_function_frequency): Use zero counts even if profile is
14870 * profile-count.h: Fix typo.
14872 2017-08-08 Julia Koval <julia.koval@intel.com>
14874 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14875 _mm512_mask_cvtsepi16_storeu_epi8,
14876 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14877 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14878 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14879 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14880 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14881 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14882 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14883 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14884 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14885 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14886 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14887 __builtin_ia32_pmovuswb256mem_mask,
14888 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14889 __builtin_ia32_pmovwb512mem_mask): New builtins.
14891 2017-08-08 Julia Koval <julia.koval@intel.com>
14893 PR target/73350,80862
14894 * config/i386/subst.md (round): Fix round pattern.
14895 * config/i386/i386.c (ix86_erase_embedded_rounding):
14896 Fix erasing rounding for the fixed pattern.
14898 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14900 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14902 2017-06-08 Martin Liska <mliska@suse.cz>
14904 PR gcov-profile/80911
14905 * gcov.c (block_info::block_info): New constructor.
14907 2017-06-07 Carl Love <cel@us.ibm.com>
14909 * config/rs6000/rs6000-c: The return type of the following
14910 built-in functions was implemented as int not long long. Fix sign
14911 of return value for the unsigned version of vec_mulo and vec_mule.
14912 vector unsigned long long vec_bperm (vector unsigned long long,
14913 vector unsigned char)
14914 vector signed long long vec_mule (vector signed int,
14916 vector unsigned long long vec_mule (vector unsigned int,
14917 vector unsigned int)
14918 vector signed long long vec_mulo (vector signed int,
14920 vector unsigned long long vec_mulo (vector unsigned int,
14921 vector unsigned int)
14922 * doc/extend.texi: Fix the documentation for the built-in
14925 2017-06-07 Carl Love <cel@us.ibm.com>
14928 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14931 2017-06-07 Carl Love <cel@us.ibm.com>
14933 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14934 support, Generate doublehv for signed int/float for BE case only.
14936 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14938 * doc/invoke.texi (mcx16): Rewrite.
14940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14942 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14943 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14944 *mov<mode>_softfloat, and an anonymous splitter): Use
14945 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14947 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14949 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14951 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14952 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14953 (rs6000_debug_reg_global): Adjust.
14954 (rs6000_init_hard_regno_mode_ok): Adjust.
14955 (rs6000_dbx_register_number): Adjust.
14956 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14957 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14958 Remove SPE_ACC and SPEFSCR.
14959 (REG_ALLOC_ORDER): Ditto.
14960 (FRAME_POINTER_REGNUM): Change to 111.
14961 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14962 (REG_CLASS_NAMES): Ditto.
14963 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14964 (REGISTER_NAMES): Ditto.
14965 (ADDITIONAL_REG_NAMES): Ditto.
14966 (rs6000_reg_names): Ditto.
14967 * config/rs6000/rs6000.md: Renumber some register number
14970 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14972 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14974 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14975 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14977 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14978 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14979 Delete the SPE high registers.
14980 (REG_ALLOC_ORDER): Ditto.
14981 (enum reg_class): Remove SPE_HIGH_REGS.
14982 (REG_CLASS_NAMES): Ditto.
14983 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14984 (REGISTER_NAMES): Ditto.
14985 (rs6000_reg_names): Ditto.
14986 * doc/tm.texi.in: Remove SPE as example.
14987 * doc/tm.texi: Regenerate.
14989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14991 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14992 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14993 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14994 * config/rs6000/rs6000.md (type): Remove "brinc".
14996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14998 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14999 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15000 * config/rs6000/linuxspe.h: Delete file.
15001 * config/rs6000/rs6000.md: Don't include spe.md.
15002 * config/rs6000/spe.h: Delete file.
15003 * config/rs6000/spe.md: Delete file.
15004 * config/rs6000/t-rs6000: Remove spe.md.
15006 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15008 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15009 (reg_or_none500mem_operand): Delete.
15010 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15011 instead of reg_or_none500mem_operand.
15013 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15016 handling of SPE flags.
15017 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15019 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15021 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15023 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15024 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15025 paired_divv2sf3): Similar.
15026 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15027 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15028 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15030 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15031 Rename the paired_* instruction patterns.
15032 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15034 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15035 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15036 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15037 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15038 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15039 PAIRED_VECTOR_MODE.
15040 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15041 (spe_func_has_64bit_regs_p): Delete.
15042 (spe_expand_predicate_builtin): Delete.
15043 (spe_expand_evsel_builtin): Delete.
15044 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15045 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15046 (invalid_e500_subreg): Delete.
15047 (rs6000_legitimize_address): Always force_reg op2 as well, for
15048 paired single memory accesses.
15049 (rs6000_member_type_forces_blk): Delete.
15050 (rs6000_spe_function_arg): Delete.
15051 (rs6000_expand_unop_builtin): Delete SPE handling.
15052 (rs6000_expand_binop_builtin): Ditto.
15053 (spe_expand_stv_builtin): Delete.
15054 (bdesc_2arg_spe): Delete.
15055 (spe_expand_builtin): Delete.
15056 (spe_expand_predicate_builtin): Delete.
15057 (spe_expand_evsel_builtin): Delete.
15058 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15059 (spe_init_builtins): Delete.
15060 (spe_func_has_64bit_regs_p): Delete.
15061 (savres_routine_name): Delete "info" parameter. Adjust callers.
15062 (rs6000_emit_stack_reset): Ditto.
15063 (rs6000_dwarf_register_span): Delete.
15064 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15065 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15066 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15068 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15070 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15071 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15072 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15073 mulv2sf3, divv2sf3): Delete expanders.
15075 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15077 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15079 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15081 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15082 * config/rs6000/rs6000.c: Ditto.
15084 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15086 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15087 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15088 comparison_operator.
15090 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15092 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15093 * config/rs6000/rs6000.opt: Ditto.
15094 * config/rs6000/t-rtems: Ditto.
15096 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15098 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15099 TARGET_E500_SINGLE by 0, simplify.
15100 * config/rs6000/rs6000.c: Ditto.
15101 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15102 (spe_build_register_parallel): Delete.
15103 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15104 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15105 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15106 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15107 (E500_CONVERT): Delete.
15108 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15112 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15113 * config/rs6000/dfp.md: Ditto.
15114 (negdd2, *negdd2_fpr): Merge.
15115 (absdd2, *absdd2_fpr): Merge.
15116 (negtd2, *negtd2_fpr): Merge.
15117 (abstd2, *abstd2_fpr): Merge.
15118 * config/rs6000/e500.h: Delete file.
15119 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15120 TARGET_FPRS by 1 and simplify.
15121 * config/rs6000/rs6000-c.c: Ditto.
15122 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15123 TARGET_DF_SPE by 0.
15124 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15126 * config/rs6000/rs6000.md: Ditto.
15127 (floatdidf2, *floatdidf2_fpr): Merge.
15128 (move_from_CR_gt_bit): Delete.
15129 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15130 (E500_CR_IOR_COMPARE): Delete.
15131 (All patterns that require !TARGET_FPRS): Delete.
15132 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15134 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15136 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15138 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15140 * graphds.c (add_edge): Intitialize edge's attached data.
15141 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15142 pointer parameter. Call pointed function on each edge during
15143 graph traversing. Skip traversing the edge when the function
15145 (graphds_dfs, graphds_scc): Ditto.
15146 (for_each_edge): New parameter. Pass the new parameter to callback
15148 * graphds.h (skip_edge_callback): New function pointer type.
15149 (graphds_dfs, graphds_scc): New function pointer parameter.
15150 (graphds_edge_callback, for_each_edge): New parameter.
15152 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15154 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15155 out code checking if runtime alias check is possible to below ...
15156 Call the new function.
15157 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15158 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15160 2017-06-07 Marek Polacek <polacek@redhat.com>
15163 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15164 TYPE_OVERFLOW_WRAPS check.
15166 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15168 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15169 if versioning is required.
15170 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15171 peeling with the check for versioning.
15173 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15175 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15176 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15177 Set true to new parameter if new ssa variable is defined.
15178 (vect_gen_vector_loop_niters): Refactor. Set range information
15179 for the new vector loop bound variable.
15180 (vect_do_peeling): Ditto.
15182 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15184 * tree-affine.c (ssa.h): Include header file.
15185 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15186 has wrapping overflow behavior.
15188 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15190 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15192 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15194 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15195 (tree_to_aff_combination): ... here.
15197 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15199 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15200 reg_pressure model function.
15201 (ivopts_global_cost_for_size): Delete.
15202 (determine_set_costs, iv_ca_recount_cost): Call new model function
15203 ivopts_estimate_reg_pressure.
15205 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15207 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15208 expensive than udiv. Remove floating point cases from mod.
15210 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15212 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15213 Increase idiv cost.
15215 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15217 * config/aarch64/aarch64.md
15218 (copysignsf3): Fix mask generation.
15220 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15222 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15224 (class dump_manager): Add register_dumps method.
15225 * dumpfile.c: Include langhooks.h.
15226 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15227 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15228 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15229 (dump_manager::dump_register): Start with 512 entries instead of 32.
15230 (dump_manager::register_dumps): New method.
15231 * toplev.c (general_init): Instead of invoking register_dumps
15232 langhook, invoke register_dumps method on the dump manager.
15233 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15236 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15238 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15239 Say that # requires an associated define_split to exist, and that
15240 the define_split must be suitable for use after register allocation.
15242 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15244 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15245 (compute_outgoing_frequencies): Also initialize zero counts.
15246 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15247 around loops; preserve more of profile when nothing changes.
15249 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15251 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15253 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15254 * config/arm/arm-cpu-cdata.h: Regenerate.
15255 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15256 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15257 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15258 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15259 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15261 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15263 * config/arm/t-rmprofile: Likewise.
15264 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15266 2017-06-06 David S. Miller <davem@davemloft.net>
15269 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15270 blockage if function uses alloca.
15272 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15274 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15275 New "uid" fields to hold pretty-print IDs of group and ref.
15276 Memory references are now identified as <group_id>:<ref_id>
15277 instead of using [random] addresses.
15278 (dump_mem_details): Simplify, no functional change.
15279 (dump_mem_ref): Simplify and make output more concise.
15280 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15281 (find_or_create_group): Initialize group uid.
15282 (record_ref): Initialize ref uid. Improve debug output.
15283 (prune_group_by_reuse, should_issue_prefetch_p,)
15284 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15285 (mark_nontemporal_store, determine_loop_nest_reuse):
15286 Improve debug output.
15288 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15290 * dbgcnt.def (prefetch): New debug counter.
15291 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15292 (schedule_prefetches): Stop issueing prefetches if debug counter
15295 2017-06-06 Tom de Vries <tom@codesourcery.com>
15297 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15298 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15300 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15302 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15303 Use aarch64_reg_or_zero predicate for operand 4.
15304 (aarch64_compare_and_swap<mode> define_insn_and_split):
15305 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15306 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15308 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15310 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15311 (arm_compute_save_core_reg_mask): This.
15312 (thumb1_compute_save_reg_mask): Rename into ...
15313 (thumb1_compute_save_core_reg_mask): This.
15314 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15315 (arm_compute_frame_layout): Likewise.
15317 2017-06-06 Richard Biener <rguenther@suse.de>
15319 PR tree-optimization/80974
15320 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15321 keep or clear leaders SSA info.
15323 2017-06-06 Tom de Vries <tom@codesourcery.com>
15325 * config/nvptx/nvptx.c (split_mode_p): New function.
15326 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15328 2017-06-06 Tom de Vries <tom@codesourcery.com>
15330 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15332 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15335 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15338 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15340 * shrink-wrap.c (handle_simple_exit): Update profile.
15341 (try_shrink_wrapping): Upate profile.
15343 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15345 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15346 (tree_guess_outgoing_edge_probabilities): New.
15347 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15348 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15350 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15352 * ipa-split.c (split_function): Initialize return bb profile.
15354 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15356 * profile.c (compute_branch_probabilities): Also initialize
15357 EXIT_BLOCK profile.
15359 2017-06-06 Richard Biener <rguenther@suse.de>
15361 PR tree-optimization/80928
15362 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15363 (vect_analyze_loop_operations): Properly guard analysis for
15365 (vect_transform_loop): Likewise.
15366 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15367 (vect_model_induction_cost): Do not cost for pure SLP.
15368 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15369 of induction in inner loop vectorization.
15370 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15371 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15372 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15374 (vect_analyze_slp_cost_1): Cost induction.
15375 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15376 (vect_get_slp_vect_defs): Likewise.
15377 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15378 (vect_transform_stmt): Handle SLP reductions.
15379 * tree-vectorizer.h (vectorizable_induction): Adjust.
15381 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15383 * config/rs6000/rs6000.c (make_resolver_func): Update
15384 init_lowered_empty_function call.
15386 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15388 * doc/invoke.texi: Document the -fprofile-abs-path option.
15389 * common.opt (fprofile-abs-path): New option.
15390 * gcov-io.h (gcov_write_filename): Declare.
15391 * gcov-io.c (gcov_write_filename): New function.
15392 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15393 * profile.c (output_location): Likewise.
15395 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15397 * shring-wrap.c: Revert accidental commit.
15399 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15401 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15403 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15405 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15407 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15408 profile in callgraph edge.
15409 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15410 (apply_scale): Likewise.
15411 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15412 Also scale profile when inlining function with zero profile.
15413 (initialize_cfun): Update exit block profile even when it is zero.
15414 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15415 when profile is read.
15417 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15419 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15420 (CLONE_*): New constants to define the processors we can generate
15421 code for with the target_clone attribute.
15422 (rs6000_clone_map): New array to identify which clone processors
15423 the current program is running on.
15424 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15425 target_clone attribute.
15426 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15427 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15428 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15429 (cpu_expand_builtin): Add support for target_clone attribute.
15430 (rs6000_valid_attribute_p): Allow "default" attribute.
15431 (get_decl_name): New debug function to simplify printing the
15432 current function name in debugging statements.
15433 (rs6000_clone_priority): New functions to support the target_clone
15434 attribute, and be able to generate code to switch between ISA 2.05
15435 through ISA 3.0 (power6 through power9).
15436 (rs6000_compare_version_priority): Likewise.
15437 (rs6000_get_function_versions_dispatcher): Likewise.
15438 (make_resolver_func): Likewise.
15439 (add_condition_to_bb): Likewise.
15440 (dispatch_function_versions): Likewise.
15441 (rs6000_generate_version_dispatcher_body): Likewise.
15442 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15443 (fusion_gpr_load_p): Fix a spacing issue.
15444 * doc/extend.texi (Common Function Attributes): Document that the
15445 PowerPC supports the target_clone attribute.
15447 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15449 * config/arm/arm.h: explain F symbol found in description of ARM
15450 register allocation in its legend.
15452 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15454 * config/mips/frame-header-opt.c: Include profile-count.h.
15455 * config/riscv/riscv.c: Include profile-count.h
15457 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15459 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15461 (sm_set_flag_if_changed): Add bbs field.
15462 (execute_sm_if_changed_flag_set): Pass BBS.
15463 (execute_sm): Update.
15465 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15467 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15470 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15472 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15473 (peephole2): New peephole2 to emit the above.
15474 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15476 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15478 * config/aarch64/aarch64.c (define_peephole2 above
15479 *sub_<shift>_<mode>): New peephole.
15481 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15483 * config/i386/i386.c (make_resolver_func): Update.
15484 * Makefile.in: Add profile-count.h and profile-count.o
15485 * auto-profile.c (afdo_indirect_call): Update to new API.
15486 (afdo_set_bb_count): Update.
15487 (afdo_propagate_edge): Update.
15488 (afdo_propagate_circuit): Update.
15489 (afdo_calculate_branch_prob): Update.
15490 (afdo_annotate_cfg): Update.
15491 * basic-block.h: Include profile-count.h
15492 (struct edge_def): Turn count to profile_count.
15493 (struct basic_block_def): Likewie.
15494 (REG_BR_PROB_BASE): Move to profile-count.h
15495 (RDIV): Move to profile-count.h
15496 * bb-reorder.c (max_entry_count): Turn to profile_count.
15497 (find_traces): Update.
15498 (rotate_loop):Update.
15499 (connect_traces):Update.
15500 (sanitize_hot_paths):Update.
15501 * bt-load.c (migrate_btr_defs): Update.
15502 * cfg.c (RDIV): Remove.
15503 (init_flow): Use alloc_block.
15504 (alloc_block): Uninitialize count.
15505 (unchecked_make_edge): Uninitialize count.
15506 (check_bb_profile): Update.
15507 (dump_edge_info): Update.
15508 (dump_bb_info): Update.
15509 (update_bb_profile_for_threading): Update.
15510 (scale_bbs_frequencies_int): Update.
15511 (scale_bbs_frequencies_gcov_type): Update.
15512 (scale_bbs_frequencies_profile_count): New.
15513 * cfg.h (update_bb_profile_for_threading): Update.
15514 (scale_bbs_frequencies_profile_count): Declare.
15515 * cfgbuild.c (compute_outgoing_frequencies): Update.
15516 (find_many_sub_basic_blocks): Update.
15517 * cfgcleanup.c (try_forward_edges): Update.
15518 (try_crossjump_to_edge): Update.
15519 * cfgexpand.c (expand_gimple_tailcall): Update.
15520 (construct_exit_block): Update.
15521 * cfghooks.c (verify_flow_info): Update.
15522 (dump_bb_for_graph): Update.
15523 (split_edge): Update.
15524 (make_forwarder_block): Update.
15525 (duplicate_block): Update.
15526 (account_profile_record): Update.
15527 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15528 (get_estimated_loop_iterations): Update.
15529 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15530 (single_likely_exit): Update.
15531 * cfgloopmanip.c (scale_loop_profile): Update.
15533 (set_zero_probability): Update.
15534 (lv_adjust_loop_entry_edge): Update.
15535 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15536 (purge_dead_edges): Update.
15537 (rtl_account_profile_record): Update.
15538 * cgraph.c (cgraph_node::create): Uninitialize count.
15539 (symbol_table::create_edge): Uninitialize count.
15540 (cgraph_update_edges_for_call_stmt_node): Update.
15541 (cgraph_edge::dump_edge_flags): Update.
15542 (cgraph_node::dump): Update.
15543 (cgraph_edge::maybe_hot_p): Update.
15544 * cgraph.h: Include profile-count.h
15545 (create_clone), create_edge, create_indirect_edge): Update.
15546 (cgraph_node): Turn count to profile_count.
15547 (cgraph_edge0: Likewise.
15548 (make_speculative, clone): Update.
15549 (create_edge): Update.
15550 (init_lowered_empty_function): Update.
15551 * cgraphclones.c (cgraph_edge::clone): Update.
15552 (duplicate_thunk_for_node): Update.
15553 (cgraph_node::create_clone): Update.
15554 * cgraphunit.c (cgraph_node::analyze): Update.
15555 (cgraph_node::expand_thunk): Update.
15556 * final.c (dump_basic_block_info): Update.
15557 * gimple-streamer-in.c (input_bb): Update.
15558 * gimple-streamer-out.c (output_bb): Update.
15559 * graphite.c (print_global_statistics): Update.
15560 (print_graphite_scop_statistics): Update.
15561 * hsa-brig.c: Include basic-block.h.
15562 * hsa-dump.c: Include basic-block.h.
15563 * hsa-gen.c (T sum_slice): Update.
15564 (convert_switch_statements):Update.
15565 * hsa-regalloc.c: Include basic-block.h.
15566 * ipa-chkp.c (chkp_produce_thunks): Update.
15567 * ipa-cp.c (struct caller_statistics): Update.
15568 (init_caller_stats): Update.
15569 (gather_caller_stats): Update.
15570 (ipcp_cloning_candidate_p): Update.
15571 (good_cloning_opportunity_p): Update.
15572 (get_info_about_necessary_edges): Update.
15573 (dump_profile_updates): Update.
15574 (update_profiling_info): Update.
15575 (update_specialized_profile): Update.
15576 (perhaps_add_new_callers): Update.
15577 (decide_about_value): Update.
15578 (ipa_cp_c_finalize): Update.
15579 * ipa-devirt.c (struct odr_type_warn_count): Update.
15580 (struct decl_warn_count): Update.
15581 (struct final_warning_record): Update.
15582 (possible_polymorphic_call_targets): Update.
15583 (ipa_devirt): Update.
15584 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15585 * ipa-icf.c (sem_function::merge): Update.
15586 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15587 * ipa-inline.c (compute_uninlined_call_time): Update.
15588 (compute_inlined_call_time): Update.
15589 (want_inline_small_function_p): Update.
15590 (want_inline_self_recursive_call_p): Update.
15591 (edge_badness): Update.
15592 (lookup_recursive_calls): Update.
15593 (recursive_inlining): Update.
15594 (inline_small_functions): Update.
15595 (dump_overall_stats): Update.
15596 (dump_inline_stats): Update.
15597 * ipa-profile.c (ipa_profile_generate_summary): Update.
15598 (ipa_propagate_frequency): Update.
15599 (ipa_profile): Update.
15600 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15601 * ipa-utils.c (ipa_merge_profiles): Update.
15602 * loop-doloop.c (doloop_modify): Update.
15603 * loop-unroll.c (report_unroll): Update.
15604 (unroll_loop_runtime_iterations): Update.
15605 * lto-cgraph.c (lto_output_edge): Update.
15606 (lto_output_node): Update.
15607 (input_node): Update.
15608 (input_edge): Update.
15609 (merge_profile_summaries): Update.
15610 * lto-streamer-in.c (input_cfg): Update.
15611 * lto-streamer-out.c (output_cfg): Update.
15612 * mcf.c (create_fixup_graph): Update.
15613 (adjust_cfg_counts): Update.
15614 (sum_edge_counts): Update.
15615 * modulo-sched.c (sms_schedule): Update.
15616 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15617 * predict.c (maybe_hot_count_p): Update.
15618 (probably_never_executed): Update.
15619 (dump_prediction): Update.
15620 (combine_predictions_for_bb): Update.
15621 (propagate_freq): Update.
15622 (handle_missing_profiles): Update.
15623 (counts_to_freqs): Update.
15624 (rebuild_frequencies): Update.
15625 (force_edge_cold): Update.
15626 * predict.h: Include profile-count.h
15627 (maybe_hot_count_p, counts_to_freqs): UPdate.
15628 * print-rtl-function.c: Do not include cfg.h
15629 * print-rtl.c: Include basic-block.h
15630 * profile-count.c: New file.
15631 * profile-count.h: New file.
15632 * profile.c (is_edge_inconsistent): Update.
15633 (correct_negative_edge_counts): Update.
15634 (is_inconsistent): Update.
15635 (set_bb_counts): Update.
15636 (read_profile_edge_counts): Update.
15637 (compute_frequency_overlap): Update.
15638 (compute_branch_probabilities): Update; Initialize and deinitialize
15640 (branch_prob): Update.
15641 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15642 (edge_gcov_count): New.
15643 (bb_gcov_count): New.
15644 * shrink-wrap.c (try_shrink_wrapping): Update.
15645 * tracer.c (better_p): Update.
15646 * trans-mem.c (expand_transaction): Update.
15647 (ipa_tm_insert_irr_call): Update.
15648 (ipa_tm_insert_gettmclone_call): Update.
15649 * tree-call-cdce.c: Update.
15650 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15651 (gimple_duplicate_sese_tail): Update.
15652 (gimple_account_profile_record): Update.
15653 (execute_fixup_cfg): Update.
15654 * tree-inline.c (copy_bb): Update.
15655 (copy_edges_for_bb): Update.
15656 (initialize_cfun): Update.
15657 (freqs_to_counts): Update.
15658 (copy_cfg_body): Update.
15659 (expand_call_inline): Update.
15660 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15661 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15662 (try_unroll_loop_completely): Update.
15663 (try_peel_loop): Update.
15664 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15665 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15666 * tree-ssa-loop-split.c (connect_loops): Update.
15667 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15668 * tree-ssa-reassoc.c (branch_fixup): Update.
15669 * tree-ssa-tail-merge.c (replace_block_by): Update.
15670 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15671 (compute_path_counts): Update.
15672 (update_profile): Update.
15673 (recompute_probabilities): Update.
15674 (update_joiner_offpath_counts): Update.
15675 (estimated_freqs_path): Update.
15676 (freqs_to_counts_path): Update.
15677 (clear_counts_path): Update.
15678 (ssa_fix_duplicate_block_edges): Update.
15679 (duplicate_thread_path): Update.
15680 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15681 (struct switch_conv_info): Update.
15682 * tree-tailcall.c (decrease_profile): Update.
15683 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15684 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15685 * value-prof.c (check_counter): Update.
15686 (gimple_divmod_fixed_value): Update.
15687 (gimple_mod_pow2): Update.
15688 (gimple_mod_subtract): Update.
15689 (gimple_ic_transform): Update.
15690 (gimple_stringop_fixed_value): Update.
15691 * value-prof.h (gimple_ic): Update.
15693 2017-06-02 Carl Love <cel@us.ibm.com>
15695 * config/rs6000/rs6000-c: Add support for built-in functions
15696 vector double vec_doublee (vector signed int);
15697 vector double vec_doublee (vector unsigned int);
15698 vector double vec_doublee (vector float);
15699 vector double vec_doubleh (vector signed int);
15700 vector double vec_doubleh (vector unsigned int);
15701 vector double vec_doubleh (vector float);
15702 vector double vec_doublel (vector signed int);
15703 vector double vec_doublel (vector unsigned int);
15704 vector double vec_doublel (vector float);
15705 vector double vec_doubleo (vector signed int);
15706 vector double vec_doubleo (vector unsigned int);
15707 vector double vec_doubleo (vector float);.
15708 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15709 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15711 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15712 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15713 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15715 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15716 vec_doublel, vec_doubleh.
15717 * doc/extend.texi: Update the built-in documentation file for the
15718 new built-in functions.
15720 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15723 * ipa-inline-analysis.c (free_growth_caches): Set
15724 edge_removal_hook_holder to NULL after removing it.
15726 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15728 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15729 comparision with zero.
15731 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15732 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15733 for early expansion of vec_min and vec_max builtins.
15734 (builtin_function_type): Add min/max unsigned variants to those
15735 identified as having unsigned arguments.
15737 2017-06-02 Olivier Hainque <hainque@adacore.com>
15739 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15741 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15743 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15744 Use VALL_F16 iterator rather than VALL.
15746 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15748 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15749 Emit CBNZ inside loop when doing a strong exchange and comparing
15750 against zero. Generate the CC flags after the loop.
15752 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15754 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15755 (dl_section_ref): New.
15756 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15757 On AIX, append an expression to subtract the size of the
15758 section length to dl_section_ref.
15760 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15762 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15763 for early expansion of vector absolute builtins.
15765 2017-06-02 Richard Biener <rguenther@suse.de>
15767 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
15768 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
15770 2017-06-02 Richard Biener <rguenther@suse.de>
15772 PR tree-optimization/80948
15773 * tree-tailcall.c (find_tail_calls): Track stmts to move in
15774 stmt order as well.
15776 2017-06-02 Richard Biener <rguenther@suse.de>
15778 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
15780 * tree-vect-stmts.c (process_use): Do not mark backedge defs
15781 for inductions as relevant.
15783 2017-06-02 Richard Biener <rguenther@suse.de>
15785 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
15786 (vectorizable_induction): ... this. Remove dead code.
15788 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
15790 * builtins. (expand_builtin_alloca): Remove second parameter and
15791 infer its value from the first parameter instead.
15792 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15794 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15796 PR rtl-optimization/80903
15797 * loop-doloop.c (add_test): Unshare sequence.
15799 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15801 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15803 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15805 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15807 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15808 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15809 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15810 (xlogue_layout::get_stub_rtx): Make static.
15811 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15812 (xlogue_layout::compute_stub_managed_regs): Rename to...
15813 (xlogue_layout::count_stub_managed_regs): ...this.
15814 (xlogue_layout::is_stub_managed_reg): New function.
15815 (xlogue_layout::m_stub_names): Rename to...
15816 (xlogue_layout::s_stub_names): ...this, make static.
15817 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15818 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15819 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15820 xlogue_layout::s_stub_names): Instantiate statics.
15821 (stub_managed_regs): Remove.
15822 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15823 (disable_call_ms2sysv_xlogues): Rename to...
15824 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15825 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15827 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15828 change after reload_completed.
15829 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15831 (ix86_expand_prologue): Likewise.
15832 (ix86_expand_epilogue): Likewise.
15833 (ix86_expand_split_stack_prologue): Likewise.
15834 (ix86_compute_frame_layout): Remove frame parameter ...
15835 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15836 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15838 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15839 (ix86_frame): Move from here ...
15840 * config/i386/i386.h (ix86_frame): ... to here.
15841 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15842 complete ix86_frame data structure instead. Remove some_ld_name.
15844 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15846 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15847 symbols that hold a DECL_VALUE_EXPR.
15849 2017-06-01 Martin Jambor <mjambor@suse.cz>
15851 PR tree-optimization/80898
15852 * tree-sra.c (process_subtree_disqualification): Removed.
15853 (disqualify_candidate): Do not acll
15854 process_subtree_disqualification.
15855 (subtree_mark_written_and_enqueue): New function.
15856 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15857 RHS has been disqualified and re-queue LHS if necessary. Apart
15858 from that, ignore disqualified RHS.
15860 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15862 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15863 address fetch for z10 or later.
15865 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15867 * config/arc/arc.md (tst_movb): Add guard when splitting.
15869 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15871 * config/arc/arc.c (arc_can_eliminate): Test against
15872 arc_frame_pointer_needed.
15874 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15876 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15877 to prevent store reordering.
15878 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15879 (type): Add block type.
15880 (stack_tie): Define special instruction to be used in
15883 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15885 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15886 constraint. It is not valid for the pattern.
15887 (noncommutative_binary_comparison): Likewise.
15889 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15891 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15894 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15896 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15897 be used by the reg-alloc.
15899 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15901 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15902 reg-alloc when having mul64 or mul32x16 instructions.
15903 (mulsidi3): Likewise.
15904 (umulsidi3): Likewise.
15905 (mulsi32x16): New pattern.
15906 (mulsi64): Likewise.
15907 (mulsidi64): Likewise.
15908 (umulsidi64): Likewise.
15909 (MUL32x16_REG): Define.
15910 (mul64_600): Use MUL32x16_REG.
15911 (mac64_600): Likewise.
15912 (umul64_600): Likewise.
15913 (umac64_600): Likewise.
15915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15917 * config/arc/arc.md (mulsi3_700): Make it commutative.
15919 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15921 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15923 (*sign_extendsidi2_insn): Likewise for movstosw.
15925 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15927 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15928 the type of the input discriminant value. Convert the
15929 discriminant value of signedness vary.
15931 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15933 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15934 Add to -Wall section.
15936 2017-06-01 Richard Biener <rguenther@suse.de>
15938 PR middle-end/66313
15939 * fold-const.c (fold_plusminus_mult_expr): If the factored
15940 factor may be zero use a wrapping type for the inner operation.
15941 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15942 and handle moved defs.
15943 (process_assignment): Properly guard the unary op case. Return a
15944 tri-state indicating that moving the stmt before the call may allow
15945 to continue. Pass through to_move.
15946 (find_tail_calls): Handle moving unrelated defs before
15949 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15952 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15953 splitter result in the canonical way.
15955 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15957 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15958 also for 32bit target. Update insn attributes.
15959 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15961 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15963 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15964 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15965 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15966 (_mm_maskz_min_ss): New intrinsics.
15968 2017-05-31 Martin Liska <mliska@suse.cz>
15970 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15971 Change comment style to one we normally use.
15972 (vectorizable_reduction): Likewise.
15973 (vectorizable_induction): Likewise.
15974 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15975 (vectorizable_call): Likewise.
15976 (vectorizable_simd_clone_call): Likewise.
15977 (vectorizable_conversion): Likewise.
15978 (vectorizable_assignment): Likewise.
15979 (vectorizable_shift): Likewise.
15980 (vectorizable_operation): Likewise.
15981 (vectorizable_store): Likewise.
15982 (vectorizable_load): Likewise.
15983 * tree-vectorizer.h: Likewise.
15985 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15987 * passes.c (emergency_dump_function): New.
15988 * tree-pass.h (emergency_dump_function): Declare.
15989 * plugin.c (plugins_internal_error_function): Remove.
15990 * plugin.h (plugins_internal_error_function): Remove declaration.
15991 * toplev.c (internal_error_function): New static function. Use it...
15992 (general_init): ...here.
15994 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15996 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15997 (arc_rtx_costs): Add costs for new patterns.
15998 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15999 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16001 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16003 * tree-ssa-strlen.c (get_next_strinfo): New function.
16004 (get_stridx_plus_constant): Use it.
16005 (zero_length_string): Likewise.
16006 (adjust_related_strinfos): Likewise.
16007 (adjust_last_stmt): Likewise.
16009 2017-05-31 Richard Biener <rguenther@suse.de>
16012 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16013 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16015 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16017 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16018 loop_vinfo argument and use of dependence distance vectors.
16019 Check instead whether the two references differ only in their
16020 initial value and assume that they have the same alignment if the
16021 difference is a multiple of the vector alignment.
16022 (vect_analyze_data_refs_alignment): Update call accordingly.
16024 2017-05-31 Martin Liska <mliska@suse.cz>
16027 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16029 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16031 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16032 (create_intersect_range_checks): Move from ...
16033 * tree-data-ref.c (create_intersect_range_checks_index)
16034 (create_intersect_range_checks): ... to here.
16035 (create_runtime_alias_checks): New function factored from ...
16036 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16037 here. Call above function.
16038 * tree-data-ref.h (create_runtime_alias_checks): New function.
16040 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16042 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16043 segment length for dr_b and compute it in wide_int.
16045 2017-05-31 Richard Biener <rguenther@suse.de>
16047 PR tree-optimization/80906
16048 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16049 and pass through iv_map.
16050 (copy_bb_and_scalar_dependences): Adjust.
16051 (translate_pending_phi_nodes): Likewise.
16052 (copy_loop_close_phi_args): Handle code-generating IVs instead
16055 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16057 * diagnostic-color.c (color_dict): Add "type-diff".
16058 (parse_gcc_colors): Update comment.
16059 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16060 -fdiagnostics-show-template-tree and -fno-elide-type.
16061 (GCC_COLORS): Add type-diff to example.
16063 (-fdiagnostics-show-template-tree): New.
16064 (-fno-elide-type): New.
16065 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16066 the pp_format_decoder callback. Call any m_format_postprocessor's
16068 (pretty_printer::pretty_printer): Initialize
16069 m_format_postprocessor.
16070 (pretty_printer::~pretty_printer): Delete any
16071 m_format_postprocessor.
16072 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16073 (class format_postprocessor): New class.
16074 (struct pretty_printer::format_decoder): Document the new parameters.
16075 (struct pretty_printer::m_format_postprocessor): New field.
16076 * tree-diagnostic.c (default_tree_printer): Update for new
16077 bool and const char ** params.
16078 * tree-diagnostic.h (default_tree_printer): Likewise.
16080 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16082 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16083 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16084 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16085 rs6000_gen_cell_microcode code.
16086 (rs6000_final_prescan_insn): Delete.
16087 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16088 "warn-cell-microcode" entries.
16089 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16090 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16091 throughout. Change cc_reg_not_micro_cr0_operand to
16092 cc_reg_not_cr0_operand throughout.
16093 (*extendhi<mode>2_noload): Delete.
16094 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16095 (mwarn-cell-microcode): Delete.
16096 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16097 -mgen-cell-microcode and -mwarn-cell-microcode.
16099 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16102 * config/i386/constraints.md (Yd): New constraint.
16104 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16105 and (?Yd, r) alternatives. Update insn attributes.
16106 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16107 and (?*Yd, r) alternatives. Update insn attributes.
16108 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16110 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16112 * gimplify.c (gimplify_modify_expr): Don't create a
16113 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16116 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16118 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16120 2017-05-30 Richard Biener <rguenther@suse.de>
16122 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16123 and reduc_def fields.
16124 (STMT_VINFO_REDUC_TYPE): New define.
16125 (STMT_VINFO_REDUC_DEF): Likewise.
16126 (vect_force_simple_reduction): Adjust prototype.
16127 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16128 (vect_is_simple_reduction): Remove check_reduction argument.
16129 (vect_force_simple_reduction): Adjust and set
16130 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16131 (vectorizable_reduction): Do not re-do reduction analysis
16132 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16133 * tree-parloops.c (gather_scalar_reductions): Adjust.
16135 2017-05-30 Richard Biener <rguenther@suse.de>
16137 PR middle-end/80901
16138 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16141 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16143 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16144 Introduce unknown_misalignment parameter and remove vf.
16145 (vect_peeling_hash_get_lowest_cost):
16146 Pass unknown_misalignment parameter.
16147 (vect_enhance_data_refs_alignment):
16148 Fix unsupportable data ref treatment.
16150 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16152 * tree-vect-data-refs.c (vect_get_data_access_cost):
16153 Workaround for SLP handling.
16154 (vect_enhance_data_refs_alignment):
16155 Compute costs for doing no peeling at all, compare to the best
16156 peeling costs so far and avoid peeling if cheaper.
16158 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16160 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16161 Return peeling info and set costs to zero for unlimited cost
16163 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16164 with unknown misalignment. Compute and costs for unknown
16165 misalignment, compare them to the costs for known misalignment
16166 and choose the cheapest for peeling.
16168 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16170 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16171 (vect_get_peeling_costs_all_drs): Create function.
16172 (vect_peeling_hash_get_lowest_cost):
16173 Use vect_get_peeling_costs_all_drs.
16174 (vect_peeling_supportable): Create function.
16175 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16177 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16179 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16180 DR_HAS_NEGATIVE_STEP.
16181 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16182 (vect_enhance_data_refs_alignment): Use.
16183 (vect_duplicate_ssa_name_ptr_info): Use.
16184 * tree-vectorizer.h (dr_misalignment): Use.
16185 (known_alignment_for_access_p): Use.
16187 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16190 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16192 (has_section_name): New function.
16194 2017-05-30 Martin Liska <mliska@suse.cz>
16197 * auto-profile.c (get_function_decl_from_block): Fix
16200 2017-05-30 Richard Biener <rguenther@suse.de>
16202 PR middle-end/80876
16203 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16205 2017-05-30 Martin Liska <mliska@suse.cz>
16207 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16208 * dumpfile.h (struct dump_file_info): Remove ctors.
16210 2017-05-30 Martin Liska <mliska@suse.cz>
16212 * predict.def: Fix GNU coding style.
16214 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16216 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16217 Mark 'to' argument with ATTRIBUTE_UNUSED.
16219 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16221 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16222 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16223 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16226 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16228 * doc/install.texi (Options specification): Restore entry of
16229 --enable-sjlj-exceptions.
16231 2017-05-27 Michael Eager <eager@eagercon.com>
16234 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16236 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16238 * config/microblaze/microblaze.h
16239 (FIXED_REGISTERS): Update in macro.
16240 (CALL_USED_REGISTERS): Update in macro.
16242 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16244 * doc/install.texi: Add links to macOS binary distributions.
16246 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16250 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16252 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16254 2017-05-26 Martin Liska <mliska@suse.cz>
16256 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16258 2017-05-26 Martin Liska <mliska@suse.cz>
16260 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16261 always leading ';; '.
16262 (dump_bb_info): Likewise.
16263 (brief_dump_cfg): Likewise.
16264 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16265 * dumpfile.c: Remove usage of TDF_VERBOSE.
16266 * dumpfile.h (enum dump_kind): Likewise.
16267 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16268 * print-tree.c (debug_verbose): Remove.
16269 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16270 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16271 * tree-diagnostic.c (default_tree_printer): Replace
16272 TDF_DIAGNOSTIC with TDF_SLIM.
16274 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16276 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16277 in parameter loop, rather than loop_vinfo.
16278 (create_intersect_range_checks): Ditto.
16279 (vect_create_cond_for_alias_checks): Update call to above functions.
16281 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16283 PR tree-optimization/80815
16284 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16285 for merging runtime alias checks. Handle negative DR_STEPs.
16287 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16289 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16291 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16292 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16293 out code pruning runtime alias checks.
16294 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16295 factored out from above.
16296 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16298 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16300 (prune_runtime_alias_test_list): New decalaration.
16302 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16304 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16305 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16306 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16307 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16308 (operator==, comp_dr_with_seg_len_pair): Ditto.
16309 (vect_prune_runtime_alias_test_list): Ditto.
16311 2017-05-26 Martin Liska <mliska@suse.cz>
16314 * params.def: Bound partial-inlining-entry-probability param.
16316 2017-05-26 Marek Polacek <polacek@redhat.com>
16319 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16322 2017-05-26 Richard Biener <rguenther@suse.de>
16324 PR tree-optimization/80842
16325 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16328 2017-05-26 Richard Biener <rguenther@suse.de>
16330 PR tree-optimization/80844
16331 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16333 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16335 * doc/md.texi (Machine Constraints): Update x86 family
16336 machine constraints section to match 'config/i386/constraints.md'.
16338 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16340 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16342 2017-05-25 Nathan Sidwell <nathan@acm.org>
16344 * doc/invoke.texi (--enable-languages): Update documentation.
16346 2017-05-25 Martin Liska <mliska@suse.cz>
16348 * dumpfile.c: Add TDF_FOLDING.
16349 * dumpfile.h (enum dump_kind): Likewise.
16350 * genmatch.c (dt_simplify::gen_1): Use it.
16352 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16354 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16356 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16358 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16359 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16361 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16363 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16364 * match.pd (X == C): Rewrite it here.
16365 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16366 with_certain_nonzero_bits2): New predicates.
16367 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16369 2017-05-24 Nathan Sidwell <nathan@acm.org>
16371 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16374 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16377 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16379 * config/powerpcspe: New port. Files are copied from the rs6000
16380 port, with "rs6000" in filenames replaced by "powerpcspe".
16382 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16384 PR rtl-optimization/80754
16385 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16387 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16389 * config/sparc/sparc.md (length): Return the correct value for -mflat
16390 sibcalls to match output_sibcall.
16392 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16396 * config/rs6000/rs6000.c (struct machine_function): Add new field
16398 (rs6000_get_separate_components): Init that field, use it.
16399 (rs6000_components_for_bb): Use the field.
16401 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16403 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16405 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16407 PR middle-end/80823
16408 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16410 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16413 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16414 address against address_operand predicate.
16415 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16417 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16419 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16420 parameters passed indirectly.
16422 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16424 * config/i386/i386.md (*movdi_internal): Remove SSE4
16425 alternative 18 (?r, *v). Update insn attributes.
16426 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16427 Update insn attributes.
16428 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16429 Update insn attributes.
16430 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16431 alternative 1 (r, v). Remove isa attribute.
16432 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16433 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16434 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16436 2017-05-23 Tom de Vries <tom@codesourcery.com>
16438 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16441 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16443 * cgraphunit.c (symbol_table::process_new_functions): Update.
16444 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16445 (inline_generate_summary): Rename to ...
16446 (ipa_fn_summary_generate): ... this one.
16447 (inline_read_summary): Rename to ...
16448 (ipa_fn_summary_read): ... this one.
16449 (inline_write_summary): Rename to ...
16450 (ipa_fn_summary_write): ... this one.
16451 (inline_free_summary): Rename to ...
16452 (ipa_free_fn_summary): ... this one.
16453 (pass_data_local_fn_summary, pass_local_fn_summary,
16454 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16455 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16456 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16457 make_pass_ipa_fn_summary): New.
16458 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16459 inline_write_summary, inline_free_summary): Remove.
16460 (ipa_free_fn_summary) : New.
16461 * ipa-inline.c (ipa_inline): Update.
16462 (pass_ipa_inline): Do not generate summaries.
16463 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16465 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16466 and add pass_ipa_fn_summary.
16467 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16469 (make_pass_inline_parameters): Remove.
16471 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16473 * omp-low.c (struct omp_context): Remove "default_kind" member.
16476 * omp-offload.c (execute_oacc_device_lower): Remove the
16477 parallelism dimensions function attributes for unparallelized
16478 OpenACC kernels constructs.
16480 2017-05-23 Martin Liska <mliska@suse.cz>
16482 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16484 (cgraph_edge::make_speculative): Likewise.
16485 (cgraph_edge::resolve_speculation): Likewise.
16486 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16487 (cgraph_node::dump): Likewise.
16488 * cgraph.h: Likewise.
16489 * cgraphunit.c (analyze_functions): Likewise.
16490 (symbol_table::compile): Likewise.
16491 * ipa-cp.c (print_all_lattices): Likewise.
16492 (determine_versionability): Likewise.
16493 (initialize_node_lattices): Likewise.
16494 (ipcp_verify_propagated_values): Likewise.
16495 (estimate_local_effects): Likewise.
16496 (update_profiling_info): Likewise.
16497 (create_specialized_node): Likewise.
16498 (perhaps_add_new_callers): Likewise.
16499 (decide_about_value): Likewise.
16500 (decide_whether_version_node): Likewise.
16501 (identify_dead_nodes): Likewise.
16502 (ipcp_store_bits_results): Likewise.
16503 * ipa-devirt.c (dump_targets): Likewise.
16504 (ipa_devirt): Likewise.
16505 * ipa-icf.c (sem_item::dump): Likewise.
