1 2017-09-29 Richard Biener <rguenther@suse.de>
3 * graphite-isl-ast-to-gimple.c
4 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5 (binary_op_to_tree): Use it.
6 (get_rename_from_scev): Likewise.
7 (copy_loop_phi_nodes): Likewise.
8 (copy_bb_and_scalar_dependences): Likewise.
9 (translate_pending_phi_nodes): Likewise.
11 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
15 for movabsq $(i32 << shift), r64.
17 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
19 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
20 index when encoding %esp as %rsp to avoid 0x67 prefix.
22 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
24 * config/i386/i386.md (*movsf_internal, *movdf_internal):
25 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
27 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
29 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
30 Extensions with more than 16 double VFP registers.
31 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
32 to_clear_mask and all code related to it. Replace the remaining
33 entry by a sbitmap and adapt code accordingly.
35 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
37 * brig-builtins.def: Change pure attributes to const.
39 2017-09-28 Joseph Myers <joseph@codesourcery.com>
41 * config.gcc (default_gnu_indirect_function): Default to yes for
42 sparc*-*-linux* with glibc.
44 2017-09-28 Joseph Myers <joseph@codesourcery.com>
46 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
47 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
48 when creating .init_array and .fini_array sections with priority
51 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
54 * config/aarch64/aarch64.c
55 (aarch64_builtin_support_vector_misalignment): Always return false
56 when misalignment is unknown.
58 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
60 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
61 this function to return false if the definition used by the swap
62 instruction is artificial, or if the memory address from which the
63 constant value is loaded is not represented by a base address held
64 in a register or if the base address register is a frame or stack
65 pointer. Additionally, return false if the base address of the
66 loaded constant is a SYMBOL_REF but is not considered to be a
68 (replace_swapped_load_constant): New function.
69 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
70 loaded constant vector with a load of a swapped constant vector.
72 2017-09-27 Carl Love <cel@us.ibm.com>
74 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
75 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
76 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
79 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
81 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
82 first, always call autopref_rank_data otherwise.
84 2017-09-27 Richard Biener <rguenther@suse.de>
86 * graphite-scop-detection.c (find_scop_parameters): Move
87 loop bound handling ...
88 (gather_bbs::before_dom_children): ... here, avoiding the need
89 to build scop_info->loop_nest.
90 (record_loop_in_sese): Remove.
91 * sese.h (sese_info_t::loop_nest): Remove.
92 * sese.c (new_sese_info): Do not allocate loop_nest.
93 (free_sese_info): Do not free loop_nest.
95 2017-09-27 Jakub Jelinek <jakub@redhat.com>
98 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
99 lhs from calls if the lhs has addressable type.
101 2017-09-27 Richard Biener <rguenther@suse.de>
103 * graphite.h (scop::max_alias_set): New member.
104 * graphite-scop-detection.c: Remove references to non-existing
106 (build_alias_sets): Record the maximum alias set used for drs.
107 (build_scops): Support zero as unlimited for
108 --param graphite-max-arrays-per-scop.
109 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
111 (build_poly_sr_1): ... here. Compute alias set based on the
112 maximum alias set used for drs rather than
113 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
115 2017-09-27 Richard Biener <rguenther@suse.de>
117 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
118 --param loop-block-tile-size=0 to disable tiling.
120 2017-09-27 Richard Biener <rguenther@suse.de>
122 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
123 (graphite-max-nb-scop-params): Document special value zero.
124 * domwalk.h (dom_walker::STOP): New symbolical constant.
125 (dom_walker::dom_walker): Add optional parameter for bb to
127 (dom_walker::~dom_walker): Declare.
128 (dom_walker::before_dom_children): Document STOP return value.
129 (dom_walker::m_user_bb_to_rpo): New member.
130 (dom_walker::m_bb_to_rpo): Likewise.
131 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
132 mapping here if not provided by the user.
133 (dom_walker::~dom_walker): Free bb to RPO mapping if not
134 provided by the user.
135 (dom_walker::STOP): Define.
136 (dom_walker::walk): Do not compute bb to RPO mapping here.
137 Support STOP return value from before_dom_children to stop
139 * graphite-optimize-isl.c (optimize_isl): If the schedule
140 is the same still generate code if -fgraphite-identity
141 or -floop-parallelize-all are given.
142 * graphite-scop-detection.c: Include cfganal.h.
143 (gather_bbs::gather_bbs): Get and pass through bb to RPO
145 (gather_bbs::before_dom_children): Return STOP for BBs
147 (build_scops): Compute bb to RPO mapping and pass it to
148 the domwalk. Treat --param graphite-max-nb-scop-params=0
149 as not limiting the number of params.
150 * graphite.c (graphite_initialize): Remove limit on the
151 number of basic-blocks in a function.
152 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
153 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
156 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
158 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
159 Adjust code to eliminate needing to do the shift right 32-bits
160 operation after XSCVDPSPN.
162 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
164 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
165 ((X / Y) != 0 -> X >= Y): Likewise.
167 2017-09-26 Carl Love <cel@us.ibm.com>
169 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
170 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
171 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
172 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
173 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
174 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
175 definitions and overloading.
176 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
177 statement for P9V_BUILTIN_XST_LEN_R.
178 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
179 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
180 define_expand and define_insn for the instructions and builtins.
181 * doc/extend.texi: Update the built-in documentation file for the new
183 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
184 define_insn for the instructions
186 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
189 * gcc/config/netbsd-protos.h: New file.
190 * gcc/config/netbsd.c: New file.
191 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
192 * gcc/config/t-netbsd: New file.
193 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
194 (tmake_file) Add t-netbsd.
195 (extra_objs) Add netbsd.o.
197 2017-09-26 Janus Weil <janus@gcc.gnu.org>
201 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
203 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
205 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
206 sign extension from a vector register to a GPR by doing a 32-bit
207 direct move and then an EXTSW.
208 (extendsi<mode>2 splitter): Likewise.
209 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
210 right or vector extract after doing XSCVDPSPN. Use
211 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
213 (movdi_from_sf_zero_ext): Likewise.
214 (reload_gpr_from_vsxsf): Likewise.
215 (p8_mfvsrd_4_disf): Delete, no longer used.
216 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
217 then needing to move the SFmode to a GPR to use the XSCVDPSP
218 instruction instead of FRSP and XSCVDPSPN.
219 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
220 it is adjacent to the other XSCVSPDP insns.
221 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
222 SFmode to be in traditional Altivec registers.
223 (vsx_xscvdpspn): Eliminate useless alternative constraint.
224 (vsx_xscvspdpn): Likewise.
225 (vsx_xscvspdpn_scalar): Likewise.
227 2017-09-26 Martin Jambor <mjambor@suse.cz>
229 * tree-sra.c (compare_access_positions): Put integral types first,
230 stabilize sorting of integral types, remove conditions putting
231 non-full-precision integers last.
232 (sort_and_splice_var_accesses): Disable scalarization if a
233 non-integert would be represented by a non-full-precision integer.
235 2017-09-26 Joseph Myers <joseph@codesourcery.com>
237 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
238 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
239 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
240 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
241 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
242 conditionals inside the function instead of around it. Call
243 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
244 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
246 2017-09-26 Richard Biener <rguenther@suse.de>
248 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
250 (scop_detection::build_scop_breadth): ... this. Removed.
251 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
252 (scop_detection::harmful_stmt_in_bb): Likewise.
253 (scop_detection::graphite_can_represent_stmt): Likewise.
254 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
255 (scop_detection::can_represent_loop): Remove recursion, fold in ...
256 (scop_detection::can_represent_loop_1): ... this. Removed.
257 (scop_detection::harmful_loop_in_region): Simplify after inlining
258 the above and remove more quadraticness.
259 (build_scops): Adjust.
260 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
263 2017-09-26 Jakub Jelinek <jakub@redhat.com>
266 * config/i386/i386.c (ix86_print_operand_address_as): Only test
267 REGNO (base) == SP_REG if base is a REG.
270 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
271 if it is different SSA_NAME.
272 (optimize_range_tests_cmp_bitwise): New function.
273 (optimize_range_tests): Call it.
275 2017-09-26 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/82321
278 * graphite.c (canonicalize_loop_closed_ssa): Properly check
279 for the def being inside the loop.
281 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
283 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
285 * config/s390/s390-builtins.def: Fix constraint on op4.
287 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
289 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
290 independent expanders.
291 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
292 ("vec_ordered", "vec_unordered"): New expanders.
294 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
296 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
299 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
301 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
302 vec_unpacks_lo_v16qi.
303 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
305 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
307 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
308 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
309 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
311 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
313 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
315 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
317 ("*vec_slb<mode>"): New insn pattern.
318 ("vec_shr_<mode>"): New expander.
319 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
320 and force the shift count operand to V16QImode.
321 ("vec_srb<mode>"): Set shift count mode to V16QI.
323 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
325 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
326 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
327 ("vec_widen_smult_hi_<mode>"): New expander definitions.
329 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
332 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
334 2017-09-26 Richard Biener <rguenther@suse.de>
336 PR tree-optimization/82320
337 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
340 2017-09-25 Jeff Law <law@redhat.com>
342 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
343 prototype for new argument.
344 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
345 mostly extracted from rs6000_emit_allocate_stack.
346 (rs6000_emit_probe_stack_range_stack_clash): New function.
347 (rs6000_emit_allocate_stack): Call
348 rs6000_emit_probe_stack_range_stack_clash as needed.
349 (rs6000_emit_probe_stack_range): Add additional argument
350 to call to gen_probe_stack_range{si,di}.
351 (output_probe_stack_range): New.
352 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
353 (output_probe_stack_range_stack_clash): New.
354 (rs6000_emit_prologue): Emit notes into dump file as requested.
355 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
356 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
357 Add additional operand and pass it to output_probe_stack_range.
359 2017-09-25 Bin Cheng <bin.cheng@arm.com>
361 PR tree-optimization/82163
362 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
363 (checking_verify_loop_closed_ssa): New parameter.
364 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
365 (check_loop_closed_ssa_stmt): Delete.
366 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
367 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
368 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
371 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
373 * brig-builtins.def: Treat HSAIL barrier builtins as
374 setjmp/longjump style functions.
376 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
378 * target.def (constant_alignment): New hook.
379 * defaults.h (CONSTANT_ALIGNMENT): Delete.
380 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
381 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
382 * doc/tm.texi: Regenerate.
383 * targhooks.h (default_constant_alignment): Declare.
384 (constant_alignment_word_strings): Likewise.
385 * targhooks.c (default_constant_alignment): New function.
386 (constant_alignment_word_strings): Likewise.
387 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
388 instead of CONSTANT_ALIGNMENT.
389 * varasm.c (align_variable, get_variable_align, build_constant_desc)
390 (force_const_mem): Likewise.
391 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
392 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
393 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
394 (TARGET_CONSTANT_ALIGNMENT): Redefine.
395 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
397 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
398 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
399 constant_alignment_word_strings.
400 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
401 (CONSTANT_ALIGNMENT): Likewise.
402 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
403 (arm_constant_alignment): New function.
404 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
405 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
406 constant_alignment_word_strings.
407 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
408 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
409 constant_alignment_word_strings.
410 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
411 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
412 (cris_constant_alignment): New function.
413 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
414 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
415 (epiphany_constant_alignment): New function.
416 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
417 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
418 constant_alignment_word_strings.
419 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
420 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
421 constant_alignment_word_strings.
422 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
423 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
424 constant_alignment_word_strings.
425 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
426 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
427 * config/i386/i386.c (ix86_constant_alignment): Make static.
428 Use the same interface as the target hook.
429 (TARGET_CONSTANT_ALIGNMENT): Redefine.
430 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
431 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
432 constant_alignment_word_strings.
433 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
434 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
435 (TARGET_CONSTANT_ALIGNMENT): Redefine.
436 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
437 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
438 constant_alignment_word_strings.
439 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
440 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
441 constant_alignment_word_strings.
442 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
443 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
444 constant_alignment_word_strings.
445 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
446 * config/microblaze/microblaze.c (microblaze_constant_alignment):
448 (TARGET_CONSTANT_ALIGNMENT): Redefine.
449 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
450 * config/mips/mips.c (mips_constant_alignment): New function.
451 (TARGET_CONSTANT_ALIGNMENT): Redefine.
452 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
453 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
454 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
455 (mmix_constant_alignment): Make static. Use the same interface
457 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
458 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
459 constant_alignment_word_strings.
460 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
461 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
462 constant_alignment_word_strings.
463 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
464 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
465 constant_alignment_word_strings.
466 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
467 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
468 (rs6000_constant_alignment): New function.
469 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
470 * config/riscv/riscv.c (riscv_constant_alignment): New function.
471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
472 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
473 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
474 (rs6000_constant_alignment): New function.
475 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
476 * config/s390/s390.c (s390_constant_alignment): New function.
477 (TARGET_CONSTANT_ALIGNMENT): Redefine.
478 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
479 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
480 constant_alignment_word_strings.
481 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
482 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
483 (sparc_constant_alignment): New function.
484 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
485 * config/spu/spu.c (spu_constant_alignment): New function.
486 (TARGET_CONSTANT_ALIGNMENT): Redefine.
487 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
488 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
489 constant_alignment_word_strings.
490 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
491 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
492 constant_alignment_word_strings.
493 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
494 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
495 constant_alignment_word_strings.
496 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
497 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
498 (visium_constant_alignment): New function.
499 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
500 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
501 (xtensa_constant_alignment): New function.
502 * system.h (CONSTANT_ALIGNMENT): Poison.
504 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
506 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
507 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
508 (rs6000_builtin_valid_without_lhs): New helper function.
509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
510 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
512 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
514 * target.h (vec_perm_indices): Use unsigned short rather than
516 (auto_vec_perm_indices): Likewise.
517 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
518 Use unsigned int rather than unsigned char.
519 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
521 2017-09-25 Richard Biener <rguenther@suse.de>
523 * cfgloop.h (sort_sibling_loops): Declare.
524 * cfgloop.c (sort_sibling_loops_cmp): New helper.
525 (sort_sibling_loops): New function sorting the sibling loop list
527 * graphite.c (graphite_transform_loops): Sort sibling loops.
529 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
531 * target.def (vec_perm_const_ok): Change sel parameter to
533 * optabs-query.c (can_vec_perm_p): Update accordingly.
534 * doc/tm.texi: Regenerate.
535 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
536 auto_vec_perm_indices and remove separate nelt field.
537 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
538 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
539 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
540 (aarch64_expand_vec_perm_const): Update accordingly.
541 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
543 * config/arm/arm.c (expand_vec_perm_d): Change perm to
544 auto_vec_perm_indices and remove separate nelt field.
545 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
546 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
547 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
549 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
551 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
552 sel to vec_perm_indices.
553 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
554 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
555 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
557 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
560 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
563 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
564 on the FUNCTION_DECL function context if it has a DIE that is a
567 2017-09-25 Richard Biener <rguenther@suse.de>
569 PR tree-optimization/82285
570 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
573 2017-09-25 Tom de Vries <tom@codesourcery.com>
577 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
580 2017-09-25 Richard Biener <rguenther@suse.de>
582 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
583 ISL errors other than isl_error_quota happen. Dump if the
584 schedule is the same.
585 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
586 errors instead of aborting inside ISL.
588 2017-09-25 Iain Sandoe <iain@codesourcery.com>
591 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
592 of libgcc_eh for m64.
593 * config/i386/darwin64.h: Likewise.
595 2017-09-25 Richard Biener <rguenther@suse.de>
598 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
599 attribute for incomplete types nor twice for complete ones.
601 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
604 * config/i386/i386.c (ix86_print_operand_address_as): Encode
605 %esp as %rsp to avoid 0x67 prefix if there is no index or base
608 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
611 * config/i386/i386.opt (mprefer-avx256): Use
612 ix86_target_flags variable.
613 * config/i386/i386.c (ix86_target_string): Move
614 -mprefer-avx256 to flag2_opts.
616 2017-09-22 Jakub Jelinek <jakub@redhat.com>
619 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
620 and x != -1 | y != -1 into (x & y) != -1.
622 2017-09-22 Steve Ellcey <sellcey@cavium.com>
624 * config.gcc: Add new case statement to set
625 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
626 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
627 s390x-*-linux* case statements. Added aarch64 to the list of
628 supported architectures.
630 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
632 PR tree-optimization/82289
633 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
634 STMT_VINFO_RELEVANT_P.
636 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
637 Alan Hayward <alan.hayward@arm.com>
638 David Sherwood <david.sherwood@arm.com>
640 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
641 for VR_RANGE only; don't allow VR_ANTI_RANGE.
642 (extract_range_from_binary_expr_1): Don't call
643 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
645 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
646 Alan Hayward <alan.hayward@arm.com>
647 David Sherwood <david.sherwood@arm.com>
649 * target.def (preferred_vector_alignment): New hook.
650 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
652 * doc/tm.texi: Regenerate.
653 * targhooks.h (default_preferred_vector_alignment): Declare.
654 * targhooks.c (default_preferred_vector_alignment): New function.
655 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
657 (DR_TARGET_ALIGNMENT): New macro.
658 (aligned_access_p): Update commentary.
659 (vect_known_alignment_in_bytes): New function.
660 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
662 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
663 Calculate the misalignment based on the target alignment rather than
665 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
666 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
667 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
668 the target alignment, rather than masking the element misalignment
669 with the number of elements in a vector. Also use the target
670 alignment when calculating the maximum number of peels.
671 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
672 instead of TYPE_ALIGN_UNIT.
673 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
674 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
675 (vect_create_addr_base_for_vector_ref): Update call accordingly.
676 (vect_create_data_ref_ptr): Likewise.
677 (vect_setup_realignment): Realign by ANDing with
678 -DR_TARGET_MISALIGNMENT.
679 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
680 the number of peels based on DR_TARGET_ALIGNMENT.
681 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
682 with the guaranteed alignment boundary when deciding whether
684 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
685 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
686 (ensure_base_align): Remove stmt_info parameter. Get the
687 target base alignment from DR_TARGET_ALIGNMENT.
688 (vectorizable_store): Update call accordingly. Interpret
689 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
691 (vectorizable_load): Likewise.
693 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
694 Alan Hayward <alan.hayward@arm.com>
695 David Sherwood <david.sherwood@arm.com>
697 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
698 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
699 (vect_enhance_data_refs_alignment): Likewise.
701 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
703 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
704 error. Only quit immediately if parsing is complete.
705 (BEGIN): Initialize fatal_err and parse_done.
706 (begin fpu, end fpu): Check number of arguments.
707 (begin arch, end arch): Likewise.
708 (begin cpu, end cpu): Likewise.
709 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
710 (optalias): Likewise.
712 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
714 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
715 * config/arm/arm-isa.h: Delete. Move definitions to ...
716 * arm-cpus.in: ... here. Use new feature and fgroup values.
717 * config/arm/arm.c (arm_option_override): Use lower case for feature
719 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
720 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
721 * config/arm/parsecpu.awk (END): Add new command 'isa'.
723 (print_isa_bits_for): New function.
724 (gen_isa): New function.
725 (gen_comm_data): Use print_isa_bits_for.
726 (define feature): New keyword.
727 (define fgroup): New keyword.
728 * config/arm/t-arm (TM_H): Remove.
729 (GTM_H): Add arm-isa.h.
730 (arm-isa.h): Add rule to generate file.
731 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
732 case for feature bit names.
734 2017-09-22 Richard Biener <rguenther@suse.de>
736 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
738 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
739 print of no dependency loops ...
740 * graphite.c (graphite_transform_loops): ... here.
741 (canonicalize_loop_closed_ssa_form): Work from inner to outer
743 (same_close_phi_node, remove_duplicate_close_phi,
744 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
745 (canonicalize_loop_closed_ssa): ... here and simplify.
746 * graphite-optimize-isl.c: Include tree-vectorizer.h.
747 (optimize_isl): Use dump_printf_loc to tell when we stopped
748 optimizing because of an ISL timeout.
750 2017-09-22 Richard Biener <rguenther@suse.de>
752 PR tree-optimization/82291
753 * tree-if-conv.c (predicate_mem_writes): Make sure to
754 remove writes in blocks predicated with false.
756 2017-09-22 Richard Biener <rguenther@suse.de>
758 * sese.c: Include cfganal.h.
759 (if_region_set_false_region): Remove.
760 (create_if_region_on_edge): Likewise.
761 (move_sese_in_condition): Re-implement without destroying
764 2017-09-22 Richard Biener <rguenther@suse.de>
766 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
767 Verify both BBs contain loop PHI nodes before dispatching to
769 (graphite_regenerate_ast_isl): Do not recompute dominators,
770 do not verify three times. Restructure for clarity.
771 * graphite-scop-detection.c (same_close_phi_node,
772 remove_duplicate_close_phi, make_close_phi_nodes_unique,
773 defined_in_loop_p, canonicalize_loop_closed_ssa,
774 canonicalize_loop_closed_ssa_form): Simplify, remove excess
775 checking and SSA rewrite, move to ...
776 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
777 (graphite_initialize): Do not pass in ctx, do not reset the
778 SCEV cache, compute only dominators.
779 (graphite_transform_loops): Allocate ISL ctx after
780 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
781 Maintain post-dominators only around build_scops.
782 * sese.c (if_region_set_false_region): Make static. Free
783 and recompute dominators.
784 (move_sese_in_condition): Assert we don't get called with
785 post-dominators computed.
786 * sese.h (if_region_set_false_region): Remove.
788 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
790 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
791 mode attribute for TARGET_AVX512VL.
793 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
795 * config/i386/i386.opt (mprefer-avx256): New option.
796 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
798 (ix86_preferred_simd_mode): Return 256-bit AVX modes
799 for TARGET_PREFER_AVX256.
800 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
802 2017-09-21 Jeff Law <law@redhat.com>
804 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
805 Fix dump output if the only stack space is for pushed registers.
807 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
809 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
812 2017-09-21 Martin Sebor <msebor@redhat.com>
815 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
816 code (in C++) or code that triggers warnings.
818 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
820 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
822 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
824 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
825 * sched-rgn.c: Ditto.
826 * sel-sched-ir.c: Ditto.
828 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
830 * toplev.h (set_random_seed): Adjust return type.
831 * toplev.c (init_local_tick): Move eager initialization of random_seed
832 to get_random_seed. Adjust comment.
833 (init_random_seed): Inline to get_random_seed, delete.
834 (get_random_seed): Initialize random_seed lazily.
835 (set_random_seed): Do not return previous value.
836 (print_switch_value): Do not call get_random_seed.
838 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
840 * cgraph.c (delete_function_version): New, broken out from...
841 (cgraph_node::delete_function_version): ...here. Rename to
842 cgraph_node::delete_function_version_by_decl. Update all uses.
843 (cgraph_node::remove): Call delete_function_version.
845 2017-09-21 Jakub Jelinek <jakub@redhat.com>
848 * tree-inline.c (expand_call_inline): Emit clobber stmts for
849 VAR_DECLs to which addressable non-volatile parameters are mapped
850 and for id->retvar after the return value assignment. Clear
851 id->retval and id->retbnd after inlining.
853 2017-09-21 Richard Biener <rguenther@suse.de>
855 PR tree-optimization/82276
856 PR tree-optimization/82244
857 * tree-vrp.c (build_assert_expr_for): Set
858 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
860 (remove_range_assertions): Revert earlier change.
862 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
865 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
867 2017-09-21 Richard Biener <rguenther@suse.de>
869 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
870 Restore valid IL after code generation errors.
871 * graphite.c (graphite_transform_loops): Diagnose code
872 generation issues as MSG_MISSED_OPTIMIZATION and continue
873 with processing SCOPs.
875 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
876 Alan Hayward <alan.hayward@arm.com>
877 David Sherwood <david.sherwood@arm.com>
879 * calls.c (compute_argument_addresses): Use simplify_gen_binary
880 rather than choosing between plus_constant and gen_rtx_<CODE>.
881 * expr.c (emit_push_insn): Likewise.
882 (expand_expr_real_2): Likewise.
884 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
885 Alan Hayward <alan.hayward@arm.com>
886 David Sherwood <david.sherwood@arm.com>
888 * loop-unroll.c (split_iv): Call copy_rtx on the step.
890 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
891 Alan Hayward <alan.hayward@arm.com>
892 David Sherwood <david.sherwood@arm.com>
894 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
895 calling tree_to_uhwi.
897 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
898 Alan Hayward <alan.hayward@arm.com>
899 David Sherwood <david.sherwood@arm.com>
901 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
902 INTEGER_CST rather than a negative test for ADDR_EXPR.
904 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
905 Alan Hayward <alan.hayward@arm.com>
906 David Sherwood <david.sherwood@arm.com>
908 * tree-vrp.c (extract_range_from_binary_expr_1): Check
909 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
911 2017-09-21 Richard Biener <rguenther@suse.de>
913 PR tree-optimization/71351
914 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
915 graphite_create_new_loop_guard): Remove, fold remaining parts
917 (translate_isl_ast_node_for): ... here and simplify.
919 2017-09-21 Jakub Jelinek <jakub@redhat.com>
922 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
923 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
924 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
925 alternative always use QI mode, for -Os imov (=R,R) alternative
926 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
929 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
930 Jeff Law <law@redhat.com>
932 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
933 (allocate_stack_space): New function, partially extracted from
935 (s390_emit_prologue): Track offset to most recent stack probe.
936 Code to allocate space moved into allocate_stack_space.
937 Dump actions when no stack is allocated.
938 (s390_prologue_plus_offset): New function.
939 (s390_emit_stack_probe): Likewise.
941 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
943 * common.opt (Wa, Wl, Wp, g, gz=): Add
945 (gno-column-info): Remove.
946 (gcolumn-info): Drop RejectNegative.
948 (gno-record-gcc-switches): Remove.
949 (grecord-gcc-switches): Drop RejectNegative.
950 (gno-split-dwarf): Remove.
951 (gsplit-dwarf): Drop RejectNegative.
952 (gno-strict-dwarf): Remove.
953 (gstrict-dwarf): Drop RejectNegative.
954 * config/darwin.opt (gfull, gused): Add RejectNegative.
955 * dwarf2out.c (gen_producer_string): Drop
956 gno-record-gcc-switches handler.
957 * optc-gen.awk: Add g to prefixes with negative forms.
958 * opts-common.c (remapping_prefix_p): New.
959 (find_opt): Check it.
960 (generate_canonical_option): Test g prefix.
961 (option_map): Add -gno- mapping.
962 (add_misspelling_candidates): Check remapping_prefix_p.
964 2017-09-20 Jeff Law <law@redhat.com>
966 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
967 thinko in stack clash protection support.
969 * explow.c (compute_stack_clash_protection_loop_data): Use
970 CONST_INT_P instead of explicit test. Verify object is a
971 CONST_INT_P before looking at INTVAL.
972 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
973 instead of explicit test.
975 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
978 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
979 address instead of to r1 and r11.
981 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
983 * config.gcc: Support "knm".
984 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
985 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
987 * config/i386/i386.c (m_KNM): Define.
988 (processor_target_table): Add "knm".
990 (ix86_option_override_internal): Add "knm".
991 (ix86_issue_rate): Add PROCESSOR_KNM.
992 (ix86_adjust_cost): Ditto.
993 (ia32_multipass_dfa_lookahead): Ditto.
994 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
995 (fold_builtin_cpu): Add M_INTEL_KNM.
996 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
997 (processor_type): Add PROCESSOR_KNM.
998 * config/i386/x86-tune.def: Add m_KNM.
999 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1001 2017-09-20 Richard Biener <rguenther@suse.de>
1003 PR tree-optimization/80213
1004 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1005 are allowed in empty BBs as well.
1006 (canonicalize_loop_closed_ssa): Also look for other complex
1008 (scop_detection::get_sese): Include the loop-closed PHI block
1010 (scop_detection::merge_sese): Remove code adding extra blocks.
1011 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1012 (build_scops): Assert the final returned scop is invalid.
1014 2017-09-20 Richard Biener <rguenther@suse.de>
1016 PR tree-optimization/82264
1017 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1019 (vn_phi_lookup): Likewise.
1020 (vn_phi_insert): Likewise.
1022 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1024 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1025 that fits into uhwi or shwi, add DW_AT_const_value regardless
1026 of early_dwarf without going through RTL, using add_AT_unsigned
1029 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1030 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1031 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1032 (DEBUG_MACRO_SECTION): Likewise.
1033 (DEBUG_LTO_MACRO_SECTION): Likewise.
1034 (DEBUG_STR_DWO_SECTION): Likewise.
1035 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1036 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1037 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1038 (DEBUG_LTO_LINE_STR_SECTION): Define.
1039 (init_sections_and_labels): Initialize debug_line_str_section
1040 variable. Initialize debug_loc_section for -gdwarf-5 to
1041 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1043 2017-09-20 Richard Biener <rguenther@suse.de>
1045 * graphite-sese-to-poly.c (extract_affine): Properly handle
1046 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1048 2017-09-20 Richard Biener <rguenther@suse.de>
1050 PR tree-optimization/81373
1051 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1052 Force SESE live-out defs to be handled even if they are
1055 2017-09-19 Jeff Law <law@redhat.com>
1057 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1058 nothing for stack adjustments with REG_STACK_CHECK.
1059 * sched-deps.c (parse_add_or_inc): Reject insns with
1060 REG_STACK_CHECK from dependency breaking.
1061 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1062 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1063 * reg-notes.def (STACK_CHECK): New note.
1065 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1066 (ix86_expand_prologue): Dump stack clash info as needed.
1067 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1069 * function.c (dump_stack_clash_frame_info): New function.
1070 * function.h (dump_stack_clash_frame_info): Prototype.
1071 (enum stack_clash_probes): New enum.
1073 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1074 flag_stack_clash_protection.
1075 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1076 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1077 (arm_frame_pointer_required): Likewise.
1078 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1079 (ia64_expand_prologue): Likewise.
1080 * config/mips/mips.c (mips_expand_prologue): Likewise.
1081 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1082 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1083 (sparc_flat_expand_prologue): Likewise.
1084 * config/spu/spu.c (spu_expand_prologue): Likewise.
1086 * explow.c: Include "params.h".
1087 (anti_adjust_stack_and_probe_stack_clash): New function.
1088 (get_stack_check_protect): Likewise.
1089 (compute_stack_clash_protection_loop_data): Likewise.
1090 (emit_stack_clash_protection_loop_start): Likewise.
1091 (emit_stack_clash_protection_loop_end): Likewise.
1092 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1093 Use anti_adjust_stack_and_probe_stack_clash.
1094 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1095 (emit_stack_clash_protection_loop_start): Likewise.
1096 (emit_stack_clash_protection_loop_end): Likewise.
1097 * rtl.h (get_stack_check_protect): Prototype.
1098 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1099 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1100 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1102 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1104 * doc/tm.texi: Rebuilt.
1105 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1106 get_stack_check_protect.
1107 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1108 * config/arm/arm.c (arm_expand_prologue): Likewise.
1109 (arm_frame_pointer_required): Likewise.
1110 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1111 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1112 * config/mips/mips.c (mips_expand_prologue): Likewise.
1113 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1114 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1115 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1116 (sparc_flat_expand_prologue): Likewise.
1118 * common.opt (-fstack-clash-protection): New option.
1119 * flag-types.h (enum stack_check_type): Note difference between
1120 -fstack-check= and -fstack-clash-protection.
1121 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1122 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1123 * toplev.c (process_options): Issue warnings/errors for cases
1124 not handled with -fstack-clash-protection.
1125 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1126 (-fstack-check): Note additional problem with -fstack-check=generic.
1127 Note that -fstack-check is primarily for Ada and refer users
1128 to -fstack-clash-protection for stack-clash-protection.
1129 Document new params for stack clash protection.
1131 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1133 * config/i386/i386.c (ix86_split_long_move): Do not handle
1134 address used for LEA in a special way.
1136 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1138 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1140 2017-09-19 Martin Sebor <msebor@redhat.com>
1143 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1144 of incompatible types.
1146 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1148 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1149 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1151 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1153 2017-09-19 Richard Biener <rguenther@suse.de>
1155 PR tree-optimization/82244
1156 * tree-vrp.c (remove_range_assertions): Do not propagate
1157 a constant to abnormals but replace the assert with a copy.
1159 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1161 PR rtl-optimization/57878
1162 PR rtl-optimization/68988
1163 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1164 avoidance test involving non_reload_pseudos. Move frequency test
1165 below the general fragmentation avoidance test.
1167 2017-09-19 Richard Biener <rguenther@suse.de>
1169 PR tree-optimization/69728
1170 * graphite-sese-to-poly.c (schedule_error): New global.
1171 (add_loop_schedule): Handle empty domain by failing the
1173 (build_original_schedule): Handle schedule_error.
1175 2017-09-19 Richard Biener <rguenther@suse.de>
1177 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1178 Do not iterate to sibling loops but only to siblings of inner
1181 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1184 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1185 that the registers are different.
1187 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1189 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1190 to processor_model and "amdfam17h" to arch_names_table.
1191 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1193 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1196 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1199 Alan Hayward <alan.hayward@arm.com>
1200 David Sherwood <david.sherwood@arm.com>
1202 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1204 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1206 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1207 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1208 vect_schedule_slp_instance.
1209 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1210 Update call to vect_slp_analyze_node_operations. Simplify return
1212 (vect_slp_analyze_bb_1): Update call accordingly.
1213 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1214 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1215 (vect_schedule_slp): Update call accordingly.
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-data-refs.c (vect_get_smallest_scalar_type): Cope
1222 with types that aren't in fact scalar.
1224 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1226 * tree-vect-slp.c (vect_record_max_nunits): New function,
1228 (vect_build_slp_tree_1): ...here.
1229 (vect_build_slp_tree_2): Call it for phis too.
1231 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1233 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1234 to vect_get_vec_def_for_operand when getting the mask operand.
1236 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1237 Alan Hayward <alan.hayward@arm.com>
1238 David Sherwood <david.sherwood@arm.com>
1240 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1243 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1244 Alan Hayward <alan.hayward@arm.com>
1245 David Sherwood <david.sherwood@arm.com>
1247 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1248 calculation for vector booleans.
1250 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1251 Alan Hayward <alan.hayward@arm.com>
1252 David Sherwood <david.sherwood@arm.com>
1254 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1256 (vect_transform_stmt): ...here.
1257 (vect_analyze_stmt): Use it instead of calling
1258 vectorizable_live_operation directly.
1260 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1262 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1263 non-SIMT targets in acc vector loops.
1265 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1267 * configure.ac: Add arc and check if assembler supports gdwarf2.
1268 * configure: Regenerate.
1270 2017-09-18 Richard Biener <rguenther@suse.de>
1272 PR tree-optimization/82220
1273 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1274 epilogue niters from the min_profitable_iters compute.
1276 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1279 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1280 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1281 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1283 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1286 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1287 switching to a new text section.
1289 2017-09-18 Richard Biener <rguenther@suse.de>
1291 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1293 (build_alias_set): Reject aliases with no access function.
1295 2017-09-18 Richard Biener <rguenther@suse.de>
1297 PR tree-optimization/79622
1298 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1300 (build_cross_bb_scalars_use): Likewise.
1302 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1304 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1306 2017-09-18 Alan Modra <amodra@gmail.com>
1309 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1310 stack_pointer_rtx for count 0. Update comments. Break up
1311 large rtl expression.
1313 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1315 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1316 Increase to 20 bytes.
1317 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1318 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1319 or avx version of the stub.
1321 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1324 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1325 compute the minimum stack alignment. Also update preferred stack
1326 boundary for leaf functions.
1328 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1330 PR tree-optimization/82228
1331 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1334 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1336 * common/config/nds32/nds32-common.c
1337 (nds32_option_optimization_table): Refine formatting.
1338 (nds32_option_optimization_table): Use -fsched-pressure and
1339 -fomit-frame-pointer for specific optimization level.
1341 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1343 * config/nds32/nds32.c: Refine formatting and comments.
1344 * config/nds32/nds32.h: Likewise.
1345 * config/nds32/nds32.md: Likewise.
1346 * config/nds32/nds32-cost.c: Likewise.
1347 * config/nds32/nds32-isr.c: Likewise.
1348 * config/nds32/nds32-md-auxiliary.c: Likewise.
1349 * config/nds32/nds32-multiple.md: Likewise.
1350 * config/nds32/nds32-predicates.c: Likewise.
1352 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1353 Jakub Jelinek <jakub@redhat.com>
1355 Add support for -std=c++2a.
1356 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1358 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1360 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1363 * doc/extend.texi (Common Function Attributes): Add
1364 references to ARM, AArch64, and S/390 specific attributes.
1365 (Function Specific Option Pragmas): Add AArch64 and S/390
1366 to list of back ends that support the target pragma.
1368 2017-09-15 Nathan Sidwell <nathan@acm.org>
1370 * doc/standards.texi: Fix C++17 description. Update URLs for
1373 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1375 * common.opt (Wcast-align=strict): New warning option.
1376 * doc/invoke.texi: Document -Wcast-align=strict.
1378 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1380 * cgraph.h (cgraph_thunk_info): Add comments.
1381 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1382 assert for VIRTUAL_* arguments stricter.
1384 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1386 PR tree-optimization/71026
1387 * match.pd: Move RDIV patterns from fold-const.c
1388 * fold-const.c (distribute_real_division): Removed.
1389 (fold_binary_loc): Remove calls to distribute_real_divison.
1391 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1393 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1394 c++1z and gnu++1z as deprecated. Change other references to
1395 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1396 Change -Wc++1z-compat to -Wc++17-compat.
1397 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1398 * dwarf2out.c (highest_c_language): Handle C++17.
1399 (gen_compile_unit_die): Likewise.
1401 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1403 PR rtl-optimization/82192
1404 * combine.c (make_extraction): Don't look through non-paradoxical
1405 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1408 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1409 Alan Hayward <alan.hayward@arm.com>
1410 David Sherwood <david.sherwood@arm.com>
1412 * target.def (function_arg_offset): New hook.
1413 * targhooks.h (default_function_arg_offset): Declare.
1414 * targhooks.c (default_function_arg_offset): New function.
1415 * function.c (locate_and_pad_parm): Use
1416 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1417 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1418 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1419 * doc/tm.texi: Regenerate.
1420 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1421 * config/spu/spu.c (spu_function_arg_offset): New function.
1422 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1423 * system.h (FUNCTION_ARG_OFFSET): Poison.
1425 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1426 Alan Hayard <alan.hayward@arm.com>
1427 David Sherwood <david.sherwood@arm.com>
1429 * target.def (truly_noop_truncation): New hook.
1430 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1431 than TRULY_NOOP_TRUNCATION.
1432 * hooks.h (hook_bool_uint_uint_true): Declare.
1433 * hooks.c (hook_bool_uint_uint_true): New function.
1434 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1435 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1436 * doc/tm.texi: Regenerate.
1437 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1438 rather than TRULY_NOOP_TRUNCATION in comments.
1439 (simplify_comparison): Likewise.
1440 (record_truncated_value): Likewise.
1441 * expmed.c (extract_bit_field_1): Likewise.
1442 (extract_split_bit_field): Likewise.
1443 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1444 instead of TRULY_NOOP_TRUNCATION.
1445 * function.c (assign_parm_setup_block): Likewise.
1446 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1447 * rtlhooks.c: Include target.h.
1448 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1449 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1450 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1451 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1452 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1453 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1454 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1455 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1456 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1457 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1458 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1459 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1460 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1461 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1462 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1463 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1464 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1465 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1466 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1467 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1468 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1469 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1470 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1471 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1472 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1473 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1474 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1475 rather than TRULY_NOOP_TRUNCATION in comments.
1476 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1477 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1478 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1479 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1480 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1481 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1482 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1483 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1484 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1485 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1486 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1487 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1488 rather than TRULY_NOOP_TRUNCATION in comments.
1489 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1490 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1491 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1492 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1493 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1494 TRULY_NOOP_TRUNCATION condition.
1495 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1496 (TRULY_NOOP_TRUNCATION): Delete.
1497 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1498 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1499 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1500 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1501 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1502 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1503 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1504 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1505 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1506 rather than TRULY_NOOP_TRUNCATION in comments.
1507 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1508 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1509 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1510 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1511 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1512 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1514 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1517 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1518 (*cmp_ior): Likewise.
1519 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1520 (*ior_scc_scc_cmp): Likewise.
1521 (*and_scc_scc): Likewise.
1522 (*and_scc_scc_cmp): Likewise.
1524 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1525 Alan Hayard <alan.hayward@arm.com>
1526 David Sherwood <david.sherwood@arm.com>
1528 * target.def (can_change_mode_class): New hook.
1529 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1530 (hard_regno_nregs): Likewise.
1531 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1532 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1533 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1534 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1535 (LOAD_EXTEND_OP): Update accordingly.
1536 * doc/tm.texi: Regenerate.
1537 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1538 CANNOT_CHANGE_MODE_CLASS.
1539 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1540 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1541 * combine.c (simplify_set): Update accordingly.
1542 * emit-rtl.c (validate_subreg): Likewise.
1543 * recog.c (general_operand): Likewise.
1544 * regcprop.c (mode_change_ok): Likewise.
1545 * reload1.c (choose_reload_regs): Likewise.
1546 (inherit_piecemeal_p): Likewise.
1547 * rtlanal.c (simplify_subreg_regno): Likewise.
1548 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1549 instead of CANNOT_CHANGE_MODE_CLASS.
1550 (reload_cse_simplify_operands): Likewise.
1551 * reload.c (push_reload): Use targetm.can_change_mode_class
1552 instead of CANNOT_CHANGE_MODE_CLASS.
1553 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1554 REG_CANNOT_CHANGE_MODE_P.
1555 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1556 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1557 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1558 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1559 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1560 (arm_can_change_mode_class): New function.
1561 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1562 than CANNOT_CHANGE_MODE_CLASS in comments.
1563 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1564 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1565 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1566 (ix86_can_change_mode_class): ...this new function, inverting the
1567 sense of the return value.
1568 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1569 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1570 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1571 (ia64_can_change_mode_class): New function.
1572 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1573 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1574 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1575 (m32c_can_change_mode_class): ...this new function, inverting the
1576 sense of the return value.
1577 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1578 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1579 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1580 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1581 (mips_can_change_mode_class): ...this new function, inverting the
1582 sense of the return value.
1583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1584 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1585 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1586 (msp430_can_change_mode_class): New function.
1587 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1588 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1589 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1590 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1591 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1592 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1593 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1594 (pa_cannot_change_mode_class): Replace with...
1595 (pa_can_change_mode_class): ...this new function, inverting the
1596 sense of the return value.
1597 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1598 than CANNOT_CHANGE_MODE_CLASS in comments.
1599 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1600 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1601 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1602 (pdp11_cannot_change_mode_class): Replace with...
1603 (pdp11_can_change_mode_class): ...this new function, inverting the
1604 sense of the return value.
1605 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1606 * config/powerpcspe/powerpcspe-protos.h
1607 (rs6000_cannot_change_mode_class_ptr): Delete.
1608 * config/powerpcspe/powerpcspe.c
1609 (rs6000_cannot_change_mode_class_ptr): Delete.
1610 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1611 (rs6000_option_override_internal): Assign to
1612 targetm.can_change_mode_class instead of
1613 rs6000_cannot_change_mode_class_ptr.
1614 (rs6000_cannot_change_mode_class): Replace with...
1615 (rs6000_can_change_mode_class): ...this new function, inverting the
1616 sense of the return value.
1617 (rs6000_debug_cannot_change_mode_class): Replace with...
1618 (rs6000_debug_can_change_mode_class): ...this new function.
1619 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1620 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1621 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1622 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1623 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1625 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1626 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1627 (rs6000_option_override_internal): Assign to
1628 targetm.can_change_mode_class instead of
1629 rs6000_cannot_change_mode_class_ptr.
1630 (rs6000_cannot_change_mode_class): Replace with...
1631 (rs6000_can_change_mode_class): ...this new function, inverting the
1632 sense of the return value.
1633 (rs6000_debug_cannot_change_mode_class): Replace with...
1634 (rs6000_debug_can_change_mode_class): ...this new function.
1635 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1636 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1637 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1638 (s390_can_change_mode_class): ...this new function, inverting the
1639 sense of the return value.
1640 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1641 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1642 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1643 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1644 (sh_cannot_change_mode_class): Replace with...
1645 (sh_can_change_mode_class): ...this new function, inverting the
1646 sense of the return value.
1647 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1648 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1649 (sparc_can_change_mode_class): New function.
1650 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1651 * config/spu/spu.c (spu_can_change_mode_class): New function.
1652 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1653 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1654 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1655 (visium_can_change_mode_class): New function.
1656 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1658 2017-09-15 Richard Biener <rguenther@suse.de>
1660 PR tree-optimization/82217
1661 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1662 but not undefined case.
1664 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1667 * postreload.c (reload_cse_simplify_operands): Skip
1668 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1670 2017-09-15 Richard Biener <rguenther@suse.de>
1672 PR tree-optimization/68823
1673 * graphite-scop-detection.c (build_alias_set): If we have a
1674 possible dependence check whether we can handle them by just
1675 looking at the DRs DR_ACCESS_FNs.
1676 (build_scops): If build_alias_set fails, fail the SCOP.
1678 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1680 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1681 to support float128 built-in functions that require the ISA 3.0
1683 (BU_FLOAT128_3_HW): Likewise.
1684 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1686 (FMAF128): Likewise.
1688 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1689 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1690 floating point instructions.
1691 (rs6000_invalid_builtin): Likewise.
1692 (rs6000_builtin_mask_names): Likewise.
1693 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1694 (RS6000_BTM_FLOAT128_HW): Likewise.
1695 (RS6000_BTM_COMMON): Likewise.
1696 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1698 * doc/extend.texi (RS/6000 built-in functions): Document the
1699 IEEE 128-bit floating point square root and fused multiply-add
1702 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1704 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1705 reg (r2) isn't in the set of registers defined in the prologue.
1707 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1708 Alan Hayward <alan.hayward@arm.com>
1709 David Sherwood <david.sherwood@arm.com>
1711 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1712 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1713 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1714 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1715 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1717 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1718 max_vectorization_factor.
1719 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1721 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1722 Alan Hayward <alan.hayward@arm.com>
1723 David Sherwood <david.sherwood@arm.com>
1725 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1726 (vect_worthwhile_without_simd_p): Declare.
1727 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1728 (vectorizable_reduction): Use it.
1729 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1730 (vectorizable_operation): Likewise.
1732 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1733 Alan Hayward <alan.hayward@arm.com>
1734 David Sherwood <david.sherwood@arm.com>
1736 * tree-vectorizer.h (vect_get_num_copies): New function.
1737 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1738 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1739 (vectorizable_induction): Likewise.
1740 (vectorizable_live_operation): Likewise.
1741 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1742 (vectorizable_bswap): Likewise.
1743 (vectorizable_call): Likewise.
1744 (vectorizable_conversion): Likewise.
1745 (vectorizable_assignment): Likewise.
1746 (vectorizable_shift): Likewise.
1747 (vectorizable_operation): Likewise.
1748 (vectorizable_store): Likewise.
1749 (vectorizable_load): Likewise.
1750 (vectorizable_condition): Likewise.
1751 (vectorizable_comparison): Likewise.
1752 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1754 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1755 Alan Hayward <alan.hayward@arm.com>
1756 David Sherwood <david.sherwood@arm.com>
1758 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1759 of vect_init_vector.
1761 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1766 an inline wrapper that provides a location.
1767 (gimple_build_vector): Likewise.
1768 * gimple-fold.c (gimple_build_vector_from_val): New function.
1769 (gimple_build_vector): Likewise.
1770 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1771 functions to build the initial value. Always return a gimple value.
1772 (get_initial_defs_for_reduction): Likewise. Only compute
1774 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1775 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1776 (vectorizable_induction): Use gimple_build_vector rather than
1779 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1780 Alan Hayward <alan.hayward@arm.com>
1781 David Sherwood <david.sherwood@arm.com>
1783 * target.h (vec_perm_indices): New typedef.
1784 (auto_vec_perm_indices): Likewise.
1785 * optabs-query.h: Include target.h
1786 (can_vec_perm_p): Take a vec_perm_indices *.
1787 * optabs-query.c (can_vec_perm_p): Likewise.
1788 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1789 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1790 * tree-vect-generic.c (lower_vec_perm): Likewise.
1791 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1792 (vect_grouped_load_supported): Likewise.
1793 (vect_shift_permute_load_chain): Likewise.
1794 (vect_permute_store_chain): Use auto_vec_perm_indices.
1795 (vect_permute_load_chain): Likewise.
1796 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
1797 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
1798 Update uses of can_vec_perm_p.
1799 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
1800 mode with a number of elements. Take a vec_perm_indices *.
1801 (vect_create_epilog_for_reduction): Update accordingly.
1802 Use auto_vec_perm_indices.
1803 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
1804 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
1805 (vect_transform_slp_perm_load): Likewise.
1806 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
1807 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
1808 (vect_gen_perm_mask_checked): Likewise.
1809 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
1810 (vect_gen_perm_mask_checked): Likewise.
1811 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
1812 (vectorizable_store): Likewise.
1813 (vectorizable_load): Likewise.
1814 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
1815 (vectorizable_bswap): Likewise.
1817 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1818 Alan Hayward <alan.hayward@arm.com>
1819 David Sherwood <david.sherwood@arm.com>
1821 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1822 * tree.c (build_vector): Likewise.
1823 (build_vector_from_ctor): Update accordingly.
1824 (build_vector_from_val): Likewise.
1825 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1826 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1827 * tree-vect-generic.c (add_rshift): Likewise.
1828 (expand_vector_divmod): Likewise.
1829 (optimize_vector_constructor): Likewise.
1830 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1831 (vect_transform_slp_perm_load): Likewise.
1832 (vect_schedule_slp_instance): Likewise.
1833 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1834 (vectorizable_call): Likewise.
1835 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1836 * expmed.c (make_tree): Likewise.
1837 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1838 a vector passed to build_vector.
1839 (fold_convert_const): Likewise.
1840 (exact_inverse): Likewise.
1841 (fold_ternary_loc): Likewise.
1842 (fold_relational_const): Likewise.
1843 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1844 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1845 (const_unop): Likewise. Store the reduction accumulator in a
1846 variable rather than an array.
1847 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1848 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1849 the new vector, rather than constructing it after the input arrays.
1850 (native_interpret_vector): Use auto_vec<tree> when building
1851 a vector passed to build_vector. Add elements in order.
1852 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1853 auto_vec<tree> when building a vector passed to build_vector.
1854 (vect_create_epilog_for_reduction): Likewise.
1855 (vectorizable_induction): Likewise.
1856 (get_initial_def_for_reduction): Likewise. Fix indentation of
1858 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1860 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1861 passed to build_vector.
1863 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1864 Alan Hayward <alan.hayward@arm.com>
1865 David Sherwood <david.sherwood@arm.com>
1867 * tree-core.h (tree_base::u): Add an "nelts" field.
1868 (tree_vector): Use VECTOR_CST_NELTS as the length.
1869 * tree.c (tree_size): Likewise.
1870 (make_vector): Initialize VECTOR_CST_NELTS.
1871 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1872 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1873 TYPE_VECTOR_SUBPARTS.