16506 (sem_function::equals): Likewise.
16507 (sem_variable::equals): Likewise.
16508 (sem_item_optimizer::read_section): Likewise.
16509 (sem_item_optimizer::execute): Likewise.
16510 (congruence_class::dump): Likewise.
16511 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16512 (dump_inline_summary): Likewise.
16513 (estimate_node_size_and_time): Likewise.
16514 (inline_analyze_function): Likewise.
16515 * ipa-inline-transform.c (inline_call): Likewise.
16516 * ipa-inline.c (report_inline_failed_reason): Likewise.
16517 (want_early_inline_function_p): Likewise.
16518 (edge_badness): Likewise.
16519 (update_edge_key): Likewise.
16520 (inline_small_functions): Likewise.
16521 * ipa-profile.c (ipa_profile): Likewise.
16522 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16523 (ipa_make_edge_direct_to_target): Likewise.
16524 (remove_described_reference): Likewise.
16525 (ipa_impossible_devirt_target): Likewise.
16526 (propagate_controlled_uses): Likewise.
16527 (ipa_print_node_params): Likewise.
16528 (ipcp_transform_function): Likewise.
16529 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16530 (propagate_pure_const): Likewise.
16531 * ipa-reference.c (generate_summary): Likewise.
16532 (read_write_all_from_decl): Likewise.
16533 (propagate): Likewise.
16534 (ipa_reference_read_optimization_summary): Likewise.
16535 * ipa-utils.c (ipa_merge_profiles): Likewise.
16536 * ipa.c (walk_polymorphic_call_targets): Likewise.
16537 (symbol_table::remove_unreachable_nodes): Likewise.
16538 (ipa_single_use): Likewise.
16539 * passes.c (execute_todo): Likewise.
16540 * predict.c (drop_profile): Likewise.
16541 * symtab.c (symtab_node::get_dump_name): New function.
16542 (symtab_node::dump_name): Likewise.
16543 (symtab_node::dump_asm_name): Likewise.
16544 (symtab_node::dump_references): Likewise.
16545 (symtab_node::dump_referring): Likewise.
16546 (symtab_node::dump_base): Likewise.
16547 (symtab_node::debug_symtab): Likewise.
16548 * tree-sra.c (convert_callers_for_node): Likewise.
16549 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16550 * value-prof.c (init_node_map): Likewise.
16552 2017-05-23 Martin Liska <mliska@suse.cz>
16554 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16555 and symtab_node::debug_symtab to symbol_table::debug.
16556 * cgraphunit.c (analyze_functions): Use the renamed function.
16557 (symbol_table::compile): Likewise.
16558 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16559 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16560 * passes.c (execute_todo): Likewise.
16561 * symtab.c (symbol_table::dump): New function.
16562 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16564 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16566 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16567 that nonconst implies exec.
16569 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16571 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16572 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16573 (inline_edge_summary_vec): Turn into ...
16574 (ipa_call_summaries): ... this one.
16575 (redirect_to_unreachable, edge_set_predicate,
16576 evaluate_properties_for_edge, inline_summary_alloc,
16577 reset_ipa_call_summary, reset_inline_summary,
16578 inline_summary_t::duplicate): Update.
16579 (inline_edge_duplication_hook): Turn to ...
16580 (ipa_call_summary_t::duplicate): ... this one.
16581 (inline_edge_removal_hook): Turn to ...
16582 (ipa_call_summary_t::remove): ... this one.
16583 (dump_inline_edge_summary): Turn to ...
16584 (dump_ipa_call_summary): ... this one.
16585 (estimate_function_body_sizes): Update.
16586 (inline_update_callee_summaries): Update.
16587 (remap_edge_change_prob): Update.
16588 (remap_edge_summaries): Update.
16589 (inline_merge_summary): Update.
16590 (do_estimate_edge_time): Update.
16591 (inline_generate_summary): Update.
16592 (inline_read_section): Update.
16593 (inline_read_summary): Update.
16594 (inline_free_summary): Update.
16595 * ipa-inline.c (can_inline_edge_p): Update.
16596 (compute_inlined_call_time): Update.
16597 (want_inline_small_function_p): Update.
16598 (edge_badness): Update.
16599 (early_inliner): Update.
16600 * ipa-inline.h (inline_edge_summary): Turn to ...
16601 (ipa_call_summary): ... this one.
16602 (ipa_call_summary_t): New class.
16603 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16604 (ipa_call_summaries): New.
16605 (inline_edge_summary): Remove.
16606 (estimate_edge_growth): Update.
16607 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16608 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16609 * ipa-split.c (execute_split_functions): Update.
16610 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16612 2017-05-23 Tom de Vries <tom@codesourcery.com>
16614 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16615 attributes): Document rdrand effective target.
16617 2017-05-23 Tom de Vries <tom@codesourcery.com>
16619 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16620 attributes): Sort alphabetically.
16622 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16624 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16626 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16629 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16630 V2DF/V2DI splat into two separate patterns, one that handles
16631 registers, and the other that only handles memory. Drop support
16632 for splatting from a GPR on ISA 2.07 and then splitting the
16633 splat into direct move and splat.
16634 (vsx_splat_<mode>_reg): Likewise.
16635 (vsx_splat_<mode>_mem): Likewise.
16637 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16639 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16641 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16643 PR middle-end/80809
16644 * omp-low.c (finish_taskreg_remap): New function.
16645 (finish_taskreg_scan): If unit size of ctx->record_type
16646 is non-constant, unshare the size expression and replace
16647 decls in it with possible outer var refs.
16649 PR middle-end/80809
16650 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16651 GOVD_SHARED rather than GOVD_PRIVATE with it.
16652 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16653 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16655 PR middle-end/80853
16656 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16657 as last argument to build_outer_var_ref for pointer bases of array
16658 section reductions.
16660 2017-05-19 Martin Sebor <msebor@redhat.com>
16662 * print-tree.c (print_node): Print DECL_READ_P flag.
16664 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16666 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16667 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16668 * cgraph.c: Likewise.
16669 * cgraphunit.c: Likewise.
16670 * gengtype.c: Likewise.
16671 * ipa-cp.c: Likewise.
16672 * ipa-devirt.c: Likewise.
16673 * ipa-icf.c: Likewise.
16674 * ipa-predicate.c: Likewise.
16675 * ipa-profile.c: Likewise.
16676 * ipa-prop.c: Likewise.
16677 * ipa-split.c: Likewise.
16679 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16680 edge_predicate_pool, dump_inline_hints,
16681 inline_summary::account_size_time, redirect_to_unreachable,
16682 edge_set_predicate, set_hint_predicate,
16683 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16684 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16685 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16686 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16687 ipa_call_summary_t::remove, initialize_growth_caches,
16688 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16689 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16690 mark_modified, unmodified_parm_1, unmodified_parm,
16691 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16692 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16693 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16694 will_be_nonconstant_predicate, record_modified_bb_info,
16695 get_minimal_bb, record_modified, param_change_prob,
16696 phi_result_unknown_predicate, predicate_for_phi_result,
16697 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16698 estimate_function_body_sizes, compute_inline_parameters,
16699 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16700 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16701 inline_update_callee_summaries, remap_edge_change_prob,
16702 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16703 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16704 inline_analyze_function, inline_summary_t::insert,
16705 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16706 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16707 inline_free_summary): Move to ipa-fnsummary.h
16708 (predicate_t): Remove.
16709 * ipa-fnsummary.c: New file.
16710 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16711 (enum inline_hints_vals, inline_hints, agg_position_info,
16712 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16713 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16714 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16715 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16716 inline_read_summary, inline_write_summary, inline_free_summary,
16717 inline_analyze_function, initialize_inline_failed,
16718 inline_merge_summary, inline_update_overall_summary,
16719 compute_inline_parameters): Move to ipa-fnsummary.h
16720 * ipa-fnsummary.h: New file.
16721 * ipa-inline-transform.h: Include ipa-inline.h.
16722 * ipa-inline.c: LIkewise.
16724 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16726 * ipa-inline.c (edge_badness): Use inlined_time instead of
16727 inline_summaries->get.
16729 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16731 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16733 2017-05-22 Nathan Sidwell <nathan@acm.org>
16735 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16736 (fdump-lang): Document 'raw' option.
16737 * dumpfile.h (TDI_tu): Delete.
16738 * dumpfile.c (dump_files): Remove translation-unit.
16739 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16741 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16743 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16744 command option from $(AWK) call.
16745 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16746 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16747 [FORMAT]: Remove handling of variable.
16748 * config/avr/t-multilib: Regenerate.
16750 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16752 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16754 (dump_inline_summary): Do not print self_time.
16755 (estimate_function_body_sizes): Do not set self_time.
16756 (compute_inline_parameters): Likewise.
16757 (inline_read_section, inline_write_summary): Do not stream self_time.
16758 * ipa-inline.h (inline_summary): Drop self_time.
16760 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16762 * ipa-inline-analysis.c (account_size_time): Rename to ...
16763 (inline_summary::account_size_time): ... this one.
16764 (reset_ipa_call_summary): Turn to ...
16765 (ipa_call_summary::reset): ... this one.
16766 (reset_inline_summary): Turn to ...
16767 (inline_summary::reset): ... this one.
16768 (inline_summary_t::remove): Update.
16769 (inline_summary_t::duplicate): Update.
16770 (ipa_call_summary_t::remove): Update.
16771 (dump_inline_summary): Update.
16772 (estimate_function_body_sizes): Update.
16773 (compute_inline_parameters): Update.
16774 (estimate_node_size_and_time): Update.
16775 (inline_merge_summary): Update.
16776 (inline_update_overall_summary): Update.
16777 (inline_read_section): Update.
16778 (inline_write_summary): Update.
16779 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
16780 add account_size_time and reset member functions.
16781 (ipa_call_summary): Add reset function.
16782 * ipa-predicate.h (predicate::operator &): Constify.
16784 2017-05-22 Richard Biener <rguenther@suse.de>
16786 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
16788 2017-05-19 Jason Merrill <jason@redhat.com>
16790 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16792 2017-05-19 Marek Polacek <polacek@redhat.com>
16795 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16796 TYPE_OVERFLOW_WRAPS checks.
16798 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16800 * tree-core.h (enum omp_clause_default_kind): Add
16801 "OMP_CLAUSE_DEFAULT_PRESENT".
16802 * tree-pretty-print.c (dump_omp_clause): Handle it.
16803 * gimplify.c (enum gimplify_omp_var_data): Add
16804 "GOVD_MAP_FORCE_PRESENT".
16805 (gimplify_adjust_omp_clauses_1): Map it to
16806 "GOMP_MAP_FORCE_PRESENT".
16807 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16809 * gimplify.c (oacc_default_clause): Clarify.
16811 2017-05-19 Nathan Sidwell <nathan@acm.org>
16813 LANG_HOOK_REGISTER_DUMPS
16814 * toplev.c (general_init): Call register dump lang hook.
16815 * doc/invoke.texi: Document -fdump-lang option family.
16816 * dumpfile.c (dump_files): Remove class dump here.
16817 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16818 * dumpfile.h (tree_dump_index): Remove TDI_class.
16819 * langhooks-def.h (lhd_register_dumps): Declare.
16820 (LANG_HOOKS_REGISTER_DUMPS): Define.
16821 (LANG_HOOKS_INITIALIZER): Add it.
16822 * langhooks.c (lhd_register_dumps): Define.
16823 * langhooks.h (struct lang_hooks): Add register_dumps.
16825 2017-05-19 Nathan Sidwell <nathan@acm.org>
16827 * context.h (context::set_passes): New.
16828 * context.c (context::context): Do not create pass manager.
16829 * toplev.c (general_init): Create pass manager here.
16831 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16833 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16834 use this splitter if two add or or instructions would also work for
16835 the constant we want to generate.
16837 2017-05-19 Richard Biener <rguenther@suse.de>
16840 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16841 predicate evaluation.
16843 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16845 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16847 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16850 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16852 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16853 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16854 (inline_edge_summary_vec): Turn into ...
16855 (ipa_call_summaries): ... this one.
16856 (redirect_to_unreachable, edge_set_predicate,
16857 evaluate_properties_for_edge, inline_summary_alloc,
16858 reset_ipa_call_summary, reset_inline_summary,
16859 inline_summary_t::duplicate): Update.
16860 (inline_edge_duplication_hook): Turn to ...
16861 (ipa_call_summary_t::duplicate): ... this one.
16862 (inline_edge_removal_hook): Turn to ...
16863 (ipa_call_summary_t::remove): ... this one.
16864 (dump_inline_edge_summary): Turn to ...
16865 (dump_ipa_call_summary): ... this one.
16866 (estimate_function_body_sizes): Update.
16867 (inline_update_callee_summaries): Update.
16868 (remap_edge_change_prob): Update.
16869 (remap_edge_summaries): Update.
16870 (inline_merge_summary): Update.
16871 (do_estimate_edge_time): Update.
16872 (inline_generate_summary): Update.
16873 (inline_read_section): Update.
16874 (inline_read_summary): Update.
16875 (inline_free_summary): Update.
16876 * ipa-inline.c (can_inline_edge_p): Update.
16877 (compute_inlined_call_time): Update.
16878 (want_inline_small_function_p): Update.
16879 (edge_badness): Update.
16880 (early_inliner): Update.
16881 * ipa-inline.h (inline_edge_summary): Turn to ...
16882 (ipa_call_summary): ... this one.
16883 (ipa_call_summary_t): New class.
16884 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16885 (ipa_call_summaries): New.
16886 (inline_edge_summary): Remove.
16887 (estimate_edge_growth): Update.
16888 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16889 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16890 * ipa-split.c (execute_split_functions): Update.
16891 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16893 2017-05-19 Richard Biener <rguenther@suse.de>
16895 PR middle-end/80764
16896 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16898 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16900 * config/rs6000/rs6000.c (struct machine_function): Add field
16901 fpr_is_wrapped_separately.
16902 (rs6000_get_separate_components): Use 64 components. Handle the
16903 new FPR components.
16904 (rs6000_components_for_bb): Handle the FPR components.
16905 (rs6000_emit_prologue_components): Handle the FPR components.
16906 (rs6000_emit_epilogue_components): Handle the FPR components.
16907 (rs6000_set_handled_components): Handle the FPR components.
16908 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16909 that are already separately shrink-wrapped.
16910 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16911 that are already separately shrink-wrapped.
16913 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16916 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16919 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16920 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16921 cases where the register allocator uses a move and an offsettable
16922 memory operation to/from a FPR register on ISA 2.06/2.07.
16923 (define_peephole2 for Altivec d-form store): Likewise.
16925 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16928 * config/i386/mmx.md (*mov<mode>_internal): Enable
16929 alternatives 11, 12, 13 and 14 also for 32bit targets.
16930 Remove alternatives 15, 16, 17 and 18.
16931 * config/i386/sse.md (vec_concatv2di): Change
16932 alternative (!x, *y) to (x, ?!*Yn).
16934 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16936 * dumpfile.h (enum dump_kind): Remove stray comma.
16938 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16940 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16941 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16942 predicate::num_conditions
16943 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16944 (CHANGED): turn into predicate::changed.
16945 (agg_position_info): Move to ipa-predicate.h
16946 (add_condition, predicate::add_clause, predicate::operator &=,
16947 predicate::or_with, predicate::evaluate, predicate::probability,
16948 dump_condition, dump_clause, predicate::dump,
16949 predicate::remap_after_duplication, predicate::remap_after_inlining,
16950 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16951 (evaluate_conditions_for_known_args): Update.
16952 (set_cond_stmt_execution_predicate): Update.
16953 * ipa-inline.h: Include ipa-predicate.h
16954 (condition, inline_param_summary, conditions, agg_position_info,
16955 predicate): Move to ipa-predicate.h
16956 * ipa-predicate.c: New file.
16957 * ipa-predicate.h: New file.
16959 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16961 * final.c (leaf_function_p): Check we are not in a sequence.
16963 2017-05-18 Martin Liska <mliska@suse.cz>
16965 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16966 * dumpfile.c (dump_register): Use new enum dump_kind.
16967 (get_dump_file_name): Likewise.
16968 (dump_enable_all): Likewise.
16969 (dump_switch_p_1): Likewise.
16970 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16971 * dumpfile.h (enum dump_kind): New enum type.
16972 (struct dump_file_info): Create constructor and
16973 format fields and comments.
16974 * passes.c (pass_manager::register_one_dump_file):
16976 * statistics.c (statistics_early_init): Likewise.
16977 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16978 TDF_TREE with TDF_SLIM.
16979 (gather_memory_references_ref): Likewise.
16981 2017-05-18 Martin Liska <mliska@suse.cz>
16983 * vec.h (struct vnull): Use it.
16985 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16987 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16988 (true_predicate, false_predicate, true_predicate_p,
16989 false_predicate_p): Remove.
16990 (single_cond_predicate, not_inlined_predicate): Turn to member function
16992 (add_condition): Update.
16993 (add_clause): Turn to...
16994 (predicate::add_clause): ... this one; update; allow passing NULL
16996 (and_predicates): Turn to ...
16997 (predicate::operator &=): ... this one.
16998 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16999 (or_predicates): Turn to ...
17000 (predicate::or_with): ... this one.
17001 (evaluate_predicate): Turn to ...
17002 (predicate::evaluate): ... this one.
17003 (predicate_probability): Turn to ...
17004 (predicate::probability): ... this one.
17005 (dump_condition): Update.
17006 (dump_predicate): Turn to ...
17007 (predicate::dump): ... this one.
17008 (account_size_time): Update.
17009 (edge_set_predicate): Update.
17010 (set_hint_predicate): UPdate.
17011 (evaluate_conditions_for_known_args): Update.
17012 (evaluate_properties_for_edge): Update.
17013 (remap_predicate_after_duplication): Turn to...
17014 (predicate::remap_after_duplication): ... this one.
17015 (remap_hint_predicate_after_duplication): Update.
17016 (inline_summary_t::duplicate): UPdate.
17017 (dump_inline_edge_summary): Update.
17018 (dump_inline_summary): Update.
17019 (set_cond_stmt_execution_predicate): Update.
17020 (set_switch_stmt_execution_predicate): Update.
17021 (compute_bb_predicates): Update.
17022 (will_be_nonconstant_expr_predicate): Update.
17023 (will_be_nonconstant_predicate): Update.
17024 (phi_result_unknown_predicate): Update.
17025 (predicate_for_phi_result): Update.
17026 (array_index_predicate): Update.
17027 (estimate_function_body_sizes): Update.
17028 (estimate_node_size_and_time): Update.
17029 (estimate_ipcp_clone_size_and_time): Update.
17030 (remap_predicate): Rename to ...
17031 (predicate::remap_after_inlining): ... this one.
17032 (remap_hint_predicate): Update.
17033 (inline_merge_summary): Update.
17034 (inline_update_overall_summary): Update.
17035 (estimate_size_after_inlining): Update.
17036 (read_predicate): Rename to ...
17037 (predicate::stream_in): ... this one.
17038 (read_inline_edge_summary): Update.
17039 (write_predicate): Rename to ...
17040 (predicate::stream_out): ... this one.
17041 (write_inline_edge_summary): Update.
17042 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17043 (clause_t): Turn to uint32_t
17044 (predicate): Turn to class; implement constructor and operators
17046 (size_time_entry): Update.
17047 (inline_summary): Update.
17048 (inline_edge_summary): Update.
17050 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17052 * fold-const.c (fold_binary_loc): Move transformation...
17053 * match.pd (C - X CMP X): ... here.
17055 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17057 * config/sparc/sparc.c (sparc_option_override): Set function
17058 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17059 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17061 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17063 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17065 2017-05-18 Marek Polacek <polacek@redhat.com>
17068 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17069 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17070 gimple_assign_load_p.
17072 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17074 PR middle-end/80692
17075 * real.c (do_compare): Give decimal_do_compare preference over
17076 comparing just the signs.
17078 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17080 * doc/md.texi (Canonicalization of Instructions): Describe the
17081 canonical form of instructions that inherently set a condition
17084 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17086 PR middle-end/80775
17087 * tree-cfg.c: Move deletion of unreachable case statements to after
17088 the merging of consecutive case labels.
17090 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17092 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17093 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17094 restoring of callee-saved registers.
17096 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17098 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17099 * config/visium/visium.c (single_set_and_flags): Likewise.
17100 * config/visium/visium.md (Substitutions): Likewise.
17102 2017-05-17 Martin Liska <mliska@suse.cz>
17104 * cfg.c: Introduce dump_flags_t type and
17105 use it instead of int type.
17107 * cfghooks.c: Likewise.
17108 * cfghooks.h (struct cfg_hooks): Likewise.
17109 * cfgrtl.c: Likewise.
17110 * cfgrtl.h: Likewise.
17111 * cgraph.c (cgraph_node::get_body): Likewise.
17112 * coretypes.h: Likewise.
17113 * domwalk.c: Likewise.
17114 * domwalk.h: Likewise.
17115 * dumpfile.c (struct dump_option_value_info): Likewise.
17116 (dump_enable_all): Likewise.
17117 (dump_switch_p_1): Likewise.
17118 (opt_info_switch_p): Likewise.
17119 * dumpfile.h (enum tree_dump_index): Likewise.
17120 (struct dump_file_info): Likewise.
17121 * genemit.c: Likewise.
17122 * generic-match-head.c: Likewise.
17123 * gengtype.c (open_base_files): Likewise.
17124 * gimple-pretty-print.c: Likewise.
17125 * gimple-pretty-print.h: Likewise.
17126 * graph.c (print_graph_cfg): Likewise.
17127 * graphite-scop-detection.c (dot_all_sese): Likewise.
17128 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17129 * loop-unroll.c (report_unroll): Likewise.
17130 * passes.c (pass_manager::register_one_dump_file): Likewise.
17131 * print-tree.c: Likewise.
17132 * statistics.c: Likewise.
17133 * tree-cfg.c: Likewise.
17134 * tree-cfg.h: Likewise.
17135 * tree-dfa.c: Likewise.
17136 * tree-dfa.h: Likewise.
17137 * tree-dump.c (dump_function): Likewise.
17138 * tree-dump.h (struct dump_info): Likewise.
17139 * tree-pretty-print.c: Likewise.
17140 * tree-pretty-print.h: Likewise.
17141 * tree-ssa-live.c: Likewise.
17142 * tree-ssa-live.h: Likewise.
17143 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17144 * tree-vect-loop.c: Likewise.
17145 * tree-vect-slp.c: Likewise.
17147 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17148 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17150 PR tree-optimization/80457
17151 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17152 of all arguments to a statement as scalar_to_vec operations.
17153 (vectorizable_call): Adjust call to vect_model_simple_cost for
17155 (vectorizable_conversion): Likewise.
17156 (vectorizable_assignment): Likewise.
17157 (vectorizable_shift): Likewise.
17158 (vectorizable_operation): Likewise.
17159 (vectorizable_comparison): Likewise.
17160 (vect_is_simple_cond): Record the def types for operands.
17161 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17162 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17163 for statement argument count.
17165 2017-05-16 Carl Love <cel@us.ibm.com>
17167 * config/rs6000/rs6000-c: Add support for built-in functions
17168 vector unsigned long long vec_bperm (vector unsigned long long,
17169 vector unsigned char)
17170 vector signed long long vec_mule (vector signed int,
17172 vector unsigned long long vec_mule (vector unsigned int,
17173 vector unsigned int)
17174 vector signed long long vec_mulo (vector signed int,
17176 vector unsigned long long vec_mulo (vector unsigned int,
17177 vector unsigned int)
17178 vector signed char vec_sldw (vector signed char,
17179 vector signed char,
17181 vector unsigned char vec_sldw (vector unsigned char,
17182 vector unsigned char,
17184 vector signed short vec_sldw (vector signed short,
17185 vector signed short,
17187 vector unsigned short vec_sldw (vector unsigned short,
17188 vector unsigned short,
17190 vector signed int vec_sldw (vector signed int,
17193 vector unsigned int vec_sldw (vector unsigned int,
17194 vector unsigned int,
17196 vector signed long long vec_sldw (vector signed long long,
17197 vector signed long long,
17199 vector unsigned long long vec_sldw (vector unsigned long long,
17200 vector unsigned long long,
17202 * config/rs6000/rs6000-c: Add support for built-in functions
17203 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17204 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17205 * doc/extend.texi: Update the built-in documentation for the
17206 new built-in functions.
17208 2017-05-16 Marek Polacek <polacek@redhat.com>
17212 * tree.c (save_expr): Don't fold the expression.
17214 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17216 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17217 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17218 and (?*y,m). Update insn attributes.
17220 2017-05-16 Martin Liska <mliska@suse.cz>
17222 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17223 flags argument of print_gimple_stmt, print_gimple_expr,
17224 print_generic_stmt and print_generic_expr.
17225 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17226 * coretypes.h: Likewise.
17227 * except.c (dump_eh_tree): Likewise.
17228 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17229 * gimple-pretty-print.h: Likewise.
17230 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17231 (backprop::push_to_worklist): Likewise.
17232 (backprop::pop_from_worklist): Likewise.
17233 (backprop::process_use): Likewise.
17234 (backprop::intersect_uses): Likewise.
17235 (note_replacement): Likewise.
17236 * gimple-ssa-store-merging.c
17237 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17238 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17239 (pass_store_merging::execute): Likewise.
17240 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17241 (ssa_base_cand_dump_callback): Likewise.
17242 (dump_incr_vec): Likewise.
17243 (replace_refs): Likewise.
17244 (replace_mult_candidate): Likewise.
17245 (create_add_on_incoming_edge): Likewise.
17246 (create_phi_basis): Likewise.
17247 (insert_initializers): Likewise.
17248 (all_phi_incrs_profitable): Likewise.
17249 (introduce_cast_before_cand): Likewise.
17250 (replace_one_candidate): Likewise.
17251 * gimplify.c (gimplify_expr): Likewise.
17252 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17253 (set_rename): Likewise.
17254 (rename_uses): Likewise.
17255 (copy_loop_phi_nodes): Likewise.
17256 (add_close_phis_to_merge_points): Likewise.
17257 (copy_loop_close_phi_args): Likewise.
17258 (copy_cond_phi_args): Likewise.
17259 (graphite_copy_stmts_from_block): Likewise.
17260 (translate_pending_phi_nodes): Likewise.
17261 * graphite-poly.c (print_pdr): Likewise.
17262 (dump_gbb_cases): Likewise.
17263 (dump_gbb_conditions): Likewise.
17264 (print_scop_params): Likewise.
17265 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17266 (build_cross_bb_scalars_use): Likewise.
17267 (gather_bbs::before_dom_children): Likewise.
17268 * hsa-dump.c (dump_hsa_immed): Likewise.
17269 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17270 (get_replacement_map): Likewise.
17271 * ipa-inline-analysis.c (dump_condition): Likewise.
17272 (estimate_function_body_sizes): Likewise.
17273 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17274 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17275 * ipa-prop.c (ipa_dump_param): Likewise.
17276 (ipa_print_node_jump_functions_for_edge): Likewise.
17277 (ipa_modify_call_arguments): Likewise.
17278 (ipa_modify_expr): Likewise.
17279 (ipa_dump_param_adjustments): Likewise.
17280 (ipa_dump_agg_replacement_values): Likewise.
17281 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17282 * ipa-pure-const.c (check_stmt): Likewise.
17283 (pass_nothrow::execute): Likewise.
17284 * ipa-split.c (execute_split_functions): Likewise.
17285 * omp-offload.c (dump_oacc_loop_part): Likewise.
17286 (dump_oacc_loop): Likewise.
17287 * trans-mem.c (tm_log_emit): Likewise.
17288 (tm_memopt_accumulate_memops): Likewise.
17289 (dump_tm_memopt_set): Likewise.
17290 (dump_tm_memopt_transform): Likewise.
17291 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17292 (print_loop): Likewise.
17293 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17294 (chkp_gather_checks_info): Likewise.
17295 (chkp_get_check_result): Likewise.
17296 (chkp_remove_check_if_pass): Likewise.
17297 (chkp_use_outer_bounds_if_possible): Likewise.
17298 (chkp_reduce_bounds_lifetime): Likewise.
17299 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17300 (chkp_mark_completed_bounds): Likewise.
17301 (chkp_register_incomplete_bounds): Likewise.
17302 (chkp_mark_invalid_bounds): Likewise.
17303 (chkp_maybe_copy_and_register_bounds): Likewise.
17304 (chkp_build_returned_bound): Likewise.
17305 (chkp_get_bound_for_parm): Likewise.
17306 (chkp_build_bndldx): Likewise.
17307 (chkp_get_bounds_by_definition): Likewise.
17308 (chkp_generate_extern_var_bounds): Likewise.
17309 (chkp_get_bounds_for_decl_addr): Likewise.
17310 * tree-chrec.c (chrec_apply): Likewise.
17311 * tree-data-ref.c (dump_data_reference): Likewise.
17312 (dump_subscript): Likewise.
17313 (dump_data_dependence_relation): Likewise.
17314 (analyze_overlapping_iterations): Likewise.
17315 * tree-inline.c (expand_call_inline): Likewise.
17316 (tree_function_versioning): Likewise.
17317 * tree-into-ssa.c (dump_defs_stack): Likewise.
17318 (dump_currdefs): Likewise.
17319 (dump_names_replaced_by): Likewise.
17320 (dump_update_ssa): Likewise.
17321 (update_ssa): Likewise.
17322 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17323 * tree-parloops.c (build_new_reduction): Likewise.
17324 (try_create_reduction_list): Likewise.
17325 (ref_conflicts_with_region): Likewise.
17326 (oacc_entry_exit_ok_1): Likewise.
17327 (oacc_entry_exit_single_gang): Likewise.
17328 * tree-pretty-print.h: Likewise.
17329 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17330 (get_scalar_evolution): Likewise.
17331 (add_to_evolution): Likewise.
17332 (get_loop_exit_condition): Likewise.
17333 (analyze_evolution_in_loop): Likewise.
17334 (analyze_initial_condition): Likewise.
17335 (analyze_scalar_evolution): Likewise.
17336 (instantiate_scev): Likewise.
17337 (number_of_latch_executions): Likewise.
17338 (gather_chrec_stats): Likewise.
17339 (final_value_replacement_loop): Likewise.
17340 (scev_const_prop): Likewise.
17341 * tree-sra.c (dump_access): Likewise.
17342 (disqualify_candidate): Likewise.
17343 (create_access): Likewise.
17344 (reject): Likewise.
17345 (maybe_add_sra_candidate): Likewise.
17346 (create_access_replacement): Likewise.
17347 (analyze_access_subtree): Likewise.
17348 (analyze_all_variable_accesses): Likewise.
17349 (sra_modify_assign): Likewise.
17350 (initialize_constant_pool_replacements): Likewise.
17351 (find_param_candidates): Likewise.
17352 (decide_one_param_reduction): Likewise.
17353 (replace_removed_params_ssa_names): Likewise.
17354 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17355 * tree-ssa-copy.c (dump_copy_of): Likewise.
17356 (copy_prop_visit_cond_stmt): Likewise.
17357 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17358 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17359 (record_equivalences_from_stmt): Likewise.
17360 * tree-ssa-dse.c (compute_trims): Likewise.
17361 (delete_dead_call): Likewise.
17362 (delete_dead_assignment): Likewise.
17363 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17364 (forward_propagate_into_cond): Likewise.
17365 (pass_forwprop::execute): Likewise.
17366 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17367 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17369 (move_computations_worker): Likewise.
17370 (execute_sm): Likewise.
17371 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17372 (remove_exits_and_undefined_stmts): Likewise.
17373 (remove_redundant_iv_tests): Likewise.
17374 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17375 (adjust_iv_update_pos): Likewise.
17376 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17377 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17378 (value_replacement): Likewise.
17379 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17380 * tree-ssa-pre.c (print_pre_expr): Likewise.
17381 (get_representative_for): Likewise.
17382 (create_expression_by_pieces): Likewise.
17383 (insert_into_preds_of_block): Likewise.
17384 (eliminate_insert): Likewise.
17385 (eliminate_dom_walker::before_dom_children): Likewise.
17386 (eliminate): Likewise.
17387 (remove_dead_inserted_code): Likewise.
17388 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17389 * tree-ssa-reassoc.c (get_rank): Likewise.
17390 (eliminate_duplicate_pair): Likewise.
17391 (eliminate_plus_minus_pair): Likewise.
17392 (eliminate_not_pairs): Likewise.
17393 (undistribute_ops_list): Likewise.
17394 (eliminate_redundant_comparison): Likewise.
17395 (update_range_test): Likewise.
17396 (optimize_range_tests_var_bound): Likewise.
17397 (optimize_vec_cond_expr): Likewise.
17398 (rewrite_expr_tree): Likewise.
17399 (rewrite_expr_tree_parallel): Likewise.
17400 (linearize_expr): Likewise.
17401 (break_up_subtract): Likewise.
17402 (linearize_expr_tree): Likewise.
17403 (attempt_builtin_powi): Likewise.
17404 (attempt_builtin_copysign): Likewise.
17405 (transform_stmt_to_copy): Likewise.
17406 (transform_stmt_to_multiply): Likewise.
17407 (dump_ops_vector): Likewise.
17408 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17409 (print_scc): Likewise.
17410 (set_ssa_val_to): Likewise.
17411 (visit_reference_op_store): Likewise.
17412 (visit_use): Likewise.
17413 (sccvn_dom_walker::before_dom_children): Likewise.
17414 (run_scc_vn): Likewise.
17415 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17417 (expr_hash_elt::print): Likewise.
17418 (const_and_copies::pop_to_marker): Likewise.
17419 (const_and_copies::record_const_or_copy_raw): Likewise.
17420 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17421 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17422 (dump_predicates): Likewise.
17423 (find_uninit_use): Likewise.
17424 (warn_uninitialized_phi): Likewise.
17425 (pass_late_warn_uninitialized::execute): Likewise.
17426 * tree-ssa.c (verify_vssa): Likewise.
17427 (verify_ssa): Likewise.
17428 (maybe_optimize_var): Likewise.
17429 * tree-vrp.c (dump_value_range): Likewise.
17430 (dump_all_value_ranges): Likewise.
17431 (dump_asserts_for): Likewise.
17432 (register_edge_assert_for_2): Likewise.
17433 (vrp_visit_cond_stmt): Likewise.
17434 (vrp_visit_switch_stmt): Likewise.
17435 (vrp_visit_stmt): Likewise.
17436 (vrp_visit_phi_node): Likewise.
17437 (simplify_cond_using_ranges_1): Likewise.
17438 (fold_predicate_in): Likewise.
17439 (evrp_dom_walker::before_dom_children): Likewise.
17440 (evrp_dom_walker::push_value_range): Likewise.
17441 (evrp_dom_walker::pop_value_range): Likewise.
17442 (execute_early_vrp): Likewise.
17444 2017-05-16 Richard Biener <rguenther@suse.de>
17446 * dwarf2out.c (loc_list_from_tree_1): Do not create
17447 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17449 2017-05-16 Richard Biener <rguenther@suse.de>
17451 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17453 (note_variable_value_in_expr): If we resolved the decl ref
17454 do not push to the stack.
17456 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17458 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17459 operations in fast-math mode.
17460 (vaddq_f16): Likewise.
17461 (vmul_f16): Likewise.
17462 (vmulq_f16): Likewise.
17463 (vsub_f16): Likewise.
17464 (vsubq_f16): Likewise.
17465 * config/arm/neon.md (add<mode>3): New.
17467 (fma:<VH:mode>3): New. Also remove outdated comment.
17470 2017-05-16 Martin Liska <mliska@suse.cz>
17474 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17475 (odr_types_equivalent_p): Likewise.
17477 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17479 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17481 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17484 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17485 non-interunit SSE move alternatives with '?'.
17486 (zero-extendsidi peephole2): New peephole to skip intermediate
17487 general register in SSE zero-extend sequence.
17489 2017-05-15 Jeff Law <law@redhat.com>
17491 * reorg.c (relax_delay_slots): Create a new variable to hold
17492 the temporary target rather than clobbering TARGET_LABEL.
17494 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17495 missing argument to extract_bit_field call.
17496 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17498 2017-05-15 Martin Liska <mliska@suse.cz>
17501 * gcc.c (process_command): Do not allow empty argument of -o option.
17503 2017-05-15 Renlin Li <renlin.li@arm.com>
17505 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17506 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17507 * config/aarch64/constraints.md (Usf): Add long call check.
17508 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17509 (call_value): Likewise.
17510 (sibcall): Likewise.
17511 (sibcall_value): Likewise.
17513 (call_value_insn): New.
17514 (sibcall_insn): Update rtx pattern.
17515 (sibcall_value_insn): Likewise.
17516 (call_internal): Remove.
17517 (call_value_internal): Likewise.
17518 (sibcall_internal): Likewise.
17519 (sibcall_value_internal): Likewise.
17520 (call_reg): Likewise.
17521 (call_symbol): Likewise.
17522 (call_value_reg): Likewise.
17523 (call_value_symbol): Likewise.
17525 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17528 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17530 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17532 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17533 compatible with CCGOCmode and with CCZmode.
17535 2017-05-14 Martin Sebor <msebor@redhat.com>
17537 PR middle-end/77671
17538 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17539 (gimple_fold_builtin_snprintf): Same.
17540 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17541 (gimple_fold_builtin_snprintf): Same.
17542 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17543 of character types.
17544 (is_call_safe): New function.
17545 (try_substitute_return_value): Call it.
17546 (try_simplify_call): New function.
17547 (pass_sprintf_length::handle_gimple_call): Call it.
17549 2017-05-14 Martin Sebor <msebor@redhat.com>
17551 PR middle-end/80669
17552 * builtins.c (expand_builtin_stpncpy): Simplify.
17554 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17556 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17557 * config/i386/i386.h
17558 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17559 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17560 (struct machine_function): Add new members call_ms2sysv,
17561 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17562 (struct machine_frame_state): New fields sp_realigned and
17563 sp_realigned_offset.
17564 * config/i386/i386.c
17565 (enum xlogue_stub): New enum.
17566 (enum xlogue_stub_sets): New enum.
17567 (class xlogue_layout): New class.
17568 (struct ix86_frame): New fields stack_realign_allocate_offset,
17569 stack_realign_offset and outlined_save_offset. Modify comments to
17570 detail stack layout when using out-of-line stubs.
17571 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17572 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17573 -mcall-ms2sysv-xlogues.
17574 (stub_managed_regs): New static variable.
17575 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17576 registers managed by out-of-line stub.
17577 (disable_call_ms2sysv_xlogues): New function.
17578 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17579 m->call_ms2sysv when appropriate and compute frame layout for
17581 (sp_valid_at, fp_valid_at): New inline functions.
17582 (choose_basereg): New function.
17583 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17585 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17586 Use align parameter of choose_baseaddr to generated aligned SSE movs
17588 (pro_epilogue_adjust_stack): Modify to track
17589 machine_frame_state::sp_realigned.
17590 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17591 (ix86_nsaved_sseregs): Likewise.
17592 (ix86_emit_save_regs): Likewise.
17593 (ix86_emit_save_regs_using_mov): Likewise.
17594 (ix86_emit_save_sse_regs_using_mov): Likewise.
17595 (get_scratch_register_on_entry): Likewise.
17596 (gen_frame_set): New function.
17597 (gen_frame_load): Likewise.
17598 (gen_frame_store): Likewise.
17599 (emit_outlined_ms2sysv_save): Likewise.
17600 (emit_outlined_ms2sysv_restore): Likewise.
17601 (ix86_expand_prologue): Modify stack re-alignment code and call
17602 emit_outlined_ms2sysv_save when appropriate.
17603 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17604 parameter rtx_insn *insn, which allows the function to be used to only
17605 generate the notes.
17606 (ix86_expand_epilogue): Modify validity checks of frame and stack
17607 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17608 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17609 * config/i386/predicates.md
17610 (save_multiple): New predicate.
17611 (restore_multiple): Likewise.
17612 * config/i386/sse.md
17613 (save_multiple<mode>): New pattern.
17614 (save_multiple_realign<mode>): Likewise.
17615 (restore_multiple<mode>): Likewise.
17616 (restore_multiple_and_return<mode>): Likewise.
17617 (restore_multiple_leave_return<mode>): Likewise.
17618 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17620 2017-05-14 Julia Koval <julia.koval@intel.com>
17622 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17623 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17624 (__builtin_ia32_xsetbv): New builtins.
17625 * config/i386/i386.c (ix86_expand_special_args_builtin):
17627 (ix86_expand_builtin): Special expand for new intrinsics.
17628 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17629 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17630 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17632 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17634 * cfganal.c (inverted_post_order_compute): Change argument type
17636 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17637 * df-core.c (rest_of_handle_df_initialize): Adjust.
17638 (rest_of_handle_df_finish): Likewise.
17639 (df_analyze_1): Likewise.
17640 (df_analyze): Likewise.
17641 (loop_inverted_post_order_compute): Change argument to be a vec *.
17642 (df_analyze_loop): Adjust.
17643 (df_get_n_blocks): Likewise.
17644 (df_get_postorder): Likewise.
17645 * df.h (struct df_d): Change field to be a vec.
17646 * lcm.c (compute_laterin): Adjust.
17647 (compute_available): Likewise.
17648 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17649 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17650 * tree-ssa-pre.c (compute_antic): Likewise.
17652 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17654 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17655 (depth_first_search::depth_first_search): Change structure init
17656 function to this constructor.
17657 (depth_first_search::add_bb): Rename function to this member.
17658 (depth_first_search::execute): Likewise.
17659 (flow_dfs_compute_reverse_finish): Adjust.
17661 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17663 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17664 (longest_simple_path): Likewise.
17665 * shrink-wrap.c (spread_components): Likewise.
17666 (disqualify_problematic_components): Likewise.
17667 (emit_common_heads_for_components): Likewise.
17668 (emit_common_tails_for_components): Likewise.
17669 (insert_prologue_epilogue_for_components): Likewise.
17671 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17673 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17678 * df-core.c (df_set_blocks): Start using auto_bitmap.
17679 (df_compact_blocks): Likewise.
17680 * df-problems.c (df_rd_confluence_n): Likewise.
17681 * df-scan.c (df_insn_rescan_all): Likewise.
17682 (df_process_deferred_rescans): Likewise.
17683 (df_update_entry_block_defs): Likewise.
17684 (df_update_exit_block_uses): Likewise.
17685 (df_entry_block_bitmap_verify): Likewise.
17686 (df_exit_block_bitmap_verify): Likewise.
17687 (df_scan_verify): Likewise.
17688 * lra-constraints.c (lra_constraints): Likewise.
17689 (undo_optional_reloads): Likewise.
17690 (lra_undo_inheritance): Likewise.
17691 * lra-remat.c (calculate_gen_cands): Likewise.
17692 (do_remat): Likewise.
17693 * lra-spills.c (assign_spill_hard_regs): Likewise.
17694 (spill_pseudos): Likewise.
17695 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17696 (bitmap_set_subtract_values): Likewise.
17698 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17700 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17701 management with auto_bitmap.
17702 (fix_inter_tick): Likewise.
17703 (fix_recovery_deps): Likewise.
17704 * ira.c (add_store_equivs): Likewise.
17705 (find_moveable_pseudos): Likewise.
17706 (split_live_ranges_for_shrink_wrap): Likewise.
17707 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17708 (rtx_reuse_manager::seen_def_p): Likewise.
17709 (rtx_reuse_manager::set_seen_def): Likewise.
17710 * print-rtl.h (class rtx_reuse_manager): Likewise.
17712 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17714 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17716 (migrate_btr_def): Likewise.
17717 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17718 * df-core.c (loop_post_order_compute): Likewise.
17719 (loop_inverted_post_order_compute): Likewise.
17720 * hsa-common.h: Likewise.
17721 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17722 * init-regs.c (initialize_uninitialized_regs): Likewise.
17723 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17724 (inline_small_functions): Likewise.
17725 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17726 * ira.c (combine_and_move_insns): Likewise.
17727 (build_insn_chain): Likewise.
17728 * loop-invariant.c (find_invariants): Likewise.
17729 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17730 * predict.c (tree_predict_by_opcode): Likewise.
17731 (predict_paths_leading_to): Likewise.
17732 (predict_paths_leading_to_edge): Likewise.
17733 (estimate_loops_at_level): Likewise.
17734 (estimate_loops): Likewise.
17735 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17736 (spread_components): Likewise.
17737 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17738 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17739 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17740 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17741 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17742 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17743 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17744 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17745 (mark_threaded_blocks): Likewise.
17746 (thread_through_all_blocks): Likewise.
17747 * tree-ssa.c (verify_ssa): Likewise.
17748 (execute_update_addresses_taken): Likewise.
17749 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17751 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17753 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17755 (post_order_compute): Likewise.
17756 (inverted_post_order_compute): Likewise.
17757 (pre_and_rev_post_order_compute_fn): Likewise.
17759 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17761 * genrecog.c (int_set::int_set): Explicitly construct our
17762 auto_vec base class.
17763 * vec.h (auto_vec::auto_vec): New constructor.
17765 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17767 * bitmap.h (class auto_bitmap): New constructor taking
17768 bitmap_obstack * argument.
17770 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17772 * bitmap.h (class auto_bitmap): Change type of m_bits to
17773 bitmap_head, and adjust ctor / dtor and member operators.