1874 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1875 as the number of units, setting it from VECTOR_CST_NELTS.
1876 (const_vector_from_tree): Likewise.
1877 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1878 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1879 (fold_negate_expr_1): Likewise.
1880 (fold_convert_const): Likewise.
1881 (const_binop): Likewise. Differentiate the number of output and
1883 (const_unop): Likewise.
1884 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1885 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1886 in cases that did the opposite.
1888 2017-09-14 Richard Biener <rguenther@suse.de>
1890 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1893 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1895 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1897 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1900 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1901 if and where to split a bb, except for splitting before debug insn
1902 sequences followed by non-label real insn. Delete debug insns
1903 in between basic blocks.
1905 * combine.c (make_compound_operation_int): Formatting fixes.
1907 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1908 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1909 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1910 * config/sol2.h (LINK_EH_SPEC): Likewise.
1911 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1912 * config/s390/linux.h (LINK_SPEC): Likewise.
1913 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1914 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1915 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1916 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1917 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1918 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1919 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1920 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1922 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1924 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1926 (ENDFILE_LINUX_SPEC): Likewise.
1927 (LINK_EH_SPEC): Likewise.
1928 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1929 (LINK_OS_LINUX_SPEC32): Likewise.
1930 (LINK_OS_LINUX_SPEC64): Likewise.
1931 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1932 (LINK_OS_LINUX_SPEC): Likewise.
1934 2017-09-13 Martin Liska <mliska@suse.cz>
1937 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1938 CASE_HIGH is NULL_TREE.
1940 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1941 Alan Hayward <alan.hayward@arm.com>
1942 David Sherwood <david.sherwood@arm.com>
1944 * target.def (secondary_memory_needed): New hook.
1945 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1946 instead of SECONDARY_MEMORY_NEEDED.
1947 (secondary_memory_needed_mode): Likewise.
1948 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1949 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1950 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1951 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1952 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1953 * doc/tm.texi: Regenerate.
1954 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1955 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1956 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1957 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1958 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1959 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1960 mode argument first and change the reg_class arguments to reg_class_t.
1961 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1962 Make static. Update the call to inline_secondary_memory_needed.
1963 (ix86_register_move_cost): Update the call to
1964 inline_secondary_memory_needed.
1965 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1966 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1968 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1969 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1971 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1972 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1973 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1974 and match hook interface. Add comment from mips.h.
1975 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1976 * config/mmix/mmix.md (truncdfsf2): Refer to
1977 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1979 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1980 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1981 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1982 (pa_secondary_memory_needed): New function.
1983 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1984 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1985 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1986 (pdp11_secondary_memory_needed): Make static and match hook interface.
1987 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1988 * config/powerpcspe/powerpcspe-protos.h
1989 (rs6000_secondary_memory_needed_ptr): Delete.
1990 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1992 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1993 (rs6000_option_override_internal): Assign to
1994 targetm.secondary_memory_needed rather than
1995 rs6000_secondary_memory_needed_ptr.
1996 (rs6000_secondary_memory_needed): Match hook interface.
1997 (rs6000_debug_secondary_memory_needed): Likewise.
1998 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1999 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2000 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2001 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2002 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2003 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2005 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2006 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2007 (rs6000_option_override_internal): Assign to
2008 targetm.secondary_memory_needed rather than
2009 rs6000_secondary_memory_needed_ptr.
2010 (rs6000_secondary_memory_needed): Match hook interface.
2011 (rs6000_debug_secondary_memory_needed): Likewise.
2012 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2013 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2014 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2015 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2016 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2017 (sparc_secondary_memory_needed): New function.
2018 * lra-constraints.c (check_and_process_move): Refer to
2019 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2021 (curr_insn_transform): Likewise.
2022 (process_alt_operands): Use targetm.secondary_memory_needed
2023 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2024 (check_secondary_memory_needed_p): Likewise.
2025 (choose_split_class): Likewise.
2026 * reload.c: Unconditionally include code that was previously
2027 conditional on SECONDARY_MEMORY_NEEDED.
2028 (push_secondary_reload): Use targetm.secondary_memory_needed
2029 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2030 (push_reload): Likewise.
2031 * reload1.c: Unconditionally include code that was previously
2032 conditional on SECONDARY_MEMORY_NEEDED.
2033 (choose_reload_regs): Use targetm.secondary_memory_needed
2034 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2035 (gen_reload): Likewise.
2036 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2038 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2039 Alan Hayward <alan.hayward@arm.com>
2040 David Sherwood <david.sherwood@arm.com>
2042 * target.def (secondary_memory_needed_mode): New hook:
2043 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2044 * targhooks.h (default_secondary_memory_needed_mode): New function.
2045 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2046 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2047 * doc/tm.texi: Regenerate.
2048 * lra-constraints.c (check_and_process_move): Use
2049 targetm.secondary_memory_needed_mode instead of
2050 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2051 (curr_insn_transform): Likewise.
2052 * reload.c (get_secondary_mem): Likewise.
2053 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2054 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2056 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2057 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2058 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2059 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2060 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2062 * config/powerpcspe/powerpcspe-protos.h
2063 (rs6000_secondary_memory_needed_mode): Delete.
2064 * config/powerpcspe/powerpcspe.c
2065 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2066 (rs6000_secondary_memory_needed_mode): Make static.
2067 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2068 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2070 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2072 (rs6000_secondary_memory_needed_mode): Make static.
2073 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2074 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2075 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2076 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2077 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2079 (sparc_secondary_memory_needed_mode): New function.
2080 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2082 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2084 * config/aarch64/constraints.md (Umq): New constraint.
2085 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2087 (mov<mode>): Update condition.
2089 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2091 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2092 when bitposition is the same.
2094 2017-09-13 Richard Biener <rguenther@suse.de>
2096 * dwarf2out.c (output_die_symbol): Remove.
2097 (output_die): Do not output a DIEs symbol.
2099 2017-09-13 Richard Biener <rguenther@suse.de>
2102 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2103 default-def to avoid breaking iterator update with the weird
2104 interaction with cgraph_update_edges_for_call_stmt_node.
2106 2017-09-13 Richard Biener <rguenther@suse.de>
2108 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2109 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2110 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2111 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2113 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2115 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2116 Disable pc relative literal load irrespective of
2117 TARGET_FIX_ERR_A53_84341 for default.
2119 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2121 * config/sparc/sparc.c (output_return): Output the source location of
2122 the insn in the delay slot, if any.
2123 (output_sibcall): Likewise.
2125 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2128 * common.opt (-static-pie): New alias.
2129 (shared): Negate static-pie.
2130 (-no-pie): Update help text.
2132 (static-pie): New option.
2133 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2134 -static-pie support.
2135 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2136 (LINK_EH_SPEC): Likewise.
2137 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2138 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2139 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2140 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2141 (init_gcc_specs): Likewise.
2142 (init_spec): Likewise.
2143 (display_help): Update help message for -pie.
2144 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2147 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2149 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2150 (movdi_aarch64): Likewise.
2151 (movti_aarch64): Likewise.
2153 2017-09-12 Simon Wright <simon@pushface.org>
2156 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2157 calculation of the minor version, always output as 0.
2159 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2162 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2163 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2164 on it early, rather than manual conversion late. For
2165 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2166 instead of performing manual conversion.
2168 2017-09-12 Carl Love <cel@us.ibm.com>
2170 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2171 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2173 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2174 VMULOSW): Add definitions.
2175 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2176 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2177 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2178 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2179 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2181 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2183 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2185 (movti_aarch64): Likewise.
2186 (movdf_aarch64): Likewise.
2187 (movtf_aarch64): Likewise.
2188 (load_pairdi): Likewise.
2189 (store_pairdi): Likewise.
2190 (load_pairdf): Likewise.
2191 (store_pairdf): Likewise.
2192 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2193 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2194 (ldr_got_small_<mode>): Likewise.
2195 (ldr_got_small_28k_<mode>): Likewise.
2196 (ldr_got_tiny): Likewise.
2197 * config/aarch64/iterators.md (ldst_sz): New.
2198 (ldpstp_sz): Likewise.
2199 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2201 (thunderx_load): Split load_8 to load_16.
2202 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2204 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2205 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2206 (xgene1_store_pair): Split store_8 to store_16.
2207 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2208 (falkor_st_0_st_sd): Split store_8 to store_16.
2210 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2212 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2213 and store1/2/3/4 to store_4/8/12/16.
2214 * config/aarch64/aarch64.md: Update for rename.
2215 * config/arm/arm.md: Likewise.: Likewise.
2216 * config/arm/arm.c: Likewise.
2217 * config/arm/thumb1.md: Likewise.
2218 * config/arm/thumb2.md: Likewise.
2219 * config/arm/vfp.md: Likewise.
2220 * config/arm/arm-generic.md: Likewise.
2221 * config/arm/arm1020e.md: Likewise.
2222 * config/arm/arm1026ejs.md: Likewise.
2223 * config/arm/arm1136jfs.md: Likewise.
2224 * config/arm/arm926ejs.md: Likewise.
2225 * config/arm/cortex-a15.md: Likewise.
2226 * config/arm/cortex-a17.md: Likewise.
2227 * config/arm/cortex-a5.md: Likewise.
2228 * config/arm/cortex-a53.md: Likewise.
2229 * config/arm/cortex-a57.md: Likewise.
2230 * config/arm/cortex-a7.md: Likewise.
2231 * config/arm/cortex-a8.md: Likewise.
2232 * config/arm/cortex-a9.md: Likewise.
2233 * config/arm/cortex-m4.md: Likewise.
2234 * config/arm/cortex-m7.md: Likewise.
2235 * config/arm/cortex-r4.md: Likewise.
2236 * config/arm/exynos-m1.md: Likewise.
2237 * config/arm/fa526.md: Likewise.
2238 * config/arm/fa606te.md: Likewise.
2239 * config/arm/fa626te.md: Likewise.
2240 * config/arm/fa726te.md: Likewise.
2241 * config/arm/fmp626.md: Likewise.
2242 * config/arm/iwmmxt.md: Likewise.
2243 * config/arm/ldmstm.md: Likewise.
2244 * config/arm/marvell-pj4.md: Likewise.
2245 * config/arm/xgene1.md: Likewise.
2246 * config/aarch64/thunderx.md: Likewise.
2247 * config/aarch64/thunderx2t99.md: Likewise.
2248 * config/aarch64/falkor.md: Likewise.
2250 2017-09-12 Martin Liska <mliska@suse.cz>
2252 * attribs.c (private_lookup_attribute): New function.
2253 * attribs.h (private_lookup_attribute): Declared here.
2254 (lookup_attribute): Called from this place.
2256 2017-09-12 Richard Biener <rguenther@suse.de>
2258 PR tree-optimization/82157
2259 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2260 stmts with side-effects.
2262 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2263 Alan Hayward <alan.hayward@arm.com>
2264 David Sherwood <david.sherwood@arm.com>
2266 * target.def (hard_regno_nregs): New hook.
2267 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2268 * targhooks.h (default_hard_regno_nregs): Declare.
2269 * targhooks.c (default_hard_regno_nregs): New function.
2270 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2271 (TARGET_HARD_REGNO_NREGS): ...this hook.
2272 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2273 (CLASS_MAX_NREGS): Likewise.
2274 * doc/tm.texi: Regenerate.
2275 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2276 instead of HARD_REGNO_NREGS.
2277 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2278 HARD_REGNO_NREGS in the comment.
2279 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2280 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2281 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2282 Return an unsigned int.
2283 (TARGET_HARD_REGNO_NREGS): Redefine.
2284 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2285 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2286 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2287 (arc_hard_regno_nregs): New function.
2288 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2289 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2290 (arm_hard_regno_nregs): New function.
2291 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2292 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2293 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2294 (TARGET_HARD_REGNO_NREGS): Redefine.
2295 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2296 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2297 (HARD_REGNO_NREGS): Delete.
2298 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2299 (cr16_hard_regno_nregs): New function.
2300 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2301 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2302 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2303 (cris_hard_regno_nregs): New function.
2304 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2305 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2306 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2307 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2308 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2309 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2310 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2311 (frv_hard_regno_nregs): Make static. Take and return an
2313 (frv_class_max_nregs): Remove outdated copy of documentation.
2314 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2315 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2316 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2317 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2318 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2319 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2320 (TARGET_HARD_REGNO_NREGS): Redefine.
2321 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2322 (CLASS_MAX_NREGS): Update comment.
2323 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2324 (ia64_hard_regno_nregs): New function.
2325 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2326 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2327 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2328 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2329 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2331 (m32c_hard_regno_nregs): Likewise. Make static.
2332 (TARGET_HARD_REGNO_NREGS): Redefine.
2333 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2334 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2335 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2336 (m68k_hard_regno_nregs): New function.
2337 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2338 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2339 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2340 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2341 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2342 Take and return an unsigned int.
2343 (TARGET_HARD_REGNO_NREGS): Redefine.
2344 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2345 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2346 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2347 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2348 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2349 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2350 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2351 (msp430_hard_regno_nregs): Make static. Take and return an
2353 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2354 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2355 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2356 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2357 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2358 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2359 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2360 (TARGET_HARD_REGNO_NREGS): Redefine.
2361 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2362 (PA_HARD_REGNO_NREGS): ...this.
2363 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2364 (PA_HARD_REGNO_NREGS): ...this.
2365 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2366 (pa_hard_regno_nregs): New function.
2367 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2368 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2369 (pdp11_hard_regno_nregs): New function.
2370 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2371 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2372 (rs6000_hard_regno_nregs_hook): New function.
2373 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2374 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2375 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2376 Take and return an unsigned int. Move earlier in file.
2377 (TARGET_HARD_REGNO_NREGS): Redefine.
2378 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2379 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2380 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2381 (rl78_hard_regno_nregs): Make static. Take and return an
2383 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2384 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2385 (rs6000_hard_regno_nregs_hook): New function.
2386 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2387 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2388 (TARGET_HARD_REGNO_NREGS): Redefine.
2389 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2390 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2391 instead of HARD_REGNO_NREGS.
2392 (s390_hard_regno_nregs): New function.
2393 (s390_hard_regno_mode_ok): Add comment from s390.h.
2394 (TARGET_HARD_REGNO_NREGS): Redefine.
2395 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2396 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2397 (sh_hard_regno_nregs): New function.
2398 (sh_pass_in_reg_p): Use it.
2399 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2400 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2401 (sparc_hard_regno_nregs): New function.
2402 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2403 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2404 (spu_function_arg_advance): Use it, supplying a valid register number.
2405 (TARGET_HARD_REGNO_NREGS): Redefine.
2406 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2407 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2408 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2409 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2410 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2411 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2412 (CLASS_MAX_NREGS): Remove copy of old documentation.
2413 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2414 (visium_hard_regno_nregs): New function.
2415 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2416 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2417 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2418 xtensa_hard_regno_nregs): New function.
2419 * system.h (HARD_REGNO_NREGS): Poison.
2421 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2423 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2424 hard_regno_nregs instead of HARD_REGNO_NREGS.
2425 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2426 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2427 (c6x_expand_epilogue): Likewise.
2428 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2429 (frv_read_iacc_argument): Likewise.
2430 * config/sh/sh.c: Include regs.h.
2431 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2432 (regs_used): Likewise.
2433 (output_stack_adjust): Likewise.
2434 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2435 * expmed.c: Include regs.h.
2436 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2437 * ree.c: Include regs.h.
2438 (combine_reaching_defs): Use hard_regno_nregs instead of
2440 (add_removable_extension): Likewise.
2442 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2444 * regs.h (hard_regno_nregs): Turn into a function.
2445 (end_hard_regno): Update accordingly.
2446 * caller-save.c (setup_save_areas): Likewise.
2447 (save_call_clobbered_regs): Likewise.
2448 (replace_reg_with_saved_mem): Likewise.
2449 (insert_restore): Likewise.
2450 (insert_save): Likewise.
2451 * combine.c (can_change_dest_mode): Likewise.
2452 (move_deaths): Likewise.
2453 (distribute_notes): Likewise.
2454 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2455 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2456 (rs6000_split_multireg_move): Likewise.
2457 (rs6000_register_move_cost): Likewise.
2458 (rs6000_memory_move_cost): Likewise.
2459 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2460 (rs6000_split_multireg_move): Likewise.
2461 (rs6000_register_move_cost): Likewise.
2462 (rs6000_memory_move_cost): Likewise.
2463 * cselib.c (cselib_reset_table): Likewise.
2464 (cselib_lookup_1): Likewise.
2465 * emit-rtl.c (set_mode_and_regno): Likewise.
2466 * function.c (aggregate_value_p): Likewise.
2467 * ira-color.c (setup_profitable_hard_regs): Likewise.
2468 (check_hard_reg_p): Likewise.
2469 (calculate_saved_nregs): Likewise.
2470 (assign_hard_reg): Likewise.
2471 (improve_allocation): Likewise.
2472 (calculate_spill_cost): Likewise.
2473 * ira-emit.c (modify_move_list): Likewise.
2474 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2475 (ira_hard_reg_in_set_p): Likewise.
2476 * ira.c (setup_reg_mode_hard_regset): Likewise.
2477 (clarify_prohibited_class_mode_regs): Likewise.
2478 (check_allocation): Likewise.
2479 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2480 (lra_setup_reg_renumber): Likewise.
2481 (setup_try_hard_regno_pseudos): Likewise.
2482 (spill_for): Likewise.
2483 (assign_hard_regno): Likewise.
2484 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2485 * lra-constraints.c (in_class_p): Likewise.
2486 (lra_constraint_offset): Likewise.
2487 (simplify_operand_subreg): Likewise.
2488 (lra_constraints): Likewise.
2489 (split_reg): Likewise.
2490 (split_if_necessary): Likewise.
2491 (invariant_p): Likewise.
2492 (inherit_in_ebb): Likewise.
2493 * lra-lives.c (process_bb_lives): Likewise.
2494 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2495 (get_hard_regs): Likewise.
2496 (do_remat): Likewise.
2497 * lra-spills.c (assign_spill_hard_regs): Likewise.
2498 * mode-switching.c (create_pre_exit): Likewise.
2499 * postreload.c (reload_combine_recognize_pattern): Likewise.
2500 * recog.c (peep2_find_free_register): Likewise.
2501 * regcprop.c (kill_value_regno): Likewise.
2502 (set_value_regno): Likewise.
2503 (copy_value): Likewise.
2504 (maybe_mode_change): Likewise.
2505 (find_oldest_value_reg): Likewise.
2506 (copyprop_hardreg_forward_1): Likewise.
2507 * regrename.c (check_new_reg_p): Likewise.
2508 (regrename_do_replace): Likewise.
2509 * reload.c (push_reload): Likewise.
2510 (combine_reloads): Likewise.
2511 (find_dummy_reload): Likewise.
2512 (operands_match_p): Likewise.
2513 (find_reloads): Likewise.
2514 (find_equiv_reg): Likewise.
2515 (reload_adjust_reg_for_mode): Likewise.
2516 * reload1.c (count_pseudo): Likewise.
2517 (count_spilled_pseudo): Likewise.
2518 (find_reg): Likewise.
2519 (clear_reload_reg_in_use): Likewise.
2520 (free_for_value_p): Likewise.
2521 (allocate_reload_reg): Likewise.
2522 (choose_reload_regs): Likewise.
2523 (reload_adjust_reg_for_temp): Likewise.
2524 (emit_reload_insns): Likewise.
2525 (delete_output_reload): Likewise.
2526 * rtlanal.c (subreg_get_info): Likewise.
2527 * sched-deps.c (sched_analyze_reg): Likewise.
2528 * sel-sched.c (init_regs_for_mode): Likewise.
2529 (mark_unavailable_hard_regs): Likewise.
2530 (choose_best_reg_1): Likewise.
2531 (verify_target_availability): Likewise.
2532 * valtrack.c (dead_debug_insert_temp): Likewise.
2533 * var-tracking.c (track_loc_p): Likewise.
2534 (emit_note_insn_var_location): Likewise.
2535 * varasm.c (make_decl_rtl): Likewise.
2536 * reginfo.c (choose_hard_reg_mode): Likewise.
2537 (init_reg_modes_target): Refer directly to
2538 this_target_regs->x_hard_regno_nregs.
2540 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2542 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2543 instead of hard_regno_nregs.
2545 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2547 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2548 end_hard_regno instead of hard_regno_nregs.
2549 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2550 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2551 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2552 * ira-color.c (improve_allocation): Likewise.
2553 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2554 * lra-lives.c (mark_regno_live): Likewise.
2555 (mark_regno_dead): Likewise.
2556 * lra-remat.c (operand_to_remat): Likewise.
2557 * lra.c (collect_non_operand_hard_regs): Likewise.
2558 * postreload.c (reload_combine_note_store): Likewise.
2559 (move2add_valid_value_p): Likewise.
2560 * reload.c (regno_clobbered_p): Likewise.
2562 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2564 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2566 * config/v850/v850.c (v850_reorg): Likewise.
2567 * reload.c (refers_to_regno_for_reload_p): Likewise.
2568 (find_equiv_reg): Likewise.
2569 * reload1.c (reload_reg_reaches_end_p): Likewise.
2571 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2573 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2575 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2576 * config/arm/arm.c (output_move_neon): Likewise.
2577 (arm_attr_length_move_neon): Likewise.
2578 (neon_split_vcombine): Likewise.
2579 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2580 (c6x_mark_reg_written): Likewise.
2581 (c6x_dwarf_register_span): Likewise.
2582 * config/i386/i386.c (ix86_save_reg): Likewise.
2583 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2584 (rws_access_reg): Likewise.
2585 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2586 * mode-switching.c (create_pre_exit): Likewise.
2587 * ree.c (combine_reaching_defs): Likewise.
2588 (add_removable_extension): Likewise.
2589 * regcprop.c (find_oldest_value_reg): Likewise.
2590 (copyprop_hardreg_forward_1): Likewise.
2591 * reload.c (reload_inner_reg_of_subreg): Likewise.
2592 (push_reload): Likewise.
2593 (combine_reloads): Likewise.
2594 (find_dummy_reload): Likewise.
2595 (reload_adjust_reg_for_mode): Likewise.
2596 * reload1.c (find_reload_regs): Likewise.
2597 (forget_old_reloads_1): Likewise.
2598 (reload_reg_free_for_value_p): Likewise.
2599 (reload_adjust_reg_for_temp): Likewise.
2600 (emit_reload_insns): Likewise.
2601 (delete_output_reload): Likewise.
2602 * sel-sched.c (choose_best_reg_1): Likewise.
2603 (choose_best_pseudo_reg): Likewise.
2605 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2606 Alan Hayward <alan.hayward@arm.com>
2607 David Sherwood <david.sherwood@arm.com>
2609 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2610 * target.def (slow_unaligned_access): New hook.
2611 * targhooks.h (default_slow_unaligned_access): Declare.
2612 * targhooks.c (default_slow_unaligned_access): New function.
2613 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2614 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2615 * doc/tm.texi: Regenerate.
2616 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2617 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2618 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2620 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2621 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2623 (rs6000_slow_unaligned_access): New function.
2624 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2625 (expand_block_compare): Likewise.
2626 (expand_strn_compare): Likewise.
2627 (rs6000_rtx_costs): Likewise.
2628 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2629 (riscv_slow_unaligned_access): Likewise.
2630 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2631 (riscv_slow_unaligned_access_p): ...this and make static.
2632 (riscv_option_override): Update accordingly.
2633 (riscv_slow_unaligned_access): New function.
2634 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2635 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2636 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2637 (rs6000_slow_unaligned_access): New function.
2638 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2639 (rs6000_rtx_costs): Likewise.
2640 * config/rs6000/rs6000-string.c (expand_block_compare)
2641 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2642 of SLOW_UNALIGNED_ACCESS.
2643 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2644 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2645 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2646 of SLOW_UNALIGNED_ACCESS.
2647 * expmed.c (simple_mem_bitfield_p): Likewise.
2648 * expr.c (alignment_for_piecewise_move): Likewise.
2649 (emit_group_load_1): Likewise.
2650 (emit_group_store): Likewise.
2651 (copy_blkmode_from_reg): Likewise.
2652 (emit_push_insn): Likewise.
2653 (expand_assignment): Likewise.
2654 (store_field): Likewise.
2655 (expand_expr_real_1): Likewise.
2656 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2657 * lra-constraints.c (simplify_operand_subreg): Likewise.
2658 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2659 * gimple-ssa-store-merging.c: Likewise in block comment at start
2661 * tree-ssa-strlen.c: Include target.h.
2662 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2663 of SLOW_UNALIGNED_ACCESS.
2664 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2666 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2668 PR rtl-optimization/82185
2669 * expmed.c (emit_store_flag_int): Only test tem if it has been
2672 2017-09-12 Richard Biener <rguenther@suse.de>
2675 * match.pd ((FTYPE) N CMP CST): Fix typo.
2677 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2679 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2681 (mips_near_type_p): Add 'short_call' attribute as a synonym
2683 * doc/extend.texi (short_call): Document new function attribute.
2685 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2688 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2689 assertion check that in the condition.
2690 (get_atomic_generic_size): Likewise. Before testing if parameter
2691 has pointer type, if it has array type, call for C++
2692 default_conversion to perform array-to-pointer conversion.
2694 2017-09-12 Richard Biener <rguenther@suse.de>
2696 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2697 for operations we cannot scalarize.
2699 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2701 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2702 vectors heap vectors. Clean up comments.
2703 Make visited_bbs a reference.
2704 (profitable_jump_thread_path): Make GC
2705 vectors heap vectors. Clean up comments.
2707 (convert_and_register_jump_thread_path): Make GC vectors heap
2709 (check_subpath_and_update_thread_path): Same. Clean up comments.
2710 Make visited_bbs a reference.
2711 (handle_phi): Abstract common code to to
2712 register_jump_thread_path_if_profitable.
2713 Rename VAR_BB to DEF_BB.
2715 Make GC vectors heap vectors.
2716 Make visited_bbs a reference.
2717 (handle_assignment): Same.
2718 (register_jump_thread_path_if_profitable): New.
2719 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2721 Make GC vectors heap vectors. Clean up comments.
2722 Make visited_bbs a reference.
2723 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2724 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2727 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2730 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2731 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2733 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2735 Revert r251800 and r251799.
2737 2017-09-11 Martin Jambor <mjambor@suse.cz>
2740 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2741 arguments in advance.
2742 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2743 use it to find predecessor edges.
2744 (naive_outof_ssa): Collect vector of predecessors.
2746 2017-09-08 Jason Merrill <jason@redhat.com>
2748 PR c++/70029 - ICE with ref-qualifier and -flto
2749 * langhooks.h (struct lang_hooks_for_types): Add
2750 copy_lang_qualifiers.
2751 * attribs.c (build_type_attribute_qual_variant): Use it.
2752 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2754 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2755 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2757 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2760 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2761 (*mulsi3_sp64): New instruction.
2762 (mulsi3): New expander.
2764 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2766 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2768 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2770 * sancov.c: Include memmodel.h.
2772 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2775 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2778 2017-09-07 Carl Love <cel@us.ibm.com>
2780 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2781 the sldi instruction.
2783 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2785 * sancov.c: Include tm_p.h.
2787 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2790 * output.h (switch_to_other_text_partition): New declaration.
2791 * varasm.c (switch_to_other_text_partition): New function.
2792 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
2793 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
2794 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
2795 to the other text partition before emitting LCL label and switch back
2796 after emitting the word after it.
2798 2017-09-07 Richard Biener <rguenther@suse.de>
2800 * passes.def (pass_split_crit_edges): Remove instance before PRE.
2801 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
2802 critical edges here, after loop init.
2803 (pass_data_pre): Remove PROP_no_crit_edges flags.
2804 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
2805 for valueization of call args to avoid leaking VN_TOP.
2806 (visit_use): Assert we do not visit default defs.
2807 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
2808 Use error_mark_node to more easily detect leaking VN_TOP.
2809 All default-defs are varying, not VN_TOP. Mark them visited.
2810 (run_scc_vn): Make code match comment.
2812 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2814 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
2815 OPTION_MASK_FLOAT128_KEYWORD.
2816 (POWERPC_MASKS): Likewise.
2817 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2818 support for the -mfloat128-type option, and make -mfloat128
2819 default on PowerPC Linux systems. Define or undefine
2820 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2821 Define __float128 to be __ieee128 if IEEE 128-bit support is
2822 enabled, or undefine it.
2823 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2824 Delete defining __FLOAT128_TYPE__.
2825 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2826 -mfloat128-type option and make -mfloat128 default on PowerPC
2828 (TARGET_FLOAT128_TYPE): Likewise.
2829 (-mfloat128-type): Likewise.
2830 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2831 Delete the -mfloat128-type option and make -mfloat128 default on
2832 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2833 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2834 128-bit floating point is enabled. Change tests from using
2835 -mfloat128-type to -mfloat128.
2836 (rs6000_mangle_type): Use the correct mangling for the __float128
2837 type even if normal long double is restricted to 64-bits.
2838 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2840 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2841 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2842 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2844 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2845 documentation for -mfloat128.
2847 2017-09-06 Olivier Hainque <hainque@adacore.com>
2849 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2851 2017-09-06 Wish Wu <wishwu007@gmail.com>
2852 Jakub Jelinek <jakub@redhat.com>
2854 * asan.c (initialize_sanitizer_builtins): Add
2855 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2856 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2857 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2858 BT_FN_VOID_UINT64_PTR variables.
2859 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2860 (BT_FN_VOID_UINT16_UINT16): Likewise.
2861 (BT_FN_VOID_UINT32_UINT32): Likewise.
2862 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2863 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2864 (BT_FN_VOID_UINT64_PTR): Likewise.
2865 * common.opt (flag_sanitize_coverage): New variable.
2866 (fsanitize-coverage=trace-pc): Remove.
2867 (fsanitize-coverage=): Add.
2868 * flag-types.h (enum sanitize_coverage_code): New enum.
2869 * fold-const.c (fold_range_test): Disable non-short-circuit
2870 optimization if flag_sanitize_coverage.
2871 (fold_truth_andor): Likewise.
2872 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2873 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2874 (coverage_sanitizer_opts): New array.
2875 (get_closest_sanitizer_option): Add OPTS argument, handle also
2876 OPT_fsanitize_coverage_.
2877 (parse_sanitizer_options): Adjusted to also handle
2878 OPT_fsanitize_coverage_.
2879 (common_handle_option): Add OPT_fsanitize_coverage_.
2880 * sancov.c (instrument_comparison, instrument_switch): New function.
2881 (sancov_pass): Add trace-cmp support.
2882 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2883 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2884 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2885 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2886 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2887 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2888 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2889 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2890 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2891 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2893 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2895 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2896 error. Only quit immediately if parsing is complete.
2897 (BEGIN): Initialize fatal_err and parse_done.
2898 (begin fpu, end fpu): Check number of arguments.
2899 (begin arch, end arch): Likewise.
2900 (begin cpu, end cpu): Likewise.
2901 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2902 (optalias): Likewise.
2904 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2906 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2907 * config/arm/arm-isa.h: Delete. Move definitions to ...
2908 * arm-cpus.in: ... here. Use new feature and fgroup values.
2909 * config/arm/arm.c (arm_option_override): Use lower case for feature
2911 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2912 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2913 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2915 (print_isa_bits_for): New function.
2916 (gen_isa): New function.
2917 (gen_comm_data): Use print_isa_bits_for.
2918 (define feature): New keyword.
2919 (define fgroup): New keyword.
2920 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2921 (arm-isa.h): Add rule to generate file.
2922 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2923 case for feature bit names.
2925 2017-09-06 Richard Biener <rguenther@suse.de>
2927 * tree-ssa-pre.c (NECESSARY): Remove.
2928 (create_expression_by_pieces): Do not touch pass-local flags.
2929 (insert_into_preds_of_block): Likewise.
2930 (do_pre_regular_insertion): Likewise.
2931 (eliminate_insert): Likewise.
2932 (eliminate_dom_walker::before_dom_children): Likewise.
2933 (fini_eliminate): Do not look at inserted_exprs.
2934 (mark_operand_necessary): Remove.
2935 (remove_dead_inserted_code): Replace with simple work-list
2936 algorithm based on inserted_exprs and SSA uses.
2937 (pass_pre::execute): Re-order fini_eliminate and
2938 remove_dead_inserted_code.
2940 2017-09-06 Olivier Hainque <hainque@adacore.com>
2942 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2943 for VxWorks 7. Adjust surrounding comments.
2945 2017-09-06 Richard Biener <rguenther@suse.de>
2947 * gimple-ssa-strength-reduction.c
2948 (find_candidates_dom_walker::before_dom_children): Also allow
2951 2017-09-06 Richard Biener <rguenther@suse.de>
2953 PR tree-optimization/82108
2954 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2955 for gap in the non-permutation SLP case.
2957 2017-09-06 Martin Jambor <mjambor@suse.cz>
2959 PR tree-optimization/82078
2960 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2961 add_access_to_work_queue...
2962 (build_accesses_from_assign): ...here.
2963 (propagate_all_subaccesses): Make sure racc is the group
2964 representative, if there is one.
2966 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2969 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2972 2017-09-06 Richard Biener <rguenther@suse.de>
2974 * gimple-ssa-strength-reduction.c
2975 (find_candidates_dom_walker::before_doom_children): Use a
2976 type and not a mode check.
2978 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2981 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2983 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2985 2017-09-05 Jeff Law <law@redhat.com>
2987 PR tree-optimization/64910
2988 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2989 cases where we have 3 or more operands.
2991 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2994 * omp-low.c (lower_omp_for): Recompute tree invariant if
2995 gimple_omp_for_initial/final is ADDR_EXPR.
2998 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2999 into gimple val before gimplification fo the COND_EXPR.
3001 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3003 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3004 REGION_COPY argument.
3005 (thread_through_all_blocks): Remove unused argument to
3006 duplicate_thread_path.
3008 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3009 Alan Hayward <alan.hayward@arm.com>
3010 David Sherwood <david.sherwood@arm.com>
3012 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3013 Take a scalar_mode rather than a machine_mode.
3014 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3015 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3016 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3017 (aarch64_gen_adjusted_ldpstp): Likewise.
3018 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3020 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3021 Alan Hayward <alan.hayward@arm.com>
3022 David Sherwood <david.sherwood@arm.com>
3024 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3025 Take a scalar_int_mode instead of a machine_mode.
3026 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3027 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3028 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3029 (aarch64_simd_attr_length_rglist): Delete.
3030 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3031 a scalar_int_mode instead of a machine_mode.
3032 (aarch64_add_offset): Likewise.
3033 (aarch64_internal_mov_immediate): Likewise
3034 (aarch64_add_constant_internal): Likewise.
3035 (aarch64_add_constant): Likewise.
3036 (aarch64_movw_imm): Likewise.
3037 (aarch64_rtx_arith_op_extract_p): Likewise.
3038 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3039 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3040 Remove assert that the mode isn't a vector.
3041 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3042 (aarch64_expand_mov_immediate): Update calls after above changes.
3043 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3044 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3045 (aarch64_move_imm): Likewise.
3046 (aarch64_can_const_movi_rtx_p): Likewise.
3047 (aarch64_strip_extend): Likewise.
3048 (aarch64_extr_rtx_p): Likewise.
3049 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3050 a CONST_INT when the mode parameter is VOIDmode.
3051 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3053 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3055 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3056 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3057 (bitwise_type_for_mode): Update accordingly.
3059 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3061 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3062 * stor-layout.c (mode_for_size_tree): Likewise.
3063 (mode_for_array): Update accordingly.
3064 (layout_decl): Likewise.
3065 (compute_record_mode): Likewise. Only set the mode once.
3067 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3069 * target.def (get_mask_mode): Change return type to opt_mode.
3071 * doc/tm.texi: Regenerate.
3072 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3073 * targhooks.c (default_get_mask_mode): Likewise.
3074 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3075 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3076 targetm.get_mask_mode.
3077 * tree.c (build_truth_vector_type): Likewise.
3079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3081 * machmode.h (mode_for_vector): Return an opt_mode.
3082 * stor-layout.c (mode_for_vector): Likewise.
3083 (mode_for_int_vector): Update accordingly.
3084 (layout_type): Likewise.
3085 * config/i386/i386.c (emit_memmov): Likewise.
3086 (ix86_expand_set_or_movmem): Likewise.
3087 (ix86_expand_vector_init): Likewise.
3088 (ix86_get_mask_mode): Likewise.
3089 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3091 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3092 * expmed.c (extract_bit_field_1): Likewise.
3093 * expr.c (expand_expr_real_2): Likewise.
3094 * optabs-query.c (can_vec_perm_p): Likewise.
3095 (can_vec_mask_load_store_p): Likewise.
3096 * optabs.c (expand_vec_perm): Likewise.
3097 * targhooks.c (default_get_mask_mode): Likewise.
3098 * tree-vect-stmts.c (vectorizable_store): Likewise.
3099 (vectorizable_load): Likewise.
3100 (get_vectype_for_scalar_type_and_size): Likewise.
3102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3104 * machmode.h (mode_for_int_vector): New function.
3105 * stor-layout.c (mode_for_int_vector): Likewise.
3106 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3107 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3108 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3109 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3110 (s390_expand_vcond): Likewise.
3112 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3114 * machmode.h (opt_machine_mode): New type.
3115 (opt_mode<T>): Allow construction from anything that can be
3117 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3118 (mode_for_size): Return an opt_machine_mode.
3119 * stor-layout.c (mode_for_size): Likewise.
3120 (mode_for_size_tree): Update call accordingly.
3121 (bitwise_mode_for_mode): Likewise.
3122 (make_fract_type): Likewise.
3123 (make_accum_type): Likewise.
3124 * caller-save.c (replace_reg_with_saved_mem): Update call
3126 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3127 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3128 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3129 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3130 * expmed.c (extract_bit_field_1): Likewise.
3131 * reload.c (get_secondary_mem): Likewise.
3132 * varasm.c (assemble_integer): Likewise.
3133 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3136 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3138 * machmode.h (decimal_float_mode_for_size): New function.
3139 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3140 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3141 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3142 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3143 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3144 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3146 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3148 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3149 (get_builtin_sync_mode): Likewise.
3150 (expand_ifn_atomic_compare_exchange): Likewise.
3151 (expand_builtin_atomic_clear): Likewise.
3152 (expand_builtin_atomic_test_and_set): Likewise.
3153 (fold_builtin_atomic_always_lock_free): Likewise.
3154 * calls.c (compute_argument_addresses): Likewise.
3155 (emit_library_call_value_1): Likewise.
3156 (store_one_arg): Likewise.
3157 * combine.c (combine_instructions): Likewise.
3158 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3159 * config/arm/arm.c (arm_function_value): Likewise.
3160 (aapcs_allocate_return_reg): Likewise.
3161 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3162 * config/i386/i386.c (construct_container): Likewise.
3163 (ix86_gimplify_va_arg): Likewise.
3164 (ix86_expand_sse_cmp): Likewise.
3165 (emit_memmov): Likewise.
3166 (emit_memset): Likewise.
3167 (expand_small_movmem_or_setmem): Likewise.
3168 (ix86_expand_pextr): Likewise.
3169 (ix86_expand_pinsr): Likewise.
3170 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3171 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3173 * config/mips/mips.c (mips_function_value_1) Likewise.
3174 (mips_block_move_straight): Likewise.
3175 (mips_expand_ins_as_unaligned_store): Likewise.
3176 * config/powerpcspe/powerpcspe.c
3177 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3178 (rs6000_darwin64_record_arg_flush): Likewise.
3179 * config/rs6000/rs6000.c
3180 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3181 (rs6000_darwin64_record_arg_flush): Likewise.
3182 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3183 (sparc_function_value_1): Likewise.
3184 * config/spu/spu.c (adjust_operand): Likewise.
3185 (spu_emit_branch_or_set): Likewise.
3186 (arith_immediate_p): Likewise.
3187 * emit-rtl.c (gen_lowpart_common): Likewise.
3188 * expr.c (expand_expr_real_1): Likewise.
3189 * function.c (assign_parm_setup_block): Likewise.
3190 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3191 * reload1.c (alter_reg): Likewise.
3192 * stor-layout.c (mode_for_vector): Likewise.
3193 (layout_type): Likewise.
3195 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3197 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3198 (spu_convert_move): Likewise.
3199 * lower-subreg.c (resolve_simple_move): Likewise.
3201 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3204 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3205 define_insn to a define_expand.
3206 (altivec_vsum2sws_direct): New define_insn.
3207 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3209 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3211 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3214 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3219 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3220 realign stack if stack alignment needed is less than incoming
3223 2017-09-05 Marek Polacek <polacek@redhat.com>
3226 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3229 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3231 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3233 2017-09-05 Richard Biener <rguenther@suse.de>
3235 PR tree-optimization/82084
3236 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3238 2017-09-05 Richard Biener <rguenther@suse.de>
3240 PR tree-optimization/82102
3241 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3243 2017-09-05 Martin Liska <mliska@suse.cz>
3245 PR tree-optimization/82032
3246 * tree-cfg.c (generate_range_test): New function.
3247 * tree-cfg.h (generate_range_test): Declared here.
3248 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3249 (cleanup_control_expr_graph): Use it.
3250 * tree-switch-conversion.c (try_switch_expansion): Remove
3252 (emit_case_nodes): Use generate_range_test.
3254 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3257 * config/i386/i386.md (*<btsc><mode>_mask): Add
3258 TARGET_USE_BT to insn constraint.
3259 (*btr<mode>_mask): Ditto.
3261 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3263 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3264 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3266 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3269 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3270 TARGET_NEON and TARGET_IWMMXT.
3271 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3272 TARGET_NEON and TARGET_IWMMXT.
3273 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3275 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3277 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3278 (ix86_rewrite_tls_address): Ditto.
3279 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3280 (ix86_rewrite_tls_address_1): Ditto.
3281 (ix86_rewrite_tls_address): Ditto.
3282 * config/i386/predicates.md (tls_address_pattern): New predicate.
3283 * config/i386/i386.md (TLS address splitter): New splitter.
3285 2017-09-04 Richard Biener <rguenther@suse.de>
3287 PR tree-optimization/82084
3288 * fold-const.h (can_native_encode_string_p): Declare.
3289 * fold-const.c (can_native_encode_string_p): Factor out from ...
3290 (native_encode_string): ... here.
3291 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3292 vectorizing stores from constants we later cannot handle.
3294 2017-09-04 Marek Polacek <polacek@redhat.com>
3297 * doc/invoke.texi: Update -Wtautological-compare documentation.
3299 2017-09-04 Jeff Law <law@redhat.com>
3301 PR tree-optimization/64910
3302 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3303 swap the first and last operand if the last is a constant.
3305 2017-09-04 Marek Polacek <polacek@redhat.com>
3308 * convert.c (do_narrow): When sanitizing signed integer overflows,
3309 bail out for signed types.
3310 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3312 2017-09-04 Richard Biener <rguenther@suse.de>
3314 PR tree-optimization/82060
3315 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3316 Move devirtualization after stmt folding and before EH/AB/noreturn
3317 cleanup to get the stmt refs canonicalized. Use a bool instead
3318 of gimple_modified_p since that doesn't work for NOPs. Schedule
3319 NOPs generated by folding for removal.
3321 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3322 Alan Hayward <alan.hayward@arm.com>
3323 David Sherwood <david.sherwood@arm.com>
3325 * coretypes.h (pad_direction): New enum.
3326 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3327 (FUNCTION_ARG_PADDING): Likewise.
3328 * target.def (function_arg_padding): New hook.
3329 * targhooks.h (default_function_arg_padding): Declare.
3330 * targhooks.c (default_function_arg_padding): New function.
3331 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3332 (TARGET_FUNCTION_ARG_PADDING): ...this.
3333 * doc/tm.texi: Regenerate.
3334 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3335 instead of direction.
3336 (compute_argument_addresses): Likewise.
3337 (load_register_parameters): Likewise.
3338 (emit_library_call_value_1): Likewise.
3339 (store_one_arg): Use targetm.calls.function_arg_padding instead
3340 of FUNCTION_ARG_PADDING.
3341 (must_pass_in_stack_var_size_or_pad): Likewise.
3342 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3343 (emit_group_store): Likewise.
3344 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3345 instead of FUNCTION_ARG_PADDING.
3346 (emit_push_insn): Likewise, and propagate enum change throughout
3348 * function.h (direction): Delete.
3349 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3351 * function.c (assign_parm_find_stack_rtl): Likewise.
3352 (assign_parm_setup_block_p): Likewise.
3353 (assign_parm_setup_block): Likewise.
3354 (gimplify_parameters): Likewise.
3355 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3356 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3358 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3359 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3360 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3361 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3362 (aarch64_function_arg_padding): ...this new function.
3363 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3364 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3365 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3366 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3367 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3368 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3369 (arm_pad_arg_upward): Replace with...
3370 (arm_function_arg_padding): ...this new function.
3371 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3373 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3374 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3375 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3376 (ia64_hpux_function_arg_padding): Replace with...
3377 (ia64_function_arg_padding): ...this new function. Use pad_direction
3378 instead of direction. Check for TARGET_HPUX.
3379 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3380 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3381 (iq2000_function_arg_padding): New function.
3382 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3383 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3384 (mips_function_arg_padding): ...this new function.
3385 (mips_pad_reg_upward): Update accordingly.
3386 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3387 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3388 targetm.calls.function_arg_padding.
3389 (FUNCTION_ARG_PADDING): Delete.
3390 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3391 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3392 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3393 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3394 (nios2_block_reg_padding): Return pad_direction instead of direction.
3395 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3396 instead of direction.
3397 (nios2_function_arg_padding): Likewise. Make static.
3398 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3399 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3400 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3401 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3402 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3403 (pa_function_arg_padding): Make static. Return pad_direction instead
3405 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3406 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3407 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3408 instead of direction. Use targetm.calls.function_arg_padding.
3409 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3410 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3411 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3412 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3413 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3415 (function_arg_padding): Rename to...
3416 (rs6000_function_arg_padding): ...this. Make static. Return
3417 pad_direction instead of direction.
3418 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3419 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3420 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3421 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3422 instead of direction. Use targetm.calls.function_arg_padding.
3423 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3424 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3425 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3426 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3427 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3428 (function_arg_padding): Rename to...
3429 (rs6000_function_arg_padding): ...this. Make static. Return
3430 pad_direction instead of direction.
3431 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3432 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3433 * config/s390/s390.c (s390_function_arg_padding): New function.
3434 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3435 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3436 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3437 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3438 (function_arg_padding): Rename to...
3439 (sparc_function_arg_padding): ...this. Make static. Return
3440 pad_direction instead of direction.
3441 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3442 * config/spu/spu.c (spu_function_arg_padding): New function.
3443 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3444 * system.h (FUNCTION_ARG_PADDING): Poison.
3446 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3447 Alan Hayward <alan.hayward@arm.com>
3448 David Sherwood <david.sherwood@arm.com>
3450 * target.def (modes_tieable_p): New hook.
3451 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3452 (TARGET_MODES_TIEABLE_P): ...this.
3453 * doc/tm.texi.in: Regenerate.
3454 * hooks.h (hook_bool_mode_mode_true): Declare.
3455 * hooks.c (hook_bool_mode_mode_true): New function.
3456 * combine.c (subst): Use targetm.modes_tieable_p instead of
3458 * dse.c (find_shift_sequence): Likewise.
3459 * expmed.c (extract_low_bits): Likewise.
3460 * lower-subreg.c: Include target.h.
3461 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3463 * rtlanal.c (rtx_cost): Likewise.
3464 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3465 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3466 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3467 (TARGET_MODES_TIEABLE_P): Redefine.
3468 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3469 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3470 (TARGET_MODES_TIEABLE_P): Redefine.
3471 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3472 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3473 (arc_modes_tieable_p): New function.
3474 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3475 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3476 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3477 (arm_modes_tieable_p): Make static.
3478 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3479 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3480 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3481 (TARGET_MODES_TIEABLE_P): Redefine.
3482 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3483 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3484 (TARGET_MODES_TIEABLE_P): Redefine.
3485 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3486 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3487 (cr16_modes_tieable_p): New function.
3488 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3489 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3490 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3491 (TRULY_NOOP_TRUNCATION): Update comment.
3492 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3493 (TRULY_NOOP_TRUNCATION): Update comment.
3494 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3495 (frv_modes_tieable_p): New function.
3496 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3497 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3498 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3499 (TARGET_MODES_TIEABLE_P): Redefine.
3500 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3501 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3502 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3503 (TARGET_MODES_TIEABLE_P): Redefine.
3504 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3505 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3506 (ia64_modes_tieable_p): New function.
3507 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3508 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3509 (iq2000_modes_tieable_p): New function.
3510 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3511 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3512 (lm32_modes_tieable_p): New function.
3513 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3514 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3515 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3516 (TARGET_MODES_TIEABLE_P): Redefine.
3517 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3518 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3519 (m32r_modes_tieable_p): New function.
3520 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3521 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3522 (m68k_modes_tieable_p): New function.
3523 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3524 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3525 (mcore_modes_tieable_p): New function.
3526 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3527 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3529 (TARGET_MODES_TIEABLE_P): Redefine.
3530 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3531 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3532 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3533 (TARGET_MODES_TIEABLE_P): Redefine.
3534 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3535 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3536 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3537 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3538 (mn10300_modes_tieable_p): ...this and make static.
3539 (TARGET_MODES_TIEABLE_P): Redefine.
3540 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3541 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3542 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3543 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3544 (msp430_modes_tieable_p): Make static.
3545 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3546 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3547 (TARGET_MODES_TIEABLE_P): Redefine.
3548 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3549 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3550 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3551 (TARGET_MODES_TIEABLE_P): Redefine.
3552 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3553 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3554 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3555 (TARGET_MODES_TIEABLE_P): Redefine.
3556 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3557 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3558 (pdp11_modes_tieable_p): New function.
3559 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3560 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3561 (rs6000_modes_tieable_p): New function.
3562 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3563 * config/powerpcspe/powerpcspe.md: Update comment.
3564 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3565 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3566 (TARGET_MODES_TIEABLE_P): Redefine.
3567 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3568 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3569 (rl78_modes_tieable_p): New function.
3570 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3571 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3572 (rs6000_modes_tieable_p): New function.
3573 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3574 * config/rs6000/rs6000.md: Update comment.
3575 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3576 * config/rx/rx.c (rx_modes_tieable_p): New function.
3577 (TARGET_MODES_TIEABLE_P): Redefine.
3578 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3579 * config/s390/s390.c (s390_modes_tieable_p): New function.
3580 (TARGET_MODES_TIEABLE_P): Redefine.
3581 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3582 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3583 (sh_modes_tieable_p): New function.
3584 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3585 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3586 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3587 (sparc_modes_tieable_p): Make static.
3588 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3589 * config/spu/spu.c (spu_modes_tieable_p): New function.
3590 (TARGET_MODES_TIEABLE_P): Redefine.
3591 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3592 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3593 (TARGET_MODES_TIEABLE_P): Redefine.
3594 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3595 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3596 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3597 * config/v850/v850.c (v850_modes_tieable_p): New function.
3598 (TARGET_MODES_TIEABLE_P): Redefine.
3599 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3600 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3601 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3602 (visium_modes_tieable_p): New function.
3603 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3604 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3605 (xtensa_modes_tieable_p): New function.