17775 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
17777 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
17778 when returned register mode doesn't match original mode.
17780 2017-05-12 Jeff Law <law@redhat.com>
17781 Jakub Jelinek <jakub@redhat.com>
17783 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
17784 we look for cc setter after the compare-elim changes.
17785 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
17786 within the vector to match what compare-elim now expects.
17787 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
17788 (xorsi3_flags, one_cmplsi2_flags): Likewise.
17790 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17791 after the compare-elim changes.
17792 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17793 the vector to match what compare-elim now expects.
17794 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17795 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17796 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17797 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17798 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17800 * config/visium/visium.c (single_set_and_flags): Fix where
17801 we look for cc setter after the compare-elim changes.
17802 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17803 with the vector to match what compare-elim now expects.
17804 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17805 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17806 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17807 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17808 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17809 (neg<mode>2_insn_set_overflow): Likewise.
17811 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17813 PR middle-end/79794
17814 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17815 maybe_expand_insn call, set ops[0].target. If still set after call,
17816 set alt_rtl. Add extra arg to recursive calls.
17817 (extract_bit_field): Add alt_rtl argument. Pass to
17819 * expmed.h (extract_bit_field): Fix prototype.
17820 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17821 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17822 to extract_bit_field_calls.
17823 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17824 Pass alt_rtl to extract_bit_field calls.
17825 * calls.c (store_unaligned_arguments_into_psuedos)
17826 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17827 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17829 * optabs.h (struct expand_operand): Add target bitfield.
17831 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17833 * compare-elim.c (try_eliminate_compare): Canonicalize
17834 operation with embedded compare to
17835 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17836 (set (reg) (operation)].
17838 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17840 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17843 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17844 cost of adding a carry flag for ADC instruction.
17845 [case MINUS]: Ignore the cost of subtracting a carry flag
17846 for SBB instruction.
17848 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17850 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17852 * config/rs6000/bmiintrin.h: New file.
17853 * config/rs6000/bmi2intrin.h: New file.
17854 * config/rs6000/x86intrin.h: New file.
17856 2017-05-12 Jeff Law <law@redhat.com>
17858 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17861 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17863 PR middle-end/80707
17864 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17866 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17869 early expansion of vector divide builtins.
17870 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17871 builtins identified as having unsigned arguments.
17873 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17875 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17876 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17877 expansion of vector logical operations (and, andc, or, xor,
17880 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17882 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17883 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17885 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17887 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17888 early GIMPLE expansion of vector multiplies.
17890 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17892 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17893 TARGET_HAVE_MOVT conditional.
17894 (movt splitter): Likewise.
17896 2017-05-12 Richard Biener <rguenther@suse.de>
17898 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17899 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17900 Fold all stmts not inplace.
17902 2017-05-12 Richard Biener <rguenther@suse.de>
17904 PR tree-optimization/80713
17905 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17906 inserted_exprs bit for not removed stmts.
17908 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17910 PR middle-end/69921
17911 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17912 parallelized" attribute for parallelized OpenACC kernels.
17913 * omp-offload.c (execute_oacc_device_lower): Use it.
17915 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17916 Set "oacc kernels" attribute.
17917 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17918 parameter. Adjust all users.
17919 (oacc_fn_attrib_kernels_p): Remove function.
17920 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17921 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17922 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17923 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17924 assert "oacc kernels" attribute is set.
17926 2017-05-11 Carl Love <cel@us.ibm.com>
17928 * config/rs6000/rs6000-c: Add support for built-in functions
17929 vector unsigned char vec_popcnt (vector signed char)
17930 vector unsigned char vec_popcnt (vector unsigned char)
17931 vector unsigned short vec_popcnt (vector signed short)
17932 vector unsigned short vec_popcnt (vector unsigned short)
17933 vector unsigned int vec_popcnt (vector signed int)
17934 vector unsigned int vec_popcnt (vector unsigned int)
17935 vector unsigned long long vec_popcnt (vector signed long long)
17936 vector unsigned long long vec_popcnt (vector unsigned long long)
17937 vector signed long long vec_slo (vector signed long long,
17938 vector signed char)
17939 vector signed long long vec_slo (vector signed long long,
17940 vector unsigned char)
17941 vector unsigned long long vec_slo (vector unsigned long long,
17942 vector signed char)
17943 vector unsigned long long vec_slo (vector unsigned long long,
17944 vector unsigned char)
17945 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17946 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17947 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17948 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17949 * doc/extend.texi: Update the built-in documentation file for the
17950 new built-in functions.
17952 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17954 * attribs.h (sorted_attr_string): Move machine independent
17955 functions for target clone support from the i386 port to common
17956 code. Rename ix86_function_versions to common_function_versions.
17957 Rename make_name to make_unique_name.
17958 (common_function_versions): Likewise.
17959 (make_unique_name): Likewise.
17960 (make_dispatcher_decl): Likewise.
17961 (is_function_default_version): Likewise.
17962 * attribs.c (attr_strcmp): Likewise.
17963 (sorted_attr_string): Likewise.
17964 (common_function_versions): Likewise.
17965 (make_unique_name): Likewise.
17966 (make_dispatcher_decl): Likewise.
17967 (is_function_default_version): Likewise.
17968 * config/i386/i386.c (attr_strcmp): Likewise.
17969 (sorted_attr_string): Likewise.
17970 (ix86_function_versions): Likewise.
17971 (make_name): Likewise.
17972 (make_dispatcher_decl): Likewise.
17973 (is_function_default_version): Likewise.
17974 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17976 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17979 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17980 Account for direct move costs for vec_construct of integer
17983 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17986 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17987 (UNSPEC_STX_ATOMIC): Ditto.
17988 (loaddi_via_sse): New insn.
17989 (storedi_via_sse): Ditto.
17990 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17991 Update corresponding peephole2 patterns.
17992 (atomic_storedi_fpu): Ditto.
17994 2017-05-11 Julia Koval <julia.koval@intel.com>
17996 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17997 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17999 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18000 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18001 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18003 2017-05-11 Nathan Sidwell <nathan@acm.org>
18005 * graphite-poly.c: Include dumpfile.h.
18007 * dumpfle.h (dump_function): Declare here ...
18008 * tree-dump.h (dump_function): ... not here.
18009 * dumpfile.c: #include tree-cfg.h.
18010 (dump_function): Move here from ...
18011 * tree-dump.c (dump_function): ... here.
18012 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18013 * graphite-poly.c: Don't include tree-dump.h.
18014 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18015 * print-tree.c: Likewise.
18016 * stor-layout.c: Likewise.
18017 * tree-nested.c: Likewise.
18019 * dumpfile.c (dump_start): Use TDF_FLAGS.
18020 (dump_enable_all): Fix TDF_KIND check thinko.
18022 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18024 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18025 array entries to represent two legal parameterizations of the
18026 overloaded __builtin_cmpb function, as represented by the
18027 P6_OV_BUILTIN_CMPB constant.
18028 (altivec_resolve_overloaded_builtin): Add special case handling
18029 for the __builtin_cmpb function, as represented by the
18030 P6_OV_BUILTIN_CMPB constant.
18031 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18032 (BU_P6_64BIT_2): New macro.
18033 (BU_P6_OVERLOAD_2): New macro
18034 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18035 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18036 (CMPB): Add overload support to represent both 32-bit and 64-bit
18037 compare-bytes function.
18038 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18039 support for TARGET_CMPB.
18040 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18041 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18042 documentation of the __builtin_cmpb overloaded built-in function.
18044 2017-05-11 Richard Biener <rguenther@suse.de>
18046 PR tree-optimization/80705
18047 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18048 bases are not vectorizable.
18050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18052 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18053 when counting register pressure.
18055 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18057 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18058 (iv_ca_more_deps): Renamed to ...
18059 (iv_ca_compare_deps): ... this.
18060 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18064 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18066 (determine_group_iv_costs): ... here.
18067 (find_inv_vars_cb): Record inv var if it's not recorded before.
18069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18071 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18072 (get_shiftadd_cost): Ditto.
18074 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18076 * tree-ssa-address.c: Include header file.
18077 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18079 (add_to_parts): Refactor.
18080 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18081 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18084 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18086 PR tree-optimization/53090
18087 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18089 (extract_cond_operands): Detect condition with IV on both sides
18090 and return COMP_IV_EXPR_2.
18091 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18092 (rewrite_use_compare): Simplify by removing call to function
18093 extract_cond_operands.
18095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18097 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18098 (extract_cond_operands): Detect condition comparing against non-
18099 invariant bound and return appropriate enum value.
18100 (find_interesting_uses_cond): Update use of extract_cond_operands.
18101 Handle its return value accordingly.
18102 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18104 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18106 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18107 nonlinear iv_use computation in loop invariant sensitive way.
18109 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18111 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18112 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18114 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18116 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18117 (dump_cand): Support iv_cand.inv_exprs.
18118 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18120 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18125 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18127 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18128 as local function. Include necessary header files.
18129 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18131 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18133 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18137 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18138 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18139 RSHIFT_EXPR and BIT_NOT_EXPR.
18141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18143 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18144 (adjust_setup_cost): New parameter supporting round up adjustment.
18145 (struct address_cost_data): Delete.
18146 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18147 (split_address_cost, ptr_difference_cost): Delete.
18148 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18149 (struct ainc_cost_data): New struct.
18150 (get_address_cost_ainc): New function.
18151 (get_address_cost, get_computation_cost): Reimplement.
18152 (determine_group_iv_cost_address): Record inv_expr for all uses of
18154 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18155 (iv_ca_has_deps): Reimplemented to ...
18156 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18158 (iv_ca_extend): Call iv_ca_more_deps.
18160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18162 * tree-ssa-address.c (struct mem_address): Move to header file.
18163 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18164 * tree-ssa-address.h (struct mem_address): Move from C file.
18165 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18169 * tree-affine.h (aff_combination_type): New interface.
18170 (aff_combination_zero_p): Remove static.
18171 (aff_combination_const_p): New interface.
18172 (aff_combination_singleton_var_p): New interfaces.
18174 2017-05-11 Richard Biener <rguenther@suse.de>
18176 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18177 Skip unreachable blocks and destinations.
18178 (eliminate): Move stmt removal and fixup ...
18179 (fini_eliminate): ... here. Skip inserted exprs.
18180 (pass_pre::execute): Move fini_pre after fini_eliminate.
18181 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18182 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18183 PRE to get rid of dead code that has invalid SSA form and
18184 split critical edges again.
18186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18188 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18190 2017-05-11 Richard Biener <rguenther@suse.de>
18192 * passes.c (execute_function_todo): Verify loops if they are
18193 said to be up-to-date.
18194 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18195 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18197 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18200 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18201 handle calling assemble_external ourself.
18204 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18205 modes with zero size. Enhance comment.
18207 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18209 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18210 built-ins for vec_xl and vec_xst with short and char pointer
18213 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18215 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18216 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18217 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18218 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18219 (_mm_maskz_min_round_ss): New intrinsics.
18220 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18221 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18222 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18223 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18224 (__builtin_ia32_minss_mask_round): New builtins.
18225 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18226 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18227 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18229 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18230 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18232 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18235 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18237 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18238 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18239 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18240 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18241 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18242 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18243 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18244 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18245 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18246 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18247 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18248 (__builtin_ia32_mulss_mask_round): New builtins.
18249 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18250 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18251 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18253 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18254 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18256 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18259 2017-05-10 Julia Koval <julia.koval@intel.com>
18261 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18262 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18263 (_mm256_setr_m128i): New intrinsics.
18265 2017-05-10 Julia Koval <julia.koval@intel.com>
18267 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18268 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18269 (_mm_maskz_rcp14_ss): New intrinsics.
18270 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18271 (__builtin_ia32_rcp14ss_mask): New builtins.
18272 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18274 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18276 PR tree-optimization/51513
18277 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18278 (assert_unreachable_fallthru_edge_p): Use it.
18279 (group_case_labels_stmt): Likewise.
18280 * tree-cfg.h: Prototype it.
18281 * stmt.c: Include cfghooks.h and tree-cfg.h.
18282 (emit_case_dispatch_table) <gap_label>: New local variable.
18283 Use it to fill dispatch table gaps.
18284 Test for default_label before updating probabilities.
18285 (expand_case) <default_label>: Remove unneeded initialization.
18286 Test for unreachable default case statement and remove its edge.
18287 Set default_label accordingly.
18288 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18290 2017-05-10 Carl Love <cel@us.ibm.com>
18292 * config/rs6000/rs6000-c: Add support for built-in functions
18293 vector signed char vec_neg (vector signed char)
18294 vector signed short int vec_neg (vector short int)
18295 vector signed int vec_neg (vector signed int)
18296 vector signed long long vec_neg (vector signed long long)
18297 vector float vec_neg (vector float)
18298 vector double vec_neg (vector double)
18299 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18301 * config/rs6000/altivec.h: Add define for vec_neg
18302 * doc/extend.texi: Update the built-in documentation for the
18303 new built-in functions.
18305 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18307 PR tree-optimization/77644
18308 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18310 2017-05-10 Nathan Sidwell <nathan@acm.org>
18312 * dumpfile.h (TDI_lang_all): New.
18313 (TDF_KIND): New. Renumber others
18314 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18316 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18318 (get_dump_file_name): Adjust suffix generation.
18319 (dump_enable_all): Use TDF_KIND.
18320 * doc/invoke.texi (-fdump-lang-all): Document.
18322 * dumpfile.h: Tabify.
18324 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18327 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18328 Move member access before delete.
18330 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18332 * tree-inline.c (expand_call_inline): Split block at stmt
18335 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18338 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18339 are now unused after splitting mov{sf,sd}_hardfloat.
18340 (f32_lr2): Likewise.
18341 (f32_lm): Likewise.
18342 (f32_lm2): Likewise.
18343 (f32_li): Likewise.
18344 (f32_li2): Likewise.
18345 (f32_lv): Likewise.
18346 (f32_sr): Likewise.
18347 (f32_sr2): Likewise.
18348 (f32_sm): Likewise.
18349 (f32_sm2): Likewise.
18350 (f32_si): Likewise.
18351 (f32_si2): Likewise.
18352 (f32_sv): Likewise.
18353 (f32_dm): Likewise.
18354 (f32_vsx): Likewise.
18355 (f32_av): Likewise.
18356 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18357 For movsf, order stores so the VSX stores occur before the GPR
18358 store which encourages the register allocator to use a traditional
18359 FPR instead of a GPR. For movsd, order the stores so that the GPR
18360 store comes before the VSX stores to allow the power6 to work.
18361 This is due to the power6 not having a 32-bit integer store
18362 instruction from a FPR.
18363 (movsf_hardfloat): Likewise.
18364 (movsd_hardfloat): Likewise.
18366 2017-05-09 Martin Sebor <msebor@redhat.com>
18368 PR translation/80280
18369 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18372 PR translation/80280
18373 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18374 data member added in r247778.
18375 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18377 2017-05-09 Nathan Sidwell <nathan@acm.org>
18379 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18381 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18384 2017-05-09 Marek Polacek <polacek@redhat.com>
18386 * doc/invoke.texi: Fix typo.
18388 2017-05-09 Richard Biener <rguenther@suse.de>
18390 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18391 (vrp_val_is_min): Likewise.
18392 (set_value_range_to_value): Likewise.
18393 (set_value_range_to_nonnegative): Likewise.
18394 (gimple_assign_nonzero_p): Likewise.
18395 (gimple_stmt_nonzero_p): Likewise.
18396 (vrp_int_const_binop): Likewise. Remove unreachable case.
18397 (adjust_range_with_scev): Adjust comments.
18398 (compare_range_with_value): Likewise.
18399 (extract_range_from_phi_node): Likewise.
18400 (test_for_singularity): Likewise.
18402 2017-05-09 Richard Biener <rguenther@suse.de>
18404 * tree-vrp.c (get_single_symbol): Add assert that we don't
18405 get overflowed constants as invariant part.
18406 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18407 checks. Use wi::cmp instead of recursing for integer constants.
18408 (compare_values): Just ignore whether we assumed undefined
18409 overflow instead of failing the compare.
18410 (extract_range_for_var_from_comparison_expr): Add comment before the
18411 TREE_NO_WARNING sets.
18412 (test_for_singularity): Likewise.
18413 (extract_range_from_comparison): Do not disable optimization
18414 when we assumed undefined overflow.
18415 (extract_range_basic): Remove init of unused var.
18417 2017-05-09 Richard Biener <rguenther@suse.de>
18419 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18420 (extract_range_from_multiplicative_op_1): Adjust.
18421 (extract_range_from_binary_expr_1): Use int_const_binop.
18423 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18426 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18427 rs6000_store_data_bypass_p in seven define_bypass directives and
18428 in several comments.
18429 * config/rs6000/rs6000-protos.h: Add prototype for
18430 rs6000_store_data_bypass_p function.
18431 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18432 function implements slightly different (rs6000-specific) semantics
18433 than store_data_bypass_p, returning false rather than aborting
18434 with assertion error when arguments do not satisfy the
18435 requirements of store data bypass.
18436 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18437 rs6000_store_data_bypass_p.
18439 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18441 * config/xtensa/xtensa-protos.h
18442 (xtensa_initial_elimination_offset): New declaration.
18443 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18444 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18445 macro definition, add case for FRAME_POINTER_REGNUM when
18446 FRAME_GROWS_DOWNWARD.
18447 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18448 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18449 xtensa_initial_elimination_offset.
18451 2017-05-08 Nathan Sidwell <nathan@acm.org>
18453 * doc/invoke.texi: Alphabetize -fdump options.
18455 2017-05-08 Martin Sebor <msebor@redhat.com>
18457 PR translation/80280
18458 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18460 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18462 * target.def (compute_frame_layout): New optional target hook.
18463 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18464 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18465 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18467 * reload1.c (verify_initial_elim_offsets): Likewise.
18468 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18469 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18470 (arm_get_frame_offsets): Split up into this ...
18471 (arm_compute_frame_layout): ... and this function.
18473 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18475 * config/aarch64/constraints.md (Usa): New constraint.
18476 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18478 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18480 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18481 with_multilib_list after it has been checked.
18483 2017-05-08 Richard Biener <rguenther@suse.de>
18485 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18486 (bitmap_set_subtract_values): Likewise.
18488 2017-05-08 Richard Biener <rguenther@suse.de>
18490 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18491 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18493 (gimple_stmt_nonzero_warnv_p): Rename to ...
18494 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18496 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18497 (extract_range_basic): Adjust, do not disable propagation on
18498 strict overflow sensitive simplification.
18499 (vrp_visit_cond_stmt): Likewise.
18501 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18503 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18504 body size unconditionally.
18506 2017-05-07 Jeff Law <law@redhat.com>
18509 2017-05-06 Jeff Law <law@redhat.com>
18510 PR tree-optimization/78496
18511 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18514 PR tree-optimization/78496
18515 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18516 (simplify_stmt_using_ranges): Call it.
18517 (vrp_dom_walker::before_dom_children): Extract equivalences
18518 from an ASSERT_EXPR with an equality comparison against a
18521 2017-05-06 Jeff Law <law@redhat.com>
18523 PR tree-optimization/78496
18524 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18527 PR tree-optimization/78496
18528 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18529 (simplify_stmt_using_ranges): Call it.
18530 (vrp_dom_walker::before_dom_children): Extract equivalences
18531 from an ASSERT_EXPR with an equality comparison against a
18534 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18536 * lra-constraints.c (lra_copy_reg_equiv): New function.
18537 (split_reg): Use it to copy equivalence information from the
18538 original register to the spill register.
18540 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18542 PR rtl-optimization/75964
18543 * simplify-rtx.c (simplify_const_relational_operation): Remove
18544 invalid handling of comparisons of integer ABS.
18546 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18548 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18549 initialize to zero.
18550 (init_regs): Remove declaration.
18551 (function_arg_advance_32): Initialize error_p as boolean variable.
18553 2017-05-05 Nathan Sidwell <nathan@acm.org>
18555 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18556 lines. Use for (;;).
18558 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18560 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18561 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18562 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18563 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18564 VF=2 that require versioning.
18566 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18568 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18571 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18573 * diagnostic.h (diagnostic_override_option_index): Convert from
18574 macro to inline function.
18576 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18578 * diagnostic.c (last_module_changed_p): New function.
18579 (set_last_module): New function.
18580 (diagnostic_report_current_module): Convert macro usage to
18581 the above functions.
18582 * diagnostic.h (diagnostic_context::last_module): Strengthen
18583 from const line_map * to const line_map_ordinary *.
18584 (diagnostic_last_module_changed): Delete macro.
18585 (diagnostic_set_last_module): Delete macro.
18587 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18589 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18590 with diagnostic_report_diagnostic.
18591 (diagnostic_n_impl_richloc): Likewise.
18592 * diagnostic.h (report_diagnostic): Delete macro.
18593 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18594 with diagnostic_report_diagnostic.
18595 * substring-locations.c (format_warning_va): Likewise.
18597 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18599 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18600 save/restor of format_spec. Move option-printing code to...
18601 (print_option_information): ...this new function, and
18602 reimplement by simply printing to the pretty_printer,
18603 rather than appending to the format string.
18605 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18607 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18608 handling logic into...
18609 (update_effective_level_from_pragmas): ...this new function.
18611 2017-05-04 Andrew Waterman <andrew@sifive.com>
18613 * config/riscv/riscv.opt (mstrict-align): New option.
18614 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18615 (SLOW_UNALIGNED_ACCESS): Define.
18616 (riscv_slow_unaligned_access): Declare.
18617 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18619 (riscv_slow_unaligned_access): New variable.
18620 (rocket_tune_info): Set slow_unaligned_access to true.
18621 (optimize_size_tune_info): Set slow_unaligned_access to false.
18622 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18623 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18624 (riscv_option_override): Set riscv_slow_unaligned_access.
18625 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18627 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18629 * config/riscv/riscv.md: Unify indentation.
18631 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18636 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18638 (extendsi<mode>2): Add support for doing sign extension via
18639 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18640 don't have ISA 3.0 instructions.
18641 (extendsi<mode>2 splitter): Likewise.
18642 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18643 generate the normal insns since SImode can now go in vector
18644 registers. Disallow the special UNSPECs needed for previous
18645 machines to hide SImode being used. Add new insns
18646 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18647 (fix_trunc<mode>si2_stfiwx): Likewise.
18648 (fix_trunc<mode>si2_internal): Likewise.
18649 (fixuns_trunc<mode>si2): Likewise.
18650 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18651 (fctiw<u>z_<mode>_smallint): Likewise.
18652 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18653 of floating point to 32-bit integer from doing a direct move to
18654 the GPR registers to do a store.
18655 (fctiwz_<mode>): Break long line.
18657 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18659 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18660 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18661 (addr_list, addr_offset_valid_p): New.
18662 (split_address_groups): Check offset validity with above function.
18663 (gt-tree-ssa-loop-ivopts.h): Include header file.
18665 2017-05-05 Nathan Sidwell <nathan@acm.org>
18667 * config.gcc (arm*-*-*): Add missing 'fi'.
18669 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18671 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18672 included in -fopt-info does not matter.
18673 * doc/optinfo.texi (-fopt-info): Fix description of default
18674 behavour. Explicitly say order of options included in -fopt-info
18677 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18679 * config.gcc: Allow combinations of aprofile and rmprofile values for
18680 --with-multilib-list.
18681 * config/arm/t-multilib: New file.
18682 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18683 variables. Remove setting of ISA and floating-point ABI in
18684 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18685 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18686 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18687 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18689 * config/arm/t-rmprofile: Likewise except for the matches changes.
18690 * doc/install.texi (--with-multilib-list): Document the combination of
18691 aprofile and rmprofile values and warn about pitfalls in doing that.
18693 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18695 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18696 (movdi_aarch64): Likewise.
18698 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18700 PR tree-optimization/80632
18701 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18703 (build_arrays): Initialize it for virtual phis.
18704 (fix_phi_nodes): Use it for virtual phis.
18706 PR tree-optimization/80558
18707 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18708 [x, y] op z into [x op, y op z] for op & or | if conditions
18711 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18712 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18715 * config/arm/arm.md (use_literal_pool): Remove.
18716 (64-bit immediate split): No longer takes cost into consideration
18717 if arm_disable_literal_pool is enabled.
18718 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18719 used when arm_disable_literal_pool is enabled.
18720 (arm_max_const_double_inline_cost): Remove use of
18721 arm_disable_literal_pool.
18722 (push_minipool_fix): Add assert.
18723 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18724 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18725 (no_literal_pool_sf_immediate): New.
18727 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18729 PR tree-optimization/80613
18730 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18731 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18733 2017-05-05 Richard Biener <rguenther@suse.de>
18735 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18737 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18739 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18740 of this flag from insn conditions due to removal from r247495.
18742 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18744 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18746 (arm_early_store_addr_dep_ptr): Likewise.
18747 * config/arm/aarch-common-protos.h
18748 (arm_early_load_addr_dep_ptr): Add prototype.
18749 (arm_early_store_addr_dep_ptr): Likewise.
18750 * config/arm/cortex-a53.md: Add new bypasses.
18752 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18754 * tree.c (next_type_uid): Change type to unsigned.
18755 (type_hash_canon): Decrement back next_type_uid if
18756 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18757 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18760 2017-05-04 Martin Sebor <msebor@redhat.com>
18762 * builtins.c: Fix a trivial typo in a comment.
18764 PR middle-end/79234
18765 * builtins.c (check_sizes): Adjust to handle reading past the end.
18766 Avoid printing excessive upper bound of ranges. Use %E to print
18767 tree nodes instead of converting them to %wu.
18768 (expand_builtin_memchr): New function.
18769 (compute_dest_size): Rename...
18770 (compute_objsize): ...to this.
18771 (expand_builtin_memcpy): Adjust.
18772 (expand_builtin_mempcpy): Adjust.
18773 (expand_builtin_strcat): Adjust.
18774 (expand_builtin_strcpy): Adjust.
18775 (check_strncat_sizes): Adjust.
18776 (expand_builtin_strncat): Adjust.
18777 (expand_builtin_strncpy): Adjust and simplify.
18778 (expand_builtin_memset): Adjust.
18779 (expand_builtin_bzero): Adjust.
18780 (expand_builtin_memcmp): Adjust.
18781 (expand_builtin): Handle memcmp.
18782 (maybe_emit_chk_warning): Check strncat just once.
18784 2017-05-04 Martin Sebor <msebor@redhat.com>
18786 PR preprocessor/79214
18787 PR middle-end/79222
18788 PR middle-end/79223
18789 * builtins.c (check_sizes): Add inlining context and issue
18790 warnings even when -Wno-system-headers is set.
18791 (check_strncat_sizes): Same.
18792 (expand_builtin_strncat): Same.
18793 (expand_builtin_memmove): New function.
18794 (expand_builtin_stpncpy): Same.
18795 (expand_builtin): Handle memmove and stpncpy.
18797 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18799 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18800 which is not used any more.
18802 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18804 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18806 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18808 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18809 (cortexa53_tunings): Likewise.
18810 (cortexa57_tunings): Likewise.
18811 (cortexa72_tunings): Likewise.
18812 (cortexa73_tunings): Likewise.
18814 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18816 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18817 Set loop alignment to 8.
18819 2017-05-04 Martin Sebor <msebor@redhat.com>
18821 PR translation/80280
18822 * builtins.c (expand_builtin_object_size): Add missing quoting to
18823 %D and like directives.
18824 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18825 (hsa_type_for_tree_type): Same.
18826 (verify_function_arguments): Same.
18827 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18828 * varasm.c (get_section): Same.
18831 2017-05-04 Martin Sebor <msebor@redhat.com>
18833 PR translation/80280
18834 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18836 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18838 * config/aarch64/aarch64.c (generic_addrcost_table):
18839 Change HI/TI mode setting.
18841 2017-05-04 Martin Jambor <mjambor@suse.cz>
18843 PR tree-optimization/80622
18844 * tree-sra.c (comes_initialized_p): New function.
18845 (build_accesses_from_assign): Only set write lazily when
18846 comes_initialized_p is false.
18847 (analyze_access_subtree): Use comes_initialized_p.
18848 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18849 instead of testing for PARM_DECL.
18851 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18853 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18854 constraint on operand 0 to allow more general addressing modes.
18855 Adjust output template.
18856 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18858 * config/aarch64/aarch64-protos.h
18859 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18860 * config/aarch64/constraints.md (Dp): New address constraint.
18861 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18864 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18866 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18867 update use of estimate_ipcp_clone_size_and_time.
18868 (estimate_local_effects): Update use of
18869 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18870 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18871 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18872 Return nonspecialized time.
18874 2017-05-04 Richard Biener <rguenther@suse.de>
18876 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18877 for the last VUSE which def dominates the PHI. Directly call
18879 (get_continuation_for_phi_1): Remove.
18881 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18883 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18884 to explain the use of truncating division. Cap the number of
18885 iterations to the maximum given by nb_iterations_upper_bound,
18888 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18890 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18891 * configure: Regenerate.
18892 * config.in: Regenerate.
18893 * config/i386/driver-mingw32.c: new file.
18894 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18895 * config.host: Link driver-mingw32.o on MinGW host.
18896 * doc/install.texi: Document new --enable-mingw-wildcard configure
18899 2017-05-04 Marek Polacek <polacek@redhat.com>
18901 PR tree-optimization/80612
18902 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18904 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18905 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18907 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18908 (movt splitter): Likewise.
18909 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18910 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18911 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18912 block for Thumb-1 with MOVT.
18913 (thumb2_legitimate_address_p): Move code block ...
18914 (can_avoid_literal_pool_for_label_p): ... into this new function.
18915 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18917 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18918 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18919 "M-profile targets with the MOVT instruction".
18921 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18923 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18924 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18925 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18927 2017-05-04 Martin Liska <mliska@suse.cz>
18929 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18930 variable cond_code.
18932 2017-05-04 Richard Biener <rguenther@suse.de>
18934 * tree.c (array_at_struct_end_p): Handle arrays at struct
18935 end with flexarrays more conservatively. Refactor and treat
18936 arrays of arrays or aggregates more strict. Fix
18937 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18938 * tree.c (array_at_struct_end_p): Adjust prototype.
18939 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18940 * gimple-fold.c (get_range_strlen): Likewise.
18941 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18943 2017-05-04 Richard Biener <rguenther@suse.de>
18945 PR tree-optimization/31130
18946 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18948 (supports_overflow_infinity): Likewise.
18949 (is_negative_overflow_infinity): Likewise.
18950 (is_positive_overflow_infinity): Likewise.
18951 (is_overflow_infinity): Likewise.
18952 (stmt_overflow_infinity): Likewise.
18953 (overflow_infinity_range_p): Likewise.
18954 (usable_range_p): Remove as always returning true.
18955 (make_overflow_infinity): Remove.
18956 (negative_overflow_infinity): Likewise.
18957 (positive_overflow_infinity): Likewise.
18958 (avoid_overflow_infinity): Likewise.
18959 (set_value_range): Adjust accordingly.
18960 (set_value_range_to_nonnegative): Likewise, remove now unused
18961 overflow_infinity arg.
18962 (vrp_operand_equal_p): Adjust.
18963 (update_value_range): Likewise.
18964 (range_int_cst_singleton_p): Likewise.
18965 (operand_less_p): Likewise.
18966 (compare_values_warnv): Likewise.
18967 (extract_range_for_var_from_comparison_expr): Likewise.
18968 (vrp_int_const_binop): Likewise.
18969 (zero_nonzero_bits_from_vr): Likewise.
18970 (extract_range_from_multiplicative_op_1): Likewise.
18971 (extract_range_from_binary_expr_1): Likewise.
18972 (extract_range_from_unary_expr): Likewise.
18973 (extract_range_from_comparison): Likewise.
18974 (extract_range_basic): Likewise.
18975 (adjust_range_with_scev): Likewise.
18976 (compare_ranges): Likewise.
18977 (compare_range_with_value): Likewise.
18978 (dump_value_range): Likewise.
18979 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18981 (simplify_cond_using_ranges): Adjust.
18983 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18985 * brig-builtins.def: Added a builtin for class_f64.
18986 * builtin-types.def: Added a builtin type needed by class_f64.
18988 2017-05-03 Jason Merrill <jason@redhat.com>
18990 * timevar.def: Add TV_CONSTEXPR.
18992 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18994 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18996 2017-05-03 Martin Jambor <mjambor@suse.cz>
18998 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18999 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19000 * ipa-cp.c (ipcp_propagate_stage): Do not call
19001 ipa_update_after_lto_read.
19002 * ipa-inline.c (ipa_inline): Likewise.
19004 2017-05-03 Martin Jambor <mjambor@suse.cz>
19006 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19007 tag. Added a default constructor and a destructor.
19008 (ipa_edge_args_sum_t): New class;
19009 (ipa_edge_args_sum): Declare.
19010 (ipa_edge_args_vector): Remove declaration.
19011 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19012 (ipa_free_edge_args_substructures): Remove declaration.
19013 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19014 (ipa_edge_args_info_available_for_edge_p): Likewise.
19015 * ipa-prop.c (ipa_edge_args_vector): Removed.
19016 (edge_removal_hook_holder): Likewise.
19017 (edge_duplication_hook_holder): Likewise.
19018 (ipa_edge_args_sum): New variable.
19019 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19020 ipa_edge_args_vector.
19021 (ipa_free_edge_args_substructures): Likewise.
19022 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19023 ipa_edge_args_vector.
19024 (ipa_edge_removal_hook): Turned into method
19025 ipa_edge_args_sum_t::remove.
19026 (ipa_edge_duplication_hook): Turned into method
19027 ipa_edge_args_sum_t::duplicate.
19028 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19029 registering edge hooks.
19030 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19031 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19032 ipa_edge_args_sum instead of ipa_edge_args_vector.
19033 * ipa-profile.c (ipa_profile): Likewise.
19035 2017-05-03 Martin Jambor <mjambor@suse.cz>
19037 * symbol-summary.h (function_summary): New method exists.
19038 (function_summary::symtab_removal): Deallocate through release.
19039 (call_summary): New class.
19040 (gt_ggc_mx): New overload.
19041 (gt_pch_nx): Likewise.
19042 (gt_pch_nx): Likewise.
19044 2017-05-03 Jeff Law <law@redhat.com>
19046 PR tree-optimization/78496
19047 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19048 from simplify_cond_using_ranges. Split off code to walk
19049 backwards through casts into ...
19050 (simplify_cond_using_ranges_2): New function.
19051 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19052 (execute_vrp): After identifying jump threads, call
19053 simplify_cond_using_ranges_2.
19055 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19058 * ipa-inline.h (inline_summary): Add ctor.
19059 (create_ggc): Do not use ggc_cleared_alloc.
19061 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19062 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19064 * gcc.c (handle_braces): Support escaping in switch matching
19066 * doc/invoke.texi (Spec Files): Document it.
19067 Remove superfluous @code markup in items.
19069 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19071 * diagnostic-show-locus.c (struct column_range): New struct.
19072 (get_affected_columns): New function.
19073 (get_printed_columns): New function.
19074 (struct correction): New struct.
19075 (correction::ensure_capacity): New function.
19076 (correction::ensure_terminated): New function.
19077 (struct line_corrections): New struct.
19078 (line_corrections::~line_corrections): New dtor.
19079 (line_corrections::add_hint): New function.
19080 (layout::print_trailing_fixits): Reimplement in terms of the new
19082 (selftest::test_overlapped_fixit_printing): New function.
19083 (selftest::diagnostic_show_locus_c_tests): Call it.
19085 2017-05-03 Nathan Sidwell <nathan@acm.org>
19087 Canonicalize canonical type hashing
19088 * tree.h (type_hash_canon_hash): Declare.
19089 * tree.c (type_hash_list, attribute_hash_list): Move into
19090 type_hash_canon_hash.
19091 (build_type_attribute_qual_variant): Break out hash code calc into
19092 type_hash_canon_hash.
19093 (type_hash_canon_hash): New. Generic type hash computation.
19094 (build_range_type_1, build_array_type_1, build_function_type,
19095 build_method_type_directly, build_offset_type, build_complex_type,
19096 make_vector_type): Call it.
19098 2017-05-03 Richard Biener <rguenther@suse.de>
19100 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19101 When all DRs have unknown misaligned do not always peel
19102 when there is a store but apply the same costing model as if
19103 there were only loads.
19105 2017-05-03 Richard Biener <rguenther@suse.de>
19108 PR tree-optimization/80492
19109 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19110 compare_base_decls returning dont-know properly.
19112 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19114 * config/arm/iterators.md (CCSI): New mode iterator.
19115 (arch): New mode attribute.
19116 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19117 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19118 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19119 code iterator for success result mode.
19120 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19121 the corresponding new insn generators.
19123 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19126 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19127 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19129 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19131 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19132 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19133 (DDR_A): Wrap DDR argument in brackets.
19134 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19135 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19136 (DDR_REVERSED_P): Likewise.
19138 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19140 PR tree-optimization/79472
19141 * tree-switch-conversion.c (struct switch_conv_info): Add
19142 contiguous_range and default_case_nonstandard fields.
19143 (collect_switch_conv_info): Compute contiguous_range and
19144 default_case_nonstandard fields, don't clear final_bb if
19145 contiguous_range and only the default case doesn't have the required
19147 (check_all_empty_except_final): Set default_case_nonstandard instead
19148 of failing if contiguous_range and the default case doesn't have empty
19150 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19151 and only the default case doesn't have the required constants. Skip
19153 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19154 if default_case_nonstandard.
19155 (build_constructors): Build constant 1 just once. Assert that default
19156 values aren't inserted in between cases if contiguous_range. Skip
19158 (build_arrays): Skip virtual phis.
19159 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19160 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19161 Handle virtual phis.
19162 (gen_inbound_check): Handle default_case_nonstandard case.
19163 (process_switch): Adjust check_final_bb caller. Call
19164 gather_default_values with the first non-default case instead of
19165 default case if default_case_nonstandard.
19167 2017-05-02 Nathan Sidwell <nathan@acm.org>
19169 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19170 check. Fix formatting.
19172 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19174 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19175 errors when comparing specialized and unspecialized times.
19177 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19179 * diagnostic-show-locus.c
19180 (layout::should_print_annotation_line_p): Make private.
19181 (layout::print_annotation_line): Make private.
19182 (layout::annotation_line_showed_range_p): Make private.
19183 (layout::show_ruler): Make private.
19184 (layout::print_source_line): Make private. Pass in line and
19185 line_width, rather than calling location_get_source_line. Drop
19187 (layout::print_leading_fixits): New method.
19188 (layout::print_any_fixits): Rename to...
19189 (layout::print_trailing_fixits): ...this, and make private.
19190 Don't print newline fixits.
19191 (diagnostic_show_locus): Move logic for printing one row into...
19192 (layout::print_line): ...this new function. Move the
19193 location_get_source_line call and error-handling from
19194 print_source_line to here. Call print_leading_fixits, and rename
19195 print_any_fixits to print_trailing_fixits.
19196 (selftest::test_fixit_insert_containing_newline): Update now that
19197 newlines are partially supported.
19198 (selftest::test_fixit_insert_containing_newline_2): New test.
19199 (selftest::test_fixit_replace_containing_newline): Update comments.
19200 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19201 * edit-context.c (class added_line): New class.
19202 (class edited_line): Describe newline handling in comment.
19203 (edited_line::actually_edited_p): New method.
19204 (edited_line::print_content): Delete redundant decl.
19205 (edited_line::m_predecessors): New field.
19206 (edited_file::print_content): Call edited_line::print_content.
19207 (edited_file::print_diff): Update to support newlines.
19208 (edited_file::print_diff_hunk): Likewise.
19209 (edited_file::print_run_of_changed_lines): New function.
19210 (edited_file::print_diff_line): Convert to...
19211 (print_diff_line): ...this.
19212 (edited_file::get_effective_line_count): New function.
19213 (edited_line::edited_line): Initialize new field m_predecessors.
19214 (edited_line::~edited_line): Clean up m_predecessors.
19215 (edited_line::apply_fixit): Handle newlines.
19216 (edited_line::get_effective_line_count): New function.
19217 (edited_line::print_content): New function.
19218 (edited_line::print_diff_lines): New function.
19219 (selftest::test_applying_fixits_insert_containing_newline): New
19221 (selftest::test_applying_fixits_replace_containing_newline): New
19223 (selftest::insert_line): New function.
19224 (selftest::test_applying_fixits_multiple_lines): Add example of
19226 (selftest::edit_context_c_tests): Call the new tests.
19228 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19230 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19231 parameter cand. Update dump information.
19232 (get_computation_cost): Update uses.
19234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19236 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19237 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19238 (get_computation_at, rewrite_use_address): Update use of
19239 get_computation_aff.
19241 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19243 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19244 (get_computation): Delete.
19245 (get_computation_cost): Implement like get_computation_cost_at.
19246 Use get_computation_at.
19247 (get_computation_cost_at): Delete.
19248 (rewrite_use_nonlinear_expr): Use get_computation_at.
19249 (rewrite_use_compare, remove_unused_ivs): Ditto.
19251 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19253 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19255 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19257 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19258 (ivopts_global_cost_for_size): Rename parameter and update uses.
19259 (iv_ca_recount_cost): Update uses.
19260 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19261 candidates seperately in n_invs and n_cands.
19262 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19264 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19266 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19267 (find_inv_vars_cb): New.
19268 (find_depends): Renamed to ...
19269 (find_inv_vars): ... this.
19270 (add_candidate_1, force_var_cost): Call find_inv_vars.
19271 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19273 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19275 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19276 inv_vars. Add inv_exprs.
19277 (struct iv_cand): Rename depends_on to inv_vars.
19278 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19279 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19280 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19281 (dump_cand): Dump inv_vars.
19282 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19283 (record_invariant, find_depends, add_candidate_1): Ditto.
19284 (set_group_iv_cost, force_var_cost): Ditto.
19285 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19286 (get_computation_cost_at, get_computation_cost): Ditto.
19287 (determine_group_iv_cost_generic): Ditto.
19288 (determine_group_iv_cost_address): Ditto.
19289 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19290 (determine_group_iv_costs): Ditto.
19291 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19292 (iv_ca_set_remove_invariants): Renamed to ...
19293 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19294 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19295 (iv_ca_set_add_invariants): Renamed to ...
19296 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19297 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19298 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19299 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19300 (create_new_ivs): Remove useless dump.
19302 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19304 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19306 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19307 (iv_ca_set_no_cp, create_new_iv): Ditto.
19309 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19311 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19313 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19315 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19316 function tree_check2.
19318 2017-05-02 Martin Liska <mliska@suse.cz>
19320 * doc/gcov.texi: Add missing preposition.
19321 * gcov.c (function_info::function_info): Properly fill up
19322 all member variables.
19324 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19326 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19328 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19330 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19332 2017-05-02 Martin Liska <mliska@suse.cz>
19335 * lto-streamer-in.c (lto_read_tree_1): Remove
19336 LTO_STREAMER_DEBUG.
19337 * lto-streamer.c (struct tree_hash_entry): Likewise.
19338 (struct tree_entry_hasher): Likewise.
19339 (tree_entry_hasher::hash): Likewise.
19340 (tree_entry_hasher::equal): Likewise.
19341 (lto_streamer_init): Likewise.
19342 (lto_orig_address_map): Likewise.
19343 (lto_orig_address_get): Likewise.
19344 (lto_orig_address_remove): Likewise.
19345 * lto-streamer.h: Likewise.
19346 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19347 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19349 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19351 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19352 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19353 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19354 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19355 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19356 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19357 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19358 (mm_maskz_sub_ss): New intrinsics.
19359 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19360 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19361 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19362 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19363 (__builtin_ia32_subss_mask_round): New builtins.
19364 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19365 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19366 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19368 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19369 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19371 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19374 2017-05-02 Martin Jambor <mjambor@suse.cz>
19376 PR tree-optimization/78687
19377 * tree-sra.c (access): New field parent.
19378 (process_subtree_disqualification): New function.
19379 (disqualify_candidate): Call it.
19380 (build_accesses_from_assign): Reset write flag if creating an
19382 (build_access_subtree): Fill in parent field and also prpagate
19383 down grp_write flag.
19384 (create_artificial_child_access): New parameter set_grp_write, set
19385 grp_write to its value.
19386 (propagate_subaccesses_across_link): Also propagate grp_write flag
19388 (propagate_all_subaccesses): Push the closest parent back to work
19389 queue if add_access_to_work_queue returned true.
19391 2017-05-02 Richard Biener <rguenther@suse.de>
19393 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19394 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19395 -fstrict-overflow documentation.
19396 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19397 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19398 flag_strict_overflow.
19399 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19400 * lto-opts.c (lto_write_options): Do not stream it.
19401 * lto-wrapper.c (merge_and_complain): Do not handle it.
19402 * opts.c (default_options_table): Do not set -fstrict-overflow.
19403 (finish_options): Likewise do not clear it when sanitizing.
19404 * simplify-rtx.c (simplify_const_relational_operation): Do not
19405 test flag_strict_overflow.
19407 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19409 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19410 using enabled attribute.
19411 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19412 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19413 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19414 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19415 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19416 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19417 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19418 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19419 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19420 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19422 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19424 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19426 2017-05-02 Richard Biener <rguenther@suse.de>
19428 PR tree-optimization/80591
19430 2017-04-10 Richard Biener <rguenther@suse.de>
19432 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19435 2017-05-02 Richard Biener <rguenther@suse.de>
19437 PR tree-optimization/80549
19438 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19439 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19440 headers if they do not have a preheader.