3606 * system.h (MODES_TIEABLE_P): Poison.
3608 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * target.def (hard_regno_mode_ok): New hook.
3613 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3614 (TARGET_HARD_REGNO_MODE_OK): ...this.
3615 * doc/tm.texi.in: Regenerate.
3616 * hooks.h (hook_bool_uint_mode_true): Declare.
3617 * hooks.c (hook_bool_uint_mode_true): New function.
3618 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3620 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3621 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3622 instead of HARD_REGNO_MODE_OK.
3623 * caller-save.c: Include target.h.
3624 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3626 * combine.c (can_combine_p): Likewise.
3627 (combinable_i3pat): Likewise.
3628 (can_change_dest_mode): Likewise.
3629 * expr.c (init_expr_target): Likewise.
3630 (convert_move): Likewise.
3631 (convert_modes): Likewise.
3632 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3633 (setup_prohibited_mode_move_regs): Likewise.
3634 * ira.h (target_ira): Likewise.
3635 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3636 * lra-constraints.c (process_alt_operands): Likewise.
3637 (split_reg): Likewise.
3638 * recog.c (peep2_find_free_register): Likewise.
3639 * ree.c (combine_reaching_defs): Likewise.
3640 * regcprop.c (maybe_mode_change): Likewise.
3641 * reginfo.c (init_reg_sets_1): Likewise.
3642 (choose_hard_reg_mode): Likewise.
3643 (simplifiable_subregs): Likewise.
3644 * regrename.c (check_new_reg_p): Likewise.
3645 * reload.c (find_valid_class): Likewise.
3646 (find_valid_class_1): Likewise.
3647 (reload_inner_reg_of_subreg): Likewise.
3648 (push_reload): Likewise.
3649 (combine_reloads): Likewise.
3650 (find_dummy_reload): Likewise.
3651 (find_reloads): Likewise.
3652 * reload1.c (find_reg): Likewise.
3653 (set_reload_reg): Likewise.
3654 (allocate_reload_reg): Likewise.
3655 (choose_reload_regs): Likewise.
3656 (reload_adjust_reg_for_temp): Likewise.
3657 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3658 (simplify_subreg_regno): Likewise.
3659 * sel-sched.c (init_regs_for_mode): Likewise.
3660 * varasm.c (make_decl_rtl): Likewise.
3661 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3662 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3664 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3665 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3667 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3668 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3669 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3670 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3671 (arc_mode_class): Delete.
3672 (HARD_REGNO_MODE_OK): Delete.
3673 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3674 (arc_hard_regno_mode_ok): Rename old array to...
3675 (arc_hard_regno_mode_ok_modes): ...this.
3676 (arc_conditional_register_usage): Update accordingly.
3677 (arc_mode_class): Make static.
3678 (arc_hard_regno_mode_ok): New function.
3679 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3680 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3681 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3682 (arm_hard_regno_mode_ok): Make static.
3683 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3685 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3686 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3687 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3690 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3691 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3692 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3693 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3695 * config/bfin/predicates.md (valid_reg_operand): Use
3696 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3697 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3698 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3700 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3701 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3702 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3703 (cr16_hard_regno_mode_ok): Make static and return a bool.
3704 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3705 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3706 (cris_hard_regno_mode_ok): New function.
3707 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3708 (epiphany_mode_class): Delete.
3709 (HARD_REGNO_MODE_OK): Delete.
3710 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3711 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3712 (hard_regno_mode_ok): Rename to...
3713 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3714 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3715 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3717 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3718 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3719 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3720 (frv_hard_regno_mode_ok): Make static and return a bool.
3721 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3723 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3724 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3725 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3726 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3729 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3730 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3731 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3734 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3735 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3736 (ia64_hard_regno_mode_ok): New function.
3737 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3738 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3739 (iq2000_hard_regno_mode_ok): New function.
3740 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3741 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3742 (lm32_hard_regno_mode_ok): New function.
3743 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3744 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3745 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3746 instead of HARD_REGNO_MODE_OK.
3747 (m32c_hard_regno_ok): Rename to...
3748 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3749 (m32c_cannot_change_mode_class): Update accordingly.
3750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3751 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3752 (m32r_mode_class): Delete.
3753 (HARD_REGNO_MODE_OK): Delete.
3754 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3755 (m32r_hard_regno_mode_ok): Rename to...
3756 (m32r_hard_regno_modes): ...this.
3757 (m32r_mode_class): Make static.
3758 (m32r_hard_regno_mode_ok): New function.
3759 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3760 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3761 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3762 (m68k_hard_regno_mode_ok): Make static.
3763 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3764 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3765 (mcore_hard_regno_mode_ok): New function.
3766 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3767 (HARD_REGNO_MODE_OK): Delete.
3768 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3770 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3771 (microblaze_hard_regno_mode_ok): New function.
3772 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3773 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3774 (mips_hard_regno_mode_ok): Delete.
3775 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3776 (mips_hard_regno_mode_ok_p): ...this and make static.
3777 (mips_hard_regno_mode_ok_p): Rename to...
3778 (mips_hard_regno_mode_ok_uncached): ...this.
3779 (mips_hard_regno_mode_ok): New function.
3780 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3781 of HARD_REGNO_MODE_OK.
3782 (mips_option_override): Update after above name changes.
3783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3784 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3785 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3786 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3787 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3788 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3789 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3790 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3791 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3792 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3793 (msp430_hard_regno_mode_ok): Make static and return a bool.
3794 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
3795 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
3796 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
3798 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3799 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
3800 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
3801 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
3802 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
3803 (PA_HARD_REGNO_MODE_OK): ...this
3804 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
3805 (PA_HARD_REGNO_MODE_OK): ...this.
3806 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3807 (pa_hard_regno_mode_ok): New function.
3808 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
3809 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3810 (pdp11_hard_regno_mode_ok): New function.
3811 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
3812 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
3814 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
3816 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3817 (rs6000_hard_regno_mode_ok): Rename to...
3818 (rs6000_hard_regno_mode_ok_uncached): ...this.
3819 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3820 (rs6000_hard_regno_mode_ok): New function.
3821 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3822 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3823 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3824 (riscv_hard_regno_mode_ok): ...this and make static.
3825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3826 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3827 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3828 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3829 (rl78_hard_regno_mode_ok): Make static and return bool.
3830 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3831 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3833 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3834 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3835 (rs6000_hard_regno_mode_ok): Rename to...
3836 (rs6000_hard_regno_mode_ok_uncached): ...this.
3837 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3838 (rs6000_hard_regno_mode_ok): New function.
3839 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3840 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3841 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3842 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3843 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3844 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3846 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3847 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3848 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3849 (sh_hard_regno_mode_ok): Make static.
3850 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3851 instead of HARD_REGNO_MODE_OK.
3852 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3853 (sparc_mode_class): Delete.
3854 (HARD_REGNO_MODE_OK): Delete.
3855 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3856 (hard_regno_mode_classes): Make static.
3857 (sparc_mode_class): Likewise.
3858 (sparc_hard_regno_mode_ok): New function.
3859 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3860 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3861 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3863 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3864 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3865 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3866 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3867 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3868 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3869 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3870 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3871 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3872 (visium_hard_regno_mode_ok): New function.
3873 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3874 instead of HARD_REGNO_MODE_OK.
3875 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3876 (HARD_REGNO_MODE_OK): Delete.
3877 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3878 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3879 (xtensa_option_override): Update accordingly.
3880 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3881 (xtensa_hard_regno_mode_ok): New function.
3882 * system.h (HARD_REGNO_MODE_OK): Poison.
3884 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3885 Alan Hayward <alan.hayward@arm.com>
3886 David Sherwood <david.sherwood@arm.com>
3888 * target.def (hard_regno_call_part_clobbered): New hook.
3889 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3890 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3891 * doc/tm.texi: Regenerate.
3892 * hooks.h (hook_bool_uint_mode_false): Declare.
3893 * hooks.c (hook_bool_uint_mode_false): New function.
3894 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3895 * cselib.c (cselib_process_insn): Use
3896 targetm.hard_regno_call_part_clobbered instead of
3897 HARD_REGNO_CALL_PART_CLOBBERED.
3898 * ira-conflicts.c (ira_build_conflicts): Likewise.
3899 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3900 * lra-constraints.c (need_for_call_save_p): Likewise.
3901 * lra-lives.c: Include target.h.
3902 (check_pseudos_live_through_calls): Use
3903 targetm.hard_regno_call_part_clobbered instead of
3904 HARD_REGNO_CALL_PART_CLOBBERED.
3905 * regcprop.c: Include target.h.
3906 (copyprop_hardreg_forward_1): Use
3907 targetm.hard_regno_call_part_clobbered instead of
3908 HARD_REGNO_CALL_PART_CLOBBERED.
3909 * reginfo.c (choose_hard_reg_mode): Likewise.
3910 * regrename.c (check_new_reg_p): Likewise.
3911 * reload.c (find_equiv_reg): Likewise.
3912 * reload1.c (emit_reload_insns): Likewise.
3913 * sched-deps.c (deps_analyze_insn): Likewise.
3914 * sel-sched.c (init_regs_for_mode): Likewise.
3915 (mark_unavailable_hard_regs): Likewise.
3916 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3917 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3918 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3921 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3922 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3924 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3926 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3927 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3928 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3930 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3931 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3932 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3934 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3935 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3937 * config/powerpcspe/powerpcspe.c
3938 (rs6000_hard_regno_call_part_clobbered): New function.
3939 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3940 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3941 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3943 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3944 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3945 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3947 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3948 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3949 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3951 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3952 Alan Hayward <alan.hayward@arm.com>
3953 David Sherwood <david.sherwood@arm.com>
3955 * rtl.h (subreg_memory_offset): Declare.
3956 * emit-rtl.c (subreg_memory_offset): New function.
3957 * expmed.c (store_bit_field_1): Use it.
3958 * expr.c (undefined_operand_subword_p): Likewise.
3959 * simplify-rtx.c (simplify_subreg): Likewise.
3961 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3963 PR rtl-optimization/57448
3966 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3967 using atomic_load pattern.
3968 (expand_atomic_store): Likewise.
3970 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3973 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3974 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3975 handling. Use replace_call_with_value with NULL instead of
3976 gsi_replace, unlink_stmt_vdef and release_defs.
3978 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3981 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3983 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3986 * rtl.h (emit_library_call_value_1): Declare.
3987 (emit_library_call): Replace declaration with a series of overloads.
3988 Remove the parameter count argument.
3989 (emit_library_call_value): Likewise.
3990 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3991 with an "rtx_mode_t *".
3992 (emit_library_call_value): Delete.
3993 (emit_library_call): Likewise.
3994 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3995 (asan_emit_allocas_unpoison): Likewise.
3996 * builtins.c (expand_builtin_powi): Likewise.
3997 (expand_asan_emit_allocas_unpoison): Likewise.
3998 * cfgexpand.c (expand_main_function): Likewise.
3999 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4000 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4001 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4002 * config/arm/arm.c (arm_trampoline_init): Likewise.
4003 (arm_call_tls_get_addr): Likewise.
4004 (arm_expand_divmod_libfunc): Likewise.
4005 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4006 (smulsi3_highpart): Likewise.
4007 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4008 (c6x_expand_compare): Likewise.
4009 (c6x_expand_movmem): Likewise.
4010 * config/frv/frv.c (frv_trampoline_init): Likewise.
4011 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4012 (ix86_expand_divmod_libfunc): Likewise.
4013 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4014 (ia64_expand_compare): Likewise.
4015 (ia64_profile_hook): Likewise.
4016 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4017 (nonlocal_goto): Likewise.
4018 (restore_stack_nonlocal): Likewise.
4019 * config/m32r/m32r.c (block_move_call): Likewise.
4020 (m32r_trampoline_init): Likewise.
4021 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4022 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4023 (m68k_call_m68k_read_tp): Likewise.
4024 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4025 (microblaze_expand_divide): Likewise.
4026 * config/mips/mips.h (mips_args): Likewise.
4027 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4028 (MIPS_ICACHE_SYNC): Likewise.
4029 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4030 (nios2_trampoline_init): Likewise.
4031 * config/pa/pa.c (hppa_tls_call): Likewise.
4032 (pa_trampoline_init): Likewise.
4033 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4034 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4035 (expand_strn_compare): Likewise.
4036 (rs6000_generate_compare): Likewise.
4037 (rs6000_expand_float128_convert): Likewise.
4038 (output_profile_hook): Likewise.
4039 (rs6000_trampoline_init): Likewise.
4040 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4041 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4042 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4043 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4044 (rs6000_generate_compare): Likewise.
4045 (rs6000_expand_float128_convert): Likewise.
4046 (output_profile_hook): Likewise.
4047 (rs6000_trampoline_init): Likewise.
4048 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4049 * config/sh/sh.c (sh_trampoline_init): Likewise.
4050 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4051 (sparc_emit_float_lib_cmp): Likewise.
4052 (sparc32_initialize_trampoline): Likewise.
4053 (sparc64_initialize_trampoline): Likewise.
4054 (sparc_profile_hook): Likewise.
4055 * config/spu/spu.c (ea_load_store): Likewise.
4056 * config/spu/spu.md (floatunssidf2): Likewise.
4057 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4058 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4059 * config/visium/visium.c (expand_block_move_4): Likewise.
4060 (expand_block_move_2): Likewise.
4061 (expand_block_move_1): Likewise.
4062 (expand_block_set_4): Likewise.
4063 (expand_block_set_2): Likewise.
4064 (expand_block_set_1): Likewise.
4065 (visium_trampoline_init): Likewise.
4066 (visium_profile_hook): Likewise.
4067 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4068 (xtensa_setup_frame_addresses): Likewise.
4069 (xtensa_trampoline_init): Likewise.
4070 * except.c (sjlj_emit_function_enter): Likewise.
4071 (sjlj_emit_function_exit): Likewise.
4072 * explow.c (allocate_dynamic_stack_space): Likewise.
4073 (probe_stack_range): Likewise.
4074 * expr.c (convert_mode_scalar): Likewise.
4075 * optabs.c (expand_binop): Likewise.
4076 (expand_twoval_binop_libfunc): Likewise.
4077 (expand_unop): Likewise.
4078 (prepare_cmp_insn): Likewise.
4079 (prepare_float_lib_cmp): Likewise.
4080 (expand_float): Likewise.
4081 (expand_fix): Likewise.
4082 (expand_fixed_convert): Likewise.
4083 (maybe_emit_sync_lock_test_and_set): Likewise.
4084 (expand_atomic_compare_and_swap): Likewise.
4085 (expand_mem_thread_fence): Likewise.
4086 (expand_atomic_fetch_op): Likewise.
4088 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4090 * doc/generic.texi (OpenACC): Adjust URL.
4091 * doc/invoke.texi (C Dialect Options): Ditto.
4093 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4095 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4096 predicate for operand 1. Add (m,<S>) constraint.
4097 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4098 Prevent memory operand 1 with register operand 2.
4100 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4102 PR rtl-optimization/82024
4103 * combine.c (try_combine): If the combination result is a PARALLEL,
4104 and we only need to retain the SET in there that would be placed
4105 at I2, check that we can place that at I3 instead, before doing so.
4107 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4110 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4112 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4113 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4116 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4117 Jeff Law <law@redhat.com>
4119 * varasm.c (bss_initializer_p): Do not put constants into .bss
4120 (categorize_decl_for_section): Handle bss_initializer_p returning
4121 false when DECL_INITIAL is NULL.
4123 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4126 * config/s390/s390.c (s390_can_inline_p): New function.
4128 2017-09-01 Jeff Law <law@redhat.com>
4130 PR tree-optimization/82052
4131 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4132 Always initialize the returned slot after a hash table miss
4133 when INSERT is true.
4135 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4137 * config/s390/s390.md (mem_signal_fence): Remove.
4138 * doc/md.texi (mem_signal_fence): Remove.
4139 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4141 * target-insns.def (mem_signal_fence): Remove.
4143 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4146 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4149 * asan.c (create_odr_indicator): Strip name encoding from assembler
4150 name before appending it after __odr_asan_.
4152 2017-09-01 Martin Liska <mliska@suse.cz>
4154 PR tree-optimization/82059
4155 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4156 frequency only when an edge is redirected.
4158 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4160 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4161 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4162 (arc_conditional_register_usage): Remove ARC600 lp_count
4164 (arc_file_start): Emit Tag_ARC_CPU_variation.
4165 (arc_can_use_doloop_p): New conditions to use ZOLs.
4166 (hwloop_fail): New function.
4167 (hwloop_optimize): Likewise.
4168 (hwloop_pattern_reg): Likewise.
4169 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4170 (arc_reorg_loops): New function, calls reorg_loops.
4171 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4172 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4174 (arc_loop_hazard): Remove function, functionality moved into
4176 (arc_hazard): Remove arc_loop_hazard call.
4177 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4178 into hwloop_optimize.
4179 (arc_label_align): Remove ZOL handling.
4180 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4181 * config/arc/arc.md (doloop_begin): Remove pattern.
4182 (doloop_begin_i): Likewise.
4183 (doloop_end_i): Likewise.
4184 (doloop_fallback): Likewise.
4185 (doloop_fallback_m): Likewise.
4186 (doloop_end): Reimplement expand.
4187 (arc_lp): New pattern for LP instruction.
4188 (loop_end): New pattern.
4189 (loop_fail): Likewise.
4190 (decrement_and_branch_until_zero): Likewise.
4191 * config/arc/arc.opt (mlpc-width): New option.
4192 * doc/invoke.texi (mlpc-width): Document option.
4194 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4196 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4197 (arc_ccfsm_advance): Fix checking for delay slots.
4198 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4200 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4202 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4204 (movhi_insn): Update store instruction constraint which are saving
4205 6-bit short immediates.
4206 (movsi_insn): Consider also short scaled load operations.
4207 (zero_extendhisi2_i): Use Usd constraint instead of T.
4208 (extendhisi2_i): Add q constraint.
4209 (arc_clzsi2): Add type and length attributes.
4210 (arc_ctzsi2): Likewise.
4211 * config/arc/constraints.md (Usc): Update constraint, the
4212 assembler can parse two relocations for a single instruction.
4214 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4216 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4217 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4219 2017-08-31 Olivier Hainque <hainque@adacore.com>
4221 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4222 match as powerpc-wrs-vxworks*.
4224 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4226 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4227 register constraint for by-element operand.
4228 (aarch64_mls_elt_merge<mode>): Likewise.
4230 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4232 * config/arc/arc.c (arc_can_follow_jump): Check for short
4235 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4237 * config.gcc: Use g.opt for arc.
4238 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4239 functionality moved to ...
4240 (legitimate_scaled_address_p): New function, ...here.
4241 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4242 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4243 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4245 (arc_override_options): Handle G option.
4246 (arc_output_pic_addr_const): Correct function definition.
4247 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4248 (arc_decl_anon_ns_mem_p): Delete.
4249 (arc_in_small_data_p): Overhaul this function to take into
4250 consideration the value given via G option.
4251 (arc_rewrite_small_data_1): Renamed and corrected old
4252 arc_rewrite_small_data function.
4253 (arc_rewrite_small_data): New function.
4254 (small_data_pattern): Don't use pic_offset_table_rtx.
4255 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4256 * config/arc/simdext.md (movmisalignv2hi): Use
4257 prepare_move_operands function.
4259 (movmisalign*): Likewise.
4260 * doc/invoke.texi (ARC options): Document -G option.
4262 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4264 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4266 * config/arc/arc.c (arc_print_operand): Output scalled address for
4267 sdata whenever is possible.
4268 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4269 load/stores are available.
4270 (compact_sda_memory_operand): Check for the alignment required by
4271 code density instructions.
4272 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4274 * config/arc/constraints.md (Usd): Update constraint.
4275 (Us0): New constraint.
4276 (Usc): Update constraint.
4278 2017-08-31 Richard Biener <rguenther@suse.de>
4281 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4284 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4286 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4287 Resize type_signature.
4289 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4290 Alan Hayward <alan.hayward@arm.com>
4291 David Sherwood <david.sherwood@arm.com>
4293 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4294 subregs whose inner modes can be stored in GPRs.
4295 (aarch64_classify_index): Likewise.
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/iterators.md (V_cmp_result): Rename to...
4302 (V_INT_EQUIV): ...this.
4303 (v_cmp_result): Rename to...
4304 (v_int_equiv): ...this.
4305 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4306 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4307 (copysign<mode>3): Likewise.
4308 (aarch64_simd_bsl<mode>_internal): Likewise.
4309 (aarch64_simd_bsl<mode>): Likewise.
4310 (vec_cmp<mode><mode>): Likewise.
4311 (vcond<mode><mode>): Likewise.
4312 (vcond<v_cmp_mixed><mode>): Likewise.
4313 (vcondu<mode><v_cmp_mixed>): Likewise.
4314 (aarch64_cm<optab><mode>): Likewise.
4315 (aarch64_cmtst<mode>): Likewise.
4316 (aarch64_fac<optab><mode>): Likewise.
4317 (vec_perm_const<mode>): Likewise.
4318 (vcond_mask_<mode><v_cmp_result>): Rename to...
4319 (vcond_mask_<mode><v_int_equiv>): ...this.
4320 (vec_cmp<mode><v_cmp_result>): Rename to...
4321 (vec_cmp<mode><v_int_equiv>): ...this.
4323 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4324 Alan Hayward <alan.hayward@arm.com>
4325 David Sherwood <david.sherwood@arm.com>
4327 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4329 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4330 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4331 (UNSPEC_LD4_DREG): New unspecs.
4332 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4333 (aarch64_ld2<mode>_dreg_be): Replace with...
4334 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4336 (aarch64_ld3<mode>_dreg_le)
4337 (aarch64_ld3<mode>_dreg_be): Replace with...
4338 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4340 (aarch64_ld4<mode>_dreg_le)
4341 (aarch64_ld4<mode>_dreg_be): Replace with...
4342 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4345 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4347 PR tree-optimization/81987
4348 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4349 insert an initializer in a location not dominated by the stride
4352 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4354 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4355 on the entire header of the finally block in the fallthru case.
4357 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4359 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4361 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4363 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4364 rs6000_emit_move_from_cr and call renamed function.
4365 (rs6000_emit_prologue): Call renamed functions.
4366 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4367 movesi_from_cr, remove volatile CRs.
4369 2017-08-30 Jon Beniston <jon@beniston.com>
4370 Richard Biener <rguenther@suse.de>
4372 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4373 of VECTOR_MODE_P check.
4374 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4375 element vector types.
4377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4379 * df.h (df_read_modify_subreg_p): Remove in favor of...
4380 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4381 const_rtx instead of an rtx.
4382 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4383 * df-problems.c (df_word_lr_mark_ref): Likewise.
4384 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4385 (mark_pseudo_reg_dead): Likewise.
4386 (mark_ref_dead): Likewise.
4387 * reginfo.c (init_subregs_of_mode): Likewise.
4388 * sched-deps.c (sched_analyze_1): Likewise.
4389 * df-scan.c (df_def_record_1): Likewise.
4390 (df_uses_record): Likewise.
4391 (df_read_modify_subreg_p): Remove in favor of...
4392 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4393 const_rtx instead of an rtx.
4395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4396 Alan Hayward <alan.hayward@arm.com>
4397 David Sherwood <david.sherwood@arm.com>
4399 * rtl.h (partial_subreg_p): New function.
4400 * caller-save.c (save_call_clobbered_regs): Use it.
4401 * calls.c (expand_call): Likewise.
4402 * combine.c (combinable_i3pat): Likewise.
4403 (simplify_set): Likewise.
4404 (make_extraction): Likewise.
4405 (make_compound_operation_int): Likewise.
4406 (gen_lowpart_or_truncate): Likewise.
4407 (force_to_mode): Likewise.
4408 (make_field_assignment): Likewise.
4409 (reg_truncated_to_mode): Likewise.
4410 (record_truncated_value): Likewise.
4411 (move_deaths): Likewise.
4412 * cse.c (record_jump_cond): Likewise.
4413 (cse_insn): Likewise.
4414 * cselib.c (cselib_lookup_1): Likewise.
4415 * expmed.c (extract_bit_field_using_extv): Likewise.
4416 * function.c (assign_parm_setup_reg): Likewise.
4417 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4418 * ira-build.c (create_insn_allocnos): Likewise.
4419 * lra-coalesce.c (merge_pseudos): Likewise.
4420 * lra-constraints.c (match_reload): Likewise.
4421 (simplify_operand_subreg): Likewise.
4422 (curr_insn_transform): Likewise.
4423 * lra-lives.c (process_bb_lives): Likewise.
4424 * lra.c (new_insn_reg): Likewise.
4425 (lra_substitute_pseudo): Likewise.
4426 * regcprop.c (mode_change_ok): Likewise.
4427 (maybe_mode_change): Likewise.
4428 (copyprop_hardreg_forward_1): Likewise.
4429 * reload.c (push_reload): Likewise.
4430 (find_reloads): Likewise.
4431 (find_reloads_subreg_address): Likewise.
4432 * reload1.c (alter_reg): Likewise.
4433 (eliminate_regs_1): Likewise.
4434 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4436 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4438 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4439 back to if statements, including unpack.
4441 2017-08-30 Martin Liska <mliska@suse.cz>
4444 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4446 (func_checker::compare_asm_inputs_outputs): ... this function.
4447 (func_checker::compare_gimple_asm): Use the function to compare
4448 also ASM constrains.
4449 * ipa-icf-gimple.h: Rename the function.
4451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4452 Alan Hayward <alan.hayward@arm.com>
4453 David Sherwood <david.sherwood@arm.com>
4455 * coretypes.h (complex_mode): New type.
4456 * gdbhooks.py (build_pretty_printer): Handle it.
4457 * machmode.h (complex_mode): New class.
4458 (complex_mode::includes_p): New function.
4459 (is_complex_int_mode): Likewise.
4460 (is_complex_float_mode): Likewise.
4461 * genmodes.c (get_mode_class): Handle complex mode classes.
4462 * function.c (expand_function_end): Use is_complex_int_mode.
4464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4465 Alan Hayward <alan.hayward@arm.com>
4466 David Sherwood <david.sherwood@arm.com>
4468 * coretypes.h (scalar_mode_pod): New typedef.
4469 * gdbhooks.py (build_pretty_printer): Handle it.
4470 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4471 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4472 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4473 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4477 Alan Hayward <alan.hayward@arm.com>
4478 David Sherwood <david.sherwood@arm.com>
4480 * machmode.h (mode_for_vector): Take a scalar_mode instead
4482 * stor-layout.c (mode_for_vector): Likewise.
4483 * explow.c (promote_mode): Use as_a <scalar_mode>.
4484 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
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 (preferred_simd_mode): Take a scalar_mode
4491 instead of a machine_mode.
4492 * targhooks.h (default_preferred_simd_mode): Likewise.
4493 * targhooks.c (default_preferred_simd_mode): Likewise.
4494 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4495 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4496 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4497 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4498 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4499 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4500 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4501 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4503 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4504 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4505 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4506 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4507 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4508 * doc/tm.texi: Regenerate.
4509 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4513 Alan Hayward <alan.hayward@arm.com>
4514 David Sherwood <david.sherwood@arm.com>
4516 * target.def (scalar_mode_supported_p): Take a scalar_mode
4517 instead of a machine_mode.
4518 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4519 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4520 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4521 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4522 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4523 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4524 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4525 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4526 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4527 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4528 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4529 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4530 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4531 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4533 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4534 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4535 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4536 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4537 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4539 * doc/tm.texi: Regenerate.
4541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4542 Alan Hayward <alan.hayward@arm.com>
4543 David Sherwood <david.sherwood@arm.com>
4545 * coretypes.h (opt_scalar_mode): New typedef.
4546 * gdbhooks.py (build_pretty_printers): Handle it.
4547 * machmode.h (mode_iterator::get_2xwider): Add overload for
4549 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4551 * expr.c (convert_mode_scalar): Likewise.
4552 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4553 * optabs.c (expand_float): Likewise.
4554 (expand_fix): Likewise.
4555 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4558 Alan Hayward <alan.hayward@arm.com>
4559 David Sherwood <david.sherwood@arm.com>
4561 * optabs.c (expand_float): Explicitly check for scalars before
4562 using a branching expansion.
4563 (expand_fix): Likewise.
4565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4566 Alan Hayward <alan.hayward@arm.com>
4567 David Sherwood <david.sherwood@arm.com>
4569 * expr.c (convert_mode): Split scalar handling out into...
4570 (convert_mode_scalar): ...this new function. Treat the modes
4573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4574 Alan Hayward <alan.hayward@arm.com>
4575 David Sherwood <david.sherwood@arm.com>
4577 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4579 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4582 Alan Hayward <alan.hayward@arm.com>
4583 David Sherwood <david.sherwood@arm.com>
4585 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4586 rather than a machine_mode.
4587 (fixed_from_string): Likewise.
4588 (fixed_convert): Likewise.
4589 (fixed_convert_from_int): Likewise.
4590 (fixed_convert_from_real): Likewise.
4591 (real_convert_from_fixed): Likewise.
4592 * fixed-value.c (fixed_from_double_int): Likewise.
4593 (fixed_from_string): Likewise.
4594 (fixed_convert): Likewise.
4595 (fixed_convert_from_int): Likewise.
4596 (fixed_convert_from_real): Likewise.
4597 (real_convert_from_fixed): Likewise.
4598 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4601 Alan Hayward <alan.hayward@arm.com>
4602 David Sherwood <david.sherwood@arm.com>
4604 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4605 of separate mode class checks. Do not allow vector modes here.
4606 (immed_wide_int_const): Use as_a <scalar_mode>.
4607 * explow.c (trunc_int_for_mode): Likewise.
4608 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4609 (wi::shwi): Likewise.
4610 (wi::min_value): Likewise.
4611 (wi::max_value): Likewise.
4612 * dwarf2out.c (loc_descriptor): Likewise.
4613 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4617 Alan Hayward <alan.hayward@arm.com>
4618 David Sherwood <david.sherwood@arm.com>
4620 * tree.h (SCALAR_TYPE_MODE): New macro.
4621 * expr.c (expand_expr_addr_expr_1): Use it.
4622 (expand_expr_real_2): Likewise.
4623 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4624 (fold_convert_const_fixed_from_int): Likewise.
4625 (fold_convert_const_fixed_from_real): Likewise.
4626 (native_encode_fixed): Likewise
4627 (native_encode_complex): Likewise
4628 (native_encode_vector): Likewise.
4629 (native_interpret_fixed): Likewise.
4630 (native_interpret_real): Likewise.
4631 (native_interpret_complex): Likewise.
4632 (native_interpret_vector): Likewise.
4633 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4634 (simd_clone_adjust_argument_types): Likewise.
4635 (simd_clone_init_simd_arrays): Likewise.
4636 (simd_clone_adjust): Likewise.
4637 * stor-layout.c (layout_type): Likewise.
4638 * tree.c (build_minus_one_cst): Likewise.
4639 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4640 * tree-inline.c (estimate_move_cost): Likewise.
4641 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4642 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4643 (vectorizable_reduction): Likewise.
4644 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4645 (vect_recog_mixed_size_cond_pattern): Likewise.
4646 (check_bool_pattern): Likewise.
4647 (adjust_bool_pattern): Likewise.
4648 (search_type_for_mask_1): Likewise.
4649 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4650 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4651 (vectorizable_load): Likewise.
4652 (vectorizable_store): Likewise.
4653 * ubsan.c (ubsan_encode_value): Likewise.
4654 * varasm.c (output_constant): Likewise.
4656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4657 Alan Hayward <alan.hayward@arm.com>
4658 David Sherwood <david.sherwood@arm.com>
4660 * coretypes.h (scalar_mode): New class.
4661 * machmode.h (scalar_mode): Likewise.
4662 (scalar_mode::includes_p): New function.
4663 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4664 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4665 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4666 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4667 * expmed.c (store_bit_field_1): Likewise.
4668 (extract_bit_field_1): Likewise.
4669 * expr.c (write_complex_part): Likewise.
4670 (read_complex_part): Likewise.
4671 (emit_move_complex_push): Likewise.
4672 (expand_expr_real_2): Likewise.
4673 * function.c (assign_parm_setup_reg): Likewise.
4674 (assign_parms_unsplit_complex): Likewise.
4675 * optabs.c (expand_binop): Likewise.
4676 * rtlanal.c (subreg_get_info): Likewise.
4677 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4678 * varasm.c (output_constant_pool_2): Likewise.
4680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4681 Alan Hayward <alan.hayward@arm.com>
4682 David Sherwood <david.sherwood@arm.com>
4684 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4686 (expmed_mult_highpart_optab): Likewise.
4687 (expmed_mult_highpart): Likewise.
4689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4690 Alan Hayward <alan.hayward@arm.com>
4691 David Sherwood <david.sherwood@arm.com>
4693 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4694 instead of a machine_mode.
4695 (builtin_memset_read_str): Likewise.
4696 * builtins.c (c_readstr): Likewise.
4697 (builtin_memcpy_read_str): Likewise.
4698 (builtin_strncpy_read_str): Likewise.
4699 (builtin_memset_read_str): Likewise.
4700 (builtin_memset_gen_str): Likewise.
4701 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4702 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4703 instead of a machine_mode.
4704 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4706 (make_extraction): Likewise.
4707 (try_widen_shift_mode): Take and return scalar_int_modes instead
4709 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4710 a scalar_int_mode instead of a machine_mode.
4711 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4712 (avr_addr_space_pointer_mode): Likewise.
4713 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4714 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4715 (msp430_unwind_word_mode): Likewise.
4716 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4717 (spu_addr_space_pointer_mode): Likewise.
4718 (spu_addr_space_address_mode): Likewise.
4719 (spu_libgcc_cmp_return_mode): Likewise.
4720 (spu_libgcc_shift_count_mode): Likewise.
4721 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4722 (rl78_addr_space_pointer_mode): Likewise.
4723 (fl78_unwind_word_mode): Likewise.
4724 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4726 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4727 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4728 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4729 (mips_valid_pointer_mode): Likewise.
4730 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4731 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4732 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4734 (ft32_addr_space_address_mode): Likewise.
4735 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4736 scalar_int_mode instead of a machine_mode.
4737 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4739 (m32c_addr_space_address_mode): Likewise.
4740 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4741 (rs6000_eh_return_filter_mode): Likewise.
4742 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4743 (rs6000_eh_return_filter_mode): Likewise.
4744 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4745 (s390_libgcc_shift_count_mode): Likewise.
4746 (s390_unwind_word_mode): Likewise.
4747 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4749 * target.def (mode_rep_extended): Likewise.
4750 (valid_pointer_mode): Likewise.
4751 (addr_space.valid_pointer_mode): Likewise.
4752 (eh_return_filter_mode): Return a scalar_int_mode rather than
4754 (libgcc_cmp_return_mode): Likewise.
4755 (libgcc_shift_count_mode): Likewise.
4756 (unwind_word_mode): Likewise.
4757 (addr_space.pointer_mode): Likewise.
4758 (addr_space.address_mode): Likewise.
4759 * doc/tm.texi: Regenerate.
4760 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4762 (do_jump): Use scalar_int_mode for local variables.
4763 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4764 rather than a machine_mode.
4765 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4766 (scompare_loc_descriptor_wide): Likewise.
4767 (scompare_loc_descriptor_narrow): Likewise.
4768 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4770 * except.c (sjlj_emit_dispatch_table): Likewise.
4771 (expand_builtin_eh_copy_values): Likewise.
4772 * explow.c (convert_memory_address_addr_space_1): Likewise.
4773 Take a scalar_int_mode rather than a machine_mode.
4774 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4775 than a machine_mode.
4776 (memory_address_addr_space): Use scalar_int_mode for local variables.
4777 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4778 rather than a machine_mode.
4779 * expmed.c (mask_rtx): Likewise.
4780 (init_expmed_one_conv): Likewise.
4781 (expand_mult_highpart_adjust): Likewise.
4782 (extract_high_half): Likewise.
4783 (expmed_mult_highpart_optab): Likewise.
4784 (expmed_mult_highpart): Likewise.
4785 (expand_smod_pow2): Likewise.
4786 (expand_sdiv_pow2): Likewise.
4787 (emit_store_flag_int): Likewise.
4788 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4790 (extract_low_bits): Likewise.
4791 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
4793 * expr.c (pieces_addr::adjust): Likewise.
4794 (can_store_by_pieces): Likewise.
4795 (store_by_pieces): Likewise.
4796 (clear_by_pieces_1): Likewise.
4797 (expand_expr_addr_expr_1): Likewise.
4798 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
4799 (expand_expr_real_1): Likewise.
4800 (try_casesi): Likewise.
4801 * final.c (shorten_branches): Likewise.
4802 * fold-const.c (fold_convert_const_int_from_fixed): Change the
4803 type of "mode" to machine_mode.
4804 * internal-fn.c (expand_arith_overflow_result_store): Take a
4805 scalar_int_mode rather than a machine_mode.
4806 (expand_mul_overflow): Use scalar_int_mode for local variables.
4807 * loop-doloop.c (doloop_modify): Likewise.
4808 (doloop_optimize): Likewise.
4809 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
4810 than a machine_mode.
4811 (expand_doubleword_shift_condmove): Likewise.
4812 (expand_doubleword_shift): Likewise.
4813 (expand_doubleword_clz): Likewise.
4814 (expand_doubleword_popcount): Likewise.
4815 (expand_doubleword_parity): Likewise.
4816 (expand_absneg_bit): Use scalar_int_mode for local variables.
4817 (prepare_float_lib_cmp): Likewise.
4818 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4819 rather than a machine_mode.
4820 (convert_memory_address_addr_space): Likewise.
4821 (get_mode_bounds): Likewise.
4822 (get_address_mode): Return a scalar_int_mode rather than a
4824 * rtlanal.c (get_address_mode): Likewise.
4825 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4826 than a machine_mode.
4827 * targhooks.c (default_mode_rep_extended): Likewise.
4828 (default_valid_pointer_mode): Likewise.
4829 (default_addr_space_valid_pointer_mode): Likewise.
4830 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4831 than a machine_mode.
4832 (default_libgcc_cmp_return_mode): Likewise.
4833 (default_libgcc_shift_count_mode): Likewise.
4834 (default_unwind_word_mode): Likewise.
4835 (default_addr_space_pointer_mode): Likewise.
4836 (default_addr_space_address_mode): Likewise.
4837 * targhooks.h (default_eh_return_filter_mode): Likewise.
4838 (default_libgcc_cmp_return_mode): Likewise.
4839 (default_libgcc_shift_count_mode): Likewise.
4840 (default_unwind_word_mode): Likewise.
4841 (default_addr_space_pointer_mode): Likewise.
4842 (default_addr_space_address_mode): Likewise.
4843 (default_mode_rep_extended): Take a scalar_int_mode rather than
4845 (default_valid_pointer_mode): Likewise.
4846 (default_addr_space_valid_pointer_mode): Likewise.
4847 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4849 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4850 rather than a machine_mode.
4851 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4852 for local variables.
4853 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4854 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4855 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4862 the mode argument to scalar_int_mode.
4863 (do_jump_by_parts_zero_rtx): Likewise.
4864 (do_jump_by_parts_equality_rtx): Likewise.
4865 (do_jump_by_parts_greater): Take a mode argument.
4866 (do_jump_by_parts_equality): Likewise.
4867 (do_jump_1): Update calls accordingly.
4869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4870 Alan Hayward <alan.hayward@arm.com>
4871 David Sherwood <david.sherwood@arm.com>
4873 * is-a.h (safe_dyn_cast): New function.
4874 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4875 (jump_table_for_label): Likewise.
4876 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4877 instead of an rtx_insn *.
4878 (shorten_branches): Use dyn_cast instead of LABEL_P and
4879 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4880 rtx_jump_table_data::get_data_mode.
4881 (final_scan_insn): Likewise.
4883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4884 Alan Hayward <alan.hayward@arm.com>
4885 David Sherwood <david.sherwood@arm.com>
4887 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4888 trying to combine a full-register integer set with a subreg
4891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4892 Alan Hayward <alan.hayward@arm.com>
4893 David Sherwood <david.sherwood@arm.com>
4895 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4896 that is always either address_mode or pointer_mode.
4898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4899 Alan Hayward <alan.hayward@arm.com>
4900 David Sherwood <david.sherwood@arm.com>
4902 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4903 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
4910 is_a <scalar_int_mode> instead of checking for a nonzero
4911 precision. Forcibly convert op_mode to a scalar_int_mode
4912 in that case. More clearly differentiate the operand and
4913 result modes and use the former when deciding what the value
4914 of a count-bits operation should be. Use is_int_mode instead
4915 of checking for a MODE_INT. Remove redundant check for whether
4916 this mode has a zero precision.
4918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4919 Alan Hayward <alan.hayward@arm.com>
4920 David Sherwood <david.sherwood@arm.com>
4922 * optabs.c (widen_leading): Change the type of the mode argument
4923 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4924 (widen_bswap): Likewise.
4925 (expand_parity): Likewise.
4926 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4927 (expand_ffs): Likewise.
4928 (epand_unop): Check for scalar integer modes before calling the
4931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4932 Alan Hayward <alan.hayward@arm.com>
4933 David Sherwood <david.sherwood@arm.com>
4935 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4937 (expand_expr_real_1): Update call accordingly.
4939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4940 Alan Hayward <alan.hayward@arm.com>
4941 David Sherwood <david.sherwood@arm.com>
4943 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4944 value_mode arguments. Use scalar_int_mode internally.
4945 (store_bit_field_1): Rename the new integer mode from imode
4946 to op0_mode and use it instead of GET_MODE (op0). Update calls
4947 to store_split_bit_field, store_bit_field_using_insv and
4948 store_fixed_bit_field.
4949 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4950 Use scalar_int_mode internally. Use a bit count rather than a mode
4951 when calculating the largest bit size for get_best_mode.
4952 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4953 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4954 Remove assertion that OP0 has a scalar integer mode.
4955 (store_split_bit_field): Add op0_mode and value_mode arguments.
4956 Update calls to extract_fixed_bit_field.
4957 (extract_bit_field_using_extv): Add an op0_mode argument.
4958 Use scalar_int_mode internally.
4959 (extract_bit_field_1): Rename the new integer mode from imode to
4960 op0_mode and use it instead of GET_MODE (op0). Update calls to
4961 extract_split_bit_field, extract_bit_field_using_extv and
4962 extract_fixed_bit_field.
4963 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4964 to extract_split_bit_field and extract_fixed_bit_field_1.
4965 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4966 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4968 (extract_split_bit_field): Add an op0_mode argument. Update call
4969 to extract_fixed_bit_field.
4971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4972 Alan Hayward <alan.hayward@arm.com>
4973 David Sherwood <david.sherwood@arm.com>
4975 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4976 * explow.c (hard_function_value): Likewise.
4977 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4978 convert_to_mode call outside the loop.
4979 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4980 for the mode iterator. Require the mode specified by max_pieces
4982 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4984 (copy_blkmode_to_reg): Likewise.
4985 (set_storage_via_setmem): Likewise.
4986 * optabs.c (prepare_cmp_insn): Likewise.
4987 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4988 * stor-layout.c (finish_bitfield_representative): Likewise.
4990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4991 Alan Hayward <alan.hayward@arm.com>
4992 David Sherwood <david.sherwood@arm.com>
4994 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4995 * expr.c (convert_move): Use them.
4996 (convert_modes): Likewise.
4997 (store_expr_with_bounds): Likewise.
4999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5000 Alan Hayward <alan.hayward@arm.com>
5001 David Sherwood <david.sherwood@arm.com>
5003 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5004 parameter for the mode of "x". Remove the "known_x", "known_mode"
5005 and "known_ret" arguments. Change the type of the mode argument
5007 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5008 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5009 (reg_num_sign_bit_copies_for_combine): Likewise.
5010 * rtlanal.c (nonzero_bits1): Likewise.
5011 (num_sign_bit_copies1): Likewise.
5012 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5013 (reg_num_sign_bit_copies_general): Likewise.
5014 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5015 (reg_nonzero_bits_general): Likewise.
5017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5018 Alan Hayward <alan.hayward@arm.com>
5019 David Sherwood <david.sherwood@arm.com>
5021 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5022 than in subroutines. Return 1 for non-integer modes.
5023 (cached_num_sign_bit_copies): Change the type of the mode parameter
5025 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5026 classes. Handle CONST_INT_P first and then check whether X also
5027 has a scalar integer mode. Check the same thing for inner registers
5028 of a SUBREG and for values that are being extended or truncated.
5030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5031 Alan Hayward <alan.hayward@arm.com>
5032 David Sherwood <david.sherwood@arm.com>
5034 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5035 in subroutines. Return the mode mask for non-integer modes.
5036 (cached_nonzero_bits): Change the type of the mode parameter
5038 (nonzero_bits1): Likewise. Remove early exit for other mode
5039 classes. Handle CONST_INT_P first and then check whether X
5040 also has a scalar integer mode.
5042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5043 Alan Hayward <alan.hayward@arm.com>
5044 David Sherwood <david.sherwood@arm.com>
5046 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5047 Return a scalar_int_mode and assert that the size is greater than
5049 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5050 (op_by_pieces_d::op_by_pieces_d): Likewise.
5051 (op_by_pieces_d::run): Likewise.
5052 (can_store_by_pieces): Likewise.
5054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5055 Alan Hayward <alan.hayward@arm.com>
5056 David Sherwood <david.sherwood@arm.com>
5058 * combine.c (extract_left_shift): Add a mode argument and update
5060 (make_compound_operation_int): Change the type of the mode parameter
5061 to scalar_int_mode and update the call to extract_left_shift.
5063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5064 Alan Hayward <alan.hayward@arm.com>
5065 David Sherwood <david.sherwood@arm.com>
5067 * combine.c (simplify_and_const_int): Change the type of the mode
5068 parameter to scalar_int_mode.
5069 (simplify_and_const_int_1): Likewise. Update recursive call.
5071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5072 Alan Hayward <alan.hayward@arm.com>
5073 David Sherwood <david.sherwood@arm.com>
5075 * combine.c (simplify_compare_const): Check that the mode is a
5076 scalar_int_mode (rather than VOIDmode) before testing its
5078 (simplify_comparison): Move COMPARISON_P handling out of the
5079 loop and restrict the latter part of the loop to scalar_int_modes.
5080 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5081 and when considering SUBREG_REGs. Use is_int_mode instead of
5082 checking GET_MODE_CLASS against MODE_INT.
5084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5085 Alan Hayward <alan.hayward@arm.com>
5086 David Sherwood <david.sherwood@arm.com>
5088 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5089 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5090 shift_unit_mode and for modes involved in scalar shifts.
5092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5093 Alan Hayward <alan.hayward@arm.com>
5094 David Sherwood <david.sherwood@arm.com>
5096 * combine.c (force_int_to_mode): New function, split out from...
5097 (force_to_mode): ...here. Keep xmode up-to-date and use it
5098 instead of GET_MODE (x).
5100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5101 Alan Hayward <alan.hayward@arm.com>
5102 David Sherwood <david.sherwood@arm.com>
5104 * optabs-query.h (extraction_insn::struct_mode): Change type to
5105 opt_scalar_int_mode and update comment.
5106 (extraction_insn::field_mode): Change type to scalar_int_mode.
5107 (extraction_insn::pos_mode): Likewise.
5108 * combine.c (make_extraction): Update accordingly.
5109 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5110 (get_optab_extraction_insn): Likewise.
5111 * recog.c (simplify_while_replacing): Likewise.
5112 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5113 parameter to opt_scalar_int_mode.
5115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5116 Alan Hayward <alan.hayward@arm.com>
5117 David Sherwood <david.sherwood@arm.com>
5119 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5120 to a scalar_int_mode instead of a machine_mode.
5121 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5122 (get_best_mode): Return a boolean and use a pointer argument to store
5123 the selected mode. Replace the limit mode parameter with a bit limit.
5124 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5125 for the values returned by bit_field_mode_iterator::next_mode.
5126 (store_bit_field): Update call to get_best_mode.
5127 (store_fixed_bit_field): Likewise.
5128 (extract_fixed_bit_field): Likewise.
5129 * expr.c (optimize_bitfield_assignment_op): Likewise.
5130 * fold-const.c (optimize_bit_field_compare): Likewise.
5131 (fold_truth_andor_1): Likewise.
5132 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5133 Update for new type of m_mode.
5134 (get_best_mode): As above.
5136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5137 Alan Hayward <alan.hayward@arm.com>
5138 David Sherwood <david.sherwood@arm.com>
5140 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5141 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5142 (store_bit_field): Check is_a <scalar_int_mode> before calling
5143 strict_volatile_bitfield_p.
5144 (extract_bit_field): Likewise.
5146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5147 Alan Hayward <alan.hayward@arm.com>
5148 David Sherwood <david.sherwood@arm.com>
5150 * target.def (cstore_mode): Return a scalar_int_mode.
5151 * doc/tm.texi: Regenerate.
5152 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5153 * targhooks.h (default_cstore_mode): Likewise.
5154 * targhooks.c (default_cstore_mode): Likewise, using a forced
5156 * expmed.c (emit_cstore): Expect the target of the cstore to be
5159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5160 Alan Hayward <alan.hayward@arm.com>
5161 David Sherwood <david.sherwood@arm.com>
5163 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5165 (niter_desc): Likewise mode.
5166 (iv_analyze): Add a mode parameter.
5168 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5169 and change its type to scalar_int_mode.
5170 * loop-iv.c: Update commentary at head of file.
5171 (iv_constant): Pass the mode paraeter before the rtx it describes
5172 and change its type to scalar_int_mode. Remove VOIDmode handling.
5173 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5174 (iv_extend): Likewise.
5175 (shorten_into_mode): Likewise.
5176 (iv_add): Use scalar_int_mode.
5177 (iv_mult): Likewise.
5178 (iv_shift): Likewise.
5179 (canonicalize_iv_subregs): Likewise.
5180 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5181 it describes and change its mode to scalar_int_mode. Also change
5182 the type of the returned inner_mode to scalar_int_mode.
5183 (get_biv_step): Likewise, turning outer_mode from a pointer
5184 into a direct parameter. Update call to get_biv_step_1.
5185 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5186 iv_constant and get_biv_step.
5187 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5188 and change its type to scalar_int_mode. Don't initialise iv->mode
5189 to VOIDmode and remove later checks for its still being VOIDmode.
5190 Update calls to iv_analyze_op and iv_analyze_expr. Check
5191 is_a <scalar_int_mode> when changing the mode under consideration.
5192 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5193 Update call to iv_analyze_expr.
5194 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5195 inner register is not also a scalar_int_mode. Update call to
5197 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5198 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5199 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5200 separate mode class checks. Update calls to iv_analyze. Remove
5201 fix-up of VOIDmodes after iv_analyze_biv.
5202 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5203 don't have a scalar_int_mode. Update call to biv_p.
5205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5206 Alan Hayward <alan.hayward@arm.com>
5207 David Sherwood <david.sherwood@arm.com>
5209 * cfgexpand.c (convert_debug_memory_address): Use
5210 as_a <scalar_int_mode>.
5211 * combine.c (expand_compound_operation): Likewise.
5212 (make_extraction): Likewise.
5213 (change_zero_ext): Likewise.