19442 2017-05-02 Martin Liska <mliska@suse.cz>
19445 * common.opt: Fix typo.
19446 * doc/invoke.texi: Likewise.
19448 2017-05-01 Jan Beulich <jbeulich@suse.com>
19450 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19451 swapping, add (x,x,m,x,n) alternative.
19453 2017-05-01 Nathan Sidwell <nathan@acm.org>
19455 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19456 unnecessary unadjusted_alignment check.
19458 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19461 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19463 * gimplify.c (gimplify_cilk_detach): New function.
19464 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19465 * tree-core.h: Document EXPR_CILK_SPAWN.
19466 * tree.h (EXPR_CILK_SPAWN): Define.
19468 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19470 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19471 to use new fixit_hint representation, using the "replace" logic.
19472 (get_line_span_for_fixit_hint): Likewise.
19473 (layout::print_any_fixits): Likewise.
19474 (selftest::test_one_liner_many_fixits): Rename to...
19475 (selftest::test_one_liner_many_fixits_1): ...this, and update
19476 comment and expected output to reflect that the multiple fix-it
19477 hints are now consolidated into one insertion.
19478 (selftest::test_one_liner_many_fixits_2): New test.
19479 (selftest::test_diagnostic_show_locus_one_liner): Update for
19481 (selftest::test_fixit_consolidation): Update for fix-it API
19483 * diagnostic.c (print_parseable_fixits): Likewise.
19484 * edit-context.c (edited_line::m_line_events): Convert from
19485 auto_vec <line_event *> to auto_vec <line_event>.
19486 (class line_event): Convert from abstract base class to a concrete
19487 class, taking over the role of replace_event.
19488 (class insert_event): Delete.
19489 (class replace_event): Rename to class line_event. Convert to
19491 (edit_context::add_fixits): Reimplement.
19492 (edit_context::apply_insert): Delete.
19493 (edit_context::apply_replace): Rename to...
19494 (edit_context::apply_fixit): ...this. Convert to half-open range.
19495 (edited_file::apply_insert): Delete.
19496 (edited_file::apply_replace): Rename to...
19497 (edited_file::apply_fixit): ...this.
19498 (edited_line::~edited_line): Drop deletion of events.
19499 (edited_line::apply_insert): Delete.
19500 (edited_line::apply_replace): Rename to...
19501 (edited_line::apply_fixit): ...this. Convert to half-open range.
19502 Update for change to type of m_line_events.
19503 * edit-context.h (edit_context::apply_insert): Delete.
19504 (edit_context::apply_replace): Rename to...
19505 (edit_context::apply_fixit): ...this.
19507 2017-05-01 Martin Sebor <msebor@redhat.com>
19509 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19512 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19515 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19516 __get_cpuid_max returns 0.
19517 (__get_cpuid_count): Ditto.
19519 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19521 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19522 replacement expression is another instance of one of its arguments.
19524 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19527 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19528 check for stack push/pop autoinc.
19529 * config/i386/i386.c (ix86_agi_dependent): Return false
19530 if the only reason why modified_in_p returned true is that
19531 addr is SP based and set_insn is a push or pop.
19533 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19535 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19538 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19541 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19542 (account_size_time): Use two predicates - exec_pred and
19544 (evaluate_conditions_for_known_args): Compute both clause and
19546 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19547 (inline_summary_t::duplicate): Update.
19548 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19550 (compute_inline_parameters): Likewise.
19551 (estimate_edge_size_and_time): Update caluclation of time.
19552 (estimate_node_size_and_time): Compute both time and nonspecialized
19554 (estimate_ipcp_clone_size_and_time): Update.
19555 (inline_merge_summary): Update.
19556 (do_estimate_edge_time): Update.
19557 (do_estimate_edge_size): Update.
19558 (do_estimate_edge_hints): Update.
19559 (inline_read_section, inline_write_summary): Stream both new predicates.
19560 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19562 (compute_inlined_call_time): Cleanup.
19563 (big_speedup_p): Update.
19564 (edge_badness): Update.
19565 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19566 (size_time_entry): Replace predicate by exec_predicate and
19567 nonconst_predicate.
19568 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19569 (estimate_edge_time): Return also nonspec_time.
19570 (reset_edge_growth_cache): Update.
19572 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19574 PR rtl-optimization/80491
19575 * ifcvt.c (noce_process_if_block): When looking for x setter
19576 with missing else_bb, don't check only the insn right before
19577 cond_earliest, but look for the last insn that x is modified in
19578 within the same bb.
19580 PR rtl-optimization/80491
19581 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19583 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19585 PR tree-optimization/80487
19586 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19588 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19590 PR tree-optimization/79697
19591 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19592 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19593 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19595 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19596 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19598 2017-04-28 Martin Sebor <msebor@redhat.com>
19600 PR tree-optimization/80523
19601 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19602 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19604 (maybe_warn, format_directive, parse_directive): Use new functions.
19605 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19607 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19609 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19611 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19613 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19614 target_header_dir): Set correctly.
19615 * configure: Regenerated.
19616 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19617 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19618 instead of SYSTEM_HEADER_DIR.
19620 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19622 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19623 (estimate_local_effects): Likewise.
19624 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19625 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19626 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19627 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19628 do_estimate_edge_time, estimate_edge_time): Likewise.
19629 * ipa-inline-analysis.c (estimate_node_size_and_time,
19630 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19631 (estimate_time_after_inlining): Remove.
19633 2017-04-28 Martin Liska <mliska@suse.cz>
19635 * doc/gcov.texi: Enhance documentation of gcov.
19637 2017-04-28 Martin Liska <mliska@suse.cz>
19639 * doc/gcov.texi: Sort options in alphabetic order.
19640 * doc/gcov-dump.texi: Likewise.
19641 * doc/gcov-tool.texi: Likewise.
19642 * gcov.c (print_usage): Likewise.
19643 * gcov-dump.c (print_usage): Likewise.
19644 * gcov-tool.c (print_merge_usage_message): Likewise.
19645 (print_rewrite_usage_message): Likewise.
19646 (print_overlap_usage_message): Likewise.
19648 2017-04-28 Martin Liska <mliska@suse.cz>
19650 PR gcov-profile/53915
19651 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19653 2017-04-28 Martin Liska <mliska@suse.cz>
19655 PR gcov-profile/79891
19656 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19657 is marked by compiler as living on a line.
19658 (get_cycles_count): Remove usage of the union.
19659 (output_intermediate_file): Likewise.
19660 (find_source): Fix GNU coding style.
19661 (accumulate_line_counts): Remove old non-all block mode.
19662 (output_lines): Remove usage of the union.
19663 * profile.c (output_location): Include all BBs, even if
19664 belonging to a same line (and file) as a previous BB.
19666 2017-04-28 Martin Liska <mliska@suse.cz>
19668 * gcov.c (process_args): Handle new argument 'w'.
19669 (read_graph_file): Assign ID to BBs.
19670 (output_branch_count): Display BB # if verbose flag is set.
19671 (output_lines): Likewise for arcs.
19672 (print_usage): Add '--verbose' option help.
19673 * doc/gcov.texi: Document --verbose (-w) option.
19675 2017-04-28 Martin Liska <mliska@suse.cz>
19677 * gcov.c (struct block_location_info): New struct.
19678 (process_file): Fill up the new structure.
19679 (read_graph_file): Replace usage of encoding by the newly added
19681 (add_line_counts): Likewise.
19682 (accumulate_line_counts): Remove usage of the union.
19683 (function_info::function_info): New function.
19684 (function_info::~function_info): Likewise.
19685 (process_file): Call delete instead of release_function.
19686 (release_function): Release the function.
19687 (release_structures): Call delete instead of release_function.
19688 (solve_flow_graph): Replace usage of num_blocks.
19689 (find_exception_blocks): Likewise.
19690 (output_lines): Fix GNU coding style.
19692 2017-04-28 Martin Liska <mliska@suse.cz>
19695 * coverage.c (coverage_remove_note_file): New function.
19696 * coverage.h: Declare the function.
19697 * toplev.c (finalize): Clean if an error has been seen.
19699 2017-04-28 Martin Liska <mliska@suse.cz>
19701 PR gcov-profile/80031
19702 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19703 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19704 * gcov.c (read_graph_file): Read just number of blocks.
19705 * profile.c (branch_prob): Do not stream 0 flags per a basic
19708 2017-04-28 Martin Liska <mliska@suse.cz>
19710 * gcov-dump.c (tag_*): Add new argument to declarations.
19711 (dump_gcov_file): Likewise.
19712 (tag_blocks): Add and use new argument depth.
19713 (tag_arcs): Likewise.
19714 (tag_lines): Likewise.
19715 (tag_counters): Likewise.
19716 (tag_summary): Likewise.
19717 (dump_working_sets): Use depth to do a proper indentation.
19719 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19722 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19723 * symtab.c (symtab_node::debug_symtab): Move definition here.
19725 2017-04-28 Richard Biener <rguenther@suse.de>
19727 * lto-streamer.h (LTO_major_version): Bump to 7.
19729 2017-04-28 Richard Biener <rguenther@suse.de>
19731 * tree-vrp.c (assert_info): New struct.
19732 (add_assert_info): New helper.
19733 (register_edge_assert_for_2): Refactor to add asserts to a vector
19735 (register_edge_assert_for_1): Likewise.
19736 (register_edge_assert_for): Likewise.
19737 (finish_register_edge_assert_for): New helper actually registering
19738 asserts where live on edge.
19739 (find_conditional_asserts): Adjust.
19740 (find_switch_asserts): Likewise.
19741 (evrp_dom_walker::try_find_new_range): Generalize.
19742 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19744 2017-04-27 Marek Polacek <polacek@redhat.com>
19747 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19748 arg10 and arg11 to itype.
19750 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19752 * doc/extend.texi (Object Size Checking): Improve grammar.
19754 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19757 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19758 that the logic for permitting reciprocal estimates matches that
19761 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19764 * tree.c (type_cache_hasher::equal): Only compare
19765 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19766 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
19767 non-aggregate element types.
19768 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
19769 about the flag on ARRAY_TYPEs in the comment, formatting fix.
19771 2017-04-27 Richard Biener <rguenther@suse.de>
19773 PR middle-end/80533
19774 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
19775 stripping ARRAY_REFs from MEM_EXPR make sure we're not
19776 keeping a reference to a trailing array.
19778 2017-04-27 Richard Biener <rguenther@suse.de>
19780 PR middle-end/80539
19781 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
19782 being in loop-closed SSA form conservatively.
19783 (chrec_fold_multiply_poly_poly): Likewise.
19785 2017-04-27 Tamar Christina <tamar.christina@arm.com>
19787 PR middle-end/79665
19788 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
19789 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19791 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19794 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19795 (aarch64_function_arg_alignment): Return unsigned int again, but still
19796 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19797 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19798 Don't emit -Wpsabi note.
19799 (aarch64_function_arg_boundary): Likewise.
19800 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19803 2017-04-26 Nathan Sidwell <nathan@acm.org>
19805 * tree.h (crc32_unsigned_n): Declare.
19806 (crc32_unsigned, crc32_unsigned): Make inline.
19807 * tree.c (crc32_unsigned_bits): Replace with ...
19808 (crc32_unsigned_n): ... this.
19809 (crc32_unsigned, crc32_byte): Remove.
19810 (crc32_string): Remove unnecessary braces.
19812 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19814 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19815 * ipa-inline-analysis.c (MAX_TIME): Remove.
19816 (account_size_time): Use sreal for time.
19817 (dump_inline_summary): Update.
19818 (estimate_function_body_sizes): Update.
19819 (estimate_edge_size_and_time): Update.
19820 (estimate_calls_size_and_time): Update.
19821 (estimate_node_size_and_time): Update.
19822 (inline_merge_summary): Update.
19823 (inline_update_overall_summary): Update.
19824 (estimate_time_after_inlining): Update.
19825 (inline_read_section): Update.
19826 (inline_write_summary): Update.
19827 * ipa-inline.c (compute_uninlined_call_time): Update.
19828 (compute_inlined_call_time): Update.
19829 (recursive_inlining): Update.
19830 (inline_small_functions): Update.
19831 (dump_overall_stats): Update.
19832 * ipa-inline.h: Include sreal.h.
19833 (size_time_entry): Turn time to sreal.
19834 (inline_summary): Turn self_time nad time to sreal.
19836 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19838 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19840 (sreal::stream_out, sreal::stream_in): New.
19841 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19843 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19845 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19848 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19851 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19853 (dimode_scalar_chain::compute_convert_gain): Ditto.
19854 (dimode_scalar_chain::make_vector_copies): Ditto.
19855 (dimode_scalar_chain::convert_reg): Ditto.
19856 (dimode_scalar_chain::convert_insn): Ditto.
19857 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19858 (VI248_AVX512BW_1): New mode iterator.
19859 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19860 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19863 2017-04-25 Martin Sebor <msebor@redhat.com>
19865 PR tree-optimization/80497
19866 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19867 constants are representable in HOST_WIDE_INT.
19868 (parse_directive): Ditto.
19870 2017-04-25 Martin Sebor <msebor@redhat.com>
19873 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19874 (new_zero_array): Adjust signature.
19875 (dom_info::dom_init): Used unsigned rather that size_t.
19876 (dom_info::dom_info): Same.
19878 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19879 Jakub Jelinek <jakub@redhat.com>
19882 * config/arm/arm.c: Include gimple.h.
19883 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19884 returns negative, increment ncrn only if it returned positive.
19885 (arm_needs_doubleword_align): Return int instead of bool,
19886 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19887 members, but if there is any such non-FIELD_DECL
19888 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19889 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19890 returns negative, increment nregs only if it returned positive.
19891 (arm_setup_incoming_varargs): Likewise.
19892 (arm_function_arg_boundary): Emit -Wpsabi note if
19893 arm_needs_doubleword_align returns negative, return
19894 DOUBLEWORD_ALIGNMENT only if it returned positive.
19896 2017-04-25 Marek Polacek <polacek@redhat.com>
19899 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19900 first argument to type.
19902 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19905 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19906 type checks to test for compatibility instead of equality.
19908 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19909 Jakub Jelinek <jakub@redhat.com>
19912 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19914 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19915 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19916 the alignment computation, but return their maximum in warn_alignment.
19917 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19918 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19920 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19921 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19926 * config/arc/simdext.md (dmpyh): Fix typo.
19928 2017-04-25 Richard Biener <rguenther@suse.de>
19930 PR tree-optimization/80492
19931 * alias.c (compare_base_decls): Handle registers with asm
19932 specification conservatively.
19933 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19934 compare_base_decls returning dont-know properly.
19936 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19938 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19939 (legitimate_offset_address_p): New function.
19940 (arc_legitimate_address_p): Use above function.
19942 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19944 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19946 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19948 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19949 ACCH registers whenever they are available.
19951 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19953 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19954 double regs fix when not used.
19956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19958 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19960 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19961 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19963 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19965 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19966 when emitting short ADD instructions.
19968 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19970 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19972 (cmpsi_cc_c_insn): Likewise.
19973 (cbranchsi4_scratch): Compute proper instruction length using
19974 compact_hreg_operand.
19975 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19977 2017-04-25 Richard Biener <rguenther@suse.de>
19979 PR middle-end/80509
19980 * passes.c (pass_manager::pass_manager): Initialize
19981 m_name_to_pass_map.
19983 2017-04-25 Richard Biener <rguenther@suse.de>
19985 PR tree-optimization/79201
19986 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19988 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19991 * config/s390/vector.md: Split MEM->GPR vector moves for
19992 non-s_operand addresses.
19994 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19997 * config/s390/predicates.md (reload_const_wide_int_operand): New
19999 * config/s390/s390.md ("movti"): Remove d/P alternative.
20000 ("movti_bigconst"): New pattern definition.
20002 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20005 * s390-protos.h (s390_expand_cs_hqi): Removed.
20006 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20007 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20008 modes as well as CCZ1mode and CCZmode.
20009 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20010 signature of s390_emit_compare_and_swap.
20011 (s390_expand_cs_hqi): Likewise, make static.
20012 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20013 compare-and-swap, in some cases.
20014 (s390_expand_cs): Wrapper function.
20015 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20017 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20018 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20019 patterns for small and large integers. Forbid symref memory operands.
20020 Move expander to s390.c. Require cc register.
20021 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20022 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20023 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20024 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20025 symref memory operands. Remove CC mode and call s390_match_ccmode
20027 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20029 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20031 * config/s390/s390.md (define_peephole2): New peephole to help
20032 combining the load-and-test pattern with volatile memory.
20034 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20036 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20037 with CCZmode for TARGET_Z196.
20039 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20041 PR rtl-optimization/80501
20042 * combine.c (make_compound_operation_int): Set subreg_code to SET
20043 even for AND with mask of the sign bit of mode.
20045 PR rtl-optimization/80500
20046 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20047 sum's initial value.
20049 2017-04-25 Julian Brown <julian@codesourcery.com>
20050 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20052 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20054 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20056 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20058 2017-04-25 Julian Brown <julian@codesourcery.com>
20059 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20061 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20062 (thunderx2t99_sha): New Reservation.
20064 2017-04-25 Julian Brown <julian@codesourcery.com>
20065 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20067 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20068 type for 1-element load.
20070 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20072 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20074 2017-04-24 Martin Jambor <mjambor@suse.cz>
20076 PR tree-optimization/80293
20077 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20078 char arrays not totally scalarizable if it is false.
20079 (analyze_all_variable_accesses): Pass correct value in the new
20080 parameter. Add a statistics counter.
20082 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20084 PR middle-end/79931
20085 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20087 2017-04-24 Richard Biener <rguenther@suse.de>
20089 PR tree-optimization/80494
20090 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20091 out for complex types.
20093 2017-04-24 Richard Biener <rguenther@suse.de>
20095 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20096 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20097 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20098 (DFS): Adjust and never fail.
20099 (sccvn_dom_walker::fail): Remove.
20100 (sccvn_dom_walker::before_dom_children): Adjust.
20101 (run_scc_vn): Likewise and never fail.
20102 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20103 (pass_fre::execute): Likewise.
20105 2017-04-24 Richard Biener <rguenther@suse.de>
20107 PR tree-optimization/79725
20108 * tree-ssa-sink.c (statement_sink_location): Return whether
20109 failure reason was zero uses. Move that check later.
20110 (sink_code_in_bb): Deal with zero uses by removing the stmt
20113 2017-04-24 Richard Biener <rguenther@suse.de>
20116 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20117 pointer-based references.
20119 2017-04-24 Richard Biener <rguenther@suse.de>
20122 * pass_manager.h (pass_manager::operator new): Remove.
20123 (pass_manager::operator delete): Likewise.
20124 * passes.c (pass_manager::operator new): Remove.
20125 (pass_manager::operator delete): Likewise.
20126 (pass_manager::pass_manager): Zero individual pass members.
20128 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20131 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20132 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20133 Check "XEXP (src, 1)" operand here.
20134 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20135 Check "XEXP (src, 1)" operand here.
20136 (dimode_scalar_chain::make_vector_copies): Detect count register
20137 of a shift instruction. Zero extend count register from QImode
20138 to DImode to satisfy vector shift pattern count operand predicate.
20139 Substitute vector shift count operand with a DImode copy.
20140 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20143 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20145 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20146 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20147 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20148 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20149 (UNSPEC_NOREX_MEM): Remove definition.
20151 2017-04-21 Richard Biener <rguenther@suse.de>
20153 PR tree-optimization/79547
20154 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20155 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20156 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20157 without any constraints.
20159 2017-04-21 Richard Biener <rguenther@suse.de>
20161 PR tree-optimization/78847
20162 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20164 2017-04-21 Richard Biener <rguenther@suse.de>
20166 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20167 (build_distinct_type_copy): Likewise.
20168 (build_variant_type_copy): Likewise.
20169 * tree.c (build_qualified_type): Pass down mem-stat info.
20170 (build_distinct_type_copy): Likewise.
20171 (build_variant_type_copy): Likewise.
20173 2017-04-21 Richard Biener <rguenther@suse.de>
20175 PR tree-optimization/80237
20176 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20178 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20179 for a simplified result.
20181 2016-04-21 Richard Biener <rguenther@suse.de>
20183 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20184 sth as strict as a simple_iv but a chrec without symbols and an
20185 operand defined in the loop we are peeling (and not some subloop).
20186 (propagate_constants_for_unrolling): Propagate all constants.
20188 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20191 * config/i386/i386.c (print_reg): Remove assert for disalowed
20192 regno values, call output_operand_lossage instead.
20194 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20197 * config/i386/constraints.md (Yc): New register constraint.
20198 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20199 Use Yc constraint for alternative 2 of operand 0. Remove
20200 preferred_for_speed attribute.
20202 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20204 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20205 lastprivate clauses in SIMT case.
20207 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20209 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20211 2017-04-20 Richard Biener <rguenther@suse.de>
20213 PR tree-optimization/57796
20214 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20215 as N scalar stores.
20216 (vect_model_load_cost): Cost gathers as N scalar loads.
20218 2017-04-20 Richard Biener <rguenther@suse.de>
20220 * ggc-page.c (ggc_allocated_p): Rename to ...
20221 (safe_lookup_page_table_entry): ... this and return the lookup
20223 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20225 2017-04-20 Richard Biener <rguenther@suse.de>
20227 PR tree-optimization/80453
20228 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20229 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20230 from the conditions.
20231 (vn_phi_eq): Pass them down.
20232 (vn_phi_lookup): Record them.
20233 (vn_phi_insert): Likewise.
20235 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20237 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20238 uninitialized variable warning to avoid buffer overrun.
20240 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20243 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20244 is suppressed for '{ 0 }' in C.
20246 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20248 * BASE-VER: Set to 8.0.0.
20250 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20252 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20253 priority .init_array and .fini_array section with SECTION_NOTYPE
20256 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20258 PR middle-end/80423
20259 * tree.h (build_array_type): Add typeless_storage default argument.
20260 * tree.c (type_cache_hasher::equal): Also compare
20261 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20262 (build_array_type): Add typeless_storage argument, set
20263 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20265 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20266 (build_array_type): Likewise. Add typeless_storage argument.
20268 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20269 Jakub Jelinek <jakub@redhat.com>
20271 PR tree-optimization/80426
20272 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20273 operation on symbolic operands, also compute the overflow for the
20274 invariant part when the operation degenerates into a negation.
20276 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20279 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20280 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20283 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20285 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20288 * config/avr/avr.c (tree.h): Include it.
20289 (cgraph.h): Include it.
20290 (avr_encode_section_info): Don't warn for uninitialized progmem
20291 variable if it's just an alias.
20293 2017-04-19 Richard Biener <rguenther@suse.de>
20296 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20297 when needed by AutoPGO.
20299 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20302 * doc/lto.texi: Remove an extra 'that'.
20304 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20306 PR rtl-optimization/80429
20307 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20308 are only used in debug insns.
20310 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20311 Vladimir Makarov <vmakarov@redhat.com>
20313 * config/sparc/predicates.md (input_operand): Add comment. Return
20314 true for any memory operand when LRA is in progress.
20315 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20317 2017-04-18 Jeff Law <law@redhat.com>
20320 * mips.md ({return,simple_return}_internal): Do not overwrite
20323 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20325 PR tree-optimization/80443
20326 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20327 instead of adding 1, subtract -1 and similarly instead of subtracting
20330 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20332 PR rtl-optimization/80357
20333 * haifa-sched.c (tmp_bitmap): New variable.
20334 (model_recompute): Handle duplicate use records.
20335 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20336 (free_global_sched_pressure_data): Free it.
20338 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20341 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20342 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20343 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20344 instead of SYSTEM_HEADER_DIR.
20346 2017-04-18 Jeff Law <law@redhat.com>
20348 PR middle-end/80422
20349 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20350 predecessors after walking up the insn chain.
20352 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20355 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20356 sizetype type into debug info.
20358 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20361 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20362 unneeded test for TARGET_UPPER_REGS_SF.
20363 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20365 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20368 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20369 instead of gsi_after_labels.
20371 2017-04-18 Jeff Law <law@redhat.com>
20373 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20377 2017-04-13 Jeff Law <law@redhat.com>
20378 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20379 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20381 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20384 * config/avr/avr.c (intl.h): Include it.
20385 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20387 2017-04-18 Martin Liska <mliska@suse.cz>
20389 PR gcov-profile/78783
20390 * gcov-tool.c (gcov_output_files): Validate that destination
20391 file is either removed by the tool or by a user.
20393 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20394 Guy Benyei <guybe@mellanox.com>
20396 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20397 block, and do not negate it, the stored id is already negative.
20399 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20401 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20403 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20406 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20407 masks of options that should be turned off if the VSX vector
20408 options are turned off.
20409 (OTHER_P8_VECTOR_MASKS): Likewise.
20410 (OTHER_VSX_VECTOR_MASKS): Likewise.
20411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20412 rs6000_disable_incompatible_switches to validate no type switches
20414 (rs6000_incompatible_switch): New function to disallow turning on
20415 other vector options if -mno-vsx, -mno-power8-vector, or
20416 -mno-power9-vector are specified.
20418 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20420 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20422 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20424 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20425 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20426 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20427 (ARG_POINTER_CFA_OFFSET): Likewise.
20429 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20431 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20432 conditions to take advantage of various optimizations.
20434 2017-04-13 Jeff Law <law@redhat.com>
20436 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20437 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20438 (zero_extendsidi2_dext): Likewise.
20440 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20443 * fold-const.c (fold_ternary_loc): Revert
20444 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20447 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20449 PR rtl-optimization/80343
20450 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20451 new scratch pseudo.
20453 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20456 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20457 to ubsan_encode_value.
20459 2017-04-13 Jeff Law <law@redhat.com>
20461 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20462 appearing in DEBUG_INSNs.
20464 2017-04-13 Martin Liska <mliska@suse.cz>
20466 PR gcov-profile/80413
20467 * gcov-io.c (gcov_write_string): Copy to buffer just when
20468 allocated size is greater than zero.
20470 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20473 * dwarf2out.c (decls_for_scope): Ignore declarations of
20474 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20476 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20479 * ipa-visibility.c (non_local_p): Fix typos.
20480 (localize_node): When localizing symbol in same comdat group,
20481 dissolve the group only when we know external symbols are going
20483 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20485 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20487 PR tree-optimization/79390
20488 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20489 order does not result in usable sequence, retry with reversed operand
20495 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20496 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20497 op0 instead of fold_convert_loc (loc, type, arg0).
20499 2017-04-12 Jeff Law <law@redhat.com>
20501 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20502 has a delay slot in the generated code.
20504 * config/cris/cris.md (cris_preferred_reload_class): Return
20505 GENNONACR_REGS rather than GENERAL_REGS.
20507 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20510 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20511 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20512 signedness of the result type.
20514 2017-04-12 Richard Biener <rguenther@suse.de>
20515 Jeff Law <law@redhat.com>
20517 PR tree-optimization/80359
20518 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20519 trim stores to TARGET_MEM_REFs.
20521 2017-04-12 Richard Biener <rguenther@suse.de>
20523 PR tree-optimization/79390
20524 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20525 threading case even more.
20527 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20530 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20531 for quad_address_p for TImode, instead of just not indexed_address.
20533 2017-04-12 Richard Biener <rguenther@suse.de>
20534 Bernd Edlinger <bernd.edlinger@hotmail.de>
20536 PR middle-end/79671
20537 * alias.c (component_uses_parent_alias_set_from): Handle
20538 TYPE_TYPELESS_STORAGE.
20539 (get_alias_set): Likewise.
20540 * tree-core.h (tree_type_common): Add typeless_storage flag.
20541 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20542 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20543 for types containing members with TYPE_TYPELESS_STORAGE.
20544 (place_field): Likewise.
20545 (layout_type): Likewise for ARRAY_TYPE.
20546 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20547 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20548 TYPE_TYPELESS_STORAGE.
20549 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20551 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20554 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20555 first argument to type.
20557 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20561 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20562 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20563 (rs6000_expand_binop_builtin): Likewise.
20564 (rs6000_expand_ternop_builtin): Likewise; also add missing
20565 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20566 vshasigma built-ins.
20567 * doc/extend.texi: Document that vec_xxpermdi's third argument
20568 must be a constant.
20570 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20572 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20573 Use shift_const cost parameter when calculating gain of STV shifts.
20575 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20577 PR rtl-optimization/70478
20578 * lra-constraints.c (process_alt_operands): Check memory for
20579 disfavoring memory insn operand.
20581 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20583 PR middle-end/80100
20584 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20585 left shift in unsigned HOST_WIDE_INT type.
20587 PR rtl-optimization/80385
20588 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20589 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20592 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20593 if they have any depend clauses.
20595 2017-04-11 Martin Liska <mliska@suse.cz>
20598 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20599 * ipa-split.c (split_function): Create a local comdat symbol
20600 if caller is in a comdat group.
20602 2017-04-11 Martin Liska <mliska@suse.cz>
20605 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20608 2017-04-11 Martin Sebor <msebor@redhat.com>
20610 PR middle-end/80364
20611 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20612 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20614 (directive::set_width, directive::set_precision, format_character):
20616 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20619 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20622 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20623 conflict, set target->arch_name instead of target->cpu_name.
20625 2017-04-11 Richard Biener <rguenther@suse.de>
20627 PR tree-optimization/80374
20628 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20629 build_zero_cst, remove fold_convertible_p check again.
20631 2017-04-11 Martin Liska <mliska@suse.cz>
20634 * ubsan.c (instrument_object_size): Do not instrument register
20637 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20640 * config/i386/i386-builtin-types.def
20641 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20642 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20643 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20644 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20645 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20646 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20647 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20648 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20649 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20650 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20651 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20652 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20653 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20654 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20655 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20656 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20657 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20658 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20659 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20660 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20661 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20662 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20663 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20664 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20665 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20666 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20667 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20668 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20669 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20670 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20671 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20672 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20673 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20674 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20675 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20676 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20677 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20678 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20679 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20680 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20681 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20682 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20683 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20684 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20685 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20686 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20688 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20689 flag to second_arg_count, handle 4 argument function type _COUNT
20690 aliases, handle second_arg_count on second argument rather than last.
20692 2017-04-10 Jeff Law <law@redhat.com>
20694 PR tree-optimization/80374
20695 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20696 record anything if we can not convert integer_zero_node to the
20699 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20702 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20703 Enhance special handling given to the TARGET_P9_MINMAX option in
20704 relation to certain other options.
20706 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20708 PR tree-optimization/80153
20709 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20710 remove POINTER_PLUS_EXPR's base part directly, rather than through
20713 2017-04-10 Richard Biener <rguenther@suse.de>
20714 Bin Cheng <bin.cheng@arm.com>
20716 PR tree-optimization/80153
20717 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20718 the first element of pointer type aff_tree. Build result expr in
20720 (add_elt_to_tree): Convert to type unconditionally. Remove other
20721 fold_convert calls.
20722 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20723 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20725 2017-04-10 Richard Biener <rguenther@suse.de>
20727 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20730 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20732 PR rtl-optimization/70478
20733 * lra-constraints.c (curr_small_class_check): New.
20734 (update_and_check_small_class_inputs): New.
20735 (process_alt_operands): Update curr_small_class_check. Disfavor
20736 alternative insn memory operands. Check available regs for small
20739 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20742 * config/mips/mips.opt (-mvirt): Update description.
20743 * doc/invoke.texi (-mvirt): Likewise.
20745 2017-04-10 Richard Biener <rguenther@suse.de>
20747 PR middle-end/80362
20748 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20749 looking for NEGATE_EXPR in -A / -B to A / B folding.
20751 2017-04-10 Martin Liska <mliska@suse.cz>
20753 PR gcov-profile/80224
20754 * gcov.c (print_usage): Fix usage string.
20755 (get_gcov_intermediate_filename): Remove.
20756 (output_gcov_file): Use both for normal and intermediate format.
20757 (generate_results): Do not initialize special file for
20758 intermediate format.
20760 2017-04-10 Richard Biener <rguenther@suse.de>
20762 PR tree-optimization/80304
20763 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20766 2017-04-10 Nathan Sidwell <nathan@acm.org>
20769 * config/rs6000/rs6000.c (rs6000_vector_type): New.
20770 (rs6000_init_builtins): Use it.
20772 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20774 * config/arm/arm.md (<mrc>): Add mode to SET source.
20775 (<mrrc>): Likewise.
20777 2017-04-10 Richard Biener <rguenther@suse.de>
20779 PR middle-end/80344
20780 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
20782 2017-04-10 Jakub Jelinek <jakub@redhat.com>
20785 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
20786 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
20787 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
20788 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
20789 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
20790 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20791 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20792 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20793 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20794 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20795 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20796 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20797 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20798 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20799 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20800 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20801 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20802 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20803 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20804 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20805 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20806 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20807 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20809 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20811 PR rtl-optimization/70478
20812 * lra-constraints.c: Reverse the last patch.
20814 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20816 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20817 Add comment for WCHAR_T.
20819 2017-04-08 Martin Liska <mliska@suse.cz>
20822 2017-04-07 Martin Liska <mliska@suse.cz>
20825 * ipa-split.c (split_function): Add function part to a same comdat
20828 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20831 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20833 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20835 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20836 (vec_pairing): ...to this.
20837 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20838 (rs6000_sched_init): Adjust for name change.
20839 (struct rs6000_sched_context): Likewise.
20840 (rs6000_init_sched_context): Likewise.
20841 (rs6000_set_sched_context): Likewise.
20843 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20849 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20851 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20852 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20853 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20855 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20857 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20859 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20861 PR rtl-optimization/70703
20862 * ira-color.c (update_conflict_hard_regno_costs): Use
20863 int64_t instead of HOST_WIDE_INT.
20865 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20867 PR rtl-optimization/70478
20868 * lra-constraints.c (process_alt_operands): Disfavor alternative
20869 insn memory operands.
20871 2017-04-07 Jeff Law <law@redhat.com>
20873 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20874 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20876 2017-04-07 Martin Liska <mliska@suse.cz>
20879 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20880 Show error message instead of an ICE.
20882 2017-04-07 Martin Liska <mliska@suse.cz>
20885 * ipa-split.c (split_function): Add function part to a same comdat
20888 2017-04-07 Richard Biener <rguenther@suse.de>
20890 PR middle-end/80341
20891 * tree.c (get_unwidened): Also handle ! for_type case for
20893 * convert.c (do_narrow): Split out from ...
20894 (convert_to_integer_1): ... here. Do not pass final truncation
20895 type to get_unwidened for TRUNC_DIV_EXPR.
20897 2017-04-07 Richard Biener <rguenther@suse.de>
20899 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20901 (aff_combination_const): Adjust.
20902 (aff_combination_scale): Likewise.
20903 (aff_combination_add_elt): Likewise.
20904 (aff_combination_add_cst): Likewise.
20905 (aff_combination_convert): Likewise.
20906 (add_elt_to_tree): Likewise. Remove unused argument.
20907 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20909 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20911 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20913 * config/arm/arm.c (arm_default_short_enums): Use
20914 ARM_DEFAULT_SHORT_ENUMS.
20915 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20917 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20920 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20921 members with redundant out-of-class redeclaration.
20923 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20926 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20927 * config/i386/i386.md (*zero_extendsidi2):
20928 Add (?*x,*x) and (?*v,*v) alternatives.
20930 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20933 * config/i386/i386.c (ix86_expand_builtin)
20934 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20935 mode from insn data. Convert operands to insn operand mode.
20936 Copy operands that don't satisfy insn predicate to a register.
20938 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20940 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20943 2017-04-06 Richard Biener <rguenther@suse.de>
20945 PR tree-optimization/80334
20946 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20947 preserve alignment of accesses.
20949 2017-04-06 Richard Biener <rguenther@suse.de>
20951 PR tree-optimization/80262
20952 * tree-sra.c (build_ref_for_offset): Preserve address-space
20954 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20955 Drop useless address-space information on MEM_REF offsets.
20957 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20959 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20961 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20963 PR rtl-optimization/70703
20964 * ira-color.c (update_conflict_hard_regno_costs): Use
20965 HOST_WIDE_INT instead of long.
20967 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20970 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20971 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20973 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20974 for x86_64 target. Handle -m3dnowa option.
20976 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20978 PR rtl-optimization/70703
20979 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20980 (update_conflict_hard_regno_costs): Use long instead of unsigned
20981 arithmetic for cost calculation.
20983 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20984 Bernd Edlinger <bernd.edlinger@hotmail.de>
20987 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20990 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20993 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20994 ptr_mode with Pmode throughout.
20995 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20996 into probe_stack_range and use DImode.
20998 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21001 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21002 call_eh_return is true.
21004 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21006 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21007 Initialize last_match_fntype_index.
21009 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21012 * tree-nvr.c: Include internal-fn.h.
21013 (pass_return_slot::execute): Ignore internal calls without
21016 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21017 Richard Biener <rguenther@suse.de>
21020 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21021 captures used multiple times, except for the last use.
21022 * generic-match-head.c: Include gimplify.h.
21024 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21026 PR tree-optimization/79390
21027 * target.h (struct noce_if_info): Declare.
21028 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21029 * target.def (noce_conversion_profitable_p): New target hook.
21030 * ifcvt.h (struct noce_if_info): New type, moved from ...
21031 * ifcvt.c (struct noce_if_info): ... here.
21032 (noce_conversion_profitable_p): Renamed to ...
21033 (default_noce_conversion_profitable_p): ... this. No longer
21035 (noce_try_store_flag_constants, noce_try_addcc,
21036 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21037 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21038 instead of noce_conversion_profitable_p.
21039 * config/i386/i386.c: Include ifcvt.h.
21040 (ix86_option_override_internal): Don't override
21041 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21042 (ix86_noce_conversion_profitable_p): New function.
21043 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21044 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21045 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21046 * doc/tm.texi: Regenerated.
21048 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21050 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21053 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21056 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21057 instructions for small multiply cores.
21059 2017-04-04 Jeff Law <law@redhat.com>
21061 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21063 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21064 that are not set by the loop over the elements.
21066 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21069 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21070 int mode, convert_modes it to mode as unsigned, otherwise use
21071 lowpart_subreg to mode rather than SImode.
21072 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21073 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21074 Use DImode instead of SImode for the shift count operand.
21075 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21078 2017-04-04 Richard Biener <rguenther@suse.de>
21080 PR middle-end/80281
21081 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21082 arithmetic done for the negate or the plus. Simplify.
21083 (A - (-B) -> A + B): Likewise.
21084 * fold-const.c (split_tree): Make sure to not negate pointers.
21086 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21088 PR rtl-optimization/60818
21089 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21090 a compare of comparisons with the thing compared if this results
21091 in a different machine mode.
21093 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21095 * alias.c (base_alias_check): Fix typo in comment.
21096 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21097 * cgraphunit.c (symbol_table::compile): Likewise.
21098 * collect2.c (maybe_run_lto_and_relink): Likewise.
21099 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21100 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21101 * config/avr/avr.c (avr_map_op_t): Likewise.
21102 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21103 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21104 * config/epiphany/epiphany.md (movcc): Likewise.
21105 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21106 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21108 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21109 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21110 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21111 * config/sh/sh.c (sh_rtx_costs): Likewise.
21112 * fold-const.c (fold_truth_andor): Likewise.
21113 * genautomata.c (collapse_flag): Likewise.
21114 * gengtype.h (struct type::u::s): Likewise.
21115 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21116 * input.c (FORMAT_AMOUNT): Likewise.
21117 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21118 (known_aggs_to_agg_replacement_list): Likewise.
21119 * ipa-inline-analysis.c: Likewise.
21120 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21121 * ipa-polymorphic-call.c
21122 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21123 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21124 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21126 * modulo-sched.c (apply_reg_moves): Likewise.
21127 * omp-expand.c (build_omp_regions_1): Likewise.
21128 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21129 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21130 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21131 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21132 * value-prof.c: Likewise.
21133 * var-tracking.c (val_reset): Likewise.
21135 2017-04-03 Richard Biener <rguenther@suse.de>
21137 PR tree-optimization/80275
21138 * fold-const.c (split_address_to_core_and_offset): Handle
21141 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21143 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21144 descriptors is at least equal to that of functions.
21146 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21148 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21150 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21153 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21154 (mov<IMOD4:mode>): New expander.
21155 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21157 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21159 PR rtl-optimization/79405
21160 * fwprop.c (propagations_left): New variable.
21161 (forward_propagate_into): Decrement it.
21162 (fwprop_init): Initialize it.
21163 (fw_prop): If the variable has reached zero, stop propagating.
21164 (fwprop_addr): Ditto.
21166 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21169 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21170 a FUNCTION_DECL, pass it as decl instead of origin to
21173 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21175 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21178 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21181 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21182 TARGET_VSX_SMALL_INTEGER.
21184 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21186 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21187 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21189 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21191 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21192 extraction from odd-numbered MSA register.
21194 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21196 PR middle-end/80173
21197 * expmed.c (store_bit_field_1): Don't attempt to create
21198 a word subreg out of hard registers wider than word if they
21199 have HARD_REGNO_NREGS of 1 for their mode.
21201 PR middle-end/80163
21202 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21203 conversions to integer types wider than word and pointer.
21206 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21207 (rtx_equal_for_cselib_p): Pass 0 to it.
21208 * cselib.c (cselib_hasher::equal): Likewise.
21209 (rtx_equal_for_cselib_1): Add depth argument. If depth
21210 is 128, don't look up VALUE locs and punt. Increment
21211 depth in recursive calls when walking VALUE locs.
21213 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21215 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21216 (make_gcov_file_name): Use the canonical path name for generating
21218 (read_line): Fix handling of files with ascii null bytes.
21220 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21222 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21223 to initialise a vector register instead
21224 of using a const_int.
21226 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21228 PR translation/80189
21229 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21230 diagnostic messages.
21232 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21235 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21236 (dfp_diex_<mode>): Update mode of operand 1.
21237 * doc/extend.texi (dxex, dxexq): Document change to return type.
21238 (diex, diexq): Document change to argument type.
21240 2017-03-30 Martin Jambor <mjambor@suse.cz>
21243 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21244 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21245 it reflects the signature changes performed at the callee side.
21246 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21247 to cgraph_build_function_type_skip_args.
21248 (build_function_decl_skip_args): Adjust call to the above function.
21250 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21253 * config/i386/sse.md
21254 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21255 register as dest whenever it is a MEM not rtx_equal_p to the
21256 corresponding dup operand, and when forcing into reg move the
21257 reg into the memory afterwards.
21258 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21259 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21260 for the force_reg mode.
21261 (avx512vl_vextractf128<mode>): Use register as dest either
21262 always when a MEM, or when it is a MEM not rtx_equal_p to the
21263 corresponding dup operand, or even not when it is a CONST_VECTOR
21264 depending on the mode and lo vs. hi.
21265 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21267 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21268 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21269 Likewise. Require that operands[2] is even.
21270 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21271 Remove extraneous parens. Require that operands[2] is a multiple
21273 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21274 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21275 disallow memory then.
21277 2017-03-30 Richard Biener <rguenther@suse.de>
21279 PR tree-optimization/77498
21280 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21281 to non-constants over backedges.
21283 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21285 PR rtl-optimization/80233
21286 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21287 as last_combined_insn. Do not test for BARRIER_P separately.
21289 2017-03-29 Andreas Schwab <schwab@suse.de>
21292 * calls.c (prepare_call_address): Convert funexp to Pmode before
21293 copying to temp reg.
21295 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21297 PR tree-optimization/80158
21298 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21299 Handle possible future case of more than one alternate
21301 (replace_rhs_if_not_dup): Likewise.
21302 (replace_one_candidate): Likewise.
21304 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21306 PR rtl-optimization/80193
21307 * ira.c (ira): Do not check allocation for LRA.
21309 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21311 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21312 (nvptx_output_simt_exit): Declare.
21313 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21314 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21315 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21316 (nvptx_declare_function_name): ...here. Emit declaration of local
21317 memory space buffer for omp_simt_enter insn.
21318 (nvptx_output_unisimt_switch): New.
21319 (nvptx_output_softstack_switch): New.
21320 (nvptx_output_simt_enter): New.
21321 (nvptx_output_simt_exit): New.
21322 * config/nvptx/nvptx.h (struct machine_function): New fields
21323 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21324 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21325 (UNSPECV_SIMT_EXIT): Ditto.
21326 (omp_simt_enter_insn): New insn.
21327 (omp_simt_enter): New expansion.
21328 (omp_simt_exit): New insn.
21329 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21331 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21332 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21333 (expand_GOMP_SIMT_EXIT): New.
21334 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21335 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21336 (GOMP_SIMT_EXIT): Ditto.
21337 * target-insns.def (omp_simt_enter): New insn.
21338 (omp_simt_exit): Ditto.
21339 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21341 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21342 (lower_rec_input_clauses): Likewise.
21343 (lower_lastprivate_clauses): Handle SIMT privatization.
21345 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21346 (ompdevlow_adjust_simt_enter): New.
21347 (find_simtpriv_var_op): New.
21348 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21349 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21351 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21352 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21353 (copy_decl_for_dup_finish): Ditto.