5214 (simplify_comparison): Likewise.
5215 * cse.c (cse_insn): Likewise.
5216 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5217 (mem_loc_descriptor): Likewise.
5218 (loc_descriptor): Likewise.
5219 * expmed.c (init_expmed_one_mode): Likewise.
5220 (synth_mult): Likewise.
5221 (emit_store_flag_1): Likewise.
5222 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5223 of a comparison with size.
5224 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5225 (reduce_to_bit_field_precision): Likewise.
5226 * function.c (expand_function_end): Likewise.
5227 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5228 * loop-doloop.c (doloop_modify): Likewise.
5229 * optabs.c (expand_binop): Likewise.
5230 (expand_unop): Likewise.
5231 (expand_copysign_absneg): Likewise.
5232 (prepare_cmp_insn): Likewise.
5233 (maybe_legitimize_operand): Likewise.
5234 * recog.c (const_scalar_int_operand): Likewise.
5235 * rtlanal.c (get_address_mode): Likewise.
5236 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5237 (simplify_cond_clz_ctz): Likewise.
5238 * tree-nested.c (get_nl_goto_field): Likewise.
5239 * tree.c (build_vector_type_for_mode): Likewise.
5240 * var-tracking.c (use_narrower_mode): Likewise.
5242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5243 Alan Hayward <alan.hayward@arm.com>
5244 David Sherwood <david.sherwood@arm.com>
5246 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5247 * builtins.c (expand_builtin_signbit): Use it.
5248 * cfgexpand.c (expand_debug_expr): Likewise.
5249 * dojump.c (do_jump): Likewise.
5250 (do_compare_and_jump): Likewise.
5251 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5252 * expmed.c (make_tree): Likewise.
5253 * expr.c (expand_expr_real_2): Likewise.
5254 (expand_expr_real_1): Likewise.
5255 (try_casesi): Likewise.
5256 * fold-const-call.c (fold_const_call_ss): Likewise.
5257 * fold-const.c (unextend): Likewise.
5258 (extract_muldiv_1): Likewise.
5259 (fold_single_bit_test): Likewise.
5260 (native_encode_int): Likewise.
5261 (native_encode_string): Likewise.
5262 (native_interpret_int): Likewise.
5263 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5264 * internal-fn.c (expand_addsub_overflow): Likewise.
5265 (expand_neg_overflow): Likewise.
5266 (expand_mul_overflow): Likewise.
5267 (expand_arith_overflow): Likewise.
5268 * match.pd: Likewise.
5269 * stor-layout.c (layout_type): Likewise.
5270 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5271 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5272 * tree-ssanames.c (get_range_info): Likewise.
5273 * tree-switch-conversion.c (array_value_type) Likewise.
5274 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5275 (vect_recog_divmod_pattern): Likewise.
5276 (vect_recog_mixed_size_cond_pattern): Likewise.
5277 * tree-vrp.c (extract_range_basic): Likewise.
5278 (simplify_float_conversion_using_ranges): Likewise.
5279 * tree.c (int_fits_type_p): Likewise.
5280 * ubsan.c (instrument_bool_enum_load): Likewise.
5281 * varasm.c (mergeable_string_section): Likewise.
5282 (narrowing_initializer_constant_valid_p): Likewise.
5283 (output_constant): Likewise.
5285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5286 Alan Hayward <alan.hayward@arm.com>
5287 David Sherwood <david.sherwood@arm.com>
5289 * machmode.h (NARROWEST_INT_MODE): New macro.
5290 * expr.c (alignment_for_piecewise_move): Use it instead of
5291 GET_CLASS_NARROWEST_MODE (MODE_INT).
5292 (push_block): Likewise.
5293 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5295 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5298 Alan Hayward <alan.hayward@arm.com>
5299 David Sherwood <david.sherwood@arm.com>
5301 * postreload.c (move2add_valid_value_p): Change the type of the
5302 mode parameter to scalar_int_mode.
5303 (move2add_use_add2_insn): Add a mode parameter and use it instead
5305 (move2add_use_add3_insn): Likewise.
5306 (reload_cse_move2add): Update accordingly.
5308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5309 Alan Hayward <alan.hayward@arm.com>
5310 David Sherwood <david.sherwood@arm.com>
5312 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5314 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5315 * optabs.c (expand_unop): Likewise.
5317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5318 Alan Hayward <alan.hayward@arm.com>
5319 David Sherwood <david.sherwood@arm.com>
5321 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5322 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5323 (popcount_loc_descriptor): Likewise.
5324 (bswap_loc_descriptor): Likewise.
5325 (rotate_loc_descriptor): Likewise.
5326 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5327 calling the functions above.
5329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5330 Alan Hayward <alan.hayward@arm.com>
5331 David Sherwood <david.sherwood@arm.com>
5333 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5335 (try_combine): Likewise.
5336 (simplify_if_then_else): Likewise.
5337 * cse.c (cse_insn): Likewise.
5338 * dwarf2out.c (mem_loc_descriptor): Likewise.
5339 * emit-rtl.c (gen_lowpart_common): Likewise.
5340 * simplify-rtx.c (simplify_truncation): Likewise.
5341 (simplify_binary_operation_1): Likewise.
5342 (simplify_const_relational_operation): Likewise.
5343 (simplify_ternary_operation): Likewise.
5344 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5351 * reload.c (push_reload): Likewise.
5352 (find_reloads): Likewise.
5354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5355 Alan Hayward <alan.hayward@arm.com>
5356 David Sherwood <david.sherwood@arm.com>
5358 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5359 (make_compound_operation_int): Likewise.
5360 (change_zero_ext): Likewise.
5361 * expr.c (convert_move): Likewise.
5362 (convert_modes): Likewise.
5363 * fwprop.c (forward_propagate_subreg): Likewise.
5364 * loop-iv.c (get_biv_step_1): Likewise.
5365 * optabs.c (widen_operand): Likewise.
5366 * postreload.c (move2add_valid_value_p): Likewise.
5367 * recog.c (simplify_while_replacing): Likewise.
5368 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5369 (simplify_binary_operation_1): Likewise. Remove redundant
5370 mode equality check.
5372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5373 Alan Hayward <alan.hayward@arm.com>
5374 David Sherwood <david.sherwood@arm.com>
5376 * combine.c (combine_simplify_rtx): Add checks for
5377 is_a <scalar_int_mode>.
5378 (simplify_if_then_else): Likewise.
5379 (make_field_assignment): Likewise.
5380 (simplify_comparison): Likewise.
5381 * ifcvt.c (noce_try_bitop): Likewise.
5382 * loop-invariant.c (canonicalize_address_mult): Likewise.
5383 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5386 Alan Hayward <alan.hayward@arm.com>
5387 David Sherwood <david.sherwood@arm.com>
5389 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5390 is_a <scalar_int_mode> instead of != BLKmode.
5392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5393 Alan Hayward <alan.hayward@arm.com>
5394 David Sherwood <david.sherwood@arm.com>
5396 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5397 instead of != VOIDmode.
5398 * combine.c (if_then_else_cond): Likewise.
5399 (change_zero_ext): Likewise.
5400 * dwarf2out.c (mem_loc_descriptor): Likewise.
5401 (loc_descriptor): Likewise.
5402 * rtlanal.c (canonicalize_condition): Likewise.
5403 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5406 Alan Hayward <alan.hayward@arm.com>
5407 David Sherwood <david.sherwood@arm.com>
5409 * simplify-rtx.c (simplify_binary_operation_1): Use
5410 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5413 Alan Hayward <alan.hayward@arm.com>
5414 David Sherwood <david.sherwood@arm.com>
5416 * wide-int.h (int_traits<unsigned char>) New class.
5417 (int_traits<unsigned short>) Likewise.
5418 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5419 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5421 * combine.c (set_nonzero_bits_and_sign_copies): Use
5422 is_a <scalar_int_mode>.
5423 (find_split_point): Likewise.
5424 (combine_simplify_rtx): Likewise.
5425 (simplify_logical): Likewise.
5426 (expand_compound_operation): Likewise.
5427 (expand_field_assignment): Likewise.
5428 (make_compound_operation): Likewise.
5429 (extended_count): Likewise.
5430 (change_zero_ext): Likewise.
5431 (simplify_comparison): Likewise.
5432 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5433 (ucompare_loc_descriptor): Likewise.
5434 (minmax_loc_descriptor): Likewise.
5435 (mem_loc_descriptor): Likewise.
5436 (loc_descriptor): Likewise.
5437 * expmed.c (init_expmed_one_mode): Likewise.
5438 * lra-constraints.c (lra_constraint_offset): Likewise.
5439 * optabs.c (prepare_libcall_arg): Likewise.
5440 * postreload.c (move2add_note_store): Likewise.
5441 * reload.c (operands_match_p): Likewise.
5442 * rtl.h (load_extend_op): Likewise.
5443 * rtlhooks.c (gen_lowpart_general): Likewise.
5444 * simplify-rtx.c (simplify_truncation): Likewise.
5445 (simplify_unary_operation_1): Likewise.
5446 (simplify_binary_operation_1): Likewise.
5447 (simplify_const_binary_operation): Likewise.
5448 (simplify_const_relational_operation): Likewise.
5449 (simplify_subreg): Likewise.
5450 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5451 * var-tracking.c (adjust_mems): Likewise.
5452 (prepare_call_arguments): Likewise.
5454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5455 Alan Hayward <alan.hayward@arm.com>
5456 David Sherwood <david.sherwood@arm.com>
5458 * machmode.h (is_int_mode): New fuction.
5459 * combine.c (find_split_point): Use it.
5460 (combine_simplify_rtx): Likewise.
5461 (simplify_if_then_else): Likewise.
5462 (simplify_set): Likewise.
5463 (simplify_shift_const_1): Likewise.
5464 (simplify_comparison): Likewise.
5465 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5466 * cse.c (notreg_cost): Likewise.
5467 (cse_insn): Likewise.
5468 * cselib.c (cselib_lookup_1): Likewise.
5469 * dojump.c (do_jump_1): Likewise.
5470 (do_compare_rtx_and_jump): Likewise.
5471 * dse.c (get_call_args): Likewise.
5472 * dwarf2out.c (rtl_for_decl_init): Likewise.
5473 (native_encode_initializer): Likewise.
5474 * expmed.c (emit_store_flag_1): Likewise.
5475 (emit_store_flag): Likewise.
5476 * expr.c (convert_modes): Likewise.
5477 (store_field): Likewise.
5478 (expand_expr_real_1): Likewise.
5479 * fold-const.c (fold_read_from_constant_string): Likewise.
5480 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5481 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5482 * optabs.c (expand_binop): Likewise.
5483 (expand_unop): Likewise.
5484 (expand_abs_nojump): Likewise.
5485 (expand_one_cmpl_abs_nojump): Likewise.
5486 * simplify-rtx.c (mode_signbit_p): Likewise.
5487 (val_signbit_p): Likewise.
5488 (val_signbit_known_set_p): Likewise.
5489 (val_signbit_known_clear_p): Likewise.
5490 (simplify_relational_operation_1): Likewise.
5491 * tree.c (vector_type_mode): Likewise.
5493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5494 Alan Hayward <alan.hayward@arm.com>
5495 David Sherwood <david.sherwood@arm.com>
5497 * machmode.h (smallest_mode_for_size): Fix formatting.
5498 (smallest_int_mode_for_size): New function.
5499 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5500 instead of smallest_mode_for_size.
5501 * combine.c (make_extraction): Likewise.
5502 * config/arc/arc.c (arc_expand_movmem): Likewise.
5503 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5504 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5505 * config/s390/s390.c (s390_expand_insv): Likewise.
5506 * config/sparc/sparc.c (assign_int_registers): Likewise.
5507 * config/spu/spu.c (spu_function_value): Likewise.
5508 (spu_function_arg): Likewise.
5509 * coverage.c (get_gcov_type): Likewise.
5510 (get_gcov_unsigned_t): Likewise.
5511 * dse.c (find_shift_sequence): Likewise.
5512 * expmed.c (store_bit_field_1): Likewise.
5513 * expr.c (convert_move): Likewise.
5514 (store_field): Likewise.
5515 * internal-fn.c (expand_arith_overflow): Likewise.
5516 * optabs-query.c (get_best_extraction_insn): Likewise.
5517 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5518 * stor-layout.c (layout_type): Likewise.
5519 (initialize_sizetypes): Likewise.
5520 * targhooks.c (default_get_mask_mode): Likewise.
5521 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5524 Alan Hayward <alan.hayward@arm.com>
5525 David Sherwood <david.sherwood@arm.com>
5527 * machmode.h (opt_mode::else_blk): New function.
5528 (int_mode_for_mode): Declare.
5529 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5530 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5532 * cfgexpand.c (expand_debug_expr): Likewise.
5533 * combine.c (gen_lowpart_or_truncate): Likewise.
5534 (gen_lowpart_for_combine): Likewise.
5535 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5536 * config/avr/avr.c (avr_to_int_mode): Likewise.
5537 (avr_out_plus_1): Likewise.
5538 (avr_out_plus): Likewise.
5539 (avr_out_round): Likewise.
5540 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5541 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5542 (s390_expand_vcond): Likewise.
5543 * config/spu/spu.c (spu_split_immediate): Likewise.
5544 (spu_expand_mov): Likewise.
5545 * dse.c (get_stored_val): Likewise.
5546 * expmed.c (store_bit_field_1): Likewise.
5547 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5549 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5550 (extract_low_bits): Likewise.
5551 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5552 handling rather than repeating the check.
5553 (emit_group_store): Likewise.
5554 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5555 * optabs.c (expand_absneg_bit): Likewise.
5556 (expand_copysign_absneg): Likewise.
5557 (expand_copysign_bit): Likewise.
5558 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5559 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5560 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5561 * var-tracking.c (prepare_call_arguments): Likewise.
5562 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5563 int_mode_for_mode instead of mode_for_size.
5564 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5567 Alan Hayward <alan.hayward@arm.com>
5568 David Sherwood <david.sherwood@arm.com>
5570 * machmode.h (int_mode_for_size): New function.
5571 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5572 instead of mode_for_size.
5573 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5575 * combine.c (expand_field_assignment): Use int_mode_for_size
5576 instead of mode_for_size.
5577 (make_extraction): Likewise.
5578 (simplify_shift_const_1): Likewise.
5579 (simplify_comparison): Likewise.
5580 * dojump.c (do_jump): Likewise.
5581 * dwarf2out.c (mem_loc_descriptor): Likewise.
5582 * emit-rtl.c (init_derived_machine_modes): Likewise.
5583 * expmed.c (flip_storage_order): Likewise.
5584 (convert_extracted_bit_field): Likewise.
5585 * expr.c (copy_blkmode_from_reg): Likewise.
5586 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5587 * internal-fn.c (expand_mul_overflow): Likewise.
5588 * lower-subreg.c (simple_move): Likewise.
5589 * optabs-libfuncs.c (init_optabs): Likewise.
5590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5591 * tree.c (vector_type_mode): Likewise.
5592 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5593 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5594 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5595 * tree-vect-stmts.c (vectorizable_load): Likewise.
5596 (vectorizable_store): Likewise.
5598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5599 Alan Hayward <alan.hayward@arm.com>
5600 David Sherwood <david.sherwood@arm.com>
5602 * coretypes.h (pod_mode): New type.
5603 (scalar_int_mode_pod): New typedef.
5604 * machmode.h (pod_mode): New class.
5605 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5606 * genmodes.c (emit_mode_int_n): Update accordingly.
5607 * lower-subreg.h (target_lower_subreg): Change type to
5608 scalar_int_mode_pod.
5609 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5610 scalar_int_mode_pod.
5612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5613 Alan Hayward <alan.hayward@arm.com>
5614 David Sherwood <david.sherwood@arm.com>
5616 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5617 machine_mode to scalar_int_mode.
5618 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5619 (rs6000_option_override_internal): Remove cast to int.
5620 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5621 machine_mode to scalar_int_mode.
5622 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5623 (rs6000_option_override_internal): Remove cast to int.
5624 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5625 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5627 * config/s390/s390.c (s390_expand_builtin): Likewise.
5628 * coretypes.h (scalar_int_mode): New type.
5629 (opt_scalar_int_mode): New typedef.
5630 * machmode.h (scalar_int_mode): New class.
5631 (scalar_int_mode::includes_p): New function.
5632 (byte_mode): Change type to scalar_int_mode.
5633 (word_mode): Likewise.
5634 (ptr_mode): Likewise.
5635 * emit-rtl.c (byte_mode): Likewise.
5636 (word_mode): Likewise.
5637 (ptr_mode): Likewise.
5638 (init_derived_machine_modes): Update accordingly.
5639 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5640 and MODE_PARTIAL_INT.
5641 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5642 opt_scalar_int_mode.
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 * target.def (libgcc_floating_mode_supported_p): Take a
5650 * doc/tm.texi: Regenerate.
5651 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5653 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5654 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5658 Alan Hayward <alan.hayward@arm.com>
5659 David Sherwood <david.sherwood@arm.com>
5661 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5662 * doc/tm.texi: Regenerate.
5663 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5664 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5665 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5666 * targhooks.h (default_floatn_mode): Likewise.
5667 * targhooks.c (default_floatn_mode): Likewise.
5668 * tree.c (build_common_tree_nodes): Update accordingly.
5670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5671 Alan Hayward <alan.hayward@arm.com>
5672 David Sherwood <david.sherwood@arm.com>
5674 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5675 (mode_iterator::iterate_p): Likewise.
5676 (mode_iterator::get_wider): Likewise.
5677 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
5684 * machmode.h (float_mode_for_size): New function.
5685 * emit-rtl.c (double_mode): Delete.
5686 (init_emit_once): Use float_mode_for_size.
5687 * stor-layout.c (layout_type): Likewise.
5688 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5691 Alan Hayward <alan.hayward@arm.com>
5692 David Sherwood <david.sherwood@arm.com>
5694 * output.h (assemble_real): Take a scalar_float_mode.
5695 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5696 * config/arm/arm.md (consttable_4): Likewise.
5697 (consttable_8): Likewise.
5698 (consttable_16): Likewise.
5699 * config/mips/mips.md (consttable_float): Likewise.
5700 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5701 * varasm.c (assemble_real): Take a scalar_float_mode.
5702 (output_constant_pool_2): Update accordingly.
5703 (output_constant): Likewise.
5705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5706 Alan Hayward <alan.hayward@arm.com>
5707 David Sherwood <david.sherwood@arm.com>
5709 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5710 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5711 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5712 (native_encode_real): Likewise.
5713 (native_interpret_real): Likewise.
5714 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5715 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5722 <scalar_float_mode>. Simplify.
5723 (gen_extend_conv_libfunc): Likewise.
5725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5726 Alan Hayward <alan.hayward@arm.com>
5727 David Sherwood <david.sherwood@arm.com>
5729 * coretypes.h (scalar_float_mode): New type.
5730 * machmode.h (mode_traits::from_int): Use machine_mode if
5731 USE_ENUM_MODES is defined.
5732 (is_a): New function.
5734 (dyn_cast): Likewise.
5735 (scalar_float_mode): New class.
5736 (scalar_float_mode::includes_p): New function.
5737 (is_float_mode): Likewise.
5738 * gdbhooks.py (MachineModePrinter): New class.
5739 (build_pretty_printer): Use it for scalar_float_mode.
5740 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5741 (format_helper::format_helper): Turn into a template.
5742 * genmodes.c (get_mode_class): New function.
5743 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5744 or machine_mode if none.
5745 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5746 as_a <scalar_float_mode>.
5747 * dwarf2out.c (mem_loc_descriptor): Likewise.
5748 (insert_float): Likewise.
5749 (add_const_value_attribute): Likewise.
5750 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5751 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5752 (expand_unop): Update accordingly.
5753 (expand_abs_nojump): Likewise.
5754 (expand_copysign_absneg): Take a scalar_float_mode.
5755 (expand_copysign_bit): Likewise.
5756 (expand_copysign): Update accordingly.
5758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5759 Alan Hayward <alan.hayward@arm.com>
5760 David Sherwood <david.sherwood@arm.com>
5762 * coretypes.h (opt_mode): New class.
5763 * machmode.h (opt_mode): Likewise.
5764 (opt_mode::else_void): New function.
5765 (opt_mode::require): Likewise.
5766 (opt_mode::exists): Likewise.
5767 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5768 (GET_MODE_2XWIDER_MODE): Likewise.
5769 (mode_iterator::get_wider): Update accordingly.
5770 (mode_iterator::get_2xwider): Likewise.
5771 (mode_iterator::get_known_wider): Likewise, turning into a template.
5772 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5773 forcing a wider mode to exist.
5774 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5775 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5776 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5777 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5778 * lower-subreg.c (init_lower_subreg): Likewise.
5779 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5780 on the final iteration.
5781 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5782 a wider mode exists before asking for a move pattern.
5783 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5784 forcing a wider mode to exist.
5785 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5786 returning false if no such mode exists.
5787 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5788 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5789 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5790 Avoid checking for a MODE_INT if we already know the mode is not a
5792 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
5793 forcing a wider mode to exist.
5794 (expmed_mult_highpart_optab): Likewise.
5795 (expmed_mult_highpart): Likewise.
5796 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
5798 * lto-streamer-in.c (lto_input_mode_table): Likewise.
5799 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
5800 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5801 * internal-fn.c (expand_mul_overflow): Update use of
5802 GET_MODE_2XWIDER_MODE.
5803 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5804 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
5805 GET_MODE_WIDER_MODE.
5806 (convert_plusminus_to_widen): Likewise.
5807 * tree-switch-conversion.c (array_value_type): Likewise.
5808 * var-tracking.c (emit_note_insn_var_location): Likewise.
5809 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5810 Return false inside rather than outside the loop if no wider mode
5812 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
5813 and GET_MODE_2XWIDER_MODE
5814 (can_compare_p): Use else_void.
5815 * gdbhooks.py (OptMachineModePrinter): New class.
5816 (build_pretty_printer): Use it for opt_mode.
5818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5819 Alan Hayward <alan.hayward@arm.com>
5820 David Sherwood <david.sherwood@arm.com>
5822 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5823 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5826 Alan Hayward <alan.hayward@arm.com>
5827 David Sherwood <david.sherwood@arm.com>
5829 * machmode.h (mode_traits): New structure.
5830 (get_narrowest_mode): New function.
5831 (mode_iterator::start): Likewise.
5832 (mode_iterator::iterate_p): Likewise.
5833 (mode_iterator::get_wider): Likewise.
5834 (mode_iterator::get_known_wider): Likewise.
5835 (mode_iterator::get_2xwider): Likewise.
5836 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5837 (FOR_EACH_MODE): Likewise.
5838 (FOR_EACH_MODE_FROM): Likewise.
5839 (FOR_EACH_MODE_UNTIL): Likewise.
5840 (FOR_EACH_WIDER_MODE): Likewise.
5841 (FOR_EACH_2XWIDER_MODE): Likewise.
5842 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5843 * combine.c (simplify_comparison): Likewise
5844 * config/i386/i386.c (type_natural_mode): Likewise.
5845 * cse.c (cse_insn): Likewise.
5846 * dse.c (find_shift_sequence): Likewise.
5847 * emit-rtl.c (init_derived_machine_modes): Likewise.
5848 (init_emit_once): Likewise.
5849 * explow.c (hard_function_value): Likewise.
5850 * expmed.c (extract_fixed_bit_field_1): Likewise.
5851 (extract_bit_field_1): Likewise.
5852 (expand_divmod): Likewise.
5853 (emit_store_flag_1): Likewise.
5854 * expr.c (init_expr_target): Likewise.
5855 (convert_move): Likewise.
5856 (alignment_for_piecewise_move): Likewise.
5857 (widest_int_mode_for_size): Likewise.
5858 (emit_block_move_via_movmem): Likewise.
5859 (copy_blkmode_to_reg): Likewise.
5860 (set_storage_via_setmem): Likewise.
5861 (compress_float_constant): Likewise.
5862 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5863 * optabs-query.c (get_best_extraction_insn): Likewise.
5864 * optabs.c (expand_binop): Likewise.
5865 (expand_twoval_unop): Likewise.
5866 (expand_twoval_binop): Likewise.
5867 (widen_leading): Likewise.
5868 (widen_bswap): Likewise.
5869 (expand_parity): Likewise.
5870 (expand_unop): Likewise.
5871 (prepare_cmp_insn): Likewise.
5872 (prepare_float_lib_cmp): Likewise.
5873 (expand_float): Likewise.
5874 (expand_fix): Likewise.
5875 (expand_sfix_optab): Likewise.
5876 * postreload.c (move2add_use_add2_insn): Likewise.
5877 * reg-stack.c (reg_to_stack): Likewise.
5878 * reginfo.c (choose_hard_reg_mode): Likewise.
5879 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5880 * stor-layout.c (mode_for_size): Likewise.
5881 (smallest_mode_for_size): Likewise.
5882 (mode_for_vector): Likewise.
5883 (finish_bitfield_representative): Likewise.
5884 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5885 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5886 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5887 * var-tracking.c (prepare_call_arguments): Likewise.
5889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5890 Alan Hayward <alan.hayward@arm.com>
5891 David Sherwood <david.sherwood@arm.com>
5893 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5894 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5895 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5896 * machmode.h (mode_size): Move earlier in file.
5897 (mode_precision): Likewise.
5898 (mode_inner): Likewise.
5899 (mode_nunits): Likewise.
5900 (mode_unit_size): Likewise.
5901 (unit_unit_precision): Likewise.
5902 (mode_wider): Likewise.
5903 (mode_2xwider): Likewise.
5904 (machine_mode): New class.
5905 (mode_to_bytes): New function.
5906 (mode_to_bits): Likewise.
5907 (mode_to_precision): Likewise.