21355 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21357 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21360 * config/i386/i386.c (ix86_option_override_internal): Always
21361 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21363 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21365 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21367 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21369 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21370 mark new edge's irreducible flag accordign to it.
21371 (vect_do_peeling): Check loop preheader edge's irreducible flag
21372 and pass it to function slpeel_add_loop_guard.
21374 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21376 PR tree-optimization/80218
21377 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21378 Update block frequencies and counts.
21380 2017-03-28 Richard Biener <rguenther@suse.de>
21382 PR tree-optimization/78644
21383 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21384 of a simplification result we may not use it at all.
21386 2017-03-28 Richard Biener <rguenther@suse.de>
21389 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21390 without arguments, generate default definition of a SSA name.
21392 2017-03-28 Richard Biener <rguenther@suse.de>
21394 PR middle-end/80222
21395 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21396 TYPE_REF_CAN_ALIAS_ALL references.
21397 * fold-const.c (fold_indirect_ref_1): Likewise.
21399 2017-03-28 Martin Liska <mliska@suse.cz>
21402 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21403 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21405 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21406 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21408 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21409 (EXTRA_SPECS): Define.
21410 (SUBTARGET_EXTRA_SPECS): Likewise.
21411 (SUBTARGET_CPP_SPEC): Likewise.
21412 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21413 SUBTARGET_EXTRA_SPECS.
21414 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21416 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21418 * config/arc/simdext.md (vst64_insn): Update pattern.
21419 (vld32wh_insn): Likewise.
21420 (vld32wl_insn): Likewise.
21421 (vld64_insn): Likewise.
21422 (vld32_insn): Likewise.
21424 2017-03-28 Marek Polacek <polacek@redhat.com>
21427 * fold-const.c (fold_comparison): Use protected_set_expr_location
21428 instead of SET_EXPR_LOCATION.
21430 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21432 * tree.c (add_expr): Avoid name lookup warning.
21434 2017-03-27 Jeff Law <law@redhat.com>
21436 PR tree-optimization/80216
21437 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21438 function name. Limit recursion depth.
21439 (record_temporary_equivalences): Corresponding changes.
21441 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21443 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21446 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21449 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21451 * cfgcleanup.c (reg_note_cfa_p): New array.
21452 (insns_have_identical_cfa_notes): New function.
21453 (old_insns_match_p): Don't cross-jump in between /f
21454 and non-/f instructions. If both i1 and i2 are frame related,
21455 verify all CFA notes, their order and content.
21457 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21460 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21461 HImode and SImode with zero extend to DImode to one insn.
21462 (bswap<mode>2_extenddi): Likewise.
21463 (bswapsi2_extenddi): Likewise.
21464 (bswaphi2_extendsi): Likewise.
21465 (bswaphi2): Combine bswap HImode and SImode into one insn.
21466 Separate memory insns from swapping register.
21467 (bswapsi2): Likewise.
21468 (bswap<mode>2): Likewise.
21469 (bswaphi2_internal): Delete, no longer used.
21470 (bswapsi2_internal): Likewise.
21471 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21472 store, and gpr<-gpr swap insns.
21473 (bswap<mode>2_store): Likewise.
21474 (bswaphi2_reg): Register only splitter, combine with the splitter.
21475 (bswaphi2 splitter): Likewise.
21476 (bswapsi2_reg): Likewise.
21477 (bswapsi2 splitter): Likewise.
21478 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21479 the insns into load, store, and register/register insns.
21480 (bswapdi2_ldbrx): Likewise.
21481 (bswapdi2_load): Likewise.
21482 (bswapdi2_store): Likewise.
21483 (bswapdi2_reg): Likewise.
21485 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21487 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21488 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21490 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21493 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21495 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21496 special handling for target option conflicts between dform
21497 options (-mpower9-dform, -mpower9-dform-vector,
21498 -mpower9-dform-scalar) and -mno-direct-move.
21500 2017-03-27 Richard Biener <rguenther@suse.de>
21502 PR tree-optimization/80181
21503 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21505 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21507 * config/arc/predicates.md (move_double_src_operand): Replace the
21508 call to move_double_src_operand with a call to address_operand.
21510 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21512 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21513 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21514 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21516 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21518 * config/arc/predicates.md (long_immediate_loadstore_operand):
21519 Consider scaled addresses cases.
21521 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21523 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21524 restored when in interrupt.
21525 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21526 doesn't have delay slot.
21528 2017-03-27 Richard Biener <rguenther@suse.de>
21531 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21532 inlined thunk clones.
21534 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21537 * asan.c (instrument_derefs): Copy over last operand from
21538 original COMPONENT_REF to the new COMPONENT_REF with
21539 DECL_BIT_FIELD_REPRESENTATIVE.
21540 * ubsan.c (instrument_object_size): Likewise.
21542 2017-03-27 Richard Biener <rguenther@suse.de>
21544 PR tree-optimization/80170
21545 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21546 sure DR/SCEV didnt fold in constants we do not see when looking
21547 at the reference base alignment.
21549 2017-03-27 Richard Biener <rguenther@suse.de>
21551 PR middle-end/80171
21552 * gimple-fold.c (fold_ctor_reference): Properly guard against
21553 NULL return value from canonicalize_constructor_val.
21555 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21558 * config/i386/i386.c (ix86_expand_builtin)
21559 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21560 flags reg setting and flags reg using instructions.
21561 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21562 clobbering instructions to zero extend op2.
21564 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21566 * doc/install.texi (Configuration) <--with-aix-soname>:
21567 Update link to AIX ld.
21569 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21571 PR rtl-optimization/80160
21572 PR rtl-optimization/80159
21573 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21574 reg_alternate_class into account.
21576 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21579 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21580 to consider in curr_insn_transform.
21582 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21584 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21585 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21586 and emit_mode_inner.
21588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21590 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21591 argument to the overloaded builtin variants. Use the new flag to
21592 deprecate certain builtin variants.
21593 * config/s390/s390-builtin-types.def: Add new builtin types.
21594 * config/s390/s390-builtins.h: Support new flags field for
21595 overloaded builtins.
21596 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21597 (s390_macro_to_expand): Enable vector float data type.
21598 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21599 builtins by incrementing the __VEC__ version number.
21600 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21602 (s390_resolve_overloaded_builtin): Emit error messages depending
21603 on the builtin flags.
21604 * config/s390/s390.c (s390_expand_builtin): Support additional
21605 flags argument. Change error message to match the messages
21606 emitted in s390-c.c.
21607 * config/s390/s390.md: New UNSPEC_* constants.
21608 (op_type): Add new instruction types.
21609 * config/s390/vecintrin.h: Add new builtins and test data class
21611 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21612 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21613 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21614 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21615 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21616 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21618 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21619 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21620 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21621 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21623 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21624 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21625 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21626 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21628 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21629 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21630 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21631 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21632 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21633 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21634 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21636 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21637 ("vec_scatter_element<V_HW_4:mode>_DI")
21638 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21639 ("vec_fpint<mode>", "vflls")
21640 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21641 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21642 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21643 ("*vec_cmphe<mode>_cc"): ... these.
21645 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21646 mode constant instead of magic value.
21648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21650 * config/s390/s390.c (s390_expand_vec_compare): Support other
21651 vector floating point modes than just V2DF.
21652 (s390_expand_vcond): Likewise.
21653 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21654 (s390_cannot_change_mode_class): Prevent mode changes between TF
21655 and V1TF in vector registers.
21656 * config/s390/s390.md (DF, SF): New mode attributes.
21657 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21658 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21659 SFmode support for VRs.
21660 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21662 (VFT, VF_HW): New mode iterators.
21663 (vw, sdx): New mode attributes.
21664 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21665 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21666 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21667 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21668 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21669 also the new vector floating point modes. Renaming to ...
21671 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21672 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21673 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21674 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21675 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21676 ("vec_unordered<mode>"): ... these.
21678 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21679 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21680 ("*vec_extendv2df"): New insn definitions.
21682 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21684 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21685 ("mulditi3_2", "*muldi3_sign"): New patterns.
21686 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21687 rename the pattern definition.
21689 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21691 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21693 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21697 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21698 instruction if possible.
21699 * config/s390/vector.md (vec_halfnumelts): New mode
21701 ("*vec_vllezlf<mode>"): New pattern.
21703 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21705 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21706 ("popcountv4si2", "popcountv2di2"): Rename to ...
21707 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21708 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21710 ("popcount<mode>2_vxe"): New pattern.
21712 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21714 * common/config/s390/s390-common.c (processor_flags_table): Add
21716 * config.gcc: Add arch12.
21717 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21718 Default to arch12 for unknown CPU model numbers.
21719 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21720 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21721 PROCESSOR_max sanity check.
21722 * config/s390/s390-opts.h (enum processor_type): Add
21724 * config/s390/s390.c (processor_table): Add arch12.
21725 (s390_expand_builtin): Add check for B_VXE flag.
21726 (s390_issue_rate): Add PROCESSOR_ARCH12.
21727 (s390_get_sched_attrmask): Likewise.
21728 (s390_get_unit_mask): Likewise.
21729 (s390_sched_score): Enable z13 scheduling for arch12.
21730 (s390_sched_reorder): Likewise.
21731 (s390_sched_variable_issue): Likewise.
21732 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21734 (s390_tune_attr): Use z13 scheduling also for arch12.
21735 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21736 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21737 (TARGET_VXE_P): New macros.
21738 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21739 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21740 * config/s390/s390.opt: Add arch12 as processor_type.
21742 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21744 * config/s390/s390.md
21745 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21746 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21747 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21749 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21750 Rename expanders to ...
21752 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21753 ("fixuns_truncdddi2_emu"): ... these.
21755 ("fixuns_trunc<mode>si2_emu"): New expander.
21757 ("*fixuns_truncdfdi2_z13"): Rename to ...
21758 ("*fixuns_truncdfdi2_vx"): ... this.
21760 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21762 * config/s390/2964.md: Remove the single element vector compare
21763 instructions which are no longer used.
21764 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21766 (s390_canonicalize_comparison): Remove handling of DFmode
21768 (s390_expand_vec_compare_scalar): Remove function.
21769 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
21770 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
21772 ("*cmp<mode>_ccs"): Add wfcdb instruction.
21774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21776 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
21778 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
21779 will anyway by matched by mov<mode>_64dfp.
21781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21783 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
21784 vlef/vstef. Add missing operand to vleif.
21786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21788 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
21789 pair for all vector types with 64 bit elements.
21790 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21791 * config/s390/vector.md (V_HW_64): ... here.
21792 (V_128_NOSINGLE): New mode iterator.
21793 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21794 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21795 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21796 ("*vec_load_pairv2di"): Change to ...
21797 ("*vec_load_pair<mode>"): ... this one.
21799 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21801 * config/s390/constraints.md: Add comments.
21802 (jKK): Reject element sizes > 8 bytes.
21803 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21805 * config/s390/s390.md: Add the s_operand checks formerly in
21806 s390_split_ok_p to various splitters where they are still
21808 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21809 for 128 bit vectors. Plus two splitters.
21811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21813 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21816 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21819 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21820 error if the boundary argument is not constant.
21822 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21824 PR rtl-optimization/80112
21825 * loop-doloop.c (doloop_condition_get): Don't check condition
21826 if cmp isn't SET with IF_THEN_ELSE src.
21828 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21830 PR tree-optimization/80158
21831 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21832 replacing a candidate statement, also replace it for the
21833 candidate's alternate interpretation.
21834 (replace_rhs_if_not_dup): Likewise.
21835 (replace_one_candidate): Likewise.
21837 2017-03-24 Richard Biener <rguenther@suse.de>
21839 PR tree-optimization/80167
21840 * graphite-isl-ast-to-gimple.c
21841 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21843 (translate_isl_ast_to_gimple::get_rename): Likewise.
21845 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21847 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21848 handling of certain combinations of target options, including the
21849 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21850 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21852 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21855 * config/arm/arm.md (*load_multiple): Add reload_completed to
21856 matching condition.
21858 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21859 Richard Biener <rguenth@suse.de>
21861 PR tree-optimization/79908
21862 PR tree-optimization/80136
21863 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21864 been cast away, gimplify_and_add suffices.
21866 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21868 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21870 2017-03-23 Richard Biener <rguenther@suse.de>
21872 PR tree-optimization/80032
21873 * gimplify.c (gimple_push_cleanup): Forced unconditional
21874 cleanups still have to go to the conditional_cleanups
21877 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21879 PR tree-optimization/80072
21880 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21882 (next_operand_entry_id): Change type to unsigned int.
21883 (sort_by_operand_rank): Make sure to return the right return value
21884 even if unsigned fields are bigger than INT_MAX.
21885 (struct oecount): Change cnt and id type to unsigned int.
21886 (oecount_hasher::equal): Formatting fix.
21887 (oecount_cmp): Make sure to return the right return value
21888 even if unsigned fields are bigger than INT_MAX.
21889 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21892 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21893 TREE_READONLY on result if writing it more than once.
21896 * doc/invoke.texi (-fsanitize=thread): Document that with
21897 -fnon-call-exceptions atomics are not able to throw
21901 * tsan.c: Include tree-eh.h.
21902 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21903 maybe_clean_or_replace_eh_stmt where needed.
21904 (instrument_memory_accesses): Add cfg_changed argument.
21905 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21906 if it returned true.
21907 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21909 PR rtl-optimization/63191
21910 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21911 wrapper function, moved the whole old content into ...
21912 (ix86_delegitimize_address_1): ... this. New inline function.
21913 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21914 true as last argument instead of ix86_delegitimize_address.
21916 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21918 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21919 cortexa57_branch_cost.
21921 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21923 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21925 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21928 * doc/md.texi (Constraints): Document wA constraint.
21929 * config/rs6000/constraints.md (wA): New.
21930 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21931 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21932 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21933 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21935 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21938 * gimplify.c (is_oacc_declared): New function.
21939 (oacc_default_clause): Use it to set default flags for acc declared
21940 variables inside parallel regions.
21941 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21942 declared variables.
21943 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21944 declare attribute to any decl as necessary.
21946 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21949 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21950 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21951 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21952 (arm_arch_lpae): This.
21953 * config/arm/arm.c (arm_arch7ve): Rename into ...
21954 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21955 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21958 2017-03-22 Martin Liska <mliska@suse.cz>
21961 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21962 error message instead of an ICE.
21964 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21966 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21968 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21970 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21972 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21975 2017-03-21 Martin Sebor <msebor@redhat.com>
21977 * doc/extend.texi: Use "cannot" instead of "can't."
21978 * doc/hostconfig.texi: Same.
21979 * doc/install.texi: Same.
21980 * doc/invoke.texi: Same.
21981 * doc/loop.texi: Same.
21982 * doc/md.texi: Same.
21983 * doc/objc.texi: Same.
21984 * doc/rtl.texi: Same.
21985 * doc/tm.texi: Same.
21986 * doc/tm.texi.in: Same.
21987 * doc/trouble.texi: Same.
21989 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21992 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21993 (collect_checksum_attributes): Set it.
21994 (die_checksum_ordered): Use it.
21996 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21998 PR tree-optimization/79908
21999 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22000 change: For a VA_ARG whose LHS has been cast away, use
22001 force_gimple_operand to construct the side effects.
22003 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22005 PR translation/80001
22006 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22007 more amenable to translation.
22008 (oacc_loop_auto_partitions): Likewise.
22010 2017-03-21 Marek Polacek <polacek@redhat.com>
22011 Martin Sebor <msebor@redhat.com>
22013 PR tree-optimization/80109
22014 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22015 on INTEGRAL_TYPE_P.
22017 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22018 Segher Boessenkool <segher@kernel.crashing.org>
22021 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22022 check reg_used_between_p between insn and one of succ or succ2
22023 depending on if succ is artificial insn not inserted into insn
22026 2017-03-21 Martin Liska <mliska@suse.cz>
22028 PR gcov-profile/80081
22029 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22030 * doc/gcc.texi: Include gcov-dump stuff.
22031 * doc/gcov-dump.texi: New file.
22033 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22035 PR rtl-optimization/79150
22036 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22037 conditional jump, if the jump is the last insn of the loop.
22039 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22040 Richard Biener <rguenth@suse.de>
22042 PR tree-optimization/79908
22043 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22044 been cast away, use force_gimple_operand to construct the side
22047 2017-03-21 Martin Liska <mliska@suse.cz>
22049 PR libfortran/79956
22050 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22053 2017-03-21 Brad Spengler <spender@grsecurity.net>
22056 * plugin.c (htab_hash_plugin): New function.
22057 (add_new_plugin): Use it and adjust.
22058 (parse_plugin_arg_opt): Adjust.
22059 (init_one_plugin): Likewise.
22061 2017-03-21 Richard Biener <rguenther@suse.de>
22063 PR tree-optimization/80032
22064 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22065 if set force the cleanup to happen unconditionally.
22066 (gimplify_target_expr): Push inserted clobbers with force_uncond
22067 to avoid them being removed by control-dependent DCE.
22069 2017-03-21 Richard Biener <rguenther@suse.de>
22071 PR tree-optimization/80122
22072 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22073 va_arg_pack_len when the inlined call stmt requires pack
22075 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22077 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22080 * tsan.c (instrument_builtin_call): If the memory model argument
22081 is not a constant, assume it is valid.
22084 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22087 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22089 PR rtl-optimization/79910
22090 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22091 if its dest is used by an insn before I2 (other than the combined
22092 insns themselves, which are properly handled already).
22094 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22097 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22099 * combine.c (record_used_regs): New static function.
22100 (try_combine): Handle situations where there is an additional
22101 instruction between I2 and I3 which needs to have a LOG_LINK
22105 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22107 * combine.c (try_combine): Delete redundant i1 test. Call
22108 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22110 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22113 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22114 alternatives 13/14.
22116 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22118 PR tree-optimization/80054
22119 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22120 the optimization if a PHI or any of its arguments is not dominated
22121 by the candidate's basis. Use gphi* rather than gimple* as
22123 (replace_profitable_candidates): Clean up a gimple* variable that
22124 should be a gphi* variable.
22126 2017-03-20 Martin Sebor <msebor@redhat.com>
22129 * doc/extend.texi (attribute constructor): Document present limitation.
22131 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22134 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22135 __POWER9_VECTOR__ #ifdef control, change template definition to
22136 use Power9-specific built-in function.
22137 (vec_any_eq): Likewise.
22138 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22139 to control outcomes from this test.
22140 (vector_ae_<mode>p): For VEC_F modes, likewise.
22142 2017-03-20 Ian Lance Taylor <iant@google.com>
22144 * config/i386/i386.c (ix86_function_regparm): Save an extra
22145 register for -fsplit-stack with DECL_STATIC_CHAIN.
22147 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22150 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22151 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22153 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22155 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22157 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22160 2017-03-20 Marek Polacek <polacek@redhat.com>
22163 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22165 2017-03-20 Richard Biener <rguenther@suse.de>
22167 PR tree-optimization/80113
22168 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22169 allocate extra SSA name for PHI def.
22170 (add_close_phis_to_outer_loops): Likewise.
22171 (add_close_phis_to_merge_points): Likewise.
22172 (copy_loop_close_phi_args): Likewise.
22173 (copy_cond_phi_nodes): Likewise.
22175 2017-03-20 Martin Liska <mliska@suse.cz>
22177 PR middle-end/79753
22178 * tree-chkp.c (chkp_build_returned_bound): Do not build
22179 returned bounds for a LHS that's not a BOUNDED_P type.
22181 2017-03-20 Martin Liska <mliska@suse.cz>
22185 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22186 COMPLEX_CST and VECTOR_CST.
22188 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22191 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22192 target operand. A new splitter adds the clobber statement in case
22193 the target operand is dead anyway.
22195 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22197 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22198 to age-old versions of binutils and glibc.
22200 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22202 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22204 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22206 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22208 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22210 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22211 requirement for binutils 2.13.
22213 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22215 * combine.c (try_combine): Delete redundant i1 test. Call
22216 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22218 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22220 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22221 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22223 <riscv64-*-elf>: Re-arrange section
22224 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22225 <riscv32-*-linux>: Likewise.
22226 <riscv64-*-elf>: Likewise
22227 <riscv64-*-linux>: Likewise.
22229 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22232 * aarch64.opt(verbose-cost-dump): Fix typo.
22234 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22237 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22238 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22240 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22242 * reload.c (find_reloads): When reloading a nonoffsettable address,
22243 use RELOAD_OTHER for it and its address reloads.
22245 PR rtl-optimization/79910
22246 * combine.c (record_used_regs): New static function.
22247 (try_combine): Handle situations where there is an additional
22248 instruction between I2 and I3 which needs to have a LOG_LINK
22251 2017-03-17 Jeff Law <law@redhat.com>
22253 PR tree-optimization/71437
22254 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22255 conditional in the hash table first.
22256 (vrp_dom_walker::before_dom_children): Extract condition from
22257 ASSERT_EXPR. Record condition, its inverion and any implied
22258 conditions as well.
22260 2017-03-17 Marek Polacek <polacek@redhat.com>
22261 Markus Trippelsdorf <markus@trippelsdorf.de>
22263 PR tree-optimization/80079
22264 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22267 2017-03-17 Richard Biener <rguenther@suse.de>
22269 PR middle-end/80075
22270 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22271 Properly verify the LHS before the RHS possibly claims to be
22273 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22276 2017-03-17 Martin Jambor <mjambor@suse.cz>
22278 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22279 (List of -O2 options): Likewise.
22280 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22283 2017-03-17 Tom de Vries <tom@codesourcery.com>
22285 * gcov-dump.c (print_usage): Print bug_report_url.
22287 2017-03-17 Richard Biener <rguenther@suse.de>
22289 PR middle-end/80050
22290 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22291 (parser::peek): Likewise.
22293 2017-03-17 Richard Biener <rguenther@suse.de>
22295 PR tree-optimization/80048
22296 * sese.c (free_sese_info): Properly release rename_map and
22297 copied_bb_map elements.
22299 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22301 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22302 Add linked-list forward and backlinks. Insert on
22303 construction, remove on destruction.
22304 (class pass_store_merging): Add m_stores_head field.
22305 (pass_store_merging::terminate_and_process_all_chains):
22306 Iterate over m_stores_head list.
22307 (pass_store_merging::terminate_all_aliasing_chains):
22309 (pass_store_merging::execute): Check for debug stmts first.
22310 Push new chains onto the m_stores_head stack.
22312 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22315 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22316 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22317 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22319 2017-03-16 Jeff Law <law@redhat.com>
22321 PR tree-optimization/71437
22322 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22323 member function. Implementation moved into after_dom_children
22324 member function and into the threader's thread_outgoing_edges
22326 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22327 some code into new thread_outgoing_edges.
22328 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22329 definition. Simplify marker handling (do it here). Assume we always
22330 have the available expression and the const/copies tables.
22331 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22333 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22334 * tree-vrp.c (equiv_stack): No longer file scoped.
22335 (vrp_dom_walker): New class.
22336 (vrp_dom_walker::before_dom_children): New member function.
22337 (vrp_dom_walker::after_dom_children): Likewise.
22338 (identify_jump_threads): Setup domwalker. Use it rather than
22339 walking edges in a random order by hand. Simplify setup/finalization.
22340 (finalize_jump_threads): Remove.
22341 (vrp_finalize): Do not call identify_jump_threads here.
22342 (execute_vrp): Do it here instead and call thread_through_all_blocks
22345 PR tree-optimization/71437
22346 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22348 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22350 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22351 (dom_opt_dom_walker::thread_across_edge): Remove
22352 handle_dominating_asserts argument. All callers changed.
22353 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22354 changes. Remove calls to lhs_of_dominating_assert. Other
22355 uses of handle_dominating_asserts turn into unconditional code
22356 (simplify_control_stmt_condition_1): Likewise.
22357 (simplify_control_stmt_condition): Likewise.
22358 (thread_through_normal_block, thread_across_edge): Likewise.
22359 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22360 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22361 object if it is not an SSA_NAME.
22362 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22363 before calling into the VRP specific simplifiers.
22364 (identify_jump_threads): Remove handle_dominating_asserts
22367 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22370 * tree-diagnostic.c (default_tree_printer): No longer static.
22371 * tree-diagnostic.h (default_tree_printer): New prototype.
22373 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22375 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22376 Change ins into fmov.
22378 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22380 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22381 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22382 Use h_con constraint for operand 1.
22383 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22384 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22386 2017-03-15 Jeff Law <law@redhat.com>
22388 PR tree-optimization/71437
22389 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22390 (record_temporary_equivalences): Use it.
22392 PR tree-optimization/71437
22393 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22394 tree-ssa-scopedtables.
22395 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22396 (record_conditions, record_cond, vuse_eq): Likewise.
22397 (record_edge_info): Adjust to API tweak of record_conditions.
22398 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22399 (record_temporary_equivalences, optimize_stmt): Likewise.
22400 (eliminate_redundant_computations): Likewise.
22401 (record_equivalences_from_stmt): Likewise.
22402 * tree-ssa-scopedtables.c: Include options.h and params.h.
22403 (vuse_eq): New function, moved from tree-ssa-dom.c
22404 (build_and_record_new_cond): Likewise.
22405 (record_conditions): Likewise. Accept vector of conditions rather
22406 than edge_equivalence structure for first argument.
22407 for the first argument.
22408 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22409 from tree-ssa-dom.c.
22410 (avail_exprs_stack::record_cond): Likewise.
22411 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22412 from tree-ssa-dom.c.
22413 (avail_exprs_stack): Add new member functions lookup_avail_expr
22415 (record_conditions): Declare.
22417 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22420 * lra-constraints.c (process_alt_operands): Increase reject for
22421 reloading an input/output operand.
22423 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22426 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22427 insns to convert from signed/unsigned char/short to IEEE 128-bit
22429 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22431 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22434 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22435 subreg of inner mode for values already in registers.
22437 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22439 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22440 iteration reg is used after the loop.
22442 2017-03-14 Martin Sebor <msebor@redhat.com>
22444 PR tree-optimization/79800
22445 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22446 precision in negative-positive range.
22447 (format_floating): Call non-const overload with adjusted precision.
22449 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22452 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22455 2017-03-14 Martin Sebor <msebor@redhat.com>
22457 PR middle-end/80020
22458 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22459 * builtins.def (aligned_alloc): Use it.
22462 * Makefile.in (GTFILES): Add calls.c.
22463 * calls.c: Include "gt-calls.h".
22465 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22467 PR rtl-optimization/79728
22468 * regs.h (struct target_regs): New field
22469 x_contains_allocatable_regs_of_mode.
22470 (contains_allocatable_regs_of_mode): New macro.
22471 * reginfo.c (init_reg_sets_1): Initialize it, and change
22472 contains_reg_of_mode so it includes global regs as well.
22473 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22474 rather than contains_regs_of_mode.
22476 2017-03-14 Martin Liska <mliska@suse.cz>
22478 * doc/invoke.texi: Document options that can't be combined with
22479 -fcheck-pointer-bounds.
22481 2017-03-14 Martin Liska <mliska@suse.cz>
22483 PR middle-end/79831
22484 * doc/invoke.texi (-Wchkp): Document the option.
22486 2017-03-14 Martin Liska <mliska@suse.cz>
22488 * Makefile.in: Install gcov-dump.
22490 2017-03-14 Martin Liska <mliska@suse.cz>
22492 * multiple_target.c (expand_target_clones): Bail out for
22493 an invalid attribute.
22495 2017-03-14 Richard Biener <rguenther@suse.de>
22497 * alias.c (struct alias_set_entry): Pack properly.
22498 * cfgloop.h (struct loop): Likewise.
22499 * cse.c (struct set): Likewise.
22500 * ipa-utils.c (struct searchc_env): Likewise.
22501 * loop-invariant.c (struct invariant): Likewise.
22502 * lra-remat.c (struct cand): Likewise.
22503 * recog.c (struct change_t): Likewise.
22504 * rtl.h (struct address_info): Likewise.
22505 * symbol-summary.h (function_summary): Likewise.
22506 * tree-loop-distribution.c (struct partition): Likewise.
22507 * tree-object-size.c (struct object_size_info): Likewise.
22508 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22509 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22510 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22511 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22512 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22513 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22514 (struct _stmt_vec_info): Likewise.
22516 2017-03-14 Martin Liska <mliska@suse.cz>
22519 * multiple_target.c (create_dispatcher_calls): Check that
22520 a target can create a function dispatcher.
22522 2017-03-14 Martin Liska <mliska@suse.cz>
22525 * multiple_target.c (expand_target_clones): Drop local.local
22526 flag for default implementation.
22528 2017-03-14 Richard Biener <rguenther@suse.de>
22530 PR tree-optimization/80030
22531 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22533 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22535 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22536 gcc_fallthrough() instead of __attribute__((fallthrough));
22538 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22540 * doc/gcc.texi: Remove "up" link to (DIR).
22541 * doc/gccint.texi: Ditto.
22543 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22545 * doc/install.texi (Specific) <avr>: Remove reference to
22548 2017-03-13 Jeff Law <law@redhat.com>
22550 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22551 attribute rather than comments.
22553 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22554 match_scratch operand is highest.
22556 2017-03-13 Martin Liska <mliska@suse.cz>
22558 PR middle-end/78339
22559 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22560 is a CHKP clone, use original declaration.
22562 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22564 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22565 (arc_conditional_register_usage): Use a different allocation order
22566 when optimizing for size.
22567 * common/config/arc/arc-common.c (arc_option_optimization_table):
22568 Section anchors default on when optimizing for size.
22570 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22572 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22574 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22576 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22577 * config/arc/arc.md (cpu_facility): Add cd variant.
22578 (*movqi_insn): Add code density variant.
22579 (*movhi_insn): Likewise.
22580 (*movqi_insn): Likewise.
22581 (*addsi3_mixed): Likewise.
22582 (subsi3_insn): Likewise.
22584 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22586 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22588 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22590 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22591 expressions with MINUS and UNARY ops.
22593 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22596 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22598 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22599 between vec_select and vector argument.
22600 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22601 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22602 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22603 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22604 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22605 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22607 2017-03-13 Richard Biener <rguenther@suse.de>
22610 * params.def (vect-max-peeling-for-alignment): Fix typo.
22612 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22614 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22615 issue that only occurred with binutils below 2.18.
22617 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22619 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22620 refer to binutils 2.11/2.12 minimum.
22622 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22624 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22625 ftp.kernel.org and simplify binutils requirement.
22627 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22629 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22631 (Optimize Options): Ditto. Also remove redundancy.
22633 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22635 PR translation/79848
22636 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22638 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22639 to G_ to avoid double translation.
22641 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22643 PR translation/79923
22644 * auto-profile.c (get_combined_location): Convert leading
22645 character of diagnostics to lower case and remove trailing period.
22646 (read_profile): Likewise for various diagnostics.
22647 * config/arm/arm.c (arm_option_override): Remove trailing period
22648 from various diagnostics.
22649 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22650 (msp430_expand_delay_cycles): Likewise.
22652 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22655 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22656 full command-line argument, rather than just "str".
22657 (aarch64_validate_march): Likewise.
22658 (aarch64_validate_mtune): Likewise.
22660 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22662 PR rtl-optimization/78911
22663 * lra-assigns.c (must_not_spill_p): New function.
22664 (spill_for): Use it.
22666 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22668 PR tree-optimization/79981
22669 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22670 ATOMIC_COMPARE_EXCHANGE ifn result.
22671 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22672 IFN_ATOMIC_COMPARE_EXCHANGE.
22674 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22677 * opts.c (parse_sanitizer_options): Add missing question mark to
22678 "did you mean" message.
22680 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22682 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22684 (VMULEUH_UNS): Likewise.
22685 (VMULOUB_UNS): Likewise.
22686 (VMULOUH_UNS): Likewise.
22687 * config/rs6000/rs6000.c (builtin_function_type): Remove
22688 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22690 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22693 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22694 x with result of extra_parsing_for_operand_code_0.
22695 (function_reader::extra_parsing_for_operand_code_0): Convert
22696 return type from void to rtx, returning x. When reading
22697 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22698 larger size containing struct block_symbol.
22700 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22703 -mfloat128-hardware without -m64.
22705 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22708 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22709 entries to the case statement that marks unsigned arguments to
22710 overloaded functions.
22712 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22715 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22717 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22720 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22721 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22723 2017-03-10 Martin Liska <mliska@suse.cz>
22727 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22728 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22731 2017-03-10 Olivier Hainque <hainque@adacore.com>
22733 * tree-switch-conversion (array_value_type): Start by resetting
22734 candidate type to it's main variant.
22736 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22738 PR rtl-optimization/79909
22739 * combine.c (try_combine): Use simplify_replace_rtx on individual
22740 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22741 of the whole CALL_INSN_FUNCTION_USAGE.
22743 PR tree-optimization/79972
22744 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22745 get_range_info on SSA_NAMEs. Formatting fixes.
22747 2017-03-10 Richard Biener <rguenther@suse.de>
22748 Jakub Jelinek <jakub@redhat.com>
22750 PR tree-optimization/77975
22751 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22752 edge to be constant.
22753 (get_val_for): For constant x return it. Formatting fix.
22754 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22755 would use the same bases as the current one.
22757 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22759 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22760 instead of vec_select for V1TImode.
22761 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22763 (VSX_LE_128): Add V1TI to this mode iterator.
22764 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22765 (*vsx_le_perm_store_<mode>): Likewise.
22766 (pre-reload splitter for VSX stores): Likewise.
22767 (post-reload splitter for VSX stores): Likewise.
22768 (*vsx_xxpermdi2_le_<mode>): Likewise.
22769 (*vsx_lxvd2x2_le_<mode>): Likewise.
22770 (*vsx_stxvd2x2_le_<mode>): Likewise.
22772 2017-03-09 Michael Eager <eager@eagercon.com>
22774 Correct failures with --enable-checking=yes,rtl.
22776 * config/microblaze/microblaze.c (microblaze_expand_shift):
22777 Replace GET_CODE test with CONST_INT_P and INTVAL test with
22778 test for const0_rtx.
22779 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
22780 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
22782 2017-03-09 Richard Biener <rguenther@suse.de>
22784 PR tree-optimization/79977
22785 * graphite-scop-detection.c (scop_detection::merge_sese):
22786 Handle the case of extra exits to blocks dominating the entry.
22788 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
22790 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22793 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22795 PR rtl-optimization/79949
22796 * lra-constraints.c (process_alt_operands): Check memory when
22797 trying to predict a cycle. Print about the overall increase.
22799 2017-03-09 Richard Biener <rguenther@suse.de>
22801 PR middle-end/79971
22802 * gimple-expr.c (useless_type_conversion_p): Preserve
22803 TYPE_SATURATING for fixed-point types.
22805 2017-03-09 Richard Biener <rguenther@suse.de>
22808 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22809 alignment of BLKmode params.
22811 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22814 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22815 (VALL_NO_V2Q): Likewise.
22817 * config/aarch64/aarch64-simd.md
22818 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22820 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22821 VALL_NO_V2Q mode iterator.
22822 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22824 2017-03-09 Martin Liska <mliska@suse.cz>
22826 PR tree-optimization/79631
22827 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22828 tree_int_cst_sign_bit just for INTEGER constants.
22830 2017-03-09 Martin Liska <mliska@suse.cz>
22834 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22837 2017-03-09 Marek Polacek <polacek@redhat.com>
22840 * tree.c (inchash::add_expr): Handle TREE_VEC.
22842 2017-03-09 Martin Liska <mliska@suse.cz>
22845 (chkp_narrow_size_and_offset): New function.
22846 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22847 (void chkp_parse_bit_field_ref): New function.
22848 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22849 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22851 2017-03-09 Martin Liska <mliska@suse.cz>
22854 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22855 (chkp_find_bounds_1): Remove gcc_unreachable.
22857 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22860 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22861 BUILT_IN_SYNC*, determine the access type from the size suffix and
22862 always build a MEM_REF with that type. Handle forgotten
22863 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22866 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22867 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22868 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22869 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22870 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22871 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22872 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22873 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22874 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22875 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22876 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22877 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22878 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22879 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22880 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22881 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22882 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22883 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22884 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22885 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22886 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22887 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22888 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22889 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22890 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22891 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22892 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22893 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22894 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22895 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22896 definitions outside of __OPTIMIZE__ guarded section.
22899 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22900 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22901 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22902 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22905 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22907 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22908 ("vfenez<mode>"): Add missing constraints.
22910 2017-03-08 Martin Sebor <msebor@redhat.com>
22913 * config/nds32/nds32.c (nds32_option_override):
22914 Fix misspelled diagnostic.
22916 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22919 * gimplify.c (gimplify_omp_for): Replace index var in outer
22920 taskloop statement with an artificial variable and add
22921 OMP_CLAUSE_PRIVATE clause for it.
22923 2017-03-08 Richard Biener <rguenther@suse.de>
22925 PR tree-optimization/79955
22926 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22927 for accesses that are completely outside of the variable.
22929 2017-03-08 Andrew Haley <aph@redhat.com>
22931 PR tree-optimization/79943
22932 * tree-ssa-loop-split.c (compute_new_first_bound): When
22933 calculating the new upper bound, (END-BEG) should be added, not
22936 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22938 * config/avr/avr.md (setmemhi): Make sure match_dup
22939 operand number comes before match_scratch.
22941 2017-03-08 Richard Biener <rguenther@suse.de>
22943 PR tree-optimization/79920
22944 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22945 with ncopies == 1 to ...
22946 (vect_transform_slp_perm_load): ... here. Properly compute
22947 all element loads by iterating VF times over the group. Do
22948 not handle ncopies (computed in a broken way) in
22949 vect_create_mask_and_perm.
22951 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22954 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22955 is a uniform vector, use uniform_vector_p return value instead of
22956 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22958 2017-03-07 Marek Polacek <polacek@redhat.com>
22960 PR middle-end/79809
22961 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22962 (alloca_call_type): Likewise.
22964 2017-03-07 Martin Liska <mliska@suse.cz>
22966 * gcov.c (process_args): Put comment to correct location.
22968 2017-03-07 Martin Liska <mliska@suse.cz>
22970 PR middle-end/68270
22971 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22972 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22973 (chkp_narrow_bounds_for_field): Likewise.
22974 (chkp_parse_array_and_component_ref): Pass one more argument to
22977 2017-03-07 Richard Biener <rguenther@suse.de>
22979 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22982 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22984 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22985 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22987 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22990 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22991 to end of description.
22992 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22994 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22996 PR rtl-optimization/79901
22997 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22999 (*avx512f_<code><mode>3<mask_name>): ... this.
23000 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23001 iterator instead of VI8_AVX2_AVX512BW.
23003 PR rtl-optimization/79901
23004 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23005 min/max expander, expand it using expand_vec_cond_expr.
23008 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23011 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23014 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23015 to void * for PCH reasons.
23016 * dwarf2out.c (output_loc_operands, output_die): Cast
23017 v.val_vec.array to unsigned char *.
23019 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23022 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23025 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23027 PR rtl-optimization/79571
23028 * lra-constraints.c (process_alt_operands): Calculate static
23029 reject and subtract it from overall when only addresses will be
23032 2017-03-06 Julia Koval <julia.koval@intel.com>
23035 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23036 incoming stack boundary to 128 for 64-bit targets.
23038 2017-03-06 Richard Biener <rguenther@suse.de>
23040 PR tree-optimization/79894
23041 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23042 to NULL after folding it.
23044 2017-03-06 Richard Biener <rguenther@suse.de>
23046 PR tree-optimization/79824
23047 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23048 check disabling peeling for gaps.
23050 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23052 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23053 attributes): Document gettimeofday.
23055 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23057 * config/s390/s390.c (s390_option_override_internal): Set
23058 PARAM_MIN_VECT_LOOP_BOUND
23060 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23062 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23063 * config/s390/s390.md: Likewise.
23065 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23068 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23069 (<avx2_avx512>_perm<mode>): Rename to ...
23070 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23072 (<avx512>_perm<mode>_mask): Rename to ...
23073 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23075 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23076 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23077 instead of VI8F_256_512.
23078 (avx512f_perm<mode>): New define_expand.
23079 (avx512f_perm<mode>_mask): Likewise.
23080 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23081 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23083 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23085 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23086 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23088 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23090 2017-03-06 Martin Liska <mliska@suse.cz>
23093 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23094 when having a SSA NAME w/o VAR_DECL assigned to it.
23096 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23098 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23099 msa_dpsub_<su>_d): Fix MODE for vec_select.
23101 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23103 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23105 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23107 2017-03-06 Richard Biener <rguenther@suse.de>
23109 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23110 * plugin.c (register_plugin_info): Likewise.
23111 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23113 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23115 * config/i386/sse.md (sse_storehps, sse_storelps,
23116 avx_<castmode><avxsizesuffix>_<castmode>,
23117 avx512f_<castmode><avxsizesuffix>_<castmode>,
23118 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23119 in condition that at least one operand is not a MEM.
23121 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23123 PR middle-end/79805
23124 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23125 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23127 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23128 gimple_call_nothrow_p flag based on whether original builtin can throw.
23129 If it can, emit following stmts on the fallthrough edge.
23130 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23131 don't create new bb if inserting just debug stmts on the edge, try to
23132 insert them on the fallthru bb or just reset debug stmts.
23134 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23137 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23138 restore recog_data (including the operand rtxes inside it) around
23139 the call to get_insn_template.
23141 2017-03-03 Martin Sebor <msebor@redhat.com>
23143 PR tree-optimization/79699
23144 * context.c (context::~context): Free MPFR caches to avoid
23145 a memory leak on program exit.
23147 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23149 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23150 Use wide_int::ulow () instead of .elt (0).
23152 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23154 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23155 (*pushxf): Limit oF constraint to 32bit targets and add oC
23156 constraint for 64bit targets.
23157 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23158 (*pushdf): Change rmF constraint to rmC.
23160 2017-03-03 Martin Liska <mliska@suse.cz>
23162 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23163 Remove unused variable.
23165 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23168 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23169 is a memory operand, increase num_memory.
23170 (ix86_expand_args_builtin): Likewise.
23172 2017-03-03 Jan Hubicka <jh@suse.cz>
23175 * ipa-devirt.c (maybe_record_node): Properly handle
23176 __cxa_pure_virtual visibility.
23178 2017-03-03 Martin Liska <mliska@suse.cz>
23180 PR tree-optimization/79803
23181 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23183 (pass_loop_prefetch::execute): Disabled optimization if an
23184 assumption about L1 cache size is not met.
23186 2017-03-03 Martin Liska <mliska@suse.cz>
23188 PR rtl-optimization/79574
23189 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23190 (hash_scan_set): Likewise.
23191 (dump_hash_table): Likewise.
23192 (hoist_code): Likewise.
23194 2017-03-03 Richard Biener <rguenther@suse.de>
23196 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23197 in place of uhigh ().
23198 (fixed_convert_from_real): Likewise.
23200 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23203 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23205 2017-03-03 Richard Biener <rguenther@suse.de>
23207 PR middle-end/79818
23208 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23209 TYPE_OVERFLOW_UNDEFINED check.
23211 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23213 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23215 (vector_ae_<mode>_p): Likewise.
23216 (vector_nez_<mode>_p): Likewise.
23217 (vector_ne_v2di_p): Likewise.
23218 (vector_ae_v2di_p): Likewise.
23219 (vector_ne_<mode>_p): Likewise.
23220 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23222 (vsx_tsqrt<mode>2_fe): Likewise.
23224 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23227 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23229 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23231 PR rtl-optimization/79780
23232 * cprop.c (one_cprop_pass): When second and further conditional trap
23233 in a single basic block is turned into an unconditional trap, turn it
23234 into a deleted note to avoid RTL verification failures.
23236 2017-03-02 Richard Biener <rguenther@suse.de>
23238 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23240 2017-03-02 Richard Biener <rguenther@suse.de>
23242 PR tree-optimization/79345
23244 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23245 param and abort the walk, returning -1 if it is hit.
23246 (walk_aliased_vdefs): Take a limit param and pass it on.
23247 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23248 defaulting to 0 and return a signed int.
23249 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23250 (check_defs): New helper.
23251 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23252 about uninitialized memory.
23253 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23254 bogus uninitialized warning.
23255 (fixed_convert_from_real): Likewise.
23257 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23259 PR tree-optimization/66768
23260 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23261 iv_use if base object can't be determined.
23263 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23265 PR tree-optimization/79345
23266 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23267 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23268 (get_pattern_stats): Initialize it.
23269 * genemit.c (gen_expand): Verify match_scratch numbers come after
23270 match_operand/match_dup numbers.
23271 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23272 match_scratch numbers.
23273 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23275 * config/s390/s390.md (trunctdsd2): Likewise.
23277 2017-03-02 Richard Biener <rguenther@suse.de>
23279 * wide-int.h (wide_int_storage::operator=): Implement in terms
23282 2017-03-02 Richard Biener <rguenther@suse.de>
23284 PR tree-optimization/79777
23285 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23286 the to insert expression to sth existing.
23288 2017-03-01 Martin Sebor <msebor@redhat.com>
23290 PR middle-end/79692
23291 * gimple-ssa-sprintf.c
23292 (directive::known_width_and_precision): New function.
23293 (format_integer): Use it.
23294 (get_mpfr_format_length): Consider the full range of precision
23295 when computing %g output with the # flag. Set the likely byte
23296 count to 3 rather than 1 when precision is indeterminate.
23297 (format_floating): Correct the lower bound of precision.
23299 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23301 * doc/invoke.texi: Document default code model for 64-bit Linux.
23303 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23306 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23307 udiv rather than div since input pattern is unsigned.
23309 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23311 * config/i386/i386.c (print_reg): Warn for values of
23312 unsupported size in integer register.
23314 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23317 * config/rs6000/predicates.md (current_file_function_operand): Do
23318 not allow self calls to be local if the function is replaceable.
23320 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23323 * config/rs6000/altivec.h (vec_ctz and others): Change the
23324 preprocessor macro that controls conditional compilation from
23325 _ARCH_PWR9 to __POWER9_VECTOR__.
23326 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23327 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23328 control (instead of _ARCH_PWR9 control) so that template
23329 definition uses power9-specific function.
23330 (vec_any_eq): Likewise.
23331 (vec_all_ne): Change macro definition to use a power9-specific
23332 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23333 _ARCH_PWR9 control).
23334 (vec_any_eq) Likewise.
23335 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23336 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23337 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23338 support for xvcmpnedp instruction.
23339 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23340 macro expansion so that Power9 implementation of vec_all_ne does
23341 not use the AltiVec predicate framework.
23342 (VCMPNEH_P): Likewise.
23343 (VCMPNEW_P): Likewise.
23344 (VCMPNED_P): Likewise.
23345 (VCMPNEFP_P): Likewise.
23346 (VCMPNEDP_P): Likewise.
23347 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23348 implementation of vec_any_eq to not use AltiVec predicate
23350 (VCMPAEH_P): Likewise.
23351 (VCMPAEW_P): Likewise.
23352 (VCMPAED_P): Likewise.
23353 (VCMPAEFP_P): Likewise.
23354 (VCMPAEDP_P): Likewise.
23355 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23356 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23357 not use the AltiVec predicate framework.
23358 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23359 of vec_any_eq to not use AltiVec predicate framework.
23360 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23361 support for predefined __POWER9_VECTOR__ macro to indicate that
23362 Power9 instruction selection is enabled.
23363 (altivec_overloaded_builtins): Remove extraneous
23364 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23365 function argument types RS6000_BTI_bool_V16QI and
23366 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23367 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23368 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23369 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23370 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23371 Power9 for implementations of vec_cmpne. Change the signature for
23372 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23373 (representing vec_all_ne) to remove the previously described first
23374 argument of type RS6000_BTI_INTSI, as this was an artifact of
23375 reliance on the AltiVec predicate framework, which is no longer
23376 used in the implementation of these functions. Add
23377 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23378 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23379 since, unlike the AltiVec predicate framework implementation, we
23380 do not share function descriptors between vec_alle and vec_anyeq.
23381 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23382 set of modes that receive special treatment even when
23383 TARGET_P9_VECTOR is true. The special treatment emits code that
23384 does not depend on Power9 instructions.
23385 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23386 define_expand to not rely on AltiVec predicate framework.
23387 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23389 (vector_ne_v2di_p): Change this define_expand to not rely on
23390 AltiVec predicate framework.
23391 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23393 (vector_ne_<mode>_p): Change this define_expand to not rely on
23394 AltiVec predicate framework.
23395 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23397 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23398 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23399 define_insn pattern.
23400 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23401 define_insn pattern because the xvcmpne<VSs>. instruction is not
23403 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23404 instruction is not supported.
23406 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23408 * config/nvptx/nvptx.c: Include intl.h.
23410 2017-03-01 Martin Jambor <mjambor@suse.cz>
23413 * ipa-prop.h (ipa_bits): Removed field known.
23414 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23415 to pointers. Adjusted their comments to warn about their sharing.
23416 (ipcp_transformation_summary): Change bits to a vector of pointers.
23417 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23418 (ipa_get_ipa_bits_for_value): Declare.
23419 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23420 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23421 (ipa_bits_hash_table): Likewise.
23422 (ipa_vr_ggc_hash_traits): Likewise.
23423 (ipa_vr_hash_table): Likewise.
23424 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23425 being pointers and vr_known being removed.
23426 (ipa_set_jf_unknown): Likewise.
23427 (ipa_get_ipa_bits_for_value): New function.
23428 (ipa_set_jfunc_bits): Likewise.
23429 (ipa_get_value_range): New overloaded functions.
23430 (ipa_set_jfunc_vr): Likewise.
23431 (ipa_compute_jump_functions_for_edge): Use the above functions to
23432 construct bits and vr parts of jump functions.
23433 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23434 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23436 (ipcp_grow_transformations_if_necessary): Also allocate
23437 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23439 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23440 them. Fix too long lines.
23441 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23442 vr_known being removed.
23443 (ipa_read_jump_function): Use new setter functions to construct bits
23444 and vr parts of jump functions or set them to NULL.
23445 (write_ipcp_transformation_info): Adjust for bits being pointers.
23446 (read_ipcp_transformation_info): Likewise.
23447 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23449 Include gt-ipa-prop.h.
23450 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23452 (ipcp_store_bits_results): Likewise.
23453 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23454 Do not write to existing jump functions but use a temporary instead.
23456 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23459 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23460 attempt to use its first operand as BIT_FIELD_REF base.
23462 2017-03-01 Richard Biener <rguenther@suse.de>
23464 PR middle-end/79721
23465 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23466 interpolating formula in wrapping arithmetic.
23467 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23469 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23471 PR tree-optimization/79734
23472 * tree-vect-generic.c (expand_vector_condition): Optimize
23473 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23474 Handle VEC_COND_EXPR where comparison has different inner width from
23475 type's inner width.
23477 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23479 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23480 markup, and similar issues. Remove @opindex entries for things
23481 that aren't options. Add missing -mmpy-option entries.
23483 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23485 PR tree-optimization/79737
23486 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23487 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23488 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23489 instead of byte_size. Formatting fix.
23490 (shift_bytes_in_array_right): Formatting fix.
23492 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23495 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23496 condition on optimize for the leaf function test.
23498 2017-02-28 Martin Liska <mliska@suse.cz>
23501 * read-rtl-function.c (function_reader::handle_unknown_directive):
23502 Bail out when one uses -flto.
23504 2017-02-28 Martin Liska <mliska@suse.cz>
23506 * common.opt: Replace space with tabular for options of <number>
23508 * config/i386/i386.opt: Show <number> value for
23509 -mlarge-data-threshold.
23510 * opts.c (print_filtered_help): Do not display number in hexadecimal
23513 2017-02-28 Martin Liska <mliska@suse.cz>
23515 * common.opt: Fix --help=option -Q for options which are of
23518 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23520 * config/i386/i386.c (print_reg): Error out for values
23521 of 8-bit size in invalid integer register.
23523 2017-02-28 Martin Sebor <msebor@redhat.com>
23525 PR tree-optimization/79691
23526 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23528 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23531 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23532 gcc_unreachable with output_operand_lossage.
23534 2017-02-28 Richard Biener <rguenther@suse.de>
23536 PR tree-optimization/79740
23537 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23539 (visit_nary_op): Insert the nary into the hashtable if we
23540 pattern-matched sth.
23541 * tree-ssa-pre.c (eliminate_insert): Robustify.
23543 2017-02-28 Richard Biener <rguenther@suse.de>
23545 PR middle-end/79731
23546 * fold-const.c (decode_field_reference): Reject out-of-bound
23549 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23551 * config/i386/i386.c: Include intl.h.
23552 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23553 instead of just cond ? "..." : "...".
23554 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23555 * coverage.c (read_counts_file): Likewise.
23556 * omp-offload.c: Include intl.h.
23557 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23558 of just cond ? "..." : "...".
23559 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23560 of just cond ? "..." : "...".
23562 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23565 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23567 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23568 'tune for' CPU name.
23569 * config/arm/arm-cpu-data.h: Regenerated.
23571 2017-02-28 Richard Biener <rguenther@suse.de>
23573 PR tree-optimization/79732
23574 * tree-inline.c (expand_call_inline): Do not shadow var.
23576 2017-02-28 Richard Biener <rguenther@suse.de>
23578 PR tree-optimization/79723
23579 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23580 address-space properly.
23582 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23584 * doc/optinfo.texi (Optimization groups): Fix option used for
23586 * doc/invoke.texi (-fopt-info): Document "omp".
23587 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23588 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23589 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23590 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23591 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23593 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23595 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23596 OPTGROUP_OMP with "omp".
23598 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23601 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23602 for arithmetic shift of unsigned V2DI.
23604 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23606 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23607 arc/linux.h headers.
23608 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23609 (LINK_SPEC): Likewise.
23610 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23611 (EXTRA_SPECS): Likewise.
23612 (STARTFILE_SPEC): Likewise.
23613 (ENDFILE_SPEC): Likewise.
23614 (LIB_SPEC): Likewise.
23615 (TARGET_SDATA_DEFAULT): Likewise.
23616 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23617 (MULTILIB_DEFAULTS): Likewise.
23618 (DWARF2_UNWIND_INFO): Likewise.
23619 * config/arc/big.h: New file.
23620 * config/arc/elf.h: Likewise.
23621 * config/arc/linux.h: Likewise.
23622 * config/arc/t-uClibc: Remove.
23624 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23626 PR tree-optimization/77536
23627 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23628 (tree_transform_and_unroll_loop): Use above function to compute the
23629 estimated niter of unrolled loop and use it when scaling profile.
23630 Also use count info rather than frequency if it's non-zero.
23631 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23632 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23633 (vect_transform_loop): Call above function.
23635 2017-02-27 Richard Biener <rguenther@suse.de>
23637 PR tree-optimization/45397
23638 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23639 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23640 (visit_nary_op): Add pattern matching for CSEing sign-changed
23641 or truncated operations with wider ones.
23643 2017-02-27 Richard Biener <rguenther@suse.de>
23645 PR tree-optimization/79690
23646 * tree-vect-stmts.c (vectorizable_store): Use vector type
23647 built from the DR with address-space.
23649 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23651 * doc/invoke.texi (Optimize Options): Refine the description
23652 of asan-use-after-return.
23654 2017-02-25 Alan Modra <amodra@gmail.com>
23656 PR rtl-optimization/79584
23657 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23658 base, not ad->base_term, the reg within base. Remove assertion
23659 that ad->base == ad->base_term. Replace gen_int_mode using
23660 bogus mode with const0_rtx.
23662 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23664 PR middle-end/79396
23665 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23666 FMA_EXPR like tcc_binary or tcc_unary.
23668 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23671 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23673 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23674 (output_loc_operands): Handle DW_OP_call_ref and
23675 DW_OP_GNU_variable_value.
23676 (struct variable_value_struct): New type.
23677 (struct variable_value_hasher): Likewise.
23678 (variable_value_hash): New variable.
23679 (string_types): Remove.
23680 (copy_loc_descr): New function.
23681 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23682 (prepend_loc_descr_to_each): New function.
23683 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23684 instead of add_loc_descr_to_each if the first argument is single
23685 location list and the second has multiple.
23686 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23687 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23688 when looking for variable value which doesn't have other location info.
23689 (loc_list_from_tree): Formatting fix.
23690 (gen_array_type_die): Simplify DW_AT_string_length handling.
23691 (adjust_string_types): Remove.
23692 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23693 string_types. Call resolve_variable_values.
23694 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23695 (resolve_addr_in_expr): Likewise. Add A argument.
23696 (copy_deref_exprloc): Remove deref argument. Adjust for the
23697 original expression being DW_OP_GNU_variable_value with optionally
23698 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23699 optionally after it.
23700 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23701 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23702 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23703 (variable_value_hasher::hash, variable_value_hasher::equal): New
23705 (resolve_variable_value_in_expr, resolve_variable_value,
23706 resolve_variable_values, note_variable_value_in_expr,
23707 note_variable_value): New functions.
23708 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23711 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23714 * opts.h (handle_generated_option): Add GENERATED_P argument.
23715 * opts-common.c (handle_option): Adjust function comment.
23716 (handle_generated_option): Add GENERATED_P argument, pass it to
23718 (control_warning_option): Pass false to handle_generated_option
23720 * opts.c (maybe_default_option): Pass true to handle_generated_option
23722 * optc-gen.awk: Likewise.
23724 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23726 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23727 a REG, look at the REG it is a SUBREG of.
23728 (splitter for cmpeqsi_t): Ditto.
23730 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23732 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23733 the special USEs with the pattern of the insn, not the insn itself.
23735 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23738 * doc/invoke.texi: Document -mload-store-pairs.
23740 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23741 Sandra Loosemore <sandra@codesourcery.com>
23743 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23744 argument isn't a CONST_INT.
23745 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23746 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23747 (nios2_validate_compare): Bypass alternate compare logic if *op2
23748 is not a CONST_INT.
23749 (ldstwm_operation_p): Return false if first_base is not a REG or
23750 if first_offset is not a CONST_INT.
23752 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23754 * config/cris/cris.md: Use correct operand in a define_peephole2.
23756 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23758 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23760 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23762 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23763 this_insn if it is an INSN or JUMP_INSN.
23764 (force_offsettable): Look at base, not at addr.
23765 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23766 on things that aren't necessarily CONST_INTs.
23768 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
23770 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
23771 -mfpmath=sse is the default also for x86-32 targets with SSE2
23772 instruction set when @option{-ffast-math} is enabled
23774 2017-02-24 Jeff Law <law@redhat.com>
23776 PR rtl-optimizatoin/79286
23777 * ira.c (update_equiv_regs): Drop may_trap_p exception to
23780 2017-02-24 Richard Biener <rguenther@suse.de>
23782 PR tree-optimization/79389
23783 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
23786 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
23788 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
23789 function comment to reflect reality.
23790 (loop_exits_before_overflow): Fix typo in function description.
23792 2017-02-24 Richard Biener <rguenther@suse.de>
23794 PR tree-optimization/79389
23795 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23796 properly that a threading opportunity exists. Detect conditional
23797 copy/constant propagation opportunities.
23799 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23801 * config/visium/visium.md (type): Add trap.
23802 (b): New mode attribute.
23803 (*btst): Rename into...
23804 (*btst<mode>): ...this and adjust.
23805 (*cbranchsi4_btst_insn): Rename into...
23806 (*cbranch<mode>4_btst_insn): ...this and adjust.
23807 (trap): New define_insn.
23809 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23811 PR tree-optimization/79389
23812 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23813 (noce_reversed_cond_code): New function.
23814 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23815 reversed_comparison_code. Formatting fix.
23816 (noce_try_store_flag): Test rev_cond != NULL in addition to
23817 reversed_comparison_code.
23818 (noce_try_store_flag_constants): Likewise.
23819 (noce_try_store_flag_mask): Likewise.
23820 (noce_try_addcc): Use rev_cond if non-NULL instead of
23821 reversed_comparison_code.
23822 (noce_try_cmove_arith): Likewise. Formatting fixes.
23823 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23824 (noce_find_if_block): Initialize rev_cond.
23825 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23826 instead of false as last argument never attempt to reverse it
23829 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23831 PR tree-optimization/79663
23832 * tree-predcom.c (combine_chains): Process refs in reverse order
23833 only for ZERO length chains, and add explaining comment.
23835 2017-02-23 Jeff Law <law@redhat.com>
23837 PR tree-optimization/79578
23838 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23839 in call to operand_equal_p.
23841 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23844 * config/i386/cpuid.h: Fix another undefined behavior.
23846 2017-02-23 Richard Biener <rguenther@suse.de>
23848 PR tree-optimization/79683
23849 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23850 vector types for data-refs.
23852 2017-02-23 Martin Liska <mliska@suse.cz>
23854 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23856 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23858 PR middle-end/79665
23859 * internal-fn.c (get_range_pos_neg): Moved to ...
23860 * tree.c (get_range_pos_neg): ... here. No longer static.
23861 * tree.h (get_range_pos_neg): New prototype.
23862 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23863 are known to be in between 0 and signed maximum inclusive, try to
23864 expand both unsigned and signed divmod and use the cheaper one from
23867 2017-02-22 Jeff Law <law@redhat.com>
23869 PR tree-optimization/79578
23870 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23871 to compare base operands.
23873 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23876 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23877 gpc_reg_operand instead of fpr_reg_operand.
23879 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23881 * config/mips/mips.c (mips_return_in_memory): Force FP
23882 vector types to be returned in memory for o32 ABI.
23884 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23886 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23887 instead of DW_TAG_member for static data member declarations and don't
23888 set no_linkage_name for static inline data members.
23889 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23892 2017-02-22 Martin Liska <mliska@suse.cz>
23894 * doc/invoke.texi: Replace inequality signs with square brackets
23897 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23899 PR tree-optimization/68644
23900 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23902 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23905 * lra-constraints.c (simplify_operand_subreg): Handle
23906 WORD_REGISTER_OPERATIONS targets.
23908 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23911 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23912 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23913 elimination by swapping fld*.
23915 2017-02-22 Richard Biener <rguenther@suse.de>
23917 PR tree-optimization/79673
23918 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23919 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23920 irrelevant address-space qualifiers and avoiding a
23921 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23923 2017-02-22 Richard Biener <rguenther@suse.de>
23925 PR tree-optimization/79666
23926 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23927 to not symbolically negate if that may introduce undefined
23930 2017-02-22 Martin Liska <mliska@suse.cz>
23933 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23934 * data-streamer-out.c (streamer_write_gcov_count_stream):
23936 * value-prof.c (stream_out_histogram_value): Make assert more
23937 precise based on type of counter.
23939 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23942 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23943 Use nonimmediate_operand instead of memory_operand for operand 1.
23944 (float-extend standard_x87sse_constant_load splitter): Ditto.
23946 2017-02-21 Jeff Law <law@redhat.com>
23948 PR tree-optimization/79621
23949 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23950 blocks with edges to themselves.
23952 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23955 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23956 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23957 Use gimple_call_builtin_p.
23960 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23961 on temporarily removed DEBUG_INSNs.
23963 PR tree-optimization/79649
23964 * tree-loop-distribution.c (classify_partition): Give up on
23965 non-generic address space loads/stores.
23967 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23969 * doc/loop.texi (Loop manipulation): Remove nonexistent
23970 tree_ssa_loop_version from the documentation.
23971 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23973 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23976 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23977 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23978 * config/rs6000/rs6000.c: Include except.h.
23979 (rs6000_expand_split_stack_prologue): Call
23980 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23982 2017-02-21 Martin Jambor <mjambor@suse.cz>
23985 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23986 have been analyzed.
23988 2017-02-21 Martin Jambor <mjambor@suse.cz>
23990 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23991 for backward compatibility only.
23992 * doc/invoke.texi (Option Summary): Remove all references to
23993 -fipa-cp-alignment.
23995 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23999 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24001 * lra-constraints.c (curr_insn_transform): Handle
24002 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24004 2017-02-21 Martin Liska <mliska@suse.cz>
24006 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24008 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24010 PR translation/79638
24011 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24013 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24016 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24017 (arm_function_ok_for_sibcall): Return false for an indirect call by
24018 descriptor if all the argument registers are used.
24019 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24020 alignment of the function.
24022 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24024 PR tree-optimization/61441
24025 * simplify-rtx.c (simplify_const_unary_operation): For
24026 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24027 the sNaN unmodified.
24029 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24031 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24032 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24033 instead of SYSTEM_HEADER_DIR.
24035 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24036 Martin Liška <mliska@suse.cz>
24038 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24039 Fix typos and grammar, use active voice, and clarify.
24041 2017-02-20 Marek Polacek <polacek@redhat.com>
24043 PR middle-end/79537
24044 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24047 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24049 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24052 * config/i386/i386.c (ix86_expand_builtin): Handle
24053 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24054 ix86_builtins_isa[fcode].isa as a requirement of those
24055 flags and any other flag in the bitmask.
24056 (ix86_init_mmx_sse_builtins): Use 0 instead of
24057 ~OPTION_MASK_ISA_64BIT as mask.
24058 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24059 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24060 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24061 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24063 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24066 * lra-constraints.c (split_reg): Check requested split mode
24067 is supported by the register.
24069 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24071 * lra-constraints.c (simplify_operand_subreg): Remove early
24074 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24077 * lra-constraints.c (curr_insn_transform): Tighten condition
24078 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24080 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24083 * lra-constraints.c (curr_insn_transform): Handle
24084 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24086 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24089 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24091 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24093 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24096 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24099 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24101 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24104 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24106 * final.c (last_columnnum, override_columnnum): New variables.
24107 (final_start_function): Set last_columnnum, pass it to begin_prologue
24108 hook and pass 0 to dwarf2out_begin_prologue.
24109 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24110 to source_line debug hook.
24111 (notice_source_line): Compute last_columnnum and for debug_column_info
24112 return true on column changes.
24113 * debug.h (struct gcc_debug_hooks): Add column argument to
24114 source_line and begin_prologue hooks.
24115 (debug_nothing_int_charstar_int_bool): Remove prototype.
24116 (debug_nothing_int_int_charstar,
24117 debug_nothing_int_int_charstar_int_bool): New prototypes.
24118 (dwarf2out_begin_prologue): Add column argument.
24119 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24120 begin_prologue hooks.
24121 (debug_nothing_int_charstar_int_bool): Remove.
24122 (debug_nothing_int_int_charstar,
24123 debug_nothing_int_int_charstar_int_bool): New functions.
24124 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24125 through to dwarf2out_source_line.
24126 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24127 (dwarf2out_source_line): Add column argument, emit it if requested.
24128 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24130 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24131 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24132 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24133 through to dwarf2out_begin_prologue.
24134 (vmsdbgout_source_line): Add column argument, pass it through to
24135 dwarf2out_source_line.
24136 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24137 dbxout_source_line caller.
24138 (dbxout_source_line): Add column argument.
24140 * common.opt (gno-column-info, gcolumn-info): New options.
24141 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24142 (check_die): Also test for multiple DW_AT_decl_column attributes.
24143 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24144 DW_AT_decl_column if requested.
24145 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24147 (gen_variable_die): Likewise.
24148 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24149 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24152 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24153 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24154 (ix86_handle_option): Handle OPT_m3dnowa.
24155 * doc/invoke.texi (-m3dnowa): Document.
24156 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24157 -m3dnowa instead of -m3dnow -march=athlon.
24160 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24161 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24163 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24166 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24167 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24168 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24169 generator for vsx_xxpermdi_<mode>_be.
24170 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24171 force big-endian semantics.
24172 (vsx_xxpermdi_<mode>_be): New define_expand with same
24173 implementation as previous version of vsx_xxpermdi_<mode>.
24175 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24177 PR tree-optimization/79327
24178 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24179 variable, its initialization and use.
24181 2017-02-17 Julia Koval <julia.koval@intel.com>
24183 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24184 (OPTION_MASK_ISA_PKU_UNSET): New.
24185 (ix86_handle_option): Handle -mrdpid.
24186 * config/i386/cpuid.h (bit_RDPID): New.
24187 * config/i386/driver-i386.c (host_detect_local_cpu):
24188 Detect RDPID feature.
24189 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24190 * config/i386/i386-c.c (ix86_target_macros_internal):
24192 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24193 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24194 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24195 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24196 * config/i386/i386.md (define_insn "rdpid"): New.
24197 * config/i386/i386.opt Add -mrdpid.
24198 * config/i386/immintrin.h (_rdpid_u32): New.
24200 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24202 PR rtl-optimization/79541
24203 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24204 instead of transforming it into USE.
24206 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24208 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24209 If HONOR_SNANS (SFmode) force the input to a register.
24210 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24211 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24212 an frsp or similar insn.
24214 2017-02-17 Martin Liska <mliska@suse.cz>
24216 PR rtl-optimization/79577
24217 * params.def (selsched-max-sched-times): Increase minimum to 1.
24219 2017-02-17 Martin Liska <mliska@suse.cz>
24221 PR rtl-optimization/79574
24222 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24224 2017-02-17 Martin Liska <mliska@suse.cz>
24226 PR tree-optimization/79529
24227 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24228 ssa_defined_default_def_p to handle cases which are implicitly
24230 * tree-ssa.c (ssa_defined_default_def_p): New function.
24231 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24232 which are implicitly defined.
24233 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24235 2017-02-17 Richard Biener <rguenther@suse.de>
24237 PR middle-end/79576
24238 * params.def (max-ssa-name-query-depth): Limit to 10.
24240 2017-02-17 Richard Biener <rguenther@suse.de>
24242 PR tree-optimization/79552
24243 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24246 2017-02-17 Richard Biener <rguenther@suse.de>
24249 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24251 2017-02-17 Marek Polacek <polacek@redhat.com>
24253 PR middle-end/79536
24254 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24255 (fold_negate_expr): New wrapper.
24257 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24259 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24260 Correct terminology and de-emphasize pre-standard behavior.
24262 2017-02-16 Alan Modra <amodra@gmail.com>
24264 PR rtl-optimization/79286
24265 * ira.c (def_dominates_uses): New function.
24266 (update_equiv_regs): Don't create an equivalence for insns that
24267 may trap where the register def does not dominate the use.
24269 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24271 PR rtl-optimization/78127
24272 * lra.c (lra): Call lra_eliminate before finish the loop after
24275 2017-02-16 Richard Biener <rguenther@suse.de>
24277 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24279 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24280 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24281 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24282 (isl_val_int_from_wi): New function.
24283 (extract_affine_gmp): Rename to ...
24284 (extract_affine_wi): ... this, take a widest_int.
24285 (extract_affine_int): Just wrap extract_affine_wi.
24286 (add_param_constraints): Use isl_val_int_from_wi.
24287 (add_loop_constraints): Likewise, and extract_affine_wi.
24289 2017-02-15 Jeff Law <law@redhat.com>
24291 PR middle-end/79521
24292 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24293 ira_init_register_move_cost_if_necessary.
24295 2017-02-15 Martin Sebor <msebor@redhat.com>
24297 PR middle-end/32003
24298 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24299 removed in a prior commit.
24301 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24303 PR tree-optimization/79347
24304 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24305 counters during peeling.
24307 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24309 * Makefile.in (site.exp): Remove "set ISLVER".
24311 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24314 * real.c (real_from_integer): Call real_convert even for decimal.
24316 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24319 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24321 2017-02-14 Andrew Pinski <apinski@cavium.com>
24323 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24324 cores and change the partno/implementer to be correct.
24325 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24326 the 'B" as the implementer.
24327 * config/aarch64/aarch64-tune.md: Regenerate.
24329 2017-02-14 Carl Love <cel@us.ibm.com>
24331 * config/rs6000/rs6000.c: Add case statement entry to make the
24332 xvcvuxdsp built-in argument unsigned.
24333 * config/rs6000/vsx.md: Fix the source and return operand types so they
24334 match the instruction definitions from the ISA document. Fix typo
24335 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24338 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24341 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24342 member early_clobber_alts.
24343 * lra-lives.c (reg_early_clobber_p): New.
24344 (process_bb_lives): Use it.
24345 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24346 (debug_operand_data): Initialize early_clobber_alts.
24347 (setup_operand_alternative): Set up early_clobber_alts.
24348 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24349 alternatives to new_insn_reg.
24350 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24352 (lra_update_insn_regno_info): Pass the new arg.
24354 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24356 PR middle-end/79505
24357 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24358 (new_oacc_loop_raw): Don't clear already cleared fields.
24361 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24362 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24363 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24365 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24368 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24370 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24373 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24374 the extra instruction to the right place to store 128-bit constant
24377 2017-02-14 Martin Sebor <msebor@redhat.com>
24379 PR middle-end/79448
24380 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24381 warning for strings of unknown length.
24383 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24385 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24387 2017-02-14 Jeff Law <law@redhat.com>
24390 * ira-costs.c (scan_one_insn): Initialize register move costs
24391 for pseudos seen in USE/CLOBBER insns.
24393 PR tree-optimization/79095
24394 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24395 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24396 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24397 if the operands are known to be not equal, then the resulting range
24399 (intersect_ranges): If the new range is ~[0,0] and the old range is
24400 wide, then prefer ~[0,0].
24401 * tree-vrp.c (overflow_comparison_p_1): New function.
24402 (overflow_comparison_p): New function.
24403 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24404 if NAME is used in an overflow test.
24405 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24406 overflow check that can be expressed as an equality test, then adjust
24407 ops to be that equality test.
24409 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24411 * config/s390/s390-builtin-types.def: Remove flags argument.
24412 * config/s390/s390.c (s390_init_builtins): Likewise.
24414 2017-02-14 Martin Liska <mliska@suse.cz>
24416 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24417 vector. Fix trailing white spaces.
24419 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24421 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24424 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24426 PR rtl-optimization/68664
24427 * config/arm/arm.c (arm_sched_can_speculate_insn):
24428 New function. Declare prototype.
24429 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24431 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24433 PR rtl-optimization/68664
24434 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24436 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24438 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24440 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24441 max skip bytes for function, loop and jump.
24443 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24445 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24446 ABS_EXPR for gimple dump.
24448 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24451 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24453 PR tree-optimization/79408
24454 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24455 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24456 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24457 also if rhs1 is INTEGER_CST.
24459 2017-02-14 Richard Biener <rguenther@suse.de>
24461 PR middle-end/79432
24462 * tree-into-ssa.c (insert_phi_nodes): When the function can
24463 have abnormal edges rewrite SSA names with broken use-def
24464 dominance out of SSA and register them for PHI insertion.
24466 2017-02-13 Martin Sebor <msebor@redhat.com>
24468 PR middle-end/79496
24469 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24470 clearing info.nowrite flag when snprintf size argument is a range.
24472 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24474 * cprop.c (cprop_jump): Add missing space in string literal.
24475 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24476 (get_constraint_for_component_ref): Likewise.
24477 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24478 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24479 * lra-constraints.c (process_alt_operands): Likewise.
24480 * ipa-inline.c (inline_small_functions): Likewise.
24481 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24482 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24483 * trans-mem.c (diagnose_tm_1_op): Likewise.
24484 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24485 (grid_parallel_clauses_gridifiable): Likewise.
24487 * config/nvptx/mkoffload.c (process): Add space in between
24490 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24491 "MOD4_SSE_REGS" and "ALL_REGS".
24493 * spellcheck.c (test_data): Add , in between "foo" and "food".
24495 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24498 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24499 boundary crossing check and subsequent code generation agree.
24501 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24503 * config/aarch64/aarch64.c (has_memory_op): Delete.
24504 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24507 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24509 PR rtl-optimization/79388
24510 PR rtl-optimization/79450
24511 * combine.c (distribute_notes): When removing TEM_INSN for which
24512 corresponding dest has last value recorded, invalidate that last
24515 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24517 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24518 of explicit '@'. Add missing assembly comment marker on branch costs
24521 2017-02-13 Nathan Sidwell <nathan@acm.org>
24523 * gengtype-lex.l (<in_struct>): Add '/'.
24525 2017-02-13 Martin Liska <mliska@suse.cz>
24528 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24530 2017-02-13 Richard Biener <rguenther@suse.de>
24532 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24534 * configure: Re-generate.
24535 * config.in: Likewise.
24536 * graphite-dependences.c: Simplify as if
24537 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24538 * graphite-isl-ast-to-gimple.c: Likewise.
24539 * graphite-optimize-isl.c: Likewise.
24540 * graphite-poly.c: Likewise.
24541 * graphite-sese-to-poly.c: Likewise.
24542 * graphite.h: Likewise.
24543 * toplev.c: Include isl/version.h and use isl_version () for
24544 printing the ISL version.
24545 * doc/install.texi: Update ISL requirement.
24547 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24549 * doc/standards.texi (Standards): Update reference to
24552 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24554 * doc/extend.texi (Named Address Spaces): sourceware.org now
24556 * doc/install.texi (Binaries): Ditto.
24559 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24561 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24562 standard terminology "stringize"/"stringizing" throughout.
24563 * doc/cppinternals.texi: Likewise.
24565 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24567 * doc/extend.texi: Fix some spelling mistakes and typos.
24568 * doc/invoke.texi: Likewise.
24570 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24573 * params.def (inline-min-speedup) Change from 10 to 8.
24575 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24577 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24580 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24583 * ipa-inline-analysis.c (get_minimal_bb): New function.
24584 (record_modified): Use it.
24585 (remap_edge_change_prob): Handle also ancestor functions.
24587 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24589 * doc/contrib.texi (Contributors): Remove broken link into
24590 the Mauve CVS repository.
24592 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24594 PR middle-end/79454
24595 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24596 result computation whenever lhs doesn't have vector mode, not
24597 just when it has BLKmode.
24599 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24601 * doc/makefile.texi (profiledbootstrap): Refer to the
24602 installation instructions only in textual form.
24604 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24607 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24609 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24611 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24612 (Specific): Update mingw-w64 reference.
24614 (Specific): Remove broken link to Renesas RX processor.
24616 2017-02-10 Richard Biener <rguenther@suse.de>
24618 * toplev.c (process_options): Do not mention obsolete graphite
24619 options when printing sorry message about missing graphite support.
24620 Mention -floop-nest-optimize.
24622 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24624 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24625 (vtst_p16): Likewise.
24626 (vtstq_p8): Likewise.
24627 (vtstq_p16): Likewise.
24629 (vtstq_p64): Likewise.
24630 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24631 (vset_lane_p64): New.
24632 (vsetq_lane_p64): New.
24634 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24636 PR tree-optimization/79411
24637 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24638 stmt operands are SSA_NAMEs used in abnormal phis.
24639 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24642 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24645 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24648 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24650 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24652 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24654 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24658 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24659 not arbitrary TREE_CONSTANT.
24662 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24663 "omp declare target link" attribute unless is_global_var.
24664 * omp-offload.c (find_link_var_op): Likewise.
24666 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24667 Chung-Lin Tang <cltang@codesourcery.com>
24669 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24671 (gimplify_adjust_omp_clauses): Don't delete TILE.
24672 (gimplify_omp_for): Deal with TILE.
24673 * internal-fn.c (expand_GOACC_TILE): New function.
24674 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24676 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24677 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24679 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24680 avoid DIV for outermost collapse var.
24681 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24682 Remove out of date comments, fix whitespace.
24683 * omp-general.c (omp_extract_for_data): Deal with tiling.
24684 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24685 adjust OLF_DIM_BASE value.
24686 (struct omp_for_data): Add tiling field.
24687 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24688 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24689 for auto loops. Remove default auto determining, moved to
24690 oacc_loop_fixed_partitions.
24691 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24692 stmts, add e_mask field.
24693 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24694 (oacc_thread_numbers): Use oacc_dim_call.
24695 (oacc_xform_tile): New.
24696 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24697 (finish_oacc_loop): Adjust for ifns vector.
24698 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24699 add case for GOACC_TILE fns.
24700 (oacc_loop_xform_loop): Delete.
24701 (oacc_loop_process): Iterate over call list directly, and add
24702 handling for GOACC_TILE fns.
24703 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24705 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24706 vector partitioning to outer loops. Assign 2 partitions to loops
24707 when available. Add TILE handling.
24708 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24709 (execite_oacc_device_lower): Process GOACC_TILE fns,
24710 ignore unknown specs.
24711 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24712 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24713 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24715 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24717 * configure.ac (ACX_BUGURL): Update.
24718 * configure: Regenerate.
24720 2017-02-09 Richard Biener <rguenther@suse.de>
24722 PR tree-optimization/69823
24723 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24724 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24726 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24728 * config/arc/arc-c.def: Add __NPS400__ definition.
24729 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24730 (TARGET_NPS400): Define.
24732 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24734 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24736 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24737 pointer, arch_info.
24738 (arc_cpu_types): Fill the arch_info field with a pointer into the
24739 arc_arch_types table.
24740 (arc_selected_cpu): Declare.
24741 * config/arc/arc.c (arc_selected_cpu): Make global.
24742 (arc_selected_arch): Delete.
24743 (arc_base_cpu): Delete.
24744 (arc_override_options): Remove references to deleted variables,
24745 update access to arch information.
24746 (ARC_OPT): Update access to arch information.
24747 (ARC_OPTX): Likewise.
24748 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24749 (TARGET_ARC600): Update access to arch information.
24750 (TARGET_ARC601): Likewise.
24751 (TARGET_ARC700): Likewise.
24752 (TARGET_EM): Likewise.
24753 (TARGET_HS): Likewise.
24754 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24757 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24760 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24761 condition/operands for integer GE/LE/GEU/LEU operations.
24763 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24765 PR translation/79397
24766 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
24769 2017-02-08 Martin Jambor <mjambor@suse.cz>
24772 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
24773 whether allocation happened.
24774 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
24775 nothing was allocated.
24777 2017-02-08 Jakub Jelinek <jakub@redhat.com>
24779 PR tree-optimization/79408
24780 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
24781 constant, but SSA_NAME with a known integer range, use the minimum
24782 of that range instead of op1 to determine if modulo can be replaced
24783 with its first operand.
24785 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24787 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
24789 2017-02-08 Richard Biener <rguenther@suse.de>
24791 PR tree-optimization/71824
24792 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24793 Check all loops contained in the merged region.
24795 2017-02-07 Andrew Pinski <apinski@cavium.com>
24797 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24799 2017-02-07 Andrew Pinski <apinski@cavium.com>
24801 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24802 (thunderxt88): Likewise.
24803 (thunderxt81): Disable LSE and change v8.1 to v8.
24804 (thunderxt83): Likewise.
24806 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24807 Richard Biener <rguenther@suse.de>
24809 PR middle-end/79399
24810 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24811 type from int to size_t.
24812 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24814 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24816 PR rtl-optimization/79386
24817 * cprop.c (bypass_conditional_jumps): Initialize
24818 bypass_last_basic_block already before splitting bbs after
24819 unconditional traps...
24820 (bypass_conditional_jumps): ... rather than here.
24823 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24824 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24825 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24826 fix -masm=intel patterns.
24828 2017-02-07 Richard Biener <rguenther@suse.de>
24830 PR tree-optimization/79256
24831 PR middle-end/79278
24832 * builtins.c (get_object_alignment_2): Use min_align_of_type
24833 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24834 and ADJUST_FIELD_ALIGN.
24836 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24838 * doc/tm.texi: Regenerate.
24839 * stor-layout.c (layout_decl): Adjust.
24840 (update_alignment_for_field): Likewise.
24841 (place_field): Likewise.
24842 (min_align_of_type): Likewise.
24843 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24844 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24845 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24846 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24847 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24848 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24849 * config/i386/i386.c (x86_field_alignment): Likewise.
24850 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24851 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24852 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24853 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24854 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24855 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24859 2017-01-30 Richard Biener <rguenther@suse.de>
24861 PR tree-optimization/79256
24862 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24863 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24866 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24868 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24869 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24870 builtins to SImode and emit a zero-extend, if necessary.
24872 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24874 * docs/invoke.texi (RISC-V Options): Alphabetize.
24876 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24878 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24881 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24883 * config/riscv/riscv.c: New file.
24884 * common/config/riscv/riscv-common.c: Likewise.
24885 * config.gcc: Likewise.
24886 * config/riscv/constraints.md: Likewise.
24887 * config/riscv/elf.h: Likewise.
24888 * config/riscv/generic.md: Likewise.
24889 * config/riscv/linux.h: Likewise.
24890 * config/riscv/multilib-generator: Likewise.
24891 * config/riscv/peephole.md: Likewise.
24892 * config/riscv/pic.md: Likewise.
24893 * config/riscv/predicates.md: Likewise.
24894 * config/riscv/riscv-builtins.c: Likewise.
24895 * config/riscv/riscv-c.c: Likewise.
24896 * config/riscv/riscv-ftypes.def: Likewise.
24897 * config/riscv/riscv-modes.def: Likewise.
24898 * config/riscv/riscv-opts.h: Likewise.
24899 * config/riscv/riscv-protos.h: Likewise.
24900 * config/riscv/riscv.h: Likewise.
24901 * config/riscv/riscv.md: Likewise.
24902 * config/riscv/riscv.opt: Likewise.
24903 * config/riscv/sync.md: Likewise.
24904 * config/riscv/t-elf-multilib: Likewise.
24905 * config/riscv/t-linux: Likewise.
24906 * config/riscv/t-linux-multilib: Likewise.
24907 * config/riscv/t-riscv: Likewise.
24908 * configure.ac: Likewise.
24909 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24910 Waterman as RISC-V maintainers.
24911 * doc/install.texi: Add RISC-V entries.
24912 * doc/invoke.texi: Add RISC-V options section.
24913 * doc/md.texi: Add RISC-V constraints section.
24914 * configure: Regenerated.
24916 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24919 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24920 false values to be constant vectors with all 0 or all 1 bits set.
24921 (vcondu<mode><mode>): Likewise.
24922 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24924 (fpmask_comparison_operator): Update comment.
24925 (vecint_comparison_operator): New predicate.
24926 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24927 vector conditionals when the true and false values are constant
24928 vectors with all 0 bits or all 1 bits set.
24930 2017-02-06 Martin Sebor <msebor@redhat.com>
24932 PR tree-optimization/79376
24933 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24935 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24937 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24938 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24939 to simplify split condition.
24941 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24943 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24944 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24947 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24949 PR rtl-optimization/68664
24950 * target.def (can_speculate_insn): New hook.
24951 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24952 * doc/tm.texi: Regenerate.
24953 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24954 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24955 (rs6000_sched_can_speculate_insn): New function.
24957 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24959 PR tree-optimization/79284
24960 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24961 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24962 vectorizable_mask_load_store, vectorizable_operation,
24963 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24964 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24965 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24966 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24967 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24968 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24969 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24970 is_gimple_assign (stmt). Replace another such test with
24971 is_gimple_assign (stmt).
24973 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24976 * config/avr/avr.c (rtl-iter.h): Include it.
24977 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24978 (avr_legitimate_combined_insn): ...and implementation.
24980 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24982 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24983 * config/s390/s390.c (s390_const_operand_ok)
24984 (s390_canonicalize_comparison, s390_extract_part)
24985 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24986 (s390_contiguous_bitmask_p, s390_rtx_costs)
24987 (legitimize_pic_address): Likewise.
24988 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24989 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24990 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24991 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24992 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24994 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24996 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24997 REGNO($0) == REGNO($1).
24999 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25001 * config/s390/linux.h(SIZE_TYPE): Add comment.
25003 2017-02-06 Julian Brown <julian@codesourcery.com>
25004 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25005 Virendra Pathak <virendra.pathak@broadcom.com>
25007 * config/aarch64/aarch64-cores.def: Change the scheduler
25009 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25010 * config/aarch64/thunderx2t99.md: New file.
25012 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25014 * doc/standards.texi (Go Language): Update link to language
25017 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25019 * tree-eh.c (lower_resx): Sanitize profile.
25020 (cleanup_empty_eh_move_lp): Likewise.
25022 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25025 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25027 * cfgloopmanip.h (loop_version): Update prototype.
25028 * modulo-sched.c (sms_schedule): Update call of loop_version.
25029 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25030 * tree-parloops.c (gen_parallel_loop): Likewise.
25031 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25032 * tree-ssa-loop-split.c (split_loop): Likewise.
25033 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25034 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25036 2017-02-05 Martin Liska <mliska@suse.cz>
25039 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25041 (print_operand_address): Initialize a struct to zero.
25043 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25045 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25046 garbage collector only in textual form.
25048 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25050 * doc/extend.texi (x86 specific memory model extensions for
25051 transactional memory): Simplify a phrase.
25053 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25056 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25057 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25058 (atomic_storedi_1): Likewise.
25060 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25062 PR tree-optimization/79338
25063 * tree-parloops.c (gather_scalar_reductions): Don't call
25064 vect_analyze_loop_form for loop->inner before destroying loop's
25067 2017-02-03 Martin Sebor <msebor@redhat.com>
25069 PR tree-optimization/79327
25070 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25071 when precision has resulted in leading zeros.
25072 (format_integer): Adjust the likely counter to assume an unknown
25073 argument that may be zero is non-zero.
25075 2017-02-03 Jason Merrill <jason@redhat.com>
25078 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25079 avoid copying non-taken branch.
25081 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25083 PR tree-optimization/79340
25084 * tree-vect-loop.c (vectorizable_reduction): Release
25085 vec_defs elements after safe_splicing them into other vectors.
25088 PR tree-optimization/79327
25089 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25090 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25092 (format_integer): Use wide_int_to_tree instead of build_int_cst
25093 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25094 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25095 of shortest and longest sequence.
25097 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25099 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25100 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25102 2017-02-03 Walter Lee <walt@tilera.com>
25105 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25106 after initial stackframe link reg save.
25107 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25109 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25112 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25113 wu for stxssp alternative.
25115 2017-02-03 Martin Sebor <msebor@redhat.com>
25117 PR tree-optimization/79352
25118 * gimple-fold.c (get_range_strlen): Add argument.
25119 (get_range_strlen): Change return type to bool.
25120 (get_maxval_strlen): Pass in a dummy argument.
25121 * gimple-fold.h (get_range_strlen): Change return type to bool.
25122 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25123 * tree.h (array_at_struct_end_p): Add argument.
25124 * tree.c (array_at_struct_end_p): Handle it.
25126 2017-02-03 Martin Liska <mliska@suse.cz>
25129 * multiple_target.c (create_dispatcher_calls): Redirect edge
25130 from a caller of a dispatcher.
25131 (expand_target_clones): Make the clones local.
25132 (ipa_target_clone): Do both target clones and resolvers.
25133 (ipa_dispatcher_calls): Remove the pass.
25134 (pass_dispatcher_calls::gate): Likewise.
25135 (make_pass_dispatcher_calls): Likewise.
25136 * passes.def (pass_target_clone): Put as very first IPA early
25139 2017-02-03 Martin Liska <mliska@suse.cz>
25141 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25142 in case of a function with ifunc attribute.
25144 2017-02-03 Martin Liska <mliska@suse.cz>
25146 * cgraph.c (cgraph_node::dump): Dump function version info.
25147 * symtab.c (symtab_node::dump_base): Add missing new line.
25149 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25151 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25152 (ifcombine_ifandif): Use it.
25154 2017-02-03 Martin Liska <mliska@suse.cz>
25156 * doc/invoke.texi: Document default value for
25157 use-after-scope-direct-emission-threshold.
25159 2017-02-03 Martin Liska <mliska@suse.cz>
25161 PR tree-optimization/79339
25162 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25163 (format_floating): Likewise.
25165 2017-02-03 Martin Liska <mliska@suse.cz>
25168 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25170 (ipa_node_params_t::remove): Likewise.
25171 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25172 initialization from removed ipa_node_params_t::insert.
25173 (ipa_node_params::~ipa_node_params): Move from removed
25174 ipa_node_params_t::release.
25175 * symbol-summary.h (symbol_summary::m_released): New member.
25176 Do not release a summary twice. Do not allow to call finalizer
25177 for types of a summary that live in GGC memory.
25179 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25181 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25184 2017-02-02 Martin Sebor <msebor@redhat.com>
25186 PR middle-end/79275
25187 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25188 (format_string): Tighten up the range of output for non-constant
25189 strings and correct the expected range for wide non-constant strings.
25191 2017-02-02 Martin Sebor <msebor@redhat.com>
25193 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25195 PR middle-end/32003
25196 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25198 (-fdump-tree-@var): Add to index and document how to come up
25199 with pass-specific option and dump file names.
25200 (-fdump-passes): Clarify where to look for output.
25202 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25204 PR middle-end/77445
25205 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25206 statistics of the analyzed path; allow threading for speed when
25207 any of BBs along the path are optimized for speed.
25209 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25211 PR middle-end/78468
25212 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25213 settings of the virtual registers.
25216 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25218 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25219 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25222 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25224 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25225 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25227 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25229 * config/s390/s390.md: Add missing comments with the expanded
25231 * config/s390/vector.md: Likewise.
25232 * config/s390/vx-builtins.md: Likewise.
25234 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25237 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25238 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25239 conditions on a single line.
25241 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25243 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25244 __S390_VX__ to __VX__.
25246 2017-02-01 Andrew Pinski <apinski@cavium.com>
25248 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25249 stmt_info to record_stmt_cost.
25250 (vect_get_known_peeling_cost): Pass stmt_info if known to
25252 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25253 cpu_vector_cost field into
25254 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25255 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25256 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25257 splitting of scalar_stmt_cost and vec_stmt_cost.
25258 (thunderx_vector_cost): Likewise.
25259 (cortexa57_vector_cost): LIkewise.
25260 (exynosm1_vector_cost): Likewise.
25261 (xgene1_vector_cost): Likewise.
25262 (thunderx2t99_vector_cost): Improve after the splitting of the two
25264 (aarch64_builtin_vectorization_cost): Update for the splitting of
25265 scalar_stmt_cost and vec_stmt_cost.
25267 2017-02-01 Torvald Riegel <triegel@redhat.com>
25268 Richard Henderson <rth@redhat.com>
25270 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25271 conditional on existance of a fast atomic load.
25272 * optabs-query.c (can_atomic_load_p): New function.
25273 * optabs-query.h (can_atomic_load_p): Declare it.
25274 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25275 no fast atomic load is available for the particular size of access.
25276 (expand_atomic_compare_and_swap): Likewise.
25277 (expand_atomic_load): Likewise.
25278 (expand_atomic_store): Likewise.
25279 (expand_atomic_fetch_op): Likewise.
25280 * testsuite/lib/target-supports.exp
25281 (check_effective_target_sync_int_128): Remove x86 because it provides
25282 no fast atomic load.
25283 (check_effective_target_sync_int_128_runtime): Likewise.
25285 2017-02-01 Richard Biener <rguenther@suse.de>
25287 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25288 (graphite_transform_loops): Provide opt-info for optimized nests.
25289 * tree-parloop.c (parallelize_loops): Provide opt-info for
25290 parallelized loops.
25292 2017-02-01 Richard Biener <rguenther@suse.de>
25294 PR middle-end/79315
25295 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25296 was not set before.
25298 2017-02-01 Richard Biener <rguenther@suse.de>
25300 PR tree-optimization/71824
25301 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25302 Verify the loops are valid in the merged SESE region.
25303 (scop_detection::can_represent_loop_1): Check analyzing the
25304 evolution of the number of iterations in the region succeeds.
25306 2017-01-31 Ian Lance Taylor <iant@golang.org>
25308 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25309 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25311 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25313 PR preprocessor/79210
25314 * input.c (get_substring_ranges_for_loc): Replace line_width
25315 assertion with error-handling.
25317 2017-01-31 Richard Biener <rguenther@suse.de>
25319 PR tree-optimization/77318
25320 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25321 (create_pw_aff_from_tree): Take loop parameter.
25322 (add_condition_to_pbb): Pass loop of the condition to
25323 create_pw_aff_from_tree.
25325 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25327 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25328 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25330 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25334 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25336 (convert_int_to_float128): Likewise.
25337 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25338 (convert_int_to_float128): Likewise.
25339 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25340 (UNSPEC_IEEE128_CONVERT): Likewise.
25341 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25342 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25343 Use local variables for IBM extended format.
25344 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25345 (fix_trunc<mode>si2_fprs): Likewise.
25346 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25347 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25348 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25349 to know that we can now have integers of all sizes in vector
25351 (fix<uns>_<mode>di2_hw): Likewise.
25352 (float<uns>_<mode>si2_hw): Likewise.
25353 (fix_<mode>si2_hw): Likewise.
25354 (fixuns_<mode>si2_hw): Likewise.
25355 (float<uns>_<mode>di2_hw): Likewise.
25356 (float_<mode>di2_hw): Likewise.
25357 (float_<mode>si2_hw): Likewise.
25358 (floatuns_<mode>di2_hw): Likewise.
25359 (floatuns_<mode>si2_hw): Likewise.
25360 (xscvqp<su>wz_<mode>): Delete, no longer used.
25361 (xscvqp<su>dz_<mode>): Likewise.
25362 (xscv<su>dqp_<mode>): Likewise.
25363 (ieee128_mfvsrd_64bit): Likewise.
25364 (ieee128_mfvsrd_32bit): Likewise.
25365 (ieee128_mfvsrwz): Likewise.
25366 (ieee128_mtvsrw): Likewise.
25367 (ieee128_mtvsrd_64bit): Likewise.
25368 (ieee128_mtvsrd_32bit): Likewise.
25370 2017-01-31 Martin Liska <mliska@suse.cz>
25373 * ipa-prop.c (ipa_free_all_node_params): Call release method
25374 instead of ~sumbol_summary to not to trigger double times
25377 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25379 PR tree-optimization/71691
25380 * bitmap.h (class auto_bitmap): New.
25381 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25382 is_maybe_undefined instead of ssa_undefined_value_p.
25384 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25386 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25387 __S390_ARCH_LEVEL__ to __ARCH__.
25389 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25391 PR tree-optimization/79267
25392 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25393 if should_remove_lhs_p is true.
25395 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25398 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25399 (add_alignment_attribute): New.
25400 (base_type_die): Add alignment attribute.
25401 (subrange_type_die): Likewise.
25402 (modified_type_die): Likewise.
25403 (gen_array_type_die): Likewise.
25404 (gen_descr_array_type_die: Likewise.
25405 (gen_enumeration_type_die): Likewise.
25406 (gen_subprogram_die): Likewise.
25407 (gen_variable_die): Likewise.
25408 (gen_field_die): Likewise.
25409 (gen_ptr_to_mbr_type_die): Likewise.
25410 (gen_struct_or_union_type_die): Likewise.
25411 (gen_subroutine_type_die): Likewise.
25412 (gen_typedef_die): Likewise.
25413 (base_type_cmp): Compare alignment attribute.
25415 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25418 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25419 (setb_unsigned) New pattern for setb with CCUNS.
25420 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25421 subfc./subfe sequence to avoid overflow problems. Generate a
25422 shorter sequence with cmpld/setb for power9.
25423 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25424 for generating subfc. instruction.
25425 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25426 now uses this instruction.
25428 2017-01-30 Ian Lance Taylor <iant@google.com>
25431 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25432 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25434 2017-01-30 Martin Sebor <msebor@redhat.com>
25436 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25437 Move constant to the right of a relational operator.
25438 (get_mpfr_format_length, format_character, format_string): Ditto.
25439 (should_warn_p, maybe_warn): Same.
25441 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25443 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25446 * asan.c (get_translation_unit_decl): Remove function.
25447 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25449 2017-01-30 Martin Liska <mliska@suse.cz>
25451 PR gcov-profile/79259
25452 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25453 -fprofile-generate.
25455 2017-01-30 Martin Liska <mliska@suse.cz>
25458 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25459 Initialize variables with NULL value.
25461 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25464 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25466 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25468 2017-01-30 Richard Biener <rguenther@suse.de>
25470 PR tree-optimization/79276
25471 * tree-vrp.c (process_assert_insertions): Properly adjust common
25472 when removing a duplicate.
25474 * gcc.dg/torture/pr79276.c: New testcase.
25476 2017-01-30 Richard Biener <rguenther@suse.de>
25478 PR tree-optimization/79256
25479 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25480 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25482 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25484 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25487 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25488 ("*r<noxa>sbg_<mode>_sll_bitmask")
25489 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25490 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25491 Use contiguous_bitmask_nowrap_operand.
25493 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25496 * config/rs6000/altivec.h (vec_xl): Revise #define.
25497 (vec_xst): Likewise.
25499 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25501 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25503 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25505 PR rtl-optimization/79194
25506 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25507 traps before call to bypass_conditional_jumps.
25509 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25511 PR tree-optimization/71374
25512 * lra-constraints.c (check_conflict_input_operands): New.
25513 (match_reload): Use it.
25515 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25518 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25519 account to calculate conflict_set.
25521 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25523 PR rtl-optimization/78559
25524 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25525 other_insn in combine.
25527 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25529 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25530 uint16_type_node for BT_UINT16.
25532 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25534 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25535 "RTL Tests" to menu.
25536 (GIMPLE Tests): New node.
25537 (RTL Tests): New node.
25539 2017-01-27 Richard Biener <rguenther@suse.de>
25541 PR tree-optimization/79245
25542 * tree-loop-distribution.c (distribute_loop): Apply cost
25543 modeling also to detected patterns.
25545 2017-01-27 Richard Biener <rguenther@suse.de>
25547 PR tree-optimization/71433
25548 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25549 (compare_assert_loc): New function.
25550 (process_assert_insertions): Sort and optimize assert locations
25551 to remove duplicates and push down identical assertions on
25552 edges to their destination block.
25554 2017-01-27 Richard Biener <rguenther@suse.de>
25556 PR tree-optimization/79244
25557 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25558 out SSA names even if abnormal.
25560 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25562 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25563 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25564 instead of MPFR_RNDN.
25566 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25569 * arm.c (arm_option_override): Don't call build_target_option_node
25570 until after doing all option overrides.
25571 (arm_valid_target_attribute_tree): Likewise.
25573 2017-01-27 Martin Liska <mliska@suse.cz>
25575 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25576 for {cd}tors and C++ {cd}tors.
25578 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25580 * config/s390/s390.md ("*setmem_long_and")
25581 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25583 2017-01-26 Martin Sebor <msebor@redhat.com>
25585 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25588 2017-01-26 Martin Sebor <msebor@redhat.com>
25590 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25591 HAVE_DFmode before using XFmode or DFmode.
25592 (parse_directive): Avoid using the z length modifier to avoid
25593 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25595 PR middle-end/78703
25596 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25597 to accept adjustment as an array.
25598 (get_int_range): New function.
25599 (struct directive): Make width and prec arrays.
25600 (directive::set_width, directive::set_precision): Call get_int_range.
25601 (format_integer, format_floating): Handle width and precision ranges.
25602 (format_string, parse_directive): Same.
25604 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25607 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25608 comdat_type_p set, just clone them, but keep the children in the
25612 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25613 which have direct callers with -fvar-tracking-assignments enabled
25615 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25616 inside of type units.
25618 2017-01-26 Martin Sebor <msebor@redhat.com>
25620 PR middle-end/78703
25621 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25623 (struct format_result): Replace number_chars, number_chars_min,
25624 and number_chars_max with a single member of struct result_range.
25626 (format_result::operator+=): Adjust.
25627 (struct fmtresult): Remove bounded. Handle likely and unlikely
25629 (fmtresult::adjust_for_width_or_precision): New function.
25630 (fmtresult:type_max_digits): New function.
25631 (bytes_remaining): Handle likely and unlikely counters.
25632 (min_bytes_remaining): Remove.
25633 (format_percent): Simplify.
25634 (format_integer, format_floating): Set likely and unlikely counters.
25635 (get_string_length, format_character, format_string): Same.
25636 (format_plain, should_warn_p): New function.
25637 (maybe_warn): Call should_warn_p. Update diagnostic messages
25638 and handle those for all directives, including plain strings.
25639 (format_directive): Handle likely and unlikely counters.
25640 Remove unnecessary quoting from diagnostics. Add an informational
25642 (add_bytes): Remove.
25643 (pass_sprintf_length::compute_format_length): Simplify.
25644 (try_substitute_return_value): Handle likely and unlikely counters.
25646 2017-01-26 Carl Love <cel@us.ibm.com>
25648 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25649 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25651 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25654 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25655 endianess for subregs into account.
25656 * lra-constraints.c (lra_constraints): Do risky transformations
25657 always on the first iteration.
25658 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25659 last_call_used_reg_set.
25660 (process_bb_lives): Define and use last_call_used_reg_set.
25661 * lra.c (lra): Always continue after lra_constraints on the first
25664 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25666 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25668 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25670 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25672 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25673 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25674 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25675 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25676 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25677 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25678 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25679 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25680 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25682 2017-01-26 Marek Polacek <polacek@redhat.com>
25685 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25686 for the third operand.
25688 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25690 PR middle-end/79236
25691 * omp-low.c (struct omp_context): Add simt_stmt field.
25692 (scan_omp_for): Return omp_context *.
25693 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25694 context to the _simt_ SIMD stmt.
25695 (lower_omp_for): For combined SIMD with sibling _simt_
25696 SIMD, make sure to use the same decls in _looptemp_
25697 clauses as in the sibling.
25699 2017-01-26 David Sherwood <david.sherwood@arm.com>
25701 PR middle-end/79212
25702 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25705 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25708 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25709 emit fld b; fld a; if possible.
25711 * brig-builtins.def: Update copyright years.
25712 * config/arm/arm_acle_builtins.def: Update copyright years.
25714 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25717 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25718 constraint instead of o for the stxsd instruction.
25720 2017-01-25 Carl Love <cel@us.ibm.com>
25722 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25723 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25725 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25727 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25729 2017-01-25 Richard Biener <rguenther@suse.de>
25731 PR tree-optimization/69264
25732 * target.def (vector_alignment_reachable): Improve documentation.
25733 * doc/tm.texi: Regenerate.
25734 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25736 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25737 earlier changes with respect to TYPE_USER_ALIGN.
25738 (vector_alignment_reachable_p): Likewise. Improve dumping.
25740 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25743 * config/arm/arm.md (xordi3): Force constant operand into a register
25746 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25748 * doc/invoke.texi (-fstore-merging): Correct default optimization
25749 levels at which it is enabled.
25750 (-O): Move -fstore-merging from list to...
25753 2017-01-25 Richard Biener <rguenther@suse.de>
25756 * omp-expand.c: Include debug.h.
25757 (expand_omp_taskreg): Make sure to generate early debug before
25758 outlining anything from a function.
25759 (expand_omp_target): Likewise.
25760 (grid_expand_target_grid_body): Likewise.
25762 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25765 * asan.c (get_translation_unit_decl): New function.
25766 (asan_add_global): Extract modules file name from globals
25767 TRANSLATION_UNIT_DECL name.
25769 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
25772 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
25773 for long calls with APCS frame and VFP.
25775 2017-01-24 David Malcolm <dmalcolm@redhat.com>
25777 * cfg.c (original_copy_tables_initialized_p): New function.
25778 * cfg.h (original_copy_tables_initialized_p): New decl.
25779 * cfgrtl.c (relink_block_chain): Guard the call to
25780 free_original_copy_tables with a call to
25781 original_copy_tables_initialized_p.
25782 * cgraph.h (symtab_node::native_rtl_p): New decl.
25783 * cgraphunit.c (symtab_node::native_rtl_p): New function.
25784 (symtab_node::needed_p): Don't assert for early assembly output
25785 for __RTL functions.
25786 (cgraph_node::finalize_function): Set "force_output" for __RTL
25788 (cgraph_node::analyze): Bail out early for __RTL functions.
25789 (analyze_functions): Update assertion to support __RTL functions.
25790 (cgraph_node::expand): Bail out early for __RTL functions.
25791 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25793 * function.h (struct function): Update comment for field
25795 * gimple-expr.c: Include "tree-pass.h".
25796 (gimple_has_body_p): Return false for __RTL functions.
25797 * Makefile.in (OBJS): Add run-rtl-passes.o.
25798 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25800 (gcc::pass_manager::get_clean_slate): New accessor.
25801 * passes.c: Include "insn-addr.h".
25802 (should_skip_pass_p): Add logging. Update logic for running
25803 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25804 property-provider override so it is only done for gimple passes.
25806 (skip_pass): New function.
25807 (execute_one_pass): Call skip_pass when skipping passes.
25808 * read-md.c (md_reader::read_char): Support filtering
25809 the input to a subset of line numbers.
25810 (md_reader::md_reader): Initialize fields
25811 m_first_line and m_last_line.
25812 (md_reader::read_file_fragment): New function.
25813 * read-md.h (md_reader::read_file_fragment): New decl.
25814 (md_reader::m_first_line): New field.
25815 (md_reader::m_last_line): New field.
25816 * read-rtl-function.c (function_reader::create_function): Only
25817 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25818 curr_properties. Set DECL_INITIAL to a dummy block.
25819 (read_rtl_function_body_from_file_range): New function.
25820 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25822 * run-rtl-passes.c: New file.
25823 * run-rtl-passes.h: New file.
25825 2017-01-24 Jeff Law <law@redhat.com>
25827 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25830 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25832 PR tree-optimization/79159
25833 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25834 (record_nonwrapping_iv): Improve boundary using above function if no
25835 value range information.
25837 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25838 Martin Jambor <mjambor@suse.cz>
25840 * brig-builtins.def: New file.
25841 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25842 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25843 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25844 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25845 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25846 * builtin-types.def (BT_INT8): New.
25847 (BT_INT16): Likewise.
25848 (BT_UINT8): Likewise.
25849 (BT_UINT16): Likewise.
25850 (BT_FN_ULONG): Likewise.
25851 (BT_FN_UINT_INT): Likewise.
25852 (BT_FN_UINT_ULONG): Likewise.
25853 (BT_FN_UINT_LONG): Likewise.
25854 (BT_FN_UINT_PTR): Likewise.
25855 (BT_FN_ULONG_PTR): Likewise.
25856 (BT_FN_INT8_FLOAT): Likewise.
25857 (BT_FN_INT16_FLOAT): Likewise.
25858 (BT_FN_UINT32_FLOAT): Likewise.
25859 (BT_FN_UINT16_FLOAT): Likewise.
25860 (BT_FN_UINT8_FLOAT): Likewise.
25861 (BT_FN_UINT64_FLOAT): Likewise.
25862 (BT_FN_UINT16_UINT32): Likewise.
25863 (BT_FN_UINT32_UINT16): Likewise.
25864 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25865 (BT_FN_INT_PTR_INT): Likewise.
25866 (BT_FN_UINT_PTR_UINT): Likewise.
25867 (BT_FN_LONG_PTR_LONG): Likewise.
25868 (BT_FN_ULONG_PTR_ULONG): Likewise.
25869 (BT_FN_VOID_UINT64_UINT64): Likewise.
25870 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25871 (BT_FN_INT8_INT8_INT8): Likewise.
25872 (BT_FN_INT16_INT16_INT16): Likewise.
25873 (BT_FN_INT_INT_INT): Likewise.
25874 (BT_FN_UINT_FLOAT_UINT): Likewise.
25875 (BT_FN_FLOAT_UINT_UINT): Likewise.
25876 (BT_FN_ULONG_UINT_UINT): Likewise.
25877 (BT_FN_ULONG_UINT_PTR): Likewise.
25878 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25879 (BT_FN_UINT_UINT_UINT): Likewise.
25880 (BT_FN_VOID_UINT_PTR): Likewise.
25881 (BT_FN_UINT_UINT_PTR: Likewise.
25882 (BT_FN_UINT32_UINT64_PTR): Likewise.
25883 (BT_FN_INT_INT_UINT_UINT): Likewise.
25884 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25885 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25886 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25887 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25888 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25889 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25890 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25891 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25892 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25893 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25894 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25895 * doc/frontends.texi: List BRIG FE.
25896 * doc/install.texi (Testing): Add BRIG tesring requirements.
25897 * doc/invoke.texi (Overall Options): Mention BRIG.
25898 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25900 2017-01-24 Richard Biener <rguenther@suse.de>
25902 PR translation/79208
25903 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25905 2017-01-24 Martin Jambor <mjambor@suse.cz>
25908 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25909 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25910 and known_contexts.
25912 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25914 PR middle-end/79123
25915 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25916 casts from signed to unsigned really don't have a range.
25918 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25920 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25921 GMP_RNDx for compatiblity.
25923 2017-01-24 Martin Liska <mliska@suse.cz>
25926 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25927 that would prevent us to call alloca with -1 as argument.
25929 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25931 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25932 Avoid -Wformat-security warning.
25934 2017-01-23 Andrew Pinski <apinski@cavium.com>
25936 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25939 2017-01-23 Martin Sebor <msebor@redhat.com>
25941 PR middle-end/78703
25942 * gimple-ssa-sprintf.c (warn_level): New global.
25943 (format_integer): Use it here and throughout the rest of the file.
25944 Use the same switch to compute sign as base.
25945 (maybe_warn): New function.
25946 (format_directive): Factor out warnings into maybe_warn.
25947 Add debugging output. Use warn_level.
25948 (add_bytes): Use warn_level.
25949 (pass_sprintf_length::compute_format_length): Add debugging output.
25950 (try_substitute_return_value): Same.
25951 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25953 PR middle-end/78703
25954 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25955 (struct fmtresult, format_integer, format_floating): Adjust.
25956 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25957 (get_string_length, format_string,format_directive): Same.
25958 (pass_sprintf_length::compute_format_length): Same.
25959 (try_substitute_return_value): Simplify slightly.
25961 PR middle-end/78703
25962 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25963 (fmtresult::operator+=): Outlined.
25964 (struct fmtresult): Add ctors.
25965 (struct conversion_spec): Rename...
25966 (struct directive): ...to this. Add and remove data members.
25967 (directive::set_width, directive::set_precision): New functions.
25968 (format_percent): Use fmtresult ctor.
25969 (get_width_and_precision): Remove.
25970 (format_integer): Make naming changes. Avoid computing width and
25972 (format_floating): Same. Adjust indentation.
25973 (format_character, format_none): New functions.
25974 (format_string): Moved character handling to format_character.
25975 (format_directive): Remove arguments, change return type.
25976 (parse_directive): New function.
25977 (pass_sprintf_length::compute_format_length): Move directive
25978 parsing to parse_directive.
25980 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25982 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25983 (assign_assembler_name_if_needed): ... this.
25984 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25985 (assign_assembler_name_if_needed): ... this.
25986 (free_lang_data_in_cgraph): Adjust callers.
25987 * cgraphunit.c (cgraph_node::analyze): Likewise.
25988 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25991 2017-01-23 Richard Biener <rguenther@suse.de>
25993 PR tree-optimization/79088
25994 PR tree-optimization/79188
25995 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25996 resetting loop bounds after last path deletion. Reset loop
25997 bounds of the target loop, make code match the comments.
25998 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25999 Make sure loops need no fixups.
26001 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26003 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26004 exponent support with double type for first argument.
26005 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26006 type returned by __builtin_vec_extract_sig,
26007 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26008 functions from "vector int" to "vector unsigned int" or from
26009 "vector long long int" to "vector unsigned long long int".
26010 Changed type returned by __builtin_vec_extract_exp,
26011 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26012 functions from "vector int" to "vector unsigned int" or from
26013 "vector long long int" to "vector unsigned long long int".
26014 Changed return type of __builtin_vec_test_data_class,
26015 __builtin_vec_test_data_class_sp, and
26016 __builtin_vec_test_data_class_dp from "vector int" to
26017 "vector bool int" or from "vector long long int" to "vector bool
26018 long long int" and changed second argument type from "unsigned
26019 int" to "int". Added new overloaded function forms "vector float
26020 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26021 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26022 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26023 double, vector unsigned long long int)" and "vector double
26024 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26025 long int)". Changed return type of
26026 __builtin_scalar_test_data_class and
26027 __builtin_scalar_test_data_class_sp and
26028 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26029 int" and changed second argument from "unsigned int" to "int".
26030 Changed type returned by __builtin_scalar_test_neg,
26031 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26032 from "int" to "bool int". Added new overloaded function form
26033 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26034 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26035 exponent double-precision with floating point first argument.
26036 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26037 documentation of scalar_test_data_class, scalar_test_neg,
26038 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26039 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26040 vec_test_data_class built-in functions to reflect refinements in
26041 their type signatures.
26043 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26045 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26047 (aarch64_elf_asm_destructor): Likewise.
26049 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26051 PR rtl-optimization/78634
26052 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26053 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26054 * ifcvt.c (noce_try_cmove): Add missing cost check.
26056 PR rtl-optimization/71724
26057 * combine.c (if_then_else_cond): Look for situations where it is
26058 beneficial to undo the work of one of the recursive calls.
26060 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26062 PR tree-optimization/70754
26063 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26064 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26065 combined stmt before it if not NULL.
26066 (combine_chains): Process refs reversely and compute dominance point
26069 2017-01-23 Martin Liska <mliska@suse.cz>
26071 PR tree-optimization/79196
26072 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26073 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26075 (strlen_optimize_stmt): Call the renamed function.
26077 2017-01-23 Michael Matz <matz@suse.de>
26079 PR tree-optimization/78384
26080 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26082 2017-01-23 Richard Biener <rguenther@suse.de>
26084 PR tree-optimization/79186
26085 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26086 both incoming edges before moving an assert.
26088 2017-01-23 Martin Jambor <mjambor@suse.cz>
26090 * ipa-prop.c (load_from_param_1): Removed.
26091 (load_from_unmodified_param): Bits from load_from_param_1 put back
26093 (load_from_param): Removed.
26094 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26095 with stmt. Reverted back to use of load_from_unmodified_param.
26097 2017-01-23 Martin Jambor <mjambor@suse.cz>
26100 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26101 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26102 field a pointer to garbage collected vector, mark lattices and
26103 ipcp_orig_node with GTY((skip)).
26104 (ipa_get_param_count): Adjust to descriptors being a pointer.
26105 (ipa_get_param): Likewise.
26106 (ipa_get_type): Likewise.
26107 (ipa_get_param_move_cost): Likewise.
26108 (ipa_set_param_used): Likewise.
26109 (ipa_get_controlled_uses): Likewise.
26110 (ipa_set_controlled_uses): Likewise.
26111 (ipa_is_param_used): Likewise.
26112 (ipa_node_params_t): Move into garbage collector. New methods insert
26114 (ipa_node_params_sum): Annotate wth GTY(()).
26115 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26117 (ipa_load_from_parm_agg): Adjust declaration.
26118 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26119 * ipa-profile.c (ipa_profile): Likewise.
26120 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26121 (ipa_populate_param_decls): Make descriptors parameter garbage
26123 (ipa_dump_param): Adjust to descriptors being a pointer.
26124 (ipa_alloc_node_params): Likewise.
26125 (ipa_initialize_node_params): Likewise.
26126 (load_from_param_1): Make descriptors parameter garbage collected.
26127 (load_from_unmodified_param): Likewise.
26128 (load_from_param): Likewise.
26129 (ipa_load_from_parm_agg): Likewise.
26130 (ipa_node_params::~ipa_node_params): Removed.
26131 (ipa_free_all_node_params): Remove call to delete operator.
26132 (ipa_node_params_t::insert): New.
26133 (ipa_node_params_t::remove): Likewise.
26134 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26135 copy known_csts and known_contexts vectors.
26136 (ipa_read_node_info): Adjust to descriptors being a pointer.
26137 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26139 (ipcp_transform_function): Make descriptors variable garbage
26142 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26144 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26145 * config/i386/avx512dqintrin.h: Ditto.
26146 * config/i386/avx512fintrin.h: Ditto.
26147 * config/i386/i386.c: Handle new builtins.
26148 * config/i386/i386-builtin.def: Add new builtins.
26149 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26150 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26152 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26153 Martin Liska <mliska@suse.cz>
26155 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26156 * asan.c (asan_expand_poison_ifn): Support stores and use
26157 appropriate ASAN report function.
26158 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26159 * internal-fn.def (ASAN_POISON_USE): Declare.
26160 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26161 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26162 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26163 ASAN_POISON calls w/o LHS.
26164 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26165 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26166 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26167 * gimplify.c (asan_poison_variables): Add attribute
26168 use_after_scope_memory to variables that really needs to live
26170 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26171 having the attribute.
26173 2017-01-23 Martin Liska <mliska@suse.cz>
26175 * asan.c (create_asan_shadow_var): New function.
26176 (asan_expand_poison_ifn): Likewise.
26177 * asan.h (asan_expand_poison_ifn): New declaration.
26178 * internal-fn.c (expand_ASAN_POISON): Likewise.
26179 * internal-fn.def (ASAN_POISON): New builtin.
26180 * sanopt.c (pass_sanopt::execute): Expand
26181 asan_expand_poison_ifn.
26182 * tree-inline.c (copy_decl_for_dup_finish): Make function
26184 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26185 * tree-ssa.c (is_asan_mark_p): New function.
26186 (execute_update_addresses_taken): Rewrite local variables
26187 (identified just by use-after-scope as addressable) into SSA.
26189 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26191 * doc/install.texi (Specific): opensource.apple.com uses https
26192 now. Remove trailing slash.
26194 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26196 * README.Portability: Remove note on an Irix compatibility issue.
26198 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26200 * gcov.c (INCLUDE_ALGORITHM): Define.
26201 (INCLUDE_VECTOR): Define.
26202 No longer include <vector> and <algorithm> directly.
26204 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26206 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26208 * doc/invoke.texi (Code Gen Options): Ditto.
26210 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26213 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26215 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26217 rtl-optimization/79125
26218 * cprop.c (local_cprop_pass): Handle cases where we make an
26219 unconditional trap.
26221 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26225 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26226 read from, for big endian.
26228 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26230 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26231 register pauth builtins for LP64 only.
26233 2017-01-20 Marek Polacek <polacek@redhat.com>
26236 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26239 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26241 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26243 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26244 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26245 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26247 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26250 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26251 in big-endian mode when they are not a single duplicated value.
26253 2017-01-20 Richard Biener <rguenther@suse.de>
26255 * BASE-VER: Bump to 7.0.1.
26257 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26259 * omp-low.c (omplow_simd_context): New struct. Use it...
26260 (lower_rec_simd_input_clauses): ...here and...
26261 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26262 references to idx, lane, max_vf, is_simt.
26264 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26266 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26269 2017-01-20 Martin Jambor <mjambor@suse.cz>
26271 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26272 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26274 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26275 (GTFILES): Rename hsa.c to hsa-common.c.
26276 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26277 * hsa-dump.c: Likewise.
26278 * hsa-gen.c: Likewise.
26279 * hsa-regalloc.c: Likewise.
26280 * ipa-hsa.c: Likewise.
26281 * omp-expand.c: Likewise.
26282 * omp-low.c: Likewise.
26283 * toplev.c: Likewise.
26285 2017-01-20 Marek Polacek <polacek@redhat.com>
26288 * doc/invoke.texi: Document -Wduplicated-branches.
26289 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26290 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26291 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26292 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26293 return 0 only when not OEP_LEXICOGRAPHIC.
26294 (fold_build_cleanup_point_expr): Use the expression
26295 location when building CLEANUP_POINT_EXPR.
26296 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26297 * tree.c (add_expr): Handle error_mark_node.
26299 2017-01-20 Martin Liska <mliska@suse.cz>
26302 * tree-profile.c (init_ic_make_global_vars): Do not call
26304 (gimple_init_gcov_profiler): Likewise.
26306 2017-01-20 Martin Liska <mliska@suse.cz>
26309 * cgraph.h (maybe_create_reference): Remove argument and
26311 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26313 * ipa-cp.c (create_specialized_node): Likewise.
26314 * symtab.c (symtab_node::maybe_create_reference): Handle
26315 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26317 2017-01-20 Martin Liska <mliska@suse.cz>
26319 * read-rtl-function.c (function_reader::create_function): Use
26320 build_decl instread of build_decl_stat.
26322 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26324 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26325 * config/i386/avx512dqintrin.h: Ditto.
26326 * config/i386/avx512fintrin.h: Ditto.
26327 * config/i386/i386-builtin-types.def: Add new types.
26328 * config/i386/i386.c: Handle new types.
26329 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26330 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26331 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26332 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26333 (__builtin_ia32_kshiftridi): New.
26334 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26336 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26340 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26341 define to rs6000_init_stack_protect_guard.
26342 (rs6000_init_stack_protect_guard): New function.
26344 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26345 Yunqiang Su <yunqiang.su@imgtec.com>
26347 * config.gcc (supported_defaults): Add madd4.
26348 (with_madd4): Add validation.
26349 (all_defaults): Add madd4.
26350 * config/mips/mips.opt (mmadd4): New option.
26351 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26353 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26355 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26356 (ISA_HAS_FUSED_MADD4): Likewise.
26357 * doc/invoke.texi (-mmadd4): Document the new option.
26358 * doc/install.texi (--with-madd4): Document the new option.
26360 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26362 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26363 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26364 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26365 (aarch64_init_pauth_hint_builtins): New.
26366 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26367 (aarch64_expand_builtin): Expand new builtins.
26369 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26371 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26372 * combine-stack-adj.c (no_unhandled_cfa): Handle
26373 REG_CFA_TOGGLE_RA_MANGLE.
26374 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26375 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26376 info for return address signing.
26377 (aarch64_expand_epilogue): Likewise.
26379 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26381 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26382 * config/aarch64/aarch64-protos.h
26383 (aarch64_return_address_signing_enabled): New declaration.
26384 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26386 (aarch64_expand_prologue): Sign return address before it's pushed onto
26388 (aarch64_expand_epilogue): Authenticate return address fetched from
26390 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26391 (aarch64_attributes): New function attributes for "sign-return-address".
26392 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26393 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26394 ("*do_return"): Generate combined instructions according to key index.
26395 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26396 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26398 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26399 * config/aarch64/aarch64.opt (msign-return-address=): New.
26400 * doc/extend.texi (AArch64 Function Attributes): Documents
26401 "sign-return-address=".
26402 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26404 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26406 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26407 overall option summary.
26409 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26411 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26412 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26413 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26414 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26416 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26418 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26419 -mpower9-minmax by default for -mcpu=power9.
26420 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26421 128-bit floating point.
26423 2017-01-20 Alan Modra <amodra@gmail.com>
26425 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26426 optimizing for size.
26428 2017-01-20 Alan Modra <amodra@gmail.com>
26431 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26432 for strcmp and strncmp from corresponding builtin decl.
26434 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26436 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26437 instead of i386/rtems-64.h.
26438 * config/i386/rtems-64.h: Remove.
26440 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26444 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26446 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26448 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26450 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26451 Change int to HOST_WIDE_INT.
26452 * config/aarch64/aarch64-protos.h
26453 (aarch64_simd_gen_const_vector_dup): Likewise.
26454 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26456 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26458 * langhooks-def.h (lhd_type_for_size): New decl.
26459 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26460 * langhooks.c (lhd_type_for_size): New function, taken from
26463 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26465 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26466 define_bypass for CR latency.
26467 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26468 (power9-alu2): Add define_bypass for CR latency.
26470 (power9-mul): Update insn latency.
26471 (power9-mul-compare): Update insn latency, bypass latency and remove
26474 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26476 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26478 * config/aarch64/aarch64.md
26479 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26480 aarch64_nopcrelative_literal_loads.
26481 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26483 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26485 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26486 TARGET_LOONGSON_3A.
26487 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26489 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26492 * config.gcc (supported_defaults): Add lxc1-sxc1.
26493 (with_lxc1_sxc1): Add validation.
26494 (all_defaults): Add lxc1-sxc1.
26495 * config/mips/mips.opt (mlxc1-sxc1): New option.
26496 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26498 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26499 __mips_no_lxc1_sxc1.
26500 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26501 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26502 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26504 2017-01-19 Richard Biener <rguenther@suse.de>
26506 PR tree-optimization/72488
26507 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26508 sure to restore SSA info.
26509 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26511 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26513 PR rtl-optimization/79121
26514 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26515 of the inner type when shifting an extended value.
26517 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26520 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26521 interposable aliases.
26523 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26526 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26527 Use the evmergelohi instruction.
26528 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26529 (mov_sitf_e500_subreg8_2_be): Likewise.
26530 (mov_sitf_e500_subreg12_2_le): Likewise.
26531 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26532 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26533 (mov_sitf_e500_subreg8_2_le): Likewise.
26534 (mov_sitf_e500_subreg12_2_be): Likewise.
26536 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26538 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26539 attribute from vecsimple to vecperm.
26540 (altivec_vbpermq2): Likewise.
26542 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26545 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26547 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26548 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26549 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26550 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26551 case where N arg is SIZE_MAX.
26552 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26553 (cmpstrsi): Add pattern.
26555 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26557 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26558 __builtin_vec_revb builtins.
26559 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26560 built-in functions to support generation of the ISA 3.0 XXBR<x>
26561 vector byte reverse instructions.
26562 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26563 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26564 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26565 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26566 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26567 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26568 (P9V_BUILTIN_VEC_REVB): Likewise.
26569 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26570 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26571 (p9_xxbrq_v16qi): Likewise.
26572 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26573 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26574 (p9_xxbrh_v8hi): Likewise.
26575 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26576 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26577 vec_revb built-in functions.
26579 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26581 PR rtl-optimization/78952
26582 * config/i386/i386.md (any_extract): New code iterator.
26583 (*insvqi_2): Use any_extract for source operand.
26584 (*insvqi_3): Use any_shiftrt for source operand.
26586 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26588 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26590 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26592 2017-01-18 Matthias Klose <doko@ubuntu.com>
26594 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26596 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26598 * config/rs6000/altivec.h (vec_bperm): Change #define.
26599 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26600 (altivec_vbpermq2): New define_insn.
26601 (altivec_vbpermd): Likewise.
26602 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26603 function interface.
26604 (VBPERMD): Likewise.
26605 (VBPERM): New polymorphic function interface.
26606 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26607 Add entries for P9V_BUILTIN_VEC_VBPERM.
26608 * doc/extend.texi: Add interfaces for vec_bperm.
26610 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26612 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26613 first letter of error messages.
26614 (s390_resolve_overloaded_builtin): Likewise.
26615 * config/s390/s390.c (s390_expand_builtin): Likewise.
26616 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26617 (s390_valid_target_attribute_inner_p): Likewise.
26618 * config/s390/s390.md ("tabort"): Likewise.
26620 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26622 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26623 (ISA_AVOID_DIV_HILO): New macro.
26624 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26625 (ISA_HAS_DDIV): Likewise.
26627 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26629 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26631 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26633 * doc/invoke.texi (fabi-version): Spelling fix.
26635 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26638 * doc/invoke.texi (fabi-version): Mention mangling fix for
26641 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26644 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26646 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26649 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26652 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26654 (stack_protect_test): Ditto.
26655 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26656 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26658 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26659 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26660 -mstack-protector-guard-offset=.
26661 (RS/6000 and PowerPC Options): Ditto.
26663 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26665 * config/i386/i386.h (MASK_CLASS_P): New define.
26666 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26667 there are no registers from different register sets also when
26668 mask registers are used. Update function comment.
26669 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26670 to (*k/*r) and (*k/*km) alternatives.
26672 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26674 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26675 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26676 (EH_RETURN_HANDLER_RTX): New define.
26677 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26678 Force frame pointer in EH return functions.
26679 (aarch64_expand_epilogue): Add barrier for eh_return.
26680 (aarch64_final_eh_return_addr): Remove.
26681 (aarch64_eh_return_handler_rtx): New function.
26682 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26684 (aarch64_eh_return_handler_rtx): New prototype.
26686 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26688 * config/rs6000/altivec.h (vec_rlmi): New #define.
26689 (vec_vrlnm): Likewise.
26690 (vec_rlnm): Likewise.
26691 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26692 (UNSPEC_VRLNM): Likewise.
26693 (VIlong): New mode iterator.
26694 (altivec_vrl<VI_char>mi): New define_insn.
26695 (altivec_vrl<VI_char>nm): Likewise.
26696 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26698 (VRLDNM): Likewise.
26699 (RLNM): New polymorphic function entry.
26700 (VRLWMI): New monomorphic function entry.
26701 (VRLDMI): Likewise.
26702 (RLMI): New polymorphic function entry.
26703 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26704 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26705 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26708 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26711 * dwarf2out.c (field_byte_offset): Restore the
26712 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26713 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26714 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26717 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26720 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26722 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26725 * dwarf2out.c (add_data_member_location_attribute): For constant
26726 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26727 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26728 DW_AT_byte_size attributes.
26730 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26732 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26733 after forcing to constant memory when the code model is medium.
26735 2017-01-17 Julia Koval <julia.koval@intel.com>
26738 * config/i386/avx512fintrin.h
26739 (_mm512_i32gather_ps): Change __addr type to void const*.
26740 (_mm512_mask_i32gather_ps): Ditto.
26741 (_mm512_i32gather_pd): Ditto.
26742 (_mm512_mask_i32gather_pd): Ditto.
26743 (_mm512_i64gather_ps): Ditto.
26744 (_mm512_mask_i64gather_ps): Ditto.
26745 (_mm512_i64gather_pd): Ditto.
26746 (_mm512_mask_i64gather_pd): Ditto.
26747 (_mm512_i32gather_epi32): Ditto.
26748 (_mm512_mask_i32gather_epi32): Ditto.
26749 (_mm512_i32gather_epi64): Ditto.
26750 (_mm512_mask_i32gather_epi64): Ditto.
26751 (_mm512_i64gather_epi32): Ditto.
26752 (_mm512_mask_i64gather_epi32): Ditto.
26753 (_mm512_i64gather_epi64): Ditto.
26754 (_mm512_mask_i64gather_epi64): Ditto.
26755 (_mm512_i32scatter_ps): Change __addr type to void*.
26756 (_mm512_mask_i32scatter_ps): Ditto.
26757 (_mm512_i32scatter_pd): Ditto.
26758 (_mm512_mask_i32scatter_pd): Ditto.
26759 (_mm512_i64scatter_ps): Ditto.
26760 (_mm512_mask_i64scatter_ps): Ditto.
26761 (_mm512_i64scatter_pd): Ditto.
26762 (_mm512_mask_i64scatter_pd): Ditto.
26763 (_mm512_i32scatter_epi32): Ditto.
26764 (_mm512_mask_i32scatter_epi32): Ditto.
26765 (_mm512_i32scatter_epi64): Ditto.
26766 (_mm512_mask_i32scatter_epi64): Ditto.
26767 (_mm512_i64scatter_epi32): Ditto.
26768 (_mm512_mask_i64scatter_epi32): Ditto.
26769 (_mm512_i64scatter_epi64): Ditto.
26770 (_mm512_mask_i64scatter_epi64): Ditto.
26771 * config/i386/avx512pfintrin.h
26772 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
26773 (_mm512_mask_prefetch_i32gather_ps): Ditto.
26774 (_mm512_mask_prefetch_i64gather_pd): Ditto.
26775 (_mm512_mask_prefetch_i64gather_ps): Ditto.
26776 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
26777 (_mm512_prefetch_i32scatter_ps): Ditto.
26778 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
26779 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
26780 (_mm512_prefetch_i64scatter_pd): Ditto.
26781 (_mm512_prefetch_i64scatter_ps): Ditto.
26782 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
26783 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
26784 * config/i386/avx512vlintrin.h
26785 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
26786 (_mm_mmask_i32gather_ps): Ditto.
26787 (_mm256_mmask_i32gather_pd): Ditto.
26788 (_mm_mmask_i32gather_pd): Ditto.
26789 (_mm256_mmask_i64gather_ps): Ditto.
26790 (_mm_mmask_i64gather_ps): Ditto.
26791 (_mm256_mmask_i64gather_pd): Ditto.
26792 (_mm_mmask_i64gather_pd): Ditto.
26793 (_mm256_mmask_i32gather_epi32): Ditto.
26794 (_mm_mmask_i32gather_epi32): Ditto.
26795 (_mm256_mmask_i32gather_epi64): Ditto.
26796 (_mm_mmask_i32gather_epi64): Ditto.
26797 (_mm256_mmask_i64gather_epi32): Ditto.
26798 (_mm_mmask_i64gather_epi32): Ditto.
26799 (_mm256_mmask_i64gather_epi64): Ditto.
26800 (_mm_mmask_i64gather_epi64): Ditto.
26801 (_mm256_i32scatter_ps): Change __addr type to void*.
26802 (_mm256_mask_i32scatter_ps): Ditto.
26803 (_mm_i32scatter_ps): Ditto.
26804 (_mm_mask_i32scatter_ps): Ditto.
26805 (_mm256_i32scatter_pd): Ditto.
26806 (_mm256_mask_i32scatter_pd): Ditto.
26807 (_mm_i32scatter_pd): Ditto.
26808 (_mm_mask_i32scatter_pd): Ditto.
26809 (_mm256_i64scatter_ps): Ditto.
26810 (_mm256_mask_i64scatter_ps): Ditto.
26811 (_mm_i64scatter_ps): Ditto.
26812 (_mm_mask_i64scatter_ps): Ditto.
26813 (_mm256_i64scatter_pd): Ditto.
26814 (_mm256_mask_i64scatter_pd): Ditto.
26815 (_mm_i64scatter_pd): Ditto.
26816 (_mm_mask_i64scatter_pd): Ditto.
26817 (_mm256_i32scatter_epi32): Ditto.
26818 (_mm256_mask_i32scatter_epi32): Ditto.
26819 (_mm_i32scatter_epi32): Ditto.
26820 (_mm_mask_i32scatter_epi32): Ditto.
26821 (_mm256_i32scatter_epi64): Ditto.
26822 (_mm256_mask_i32scatter_epi64): Ditto.
26823 (_mm_i32scatter_epi64): Ditto.
26824 (_mm_mask_i32scatter_epi64): Ditto.
26825 (_mm256_i64scatter_epi32): Ditto.
26826 (_mm256_mask_i64scatter_epi32): Ditto.
26827 (_mm_i64scatter_epi32): Ditto.
26828 (_mm_mask_i64scatter_epi32): Ditto.
26829 (_mm256_i64scatter_epi64): Ditto.
26830 (_mm256_mask_i64scatter_epi64): Ditto.
26831 (_mm_i64scatter_epi64): Ditto.
26832 (_mm_mask_i64scatter_epi64): Ditto.
26833 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26834 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26835 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26836 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26837 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26838 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26839 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26840 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26841 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26842 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26843 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26844 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26845 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26846 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26847 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26848 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26849 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26850 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26851 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26852 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26853 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26854 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26855 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26856 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26857 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26858 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26859 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26860 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26861 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26862 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26863 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26864 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26865 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26866 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26867 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26868 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26869 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26870 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26871 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26872 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26873 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26874 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26875 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26876 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26877 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26878 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26879 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26880 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26881 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26882 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26883 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26884 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26885 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26886 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26887 definitions accordingly.
26889 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26890 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26893 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26896 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26899 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26900 hard regs for inner regno.
26902 2017-01-17 Martin Liska <mliska@suse.cz>
26905 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26906 assumption and add comment.
26908 2017-01-17 Nathan Sidwell <nathan@acm.org>
26910 * ipa-visibility.c (localize_node): New function, broken out of ...
26911 (function_and_variable_visibility): ... here. Call it.
26913 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26915 PR middle-end/77445
26916 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26917 correctly set frequency of oudgoing edge.
26918 (duplicate_thread_path): Fix profile updating.
26920 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26923 * configure.ac: Add GCC_BASE_VER.
26924 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26925 version from BASE-VER file.
26926 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26927 (gcc.o): Depend on $(BASEVER).
26928 * common.opt (dumpfullversion): New option.
26929 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26930 * doc/invoke.texi: Document -dumpfullversion.
26931 * doc/install.texi: Document --with-gcc-major-version-only.
26932 * configure: Regenerated.
26934 2017-01-17 Richard Biener <rguenther@suse.de>
26936 PR tree-optimization/71433
26937 * tree-vrp.c (register_new_assert_for): Merge same asserts
26938 on all incoming edges.
26939 (process_assert_insertions_for): Handle insertions at the
26942 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26944 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26945 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26947 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26950 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26953 2017-01-17 Alan Modra <amodra@gmail.com>
26956 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26957 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26958 symbolic stack limit when pic.
26960 2017-01-16 Martin Sebor <msebor@redhat.com>
26962 PR tree-optimization/78608
26963 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26965 2017-01-16 Jeff Law <law@redhat.com>
26968 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26969 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26970 for several include directories that may be relative to sysroot.
26971 * config/i386/x-mingw32 (gplus_includedir): Define.
26972 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26973 (native_system_includedir): Likewise.
26974 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26975 override if TARGET_SYSTEM_ROOT is defined.
26976 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26978 PR tree-optimization/79090
26979 PR tree-optimization/33562
26980 PR tree-optimization/61912
26981 PR tree-optimization/77485
26982 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26983 and computed trims into the dump file.
26985 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26987 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26989 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26992 * gimplify.c (gimplify_init_constructor): If want_value and
26993 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26997 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26998 sequence. Formatting fixes.
26999 (doloop_optimize): Formatting fixes.
27002 * gcc.c (debug_level_greater_than_spec_func): New function.
27003 (static_spec_functions): Add debug-level-gt spec function.
27004 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27006 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27007 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27008 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27009 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27010 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27011 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27013 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27015 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27016 QImode fixups to general and mask registers only.
27018 2017-01-16 Carl Love <cel@us.ibm.com>
27020 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27021 for built-in functions
27022 vector signed char vec_nabs (vector signed char)
27023 vector signed short vec_nabs (vector signed short)
27024 vector signed int vec_nabs (vector signed int)
27025 vector signed long long vec_nabs (vector signed long long)
27026 vector float vec_nabs (vector float)
27027 vector double vec_nabs (vector double)
27028 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27030 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27031 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27032 * doc/extend.texi: Update the documentation file for the new built-in
27035 2017-01-16 Martin Sebor <msebor@redhat.com>
27037 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27040 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27042 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27043 UNSPEC_VSX__XXSPLTD to require special splat handling.
27045 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27048 * system.h: Poison strndup.
27050 2017-01-16 Alan Modra <amodra@gmail.com>
27053 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27056 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27058 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27060 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27062 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27063 call recog here. Assert that INSN_CODE (insn) is non-negative.
27065 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27068 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27070 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27071 in the currently scheduled RTL fragment.
27073 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27075 PR rtl-optimization/78751
27076 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27079 2017-01-14 Jeff Law <law@redhat.com>
27081 PR tree-optimization/79090
27082 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27083 variable length stores.
27084 (compute_trims): Delete dead assignment to *trim_tail.
27085 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27088 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27090 PR rtl-optimization/78626
27091 PR rtl-optimization/78727
27092 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27093 of a block, and split such blocks after everything else is finished.
27095 2017-01-14 Alan Modra <amodra@gmail.com>
27098 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27099 target legitimate_combined_insn.
27100 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27101 (rs6000_legitimate_combined_insn): New function.
27102 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27104 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27105 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27106 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27108 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27110 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27112 2017-01-13 Jeff Law <law@redhat.com>
27114 PR tree-optimization/33562
27115 PR tree-optimization/61912
27116 PR tree-optimization/77485
27117 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27119 (delete_dead_assignment): Likewise.
27120 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27121 statement to delete_dead_call and delete_dead_assignment.
27123 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27126 * substring-locations.c (format_warning_va): Strengthen case 1 so
27127 that both endpoints of the substring must be within the format
27128 range for just the substring to be printed.
27130 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27132 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27133 * config/i386/i386.c (ix86_target_string): Add missing options
27134 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27135 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27136 flags_other and ix86_target_other to flags2_other. Display unknown
27138 (ix86_valid_target_attribute_inner_p): Add missing options and
27139 reorder options by implied ISAs, as in ix86_target_string.
27141 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27143 * hash-table.h (hash_table::too_empty_p): New function.
27144 (hash_table::expand): Use it.
27145 (hash_table::traverse): Likewise.
27146 (hash_table::empty_slot): Use sizeof (value_type) instead of
27147 sizeof (PTR) to convert bytes to elements. Shrink the table
27148 if the current size is excessive for the current number of
27151 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27153 * ira-costs.c (record_reg_classes): Break from the inner loop
27154 early once alt_fail is known to be true. Update outer loop
27155 handling accordingly.
27157 2017-01-13 Jeff Law <law@redhat.com>
27159 * tree-ssa-dse.c (decrement_count): New function.
27160 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27161 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27162 when we know the partially dead statement is a mem* function.
27164 PR tree-optimization/61912
27165 PR tree-optimization/77485
27166 * tree-ssa-dse.c: Include expr.h.
27167 (maybe_trim_constructor_store): New function.
27168 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27170 PR tree-optimization/33562
27171 PR tree-optimization/61912
27172 PR tree-optimization/77485
27173 * doc/invoke.texi: Document new dse-max-object-size param.
27174 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27175 * tree-ssa-dse.c: Include params.h.
27176 (dse_store_status): New enum.
27177 (initialize_ao_ref_for_dse): New, partially extracted from
27179 (valid_ao_ref_for_dse, normalize_ref): New.
27180 (setup_live_bytes_from_ref, compute_trims): Likewise.
27181 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27182 (maybe_trim_partially_dead_store): Likewise.
27183 (maybe_trim_complex_store): Likewise.
27184 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27185 Track what bytes live from the original store. Return tri-state
27186 for dead, partially dead or live.
27187 (dse_dom_walker): Add constructor, destructor and new private members.
27188 (delete_dead_call, delete_dead_assignment): New extracted from
27190 (dse_optimize_stmt): Make a member of dse_dom_walker.
27191 Use initialize_ao_ref_for_dse.
27193 PR tree-optimization/33562
27194 PR tree-optimization/61912
27195 PR tree-optimization/77485
27196 * sbitmap.h (bitmap_count_bits): Prototype.
27197 (bitmap_clear_range, bitmap_set_range): Likewise.
27198 * sbitmap.c (bitmap_clear_range): New function.
27199 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27201 2017-01-13 Martin Liska <mliska@suse.cz>
27204 * function.c (set_cfun): Add new argument force.
27205 * function.h (set_cfun): Likewise.
27206 * ipa-inline-transform.c (inline_call): Use the function when
27207 strict alising from is dropped for function we inline to.
27209 2017-01-13 Richard Biener <rguenther@suse.de>
27211 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27212 for dumping GIMPLE INTEGER_CSTs.
27214 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27216 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27217 to 201112L since C++17.
27219 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27222 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27223 if -fsanitize=kernel-address is present.
27225 2017-01-13 Richard Biener <rguenther@suse.de>
27227 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27228 as _Literal ( type ) number in case usual suffixes do not
27229 preserve all information.
27231 2017-01-13 Richard Biener <rguenther@suse.de>
27233 PR tree-optimization/77283
27234 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27235 and ssa-iterators.h.
27236 (is_feasible_trace): Implement a cost model based on joiner
27239 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27242 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27243 char or short to __float128/_Float128 directly.
27245 2017-01-12 Martin Sebor <msebor@redhat.com>
27247 to -Wformat-overflow.
27248 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27249 (min_bytes_remaining): Same.
27250 (get_string_length): Same.
27251 (format_string): Same.
27252 (format_directive): Same.
27254 (pass_sprintf_length::handle_gimple_call): Same.
27256 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27258 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27259 info.nowrite calls with no lhs that can't throw. Return bool
27260 whether gsi_remove has been called or not.
27261 (pass_sprintf_length::handle_gimple_call): Return bool whether
27262 try_substitute_return_value called gsi_remove. Formatting fix.
27263 (pass_sprintf_length::execute): Don't use gsi_remove if
27264 handle_gimple_call returned true.
27267 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27268 be removed due to side-effects, don't remove following barrier nor
27269 turn the successor edge into fallthru edge.
27271 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27274 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27275 element-reversing loads and stores as not swappable.
27277 2017-01-12 Nathan Sidwell <nathan@acm.org>
27278 Nicolai Stange <nicstange@gmail.com>
27280 * combine.c (try_combine): Don't ignore result of overlap checking
27281 loop. Combine overlap & asm check into single loop.
27283 2017-01-12 Richard Biener <rguenther@suse.de>
27285 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27286 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27288 2017-01-12 Richard Biener <rguenther@suse.de>
27290 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27291 and TS_TARGET_OPTION directly derive from TS_BASE.
27292 * tree-core.h (tree_optimization_option): Derive from tree_base.
27293 (tree_target_option): Likewise.
27295 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27297 * config/i386/i386.c (memory_address_length): Increase len
27298 only when rip_relative_addr_p returns false.
27300 2017-01-11 Julia Koval <julia.koval@intel.com>
27302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27303 (OPTION_MASK_ISA_SGX_SET): New.
27304 (ix86_handle_option): Handle OPT_msgx.
27305 * config.gcc: Added sgxintrin.h.
27306 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27307 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27308 * config/i386/i386.c (ix86_target_string): Add -msgx.
27310 (ix86_option_override_internal): Handle new options.
27311 (ix86_valid_target_attribute_inner_p): Add sgx.
27312 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27313 * config/i386/i386.opt: Add msgx.
27314 * config/i386/sgxintrin.h: New file.
27315 * config/i386/x86intrin.h: Add sgxintrin.h.
27317 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27320 * fold-const.c (maybe_nonzero_address): Return 1 for function
27322 (tree_single_nonzero_warnv_p): Don't handle function local objects
27326 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27329 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27332 * opts.c: Include "spellcheck.h"
27333 (struct string_fragment): New struct.
27334 (struct edit_distance_traits<const string_fragment &>): New
27336 (get_closest_sanitizer_option): New function.
27337 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27339 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27341 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27343 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27344 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27345 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27346 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27347 for initial die_offset if dwarf_split_debug_info.
27348 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27349 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27351 (output_skeleton_debug_sections): Formatting fix. Use
27352 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27353 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27355 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27357 * config/arm/cortex-a53.md: Add bypasses for
27358 cortex_a53_r2f_cvt.
27359 (cortex_a53_r2f): Only use for transfers.
27360 (cortex_a53_f2r): Likewise.
27361 (cortex_a53_r2f_cvt): Add reservation for conversions.
27362 (cortex_a53_f2r_cvt): Likewise.
27364 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27366 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27367 to all inlined functions, change static to extern.
27369 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27372 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27374 (arm_assemble_integer): Likewise.
27376 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27378 * config.gcc: Use new awk script to check CPU, FPU and architecture
27379 parameters for --with-... options.
27380 * config/arm/parsecpu.awk: New file
27381 * config/arm/arm-cpus.in: New file.
27382 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27384 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27386 * config/arm/t-arm: Update dependency rules.
27387 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27388 of processing .def files.
27389 * config/arm/genopt.sh: Deleted.
27390 * config/arm/gentune.sh: Deleted.
27391 * config/arm/arm-cores.def: Deleted.
27392 * config/arm/arm-arches.def: Deleted.
27393 * config/arm/arm-fpus.def: Deleted.
27394 * config/arm/arm-tune.md: Regenerated.
27395 * config/arm/arm-tables.opt: Regenerated.
27396 * config/arm/arm-cpu.h: New generated file.
27397 * config/arm/arm-cpu-data.h: New generated file.
27398 * config/arm/arm-cpu-cdata.h: New generated file.
27400 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27403 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27405 (input_varpool_node): Unpack dynamically_initialized bit.
27407 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27409 PR rtl-optimization/79032
27410 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27411 the alignment of the adjusted memory reference against that of MODE,
27412 instead of the alignment of the original memory reference.
27414 2017-01-11 Martin Jambor <mjambor@suse.cz>
27416 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27418 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27419 decorated functions.
27421 2017-01-11 Richard Biener <rguenther@suse.de>
27423 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27424 set range/nonnull info for PHI results. Do not set it on
27425 stmts marked for removal.
27427 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27429 * expr.c (store_field): In the bitfield case, fetch the return value
27430 from the registers before applying a single big-endian adjustment.
27431 Always do a final load for a BLKmode value not larger than a word.
27433 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27436 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27437 that we correctly handle column numbers greater than
27438 LINE_MAP_MAX_COLUMN_NUMBER.
27440 2017-01-10 Martin Sebor <msebor@redhat.com>
27442 PR middle-end/78245
27443 * gimple-ssa-sprintf.c (get_destination_size): Call
27444 {init,fini}object_sizes.
27445 * tree-object-size.c (addr_object_size): Adjust.
27446 (pass_through_call): Adjust.
27447 (pass_object_sizes::execute): Adjust.
27448 * tree-object-size.h (fini_object_sizes): Declare.
27450 2017-01-10 Martin Sebor <msebor@redhat.com>
27452 PR tree-optimization/78775
27453 * builtins.c (get_size_range): Move...
27454 * calls.c: ...to here.
27455 (alloc_max_size): Accept zero argument.
27456 (operand_signed_p): Remove.
27457 (maybe_warn_alloc_args_overflow): Call get_size_range.
27458 * calls.h (get_size_range): Declare.
27460 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27462 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27463 from TI's devices.csv file as of September 2016.
27464 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27466 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27468 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27469 * doc/invoke.texi: Likewise.
27470 * doc/md.texi: Likewise.
27471 * doc/objc.texi: Likewise.
27473 2017-01-10 Joshua Conner <joshconner@google.com>
27475 * config/arm/fuchsia-elf.h: New file.
27476 * config/fuchsia.h: New file.
27477 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27478 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27480 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27482 2016-01-10 Richard Biener <rguenther@suse.de>
27484 PR tree-optimization/79034
27485 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27486 Propagate out degenerate PHIs in the joiner.
27488 2017-01-10 Martin Liska <mliska@suse.cz>
27490 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27491 (sort_congruence_classes_by_decl_uid): Likewise.
27492 (sort_congruence_class_groups_by_decl_uid): Likewise.
27493 (sem_item_optimizer::merge_classes): Sort class, groups in these
27494 classes and members in the groups by DECL_UID of declarations.
27495 This would make merge operations stable.
27497 2017-01-10 Martin Liska <mliska@suse.cz>
27499 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27500 usage of m_classes_vec.
27501 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27502 (sem_item_optimizer::get_group_by_hash): Likewise.
27503 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27504 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27505 (sem_item_optimizer::verify_classes): Likewise.
27506 (sem_item_optimizer::process_cong_reduction): Likewise.
27507 (sem_item_optimizer::dump_cong_classes): Likewise.
27508 (sem_item_optimizer::merge_classes): Likewise.
27509 * ipa-icf.h (congruence_class_hash): Rename from
27510 congruence_class_group_hash. Remove declaration of m_classes_vec.
27512 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27514 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27515 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27516 * config.gcc: Add avx512vpopcntdqintrin.h.
27517 * config/i386/avx512vpopcntdqintrin.h: New.
27518 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27519 * config/i386/i386-builtin-types.def: Add new types.
27520 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27521 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27522 __builtin_ia32_vpopcountq_v8di_mask): New.
27523 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27524 __AVX512VPOPCNTDQ__.
27525 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27526 (PTA_AVX512VPOPCNTDQ): Define.
27527 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27528 TARGET_AVX512VPOPCNTDQ_P): Define.
27529 * config/i386/i386.opt: Add mavx512vpopcntdq.
27530 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27531 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27533 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27535 PR middle-end/77484
27536 * predict.def (PRED_CALL): Set to 67.
27538 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27540 * expr.c (store_field): In the bitfield case, if the value comes from
27541 a function call and is of an aggregate type returned in registers, do
27542 not modify the field mode; extract the value in all cases if the mode
27543 is BLKmode and the size is not larger than a word.
27545 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27548 * config/i386/cpuid.h: Fix undefined behavior.
27550 2017-01-04 Jeff Law <law@redhat.com>
27552 PR tree-optimization/79007
27553 PR tree-optimization/67955
27554 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27555 conservative for pt.null when flag_non_call_exceptions is on.
27557 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27559 PR translation/79019
27560 PR translation/79020
27561 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27562 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27563 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27565 * config/avr/avr.opt (maccumulate-args): Likewise.
27566 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27567 * common.opt (freport-bug): Likewise.
27568 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27569 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27570 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27571 translatable string.
27572 * config/i386/i386.c (function_value_32): Likewise.
27573 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27574 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27576 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27577 * common/config/msp430/msp430-common.c (msp430_handle_option):
27579 * symtab.c (symtab_node::verify_base): Likewise.
27580 * opts.c (set_debug_level): Likewise.
27581 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27582 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27583 missing whitespace to translatable strings.
27584 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27585 * config/sh/superh.h: Likewise.
27586 * config/i386/xopintrin.h: Likewise.
27587 * config/i386/znver1.md: Likewise.
27588 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27589 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27590 * double-int.h (struct double_int): Likewise.
27591 * double-int.c (div_and_round_double): Likewise.
27592 * wide-int.cc: Likewise.
27593 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27594 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27595 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27596 (crossjumps_occurred): ... this.
27597 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27600 PR tree-optimization/78899
27601 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27602 returning bool return struct loop *, NULL for failure and the new
27604 (versionable_outer_loop_p): Don't version outer loop if it has
27605 dont_vectorized bit set.
27606 (tree_if_conversion): When versioning outer loop, ensure
27607 tree_if_conversion is performed also on the inner loop of the
27608 non-vectorizable outer loop copy.
27609 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27610 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27611 prevent vectorization of it.
27612 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27613 the outer loop vectorization of the non-scalar version is attempted
27614 before vectorization of the inner loop in scalar version. If
27615 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27616 vectorization of its inner loop.
27617 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27618 has 2 inner loops, rename also on edges from bb whose single pred
27619 is outer_loop->header. Fix typo in function comment.
27621 2017-01-09 Martin Sebor <msebor@redhat.com>
27624 * asan.c (asan_emit_stack_protection): Increase local buffer size
27625 to avoid snprintf truncation warning.
27627 2017-01-09 Andrew Pinski <apinski@cavium.com>
27629 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27630 to reference thunderx2t99 for the tuning structure
27631 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27633 (thunderx2t99_extra_costs): This.
27634 * config/aarch64/aarch64-tune.md: Regenerate.
27635 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27636 (vulcan_addrcost_table): This.
27637 (vulcan_regmove_cost): Rename to ...
27638 (thunderx2t99_regmove_cost): This.
27639 (vulcan_vector_cost): Rename to ...
27640 (thunderx2t99_vector_cost): this.
27641 (vulcan_branch_cost): Rename to ...
27642 (thunderx2t99_branch_cost): This.
27643 (vulcan_tunings): Rename to ...
27644 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27645 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27647 2017-01-09 Martin Jambor <mjambor@suse.cz>
27651 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27652 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27653 (propagate_vr_accross_jump_function): Use the above function for all
27654 value range computations for pass-through jump functions and type
27655 converasion from explicit value range values.
27656 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27657 parameters from TYPE_ARG_TYPES.
27658 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27659 (ipa_write_node_info): Stream type of the actual argument.
27660 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27662 2017-01-09 Martin Liska <mliska@suse.cz>
27665 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27666 (lookup_compiler): Do not show error message with have_E.
27668 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27670 PR tree-optimization/78938
27671 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27672 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27673 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27674 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27677 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27679 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27682 2017-01-09 Richard Biener <rguenther@suse.de>
27684 PR tree-optimization/78997
27685 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27686 name condition properly.
27688 2017-01-09 Richard Biener <rguenther@suse.de>
27691 * dwarf2out.c (is_cxx): New overload with context.
27692 (is_naming_typedef_decl): Use it.
27694 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27696 * invoke.texi (Option Summary): Correct spacing in option lists
27697 and add line breaks to fix over-long lines.
27699 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27701 PR middle-end/17660
27703 * extend.texi (Common Variable Attributes): Add xref to GCC
27704 Internals manual to explain mode attribute keywords.
27706 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27709 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27710 and Preprocessor Options.
27711 (Options for Linking): Document -pthread here....
27712 (RS/6000 and PowerPC Options): ...not here.
27713 (Solaris 2 Options): ...or here.
27714 * doc/cppopts.texi: Document -pthread.
27716 2017-01-08 Martin Sebor <msebor@redhat.com>
27718 PR middle-end/77708
27719 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27720 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27721 New member functions.
27722 (format_directive): Used them.
27724 (pass_sprintf_length::handle_gimple_call): Same.
27725 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27726 to avoid truncation for any argument.
27727 (extract_affine_mul): Same.
27728 * tree.c (get_file_function_name): Same.
27730 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27732 PR middle-end/77484
27733 * predict.def (PRED_INDIR_CALL): Set to 86.
27735 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27737 PR preprocessor/54124
27738 * doc/cppopts.texi: Reformat -d subtable to list the full name
27739 of the options. Add cross-reference to the docs for the general
27740 compiler -d options.
27741 * doc/invoke.texi (Developer Options): Add cross-reference to the
27742 preprocessor-specific -d option documentation.
27744 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27746 PR preprocessor/13498
27747 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27748 redudant material, and reflect new command-line options.
27749 (System Headers): Likewise.
27751 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27753 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27754 -isystem, and -idirafter. Copy-edit.
27755 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27756 default for -ftrack-macro-expansion. Delete obsolete and
27757 badly-formatted implementation details about -fdebug-cpp output.
27758 * doc/cppwarnopts.texi: Copy-edit.
27760 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27763 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27764 that the transition from a max line width >= 1<<10 to narrower
27765 lines works correctly.
27767 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
27769 * doc/options.texi (PerFunction): New.
27770 * opt-functions.awk (switch_flags): Map both Optimization and
27771 PerFunction to CL_OPTIMIZATION.
27772 * opth-gen.awk: Test for PerFunction flag along with
27774 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
27775 it only when the latter is present. Skip those that don't in
27776 the hash function generator.
27777 * common.opt (fvar-tracking): Mark as PerFunction instead of
27779 (fvar-tracking-assignments): Likewise.
27780 (fvar-tracking-assignments-toggle): Likewise.
27781 (fvar-tracking-uninit): Likewise.
27783 2017-01-07 Jakub Jelinek <jakub@redhat.com>
27785 PR translation/79018
27786 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
27789 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27792 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27793 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27794 TARGET_LONG_JUMP_TABLE_OFFSETS.
27795 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27796 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27797 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27798 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27799 * config/m68k/m68k.md (tablejump expander): Likewise.
27800 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27801 TARGET_LONG_JUMP_TABLE_OFFSETS.
27802 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27803 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27805 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27806 David Holsgrove <david.holsgrove@xilinx.com>
27808 * common/config/microblaze/microblaze-common.c
27809 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27810 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27812 * config/microblaze/microblaze.c (microblaze_must_save_register)
27813 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27815 (microblaze_eh_return): New function.
27816 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27817 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27818 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27819 * config/microblaze/microblaze.md (eh_return): New pattern.
27821 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27823 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27824 GCC_DIAGNOSTIC_STRINGIFY): Define.
27826 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27828 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27830 * config/arm/arm.md (<mcrr>): New.
27832 * config/arm/arm.c (arm_arch5te): New.
27833 (arm_option_override): Set arm_arch5te.
27834 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27836 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27837 (arm_mcrr_qualifiers): ... this. New.
27838 (MRRC_QUALIFIERS): Define to...
27839 (arm_mrrc_qualifiers): ... this. New.
27840 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27842 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27843 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27844 (MRRCI, mrrc, MRRC): New.
27845 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27846 VUNSPEC_MRRC2): New.
27848 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27850 * config/arm/arm.md (<mcr>): New.
27852 * config/arm/arm.c (arm_coproc_builtin_available): Add
27853 support for mcr, mrc, mcr2 and mrc2.
27854 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27855 (arm_mcr_qualifiers): ... this. New.
27856 (MRC_QUALIFIERS): Define to ...
27857 (arm_mrc_qualifiers): ... this. New.
27858 (MCR_QUALIFIERS): Define to ...
27859 (arm_mcr_qualifiers): ... this. New.
27860 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27862 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27863 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27864 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27865 VUNSPEC_MRC2): New.
27867 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27869 * config/arm/arm.md (*ldc): New.
27873 * config/arm/arm.c (arm_coproc_builtin_available): Add
27874 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27875 (arm_coproc_ldc_stc_legitimate_address): New.
27876 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27877 'qualifier_const_pointer'.
27878 (LDC_QUALIFIERS): Define to...
27879 (arm_ldc_qualifiers): ... this. New.
27880 (STC_QUALIFIERS): Define to...
27881 (arm_stc_qualifiers): ... this. New.
27882 * config/arm/arm-protos.h
27883 (arm_coproc_ldc_stc_legitimate_address): New.
27884 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27885 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27886 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27887 stc2, stcl, stc2l): New.
27888 * config/arm/constraints.md (Uz): New.
27889 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27890 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27891 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27892 VUNSPEC_STC2L): New.
27894 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27896 * config/arm/arm.md (<cdp>): New.
27897 * config/arm/arm.c (neon_const_bounds): Rename this ...
27898 (arm_const_bounds): ... this.
27899 (arm_coproc_builtin_available): New.
27900 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27901 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27902 (CDP_QUALIFIERS): Define to...
27903 (arm_cdp_qualifiers): ... this. New.
27905 (arm_expand_builtin_args): Add case for 6 arguments.
27906 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27907 (arm_const_bounds): ... this.
27908 (arm_coproc_builtin_available): New.
27909 * config/arm/arm_acle.h (__arm_cdp): New.
27911 * config/arm/arm_acle_builtins.def (cdp): New.
27913 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27914 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27915 'arm_const_bounds'.
27916 * config/arm/types.md (coproc): New.
27917 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27918 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27919 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27920 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27922 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27924 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27925 (UBINOP_QUALIFIERS): New.
27927 (acle_builtin_data): New. Change comment.
27928 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27929 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27930 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27931 arm_acle_builtins.def.
27932 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27933 (arm_init_acle_builtins): New.
27934 (CRC32_BUILTIN): Remove.
27935 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27936 crc32cb, crc32ch and crc32cw.
27937 (arm_init_crc32_builtins): Remove.
27938 (arm_init_builtins): Use arm_init_acle_builtins rather
27939 than arm_init_crc32_builtins.
27940 (arm_expand_acle_builtin): New.
27941 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27942 * config/arm/arm_acle_builtins.def: New.
27944 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27946 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27947 (arm_builtin_datum): ... this.
27948 (arm_init_neon_builtin): Rename to ...
27949 (arm_init_builtin): ... this. Add a new parameters PREFIX
27950 and USE_SIG_IN_NAME.
27951 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27952 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27953 'arm_builtin_datum'.
27954 (arm_init_vfp_builtins): Likewise.
27955 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27956 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27957 (arm_expand_neon_args): Rename to ...
27958 (arm_expand_builtin_args): ... this. Rename builtin_arg
27959 enum values and differentiate between ARG_BUILTIN_MEMORY
27960 and ARG_BUILTIN_NEON_MEMORY.
27961 (arm_expand_neon_builtin_1): Rename to ...
27962 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27963 values, arm_expand_builtin_args and add bool parameter NEON.
27964 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27965 (arm_expand_vfp_builtin): Likewise.
27966 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27968 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27970 PR middle-end/77484
27971 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27972 * predict.c (tree_estimate_probability_bb): Reverse direction of
27973 polymorphic call predictor.
27975 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27977 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27978 (determine_pass_name_match): ...this new function and...
27979 (should_skip_pass_p): ...this new function.
27981 2017-01-06 Nathan Sidwell <nathan@acm.org>
27983 * ipa-visibility.c (function_and_variable_visibility): Reformat
27984 comments and long lines. Remove extrneous if.
27985 * symtab.c (symtab_node::make_decl_local): Fix code format.
27986 (symtab_node::set_section_for_node): Fix comment typo.
27988 2017-01-06 Martin Liska <mliska@suse.cz>
27991 * lra-constraints.c: Rename invariant to lra_invariant.
27992 * predict.c (set_even_probabilities): Initialize e to NULL.
27994 2017-01-05 Martin Sebor <msebor@redhat.com>
27996 PR tree-optimization/78910
27997 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27998 (format_integer): Correct off-by-one error in the handling
27999 of precision with negative numbers in signed conversions..
28001 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28003 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28005 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28007 PR tree-optimization/71016
28008 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28009 factor_out_conditional_conversion. Formatting fix.
28010 (factor_out_conditional_conversion): Add cond_stmt argument.
28011 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28012 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28015 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28017 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28018 read-rtl-function.o, and selftest-rtl.o.
28019 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28020 (selftest::aarch64_test_loading_full_dump): New function.
28021 (selftest::aarch64_run_selftests): New function.
28022 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28023 selftest::aarch64_run_selftests.
28024 * config/i386/i386.c
28025 (selftest::ix86_test_loading_dump_fragment_1): New function.
28026 (selftest::ix86_test_loading_call_insn): New function.
28027 (selftest::ix86_test_loading_full_dump): New function.
28028 (selftest::ix86_test_loading_unspec): New function.
28029 (selftest::ix86_run_selftests): Call the new functions.
28030 * emit-rtl.c (maybe_set_max_label_num): New function.
28031 * emit-rtl.h (maybe_set_max_label_num): New decl.
28032 * function.c (instantiate_decls): Guard call to
28033 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28034 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28036 * gensupport.c (gen_reader::gen_reader): Pass "false"
28037 for new "compact" param of rtx_reader.
28038 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28039 rather than an empty string for NULL strings.
28040 * read-md.c: Potentially include config.h rather than bconfig.h.
28041 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28042 (have_error): New global, copied from errors.c.
28043 (md_reader::read_name): Rename to...
28044 (md_reader::read_name_1): ...this, adding "out_loc" param,
28045 and converting "missing name or number" to returning false, rather
28047 (md_reader::read_name): Reimplement in terms of read_name_1.
28048 (md_reader::read_name_or_nil): New function.
28049 (md_reader::read_string): Handle "(nil)" by returning NULL.
28050 (md_reader::md_reader): Add new param "compact".
28051 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28052 (md_reader::read_file): New method.
28053 * read-md.h (md_reader::md_reader): Add new param "compact".
28054 (md_reader::read_file): New method.
28055 (md_reader::is_compact): New accessor.
28056 (md_reader::read_name): Convert return type from void to file_location.
28057 (md_reader::read_name_or_nil): New decl.
28058 (md_reader::read_name_1): New decl.
28059 (md_reader::m_compact): New field.
28060 (noop_reader::noop_reader): Pass "false" for new "compact" param
28062 (rtx_reader::rtx_reader): Add new "compact" param.
28063 (rtx_reader::read_rtx_operand): Make virtual and convert return
28064 type from void to rtx.
28065 (rtx_reader::read_until): New decl.
28066 (rtx_reader::handle_any_trailing_information): New virtual function.
28067 (rtx_reader::postprocess): New virtual function.
28068 (rtx_reader::finalize_string): New virtual function.
28069 (rtx_reader::m_in_call_function_usage): New field.
28070 (rtx_reader::m_reuse_rtx_by_id): New field.
28071 * read-rtl-function.c: New file.
28072 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28073 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28074 (selftest::verify_three_block_rtl_cfg): New decl.
28075 * read-rtl-function.h: New file.
28076 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28077 For host, include function.h, memmodel.h, and emit-rtl.h.
28078 (one_time_initialization): New function.
28079 (struct compact_insn_name): New struct.
28080 (compact_insn_names): New array.
28081 (find_code): Handle insn codes in compact dumps.
28082 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28083 (bind_subst_iter_and_attr): Likewise.
28084 (add_condition_to_string): Likewise.
28085 (add_condition_to_rtx): Likewise.
28086 (apply_attribute_uses): Likewise.
28087 (add_current_iterators): Likewise.
28088 (apply_iterators): Likewise.
28089 (initialize_iterators): Guard usage of apply_subst_iterator with
28090 #ifdef GENERATOR_FILE.
28091 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28092 (md_reader::read_mapping): Likewise.
28093 (add_define_attr_for_define_subst): Likewise.
28094 (add_define_subst_attr): Likewise.
28095 (read_subst_mapping): Likewise.
28096 (check_code_iterator): Likewise.
28097 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28099 (one_time_initialization): New function.
28100 (rtx_reader::read_until): New method.
28101 (read_flags): New function.
28102 (parse_reg_note_name): New function.
28103 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28104 Handle reuse_rtx ids.
28105 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28106 Add parsing support for RTL dumps, mirroring the special-cases in
28107 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28108 values, and calling handle_any_trailing_information.
28109 (rtx_reader::read_rtx_operand): Convert return type from void
28110 to rtx, returning return_rtx. Handle case 'e'. Call
28111 finalize_string on XSTR and XTMPL fields.
28112 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28113 "(nil)" values were omitted. Call the postprocess vfunc on the
28115 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28116 class ctor. Initialize m_in_call_function_usage. Call
28117 one_time_initialization.
28118 * rtl-tests.c (selftest::test_uncond_jump): Call
28119 set_new_first_and_last_insn.
28120 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28121 * selftest-rtl.c: New file.
28122 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28123 (selftest::get_insn_by_uid): New decl.
28124 * selftest-run-tests.c (selftest::run_tests): Call
28125 read_rtl_function_c_tests.
28126 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28127 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28130 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28132 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28133 operands in a special way. Assert that pos+len <= mode precision.
28135 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28137 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28138 3 argument Alias with unlimited for the negative form.
28139 (fno-vect-cost-model): Removed.
28141 2017-01-05 Martin Liska <mliska@suse.cz>
28143 * hsa-gen.c (gen_hsa_divmod): New function.
28144 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28146 2017-01-05 Martin Liska <mliska@suse.cz>
28149 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28152 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28154 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28155 small constant length operands.
28157 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28159 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28160 between loop iterations.
28162 2017-01-05 Martin Liska <mliska@suse.cz>
28165 * gimplify.c (gimplify_decl_expr): Compare to
28166 asan_poisoned_variables instread of checking flags.
28167 (gimplify_target_expr): Likewise.
28168 (gimplify_expr): Likewise.
28169 (gimplify_function_tree): Conditionally initialize
28170 asan_poisoned_variables.
28172 2017-01-04 Jeff Law <law@redhat.com>
28174 PR tree-optimizatin/78812
28175 * rtl.h (contains_mem_rtx_p): Prototype.
28176 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28177 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28178 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28179 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28180 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28182 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28184 * input.c (assert_char_at_range): Default-initialize actual_range.
28186 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28188 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28189 to match the caller.
28191 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28193 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28194 insns after final jump in test to emit dummy move.
28196 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28198 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28199 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28201 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28203 * multiple_target.c (create_dispatcher_calls): Init e_next.
28204 * tree-ssa-loop-split.c (split_loop): Init border.
28205 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28208 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28213 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28214 (altivec_register_operand): Do not return true if the operand
28215 contains a SUBREG mixing SImode and SFmode.
28216 (vsx_register_operand): Likewise.
28217 (vsx_reg_sfsubreg_ok): New predicate.
28218 (vfloat_operand): Do not return true if the operand contains a
28219 SUBREG mixing SImode and SFmode.
28220 (vint_operand): Likewise.
28221 (vlogical_operand): Likewise.
28222 (gpc_reg_operand): Likewise.
28223 (int_reg_operand): Likewise.
28224 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28225 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28226 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28228 (rs6000_emit_move_si_sf_subreg): New helper function.
28229 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28230 fixup SUBREGs involving SImode and SFmode.
28231 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28232 numbers for the new peephole2 optimization.
28233 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28234 the GLIBC math library that do AND/IOR/XOR operations on single
28235 precision floating point.
28236 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28237 target macros to say whether we need to avoid SUBREGs mixing
28239 (TARGET_ALLOW_SF_SUBREG): Likewise.
28240 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28241 (UNSPEC_SI_FROM_SF): Likewise.
28242 (iorxor): Change spacing.
28243 (and_ior_xor): New iterator for AND, IOR, and XOR.
28244 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28245 (movdi_from_sf_zero_ext): Likewise.
28246 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28247 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28248 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28249 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28250 (fms<mode>4): Likewise.
28251 (fnma<mode>4): Likewise.
28252 (fnms<mode>4): Likewise.
28253 (nfma<mode>4): Likewise.
28254 (nfms<mode>4): Likewise.
28256 2017-01-04 Marek Polacek <polacek@redhat.com>
28259 * doc/invoke.texi: Document -Wpointer-compare.
28261 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28263 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28266 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28267 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28270 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28273 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28274 documentation of the powerpc_popcntb_ok attribute.
28275 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28276 code to issue warning messages if a requested CPU configuration is
28277 not supported by the binary (assembler and loader) toolchain.
28278 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28279 made to define a built-in function that has been disabled.
28280 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28281 made to define a built-in function that has been disabled.
28282 (altivec_init_builtins): Add comment explaining why definition
28283 of the DST built-in functions is not preceded by an assertion
28284 check. Add assertions to prevent ICE if attempts are made to
28285 define an altivec predicate or an abs* built-in function that has
28287 (htm_init_builtins): Add comment explaining why definition of the
28288 htm built-in functions is not preceded by an assertion check.
28290 2017-01-04 Jeff Law <law@redhat.com>
28292 PR tree-optimizatin/67955
28293 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28294 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28295 the points-to solution does not include pt_null. Use DECL_PT_UID
28298 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28300 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28301 Use gen_int_mode instead of gen_lopwart for const_int operands.
28303 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28305 PR tree-optimization/71563
28306 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28307 out of range value - has low bits known to be zero.
28309 2017-01-04 Alan Modra <amodra@gmail.com>
28311 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28312 * configure: Regenerate.
28313 * config.in: Regenerate.
28315 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28318 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28319 a substring of the message, but strcmp with the whole message. Ifdef
28320 ENABLE_NLS, translate the message first using dgettext.
28322 2017-01-03 Jeff Law <law@redhat.com>
28324 PR tree-optimizatin/78856
28325 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28326 (mark_threaded_blocks): Remove code to truncate thread paths that
28327 cross multiple loop headers. Instead invalidate the cached loop
28328 iteration information and handle case of a thread path walking
28329 into an irreducible region.
28331 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28334 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28335 assertions. Add support for doing the signbit if the IEEE 128-bit
28336 floating point value is in a GPR.
28337 * config/rs6000/rs6000.md (Fsignbit): Delete.
28338 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28339 Update the length attribute if the value is in a GPR.
28340 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28341 the sign or zero extension instruction, since the value is always 0/1.
28342 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28345 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28346 extracting SImode to a GPR register so that we can generate a
28347 store, limit the vector to be in a traditional Altivec register
28348 for the vextuwrx instruction.
28350 2017-01-03 Ian Lance Taylor <iant@google.com>
28352 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28354 2017-01-03 Martin Sebor <msebor@redhat.com>
28356 PR tree-optimization/78696
28357 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28358 precision. Use MPFR for %f for greater fidelity. Correct handling
28360 (pass_sprintf_length::compute_format_length): Set width and precision
28361 specified by asrerisk to void_node for vararg functions.
28362 (try_substitute_return_value): Adjust dump output.
28364 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28366 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28368 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28370 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28371 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28372 -mlra/-mno-lra was passed to the compiler.
28374 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28376 PR rtl-optimization/65618
28377 * emit-rtl.c (try_split): Move initialization of "before" and
28378 "after" to just before the call to emit_insn_after_setloc.
28380 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28382 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28384 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28386 * dwarf2out.c (gen_enumeration_type_die): When
28387 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28389 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28391 PR tree-optimization/78965
28392 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28393 Change first argument from const call_info & to call_info &. For %n
28394 set info.nowrite to false.
28396 PR middle-end/78901
28397 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28398 possibly throwing calls.
28400 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28401 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28402 and fns handling, rather than in a separate case SSA_NAME.
28404 2017-01-02 Jeff Law <law@redhat.com>
28406 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28407 fixes for first_period and second_period variables.
28409 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28412 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28413 (*insvqi_1): New insn pattern.
28414 (*insvqi_1_mem_rex64): Ditto.
28415 (*insvqi_2): Ditto.
28416 (*insvqi_3): Rename from *insvqi.
28418 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28420 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28422 * doc/cfg.texi (Edges): Remove reference to Java.
28423 (Maintaining the CFG): Ditto.
28425 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28427 PR middle-end/77674
28428 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28429 transparent aliases.
28431 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28433 PR middle-end/77484
28434 * predict.def (PRED_CALL): Update hitrate.
28435 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28436 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28437 into direct/indirect/polymorphic variants.
28439 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28441 Update copyright years.
28443 * gcc.c (process_command): Update copyright notice dates.
28444 * gcov-dump.c (print_version): Ditto.
28445 * gcov.c (print_version): Ditto.
28446 * gcov-tool.c (print_version): Ditto.
28447 * gengtype.c (create_file): Ditto.
28448 * doc/cpp.texi: Bump @copying's copyright year.
28449 * doc/cppinternals.texi: Ditto.
28450 * doc/gcc.texi: Ditto.
28451 * doc/gccint.texi: Ditto.
28452 * doc/gcov.texi: Ditto.
28453 * doc/install.texi: Ditto.
28454 * doc/invoke.texi: Ditto.
28456 Copyright (C) 2017 Free Software Foundation, Inc.
28458 Copying and distribution of this file, with or without modification,
28459 are permitted in any medium without royalty provided the copyright
28460 notice and this notice are preserved.