5908 (mode_to_inner): Likewise.
5909 (mode_to_unit_size): Likewise.
5910 (mode_to_unit_precision): Likewise.
5911 (mode_to_nunits): Likewise.
5912 (GET_MODE_SIZE): Use mode_to_bytes.
5913 (GET_MODE_BITSIZE): Use mode_to_bits.
5914 (GET_MODE_PRECISION): Use mode_to_precision.
5915 (GET_MODE_INNER): Use mode_to_inner.
5916 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5917 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5918 (GET_MODE_NUNITS): Use mode_to_nunits.
5919 * system.h (ALWAYS_INLINE): New macro.
5920 * config/powerpcspe/powerpcspe-c.c
5921 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5922 int for arg1_mode and arg2_mode.
5924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5925 Alan Hayward <alan.hayward@arm.com>
5926 David Sherwood <david.sherwood@arm.com>
5928 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5929 Prefix mode names with E_ in case statements.
5930 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5931 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5932 (aarch64_split_simd_move): Likewise.
5933 (aarch64_gen_storewb_pair): Likewise.
5934 (aarch64_gen_loadwb_pair): Likewise.
5935 (aarch64_gen_store_pair): Likewise.
5936 (aarch64_gen_load_pair): Likewise.
5937 (aarch64_get_condition_code_1): Likewise.
5938 (aarch64_constant_pool_reload_icode): Likewise.
5939 (get_rsqrte_type): Likewise.
5940 (get_rsqrts_type): Likewise.
5941 (get_recpe_type): Likewise.
5942 (get_recps_type): Likewise.
5943 (aarch64_gimplify_va_arg_expr): Likewise.
5944 (aarch64_simd_container_mode): Likewise.
5945 (aarch64_emit_load_exclusive): Likewise.
5946 (aarch64_emit_store_exclusive): Likewise.
5947 (aarch64_expand_compare_and_swap): Likewise.
5948 (aarch64_gen_atomic_cas): Likewise.
5949 (aarch64_emit_bic): Likewise.
5950 (aarch64_emit_atomic_swap): Likewise.
5951 (aarch64_emit_atomic_load_op): Likewise.
5952 (aarch64_evpc_trn): Likewise.
5953 (aarch64_evpc_uzp): Likewise.
5954 (aarch64_evpc_zip): Likewise.
5955 (aarch64_evpc_ext): Likewise.
5956 (aarch64_evpc_rev): Likewise.
5957 (aarch64_evpc_dup): Likewise.
5958 (aarch64_gen_ccmp_first): Likewise.
5959 (aarch64_gen_ccmp_next): Likewise.
5960 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5961 (alpha_emit_xfloating_libcall): Likewise.
5962 (emit_insxl): Likewise.
5963 (alpha_arg_type): Likewise.
5964 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5965 (arc_preferred_simd_mode): Likewise.
5966 (arc_secondary_reload): Likewise.
5967 (get_arc_condition_code): Likewise.
5968 (arc_print_operand): Likewise.
5969 (arc_legitimate_constant_p): Likewise.
5970 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5971 * config/arc/arc.md (casesi_load): Likewise.
5972 (casesi_compact_jump): Likewise.
5973 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5974 (cc_use_register): Likewise.
5975 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5976 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5977 (arm_init_iwmmxt_builtins): Likewise.
5978 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5979 (neon_expand_vector_init): Likewise.
5980 (arm_attr_length_move_neon): Likewise.
5981 (maybe_get_arm_condition_code): Likewise.
5982 (arm_emit_vector_const): Likewise.
5983 (arm_preferred_simd_mode): Likewise.
5984 (arm_output_iwmmxt_tinsr): Likewise.
5985 (thumb1_output_casesi): Likewise.
5986 (thumb2_output_casesi): Likewise.
5987 (arm_emit_load_exclusive): Likewise.
5988 (arm_emit_store_exclusive): Likewise.
5989 (arm_expand_compare_and_swap): Likewise.
5990 (arm_evpc_neon_vuzp): Likewise.
5991 (arm_evpc_neon_vzip): Likewise.
5992 (arm_evpc_neon_vrev): Likewise.
5993 (arm_evpc_neon_vtrn): Likewise.
5994 (arm_evpc_neon_vext): Likewise.
5995 (arm_validize_comparison): Likewise.
5996 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5997 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5998 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5999 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6000 (c6x_preferred_simd_mode): Likewise.
6001 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6002 (epiphany_rtx_costs): Likewise.
6003 * config/epiphany/predicates.md (proper_comparison_operator):
6005 * config/frv/frv.c (condexec_memory_operand): Likewise.
6006 (frv_emit_move): Likewise.
6007 (output_move_single): Likewise.
6008 (output_condmove_single): Likewise.
6009 (frv_hard_regno_mode_ok): Likewise.
6010 (frv_matching_accg_mode): Likewise.
6011 * config/h8300/h8300.c (split_adds_subs): Likewise.
6012 (h8300_rtx_costs): Likewise.
6013 (h8300_print_operand): Likewise.
6014 (compute_mov_length): Likewise.
6015 (output_logical_op): Likewise.
6016 (compute_logical_op_length): Likewise.
6017 (compute_logical_op_cc): Likewise.
6018 (h8300_shift_needs_scratch_p): Likewise.
6019 (output_a_shift): Likewise.
6020 (compute_a_shift_length): Likewise.
6021 (compute_a_shift_cc): Likewise.
6022 (expand_a_rotate): Likewise.
6023 (output_a_rotate): Likewise.
6024 * config/i386/i386.c (classify_argument): Likewise.
6025 (function_arg_advance_32): Likewise.
6026 (function_arg_32): Likewise.
6027 (function_arg_64): Likewise.
6028 (function_value_64): Likewise.
6029 (ix86_gimplify_va_arg): Likewise.
6030 (ix86_legitimate_constant_p): Likewise.
6031 (put_condition_code): Likewise.
6032 (split_double_mode): Likewise.
6033 (ix86_avx256_split_vector_move_misalign): Likewise.
6034 (ix86_expand_vector_logical_operator): Likewise.
6035 (ix86_split_idivmod): Likewise.
6036 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6037 (ix86_build_const_vector): Likewise.
6038 (ix86_build_signbit_mask): Likewise.
6039 (ix86_match_ccmode): Likewise.
6040 (ix86_cc_modes_compatible): Likewise.
6041 (ix86_expand_branch): Likewise.
6042 (ix86_expand_sse_cmp): Likewise.
6043 (ix86_expand_sse_movcc): Likewise.
6044 (ix86_expand_int_sse_cmp): Likewise.
6045 (ix86_expand_vec_perm_vpermi2): Likewise.
6046 (ix86_expand_vec_perm): Likewise.
6047 (ix86_expand_sse_unpack): Likewise.
6048 (ix86_expand_int_addcc): Likewise.
6049 (ix86_split_to_parts): Likewise.
6050 (ix86_vectorize_builtin_gather): Likewise.
6051 (ix86_vectorize_builtin_scatter): Likewise.
6052 (avx_vpermilp_parallel): Likewise.
6053 (inline_memory_move_cost): Likewise.
6054 (ix86_tieable_integer_mode_p): Likewise.
6055 (x86_maybe_negate_const_int): Likewise.
6056 (ix86_expand_vector_init_duplicate): Likewise.
6057 (ix86_expand_vector_init_one_nonzero): Likewise.
6058 (ix86_expand_vector_init_one_var): Likewise.
6059 (ix86_expand_vector_init_concat): Likewise.
6060 (ix86_expand_vector_init_interleave): Likewise.
6061 (ix86_expand_vector_init_general): Likewise.
6062 (ix86_expand_vector_set): Likewise.
6063 (ix86_expand_vector_extract): Likewise.
6064 (emit_reduc_half): Likewise.
6065 (ix86_emit_i387_round): Likewise.
6066 (ix86_mangle_type): Likewise.
6067 (ix86_expand_round_sse4): Likewise.
6068 (expand_vec_perm_blend): Likewise.
6069 (canonicalize_vector_int_perm): Likewise.
6070 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6071 (expand_vec_perm_1): Likewise.
6072 (expand_vec_perm_interleave3): Likewise.
6073 (expand_vec_perm_even_odd_pack): Likewise.
6074 (expand_vec_perm_even_odd_1): Likewise.
6075 (expand_vec_perm_broadcast_1): Likewise.
6076 (ix86_vectorize_vec_perm_const_ok): Likewise.
6077 (ix86_expand_vecop_qihi): Likewise.
6078 (ix86_expand_mul_widen_hilo): Likewise.
6079 (ix86_expand_sse2_abs): Likewise.
6080 (ix86_expand_pextr): Likewise.
6081 (ix86_expand_pinsr): Likewise.
6082 (ix86_preferred_simd_mode): Likewise.
6083 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6084 * config/i386/sse.md (*andnot<mode>3): Likewise.
6085 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6086 (*<code><mode>3): Likewise.
6087 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6088 (ia64_expand_atomic_op): Likewise.
6089 (ia64_arg_type): Likewise.
6090 (ia64_mode_to_int): Likewise.
6091 (ia64_scalar_mode_supported_p): Likewise.
6092 (ia64_vector_mode_supported_p): Likewise.
6093 (expand_vec_perm_broadcast): Likewise.
6094 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6095 (iq2000_function_arg_advance): Likewise.
6096 (iq2000_function_arg): Likewise.
6097 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6098 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6099 (m68k_libcall_value): Likewise.
6100 (m68k_function_value): Likewise.
6101 (sched_attr_op_type): Likewise.
6102 * config/mcore/mcore.c (mcore_output_move): Likewise.
6103 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6105 (microblaze_function_arg): Likewise.
6106 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6107 (mips_print_operand): Likewise.
6108 (mips_mode_ok_for_mov_fmt_p): Likewise.
6109 (mips_vector_mode_supported_p): Likewise.
6110 (mips_preferred_simd_mode): Likewise.
6111 (mips_expand_vpc_loongson_even_odd): Likewise.
6112 (mips_expand_vec_unpack): Likewise.
6113 (mips_expand_vi_broadcast): Likewise.
6114 (mips_expand_vector_init): Likewise.
6115 (mips_expand_vec_reduc): Likewise.
6116 (mips_expand_msa_cmp): Likewise.
6117 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6118 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6119 (cc_flags_for_mode): Likewise.
6120 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6121 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6122 (nds32_output_casesi_pc_relative): Likewise.
6123 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6124 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6125 (nvptx_gen_unpack): Likewise.
6126 (nvptx_gen_pack): Likewise.
6127 (nvptx_gen_shuffle): Likewise.
6128 (nvptx_gen_wcast): Likewise.
6129 (nvptx_preferred_simd_mode): Likewise.
6130 * config/pa/pa.c (pa_secondary_reload): Likewise.
6131 * config/pa/predicates.md (base14_operand): Likewise.
6132 * config/powerpcspe/powerpcspe-c.c
6133 (altivec_resolve_overloaded_builtin): Likewise.
6134 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6136 (rs6000_preferred_simd_mode): Likewise.
6137 (output_vec_const_move): Likewise.
6138 (rs6000_expand_vector_extract): Likewise.
6139 (rs6000_split_vec_extract_var): Likewise.
6140 (reg_offset_addressing_ok_p): Likewise.
6141 (rs6000_legitimate_offset_address_p): Likewise.
6142 (rs6000_legitimize_address): Likewise.
6143 (rs6000_emit_set_const): Likewise.
6144 (rs6000_const_vec): Likewise.
6145 (rs6000_emit_move): Likewise.
6146 (spe_build_register_parallel): Likewise.
6147 (rs6000_darwin64_record_arg_recurse): Likewise.
6148 (swap_selector_for_mode): Likewise.
6149 (spe_init_builtins): Likewise.
6150 (paired_init_builtins): Likewise.
6151 (altivec_init_builtins): Likewise.
6152 (do_load_for_compare): Likewise.
6153 (rs6000_generate_compare): Likewise.
6154 (rs6000_expand_float128_convert): Likewise.
6155 (emit_load_locked): Likewise.
6156 (emit_store_conditional): Likewise.
6157 (rs6000_output_function_epilogue): Likewise.
6158 (rs6000_handle_altivec_attribute): Likewise.
6159 (rs6000_function_value): Likewise.
6160 (emit_fusion_gpr_load): Likewise.
6161 (emit_fusion_p9_load): Likewise.
6162 (emit_fusion_p9_store): Likewise.
6163 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6164 (fusion_gpr_mem_load): Likewise.
6165 (fusion_addis_mem_combo_load): Likewise.
6166 (fusion_addis_mem_combo_store): Likewise.
6167 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6168 (fusion_gpr_mem_load): Likewise.
6169 (fusion_addis_mem_combo_load): Likewise.
6170 (fusion_addis_mem_combo_store): Likewise.
6171 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6173 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6174 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6175 (rs6000_preferred_simd_mode): Likewise.
6176 (output_vec_const_move): Likewise.
6177 (rs6000_expand_vector_extract): Likewise.
6178 (rs6000_split_vec_extract_var): Likewise.
6179 (reg_offset_addressing_ok_p): Likewise.
6180 (rs6000_legitimate_offset_address_p): Likewise.
6181 (rs6000_legitimize_address): Likewise.
6182 (rs6000_emit_set_const): Likewise.
6183 (rs6000_const_vec): Likewise.
6184 (rs6000_emit_move): Likewise.
6185 (rs6000_darwin64_record_arg_recurse): Likewise.
6186 (swap_selector_for_mode): Likewise.
6187 (paired_init_builtins): Likewise.
6188 (altivec_init_builtins): Likewise.
6189 (rs6000_expand_float128_convert): Likewise.
6190 (emit_load_locked): Likewise.
6191 (emit_store_conditional): Likewise.
6192 (rs6000_output_function_epilogue): Likewise.
6193 (rs6000_handle_altivec_attribute): Likewise.
6194 (rs6000_function_value): Likewise.
6195 (emit_fusion_gpr_load): Likewise.
6196 (emit_fusion_p9_load): Likewise.
6197 (emit_fusion_p9_store): Likewise.
6198 * config/rx/rx.c (rx_gen_move_template): Likewise.
6199 (flags_from_mode): Likewise.
6200 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6201 (s390_slb_comparison): Likewise.
6202 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6203 (s390_vector_mode_supported_p): Likewise.
6204 (s390_cc_modes_compatible): Likewise.
6205 (s390_match_ccmode_set): Likewise.
6206 (s390_canonicalize_comparison): Likewise.
6207 (s390_emit_compare_and_swap): Likewise.
6208 (s390_branch_condition_mask): Likewise.
6209 (s390_rtx_costs): Likewise.
6210 (s390_secondary_reload): Likewise.
6211 (__SECONDARY_RELOAD_CASE): Likewise.
6212 (s390_expand_cs): Likewise.
6213 (s390_preferred_simd_mode): Likewise.
6214 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6215 * config/sh/sh.c (sh_print_operand): Likewise.
6216 (dump_table): Likewise.
6217 (sh_secondary_reload): Likewise.
6218 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6219 * config/sh/sh.md (casesi_worker_1): Likewise.
6220 (casesi_worker_2): Likewise.
6221 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6222 (fcc_comparison_operator): Likewise.
6223 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6224 (emit_soft_tfmode_cvt): Likewise.
6225 (sparc_preferred_simd_mode): Likewise.
6226 (output_cbranch): Likewise.
6227 (sparc_print_operand): Likewise.
6228 (sparc_expand_vec_perm_bmask): Likewise.
6229 (vector_init_bshuffle): Likewise.
6230 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6231 (spu_vector_mode_supported_p): Likewise.
6232 (spu_expand_insv): Likewise.
6233 (spu_emit_branch_or_set): Likewise.
6234 (spu_handle_vector_attribute): Likewise.
6235 (spu_builtin_splats): Likewise.
6236 (spu_builtin_extract): Likewise.
6237 (spu_builtin_promote): Likewise.
6238 (spu_expand_sign_extend): Likewise.
6239 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6240 (tilegx_simd_int): Likewise.
6241 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6242 (tilepro_simd_int): Likewise.
6243 * config/v850/v850.c (const_double_split): Likewise.
6244 (v850_print_operand): Likewise.
6245 (ep_memory_offset): Likewise.
6246 * config/vax/vax.c (vax_rtx_costs): Likewise.
6247 (vax_output_int_move): Likewise.
6248 (vax_output_int_add): Likewise.
6249 (vax_output_int_subtract): Likewise.
6250 * config/visium/predicates.md (visium_branch_operator): Likewise.
6251 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6252 (visium_print_operand_address): Likewise.
6253 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6254 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6255 (xtensa_expand_conditional_branch): Likewise.
6256 (xtensa_copy_incoming_a7): Likewise.
6257 (xtensa_output_literal): Likewise.
6258 * dfp.c (decimal_real_maxval): Likewise.
6259 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6262 Alan Hayward <alan.hayward@arm.com>
6263 David Sherwood <david.sherwood@arm.com>
6265 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6266 (mode_nunits_inline): Likewise.
6267 (mode_inner_inline): Likewise.
6268 (mode_unit_size_inline): Likewise.
6269 (mode_unit_precision_inline): Likewise.
6270 (emit_insn_modes_h): Likewise. Also emit a #define of the
6272 (emit_mode_wider): Add an E_ prefix to mode names.
6273 (emit_mode_complex): Likewise.
6274 (emit_mode_inner): Likewise.
6275 (emit_mode_adjustments): Likewise.
6276 (emit_mode_int_n): Likewise.
6277 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6278 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6279 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6280 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6281 (CRC32_BUILTIN, ENTRY): Likewise.
6282 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6283 (aarch64_pop_regs): Likewise.
6284 (aarch64_process_components): Likewise.
6285 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6286 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6287 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6288 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6289 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6290 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6292 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6293 (pcmpistr): Likewise.
6294 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6295 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6296 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6298 * config/rl78/rl78.c (mduc_regs): Likewise.
6299 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6300 (htm_expand_builtin): Likewise.
6301 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6302 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6303 * config/xtensa/xtensa.c (print_operand): Likewise.
6304 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6305 (NUM_MODE_VECTOR_INT): Likewise.
6306 * genoutput.c (null_operand): Likewise.
6307 (output_operand_data): Likewise.
6308 * genrecog.c (print_parameter_value): Likewise.
6309 * lra.c (debug_operand_data): Likewise.
6311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6313 * dwarf2out.c (scompare_loc_descriptor_wide)
6314 (scompare_loc_descriptor_narrow): New functions, split out from...
6315 (scompare_loc_descriptor): ...here.
6316 * expmed.c (emit_store_flag_int): New function, split out from...
6317 (emit_store_flag): ...here.
6319 2017-08-30 Richard Biener <rguenther@suse.de>
6321 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6322 (dwarf2out_early_finish): Move setting of AT_pubnames from
6323 early debug output to early finish.
6325 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6327 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6328 and -mdata-region to the assembler.
6330 2017-08-30 Richard Biener <rguenther@suse.de>
6332 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6334 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6335 (dwarf2out_early_global_decl): Only generate a DIE for the
6336 abstract origin if it doesn't already exist or is a declaration DIE.
6337 (resolve_addr): Do not add the linkage name twice when
6338 generating a stub DIE for the DW_TAG_GNU_call_site target.
6340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6342 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6343 Use machine_mode rather than int for arg1_mode.
6345 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6348 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6349 that the second argument of the built-in functions to unpack
6350 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6351 switch statement instead a lot of if statements.
6352 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6353 Allow 64-bit values to be in Altivec registers as well as
6354 traditional floating point registers.
6355 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6357 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6359 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6360 MAX_REGS_PER_ADDRESS == 1.
6362 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6364 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6365 * config/i386/i386.c (ix86_option_override_internal): Simplify
6366 setting of opts->x_flag_entry.
6368 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6369 Jakub Jelinek <jakub@redhat.com>
6370 Richard Biener <rguenther@suse.de>
6372 PR tree-optimization/81503
6373 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6374 folded constant fits in the target type; reorder tests for clarity.
6376 2017-08-29 Martin Liska <mliska@suse.cz>
6378 * passes.def: Include pass_lower_switch.
6379 * stmt.c (dump_case_nodes): Remove and move to
6380 tree-switch-conversion.
6381 (case_values_threshold): Likewise.
6382 (expand_switch_as_decision_tree_p): Likewise.
6383 (emit_case_decision_tree): Likewise.
6384 (expand_case): Likewise.
6385 (balance_case_nodes): Likewise.
6386 (node_has_low_bound): Likewise.
6387 (node_has_high_bound): Likewise.
6388 (node_is_bounded): Likewise.
6389 (emit_case_nodes): Likewise.
6390 (struct simple_case_node): New struct.
6391 (add_case_node): Remove.
6392 (emit_case_dispatch_table): Use vector instead of case_list.
6393 (reset_out_edges_aux): Remove.
6394 (compute_cases_per_edge): Likewise.
6395 (expand_case): Build list of simple_case_node.
6396 (expand_sjlj_dispatch_table): Use it.
6397 * tree-switch-conversion.c (struct case_node): Moved from
6398 stmt.c and adjusted.
6399 (emit_case_nodes): Likewise.
6400 (node_has_low_bound): Likewise.
6401 (node_has_high_bound): Likewise.
6402 (node_is_bounded): Likewise.
6403 (case_values_threshold): Likewise.
6404 (reset_out_edges_aux): Likewise.
6405 (compute_cases_per_edge): Likewise.
6406 (add_case_node): Likewise.
6407 (dump_case_nodes): Likewise.
6408 (balance_case_nodes): Likewise.
6409 (expand_switch_as_decision_tree_p): Likewise.
6410 (emit_jump): Likewise.
6411 (emit_case_decision_tree): Likewise.
6412 (try_switch_expansion): Likewise.
6413 (do_jump_if_equal): Likewise.
6414 (emit_cmp_and_jump_insns): Likewise.
6415 (fix_phi_operands_for_edge): New function.
6416 (record_phi_operand_mapping): Likewise.
6417 (class pass_lower_switch): New pass.
6418 (pass_lower_switch::execute): New function.
6419 (make_pass_lower_switch): Likewise.
6420 (conditional_probability):
6421 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6422 * tree-pass.h: Add make_pass_lower_switch.
6424 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6427 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6430 2017-08-29 Richard Biener <rguenther@suse.de>
6432 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6433 we do not add a DW_AT_inline attribute twice.
6434 (gen_subprogram_die): Remove code setting DW_AT_inline on
6435 DECL_ABSTRACT_P nodes.
6437 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6439 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6440 calls to internal functions.
6441 (gimplify_modify_expr): Likewise.
6442 * tree-call-cdce.c (use_internal_fn): Likewise.
6443 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6444 (convert_to_divmod): Set the nothrow flag.
6445 * tree-if-conv.c (predicate_mem_writes): Likewise.
6446 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6447 (vectorizable_call): Likewise.
6448 (vectorizable_store): Likewise.
6449 (vectorizable_load): Likewise.
6450 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6451 (vect_recog_mask_conversion_pattern): Likewise.
6453 2017-08-29 Martin Liska <mliska@suse.cz>
6456 * gcc.c (driver_handle_option): Add new argument.
6457 * opts-common.c (handle_option): Pass
6458 target_option_override_hook.
6459 * opts-global.c (lang_handle_option): Add new option.
6460 (set_default_handlers): Add new argument.
6461 (decode_options): Likewise.
6462 * opts.c (target_handle_option): Likewise.
6463 (common_handle_option): Call target_option_override_hook.
6464 * opts.h (struct cl_option_handler_func): Add hook for
6465 target option override.
6466 (struct cl_option_handlers): Likewise.
6467 (set_default_handlers): Add new argument.
6468 (decode_options): Likewise.
6469 (common_handle_option): Likewise.
6470 (target_handle_option): Likewise.
6471 * toplev.c (toplev::main): Pass targetm.target_option.override
6474 2017-08-29 Richard Biener <rguenther@suse.de>
6475 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6477 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6478 life to the active subtree.
6480 2017-08-28 Jeff Law <law@redhat.com>
6482 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6483 derive_equivalences.
6484 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6486 (edge_info::derive_equivalences): New private member function
6488 * tree-ssa-dom.c (class edge_info): Changed from a struct
6489 to a class. Add ctor/dtor, methods and data members.
6490 (edge_info::edge_info): Renamed from allocate_edge_info.
6491 Initialize additional members.
6492 (edge_info::~edge_info): New.
6493 (free_dom_edge_info): Delete the edge info.
6494 (record_edge_info): Use new class & associated member functions.
6495 Tighten forms for testing for edge equivalences.
6496 (record_temporary_equivalences): Iterate over the simple
6497 equivalences rather than assuming there's only one per edge.
6498 (cprop_into_successor_phis): Iterate over the simple
6499 equivalences rather than assuming there's only one per edge.
6500 (optimize_stmt): Use operand_equal_p rather than pointer
6501 equality for mini-DSE code.
6503 2017-08-28 Nathan Sidwell <nathan@acm.org>
6505 * gcc.c (execute): Fold SIGPIPE handling into switch
6506 statement. Adjust internal error message.
6508 2017-08-28 Richard Biener <rguenther@suse.de>
6511 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6512 Do nothing for removed DIEs.
6514 2017-08-28 Richard Biener <rguenther@suse.de>
6516 PR tree-optimization/81977
6517 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6520 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6523 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6524 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6525 using targetm.gen_mem_thread_fence.
6527 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6530 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6531 predicate to register_operand. Reorder operands.
6532 (*btr<mode>): Ditto.
6533 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6534 (*btr<mode>_mask): Ditto.
6536 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6538 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6539 * config/rs6000/xmmintrin.h: New file.
6540 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6542 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6545 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6546 parameter and_insn and return it.
6547 (recombine_lvx_pattern): Insert a copy to ensure availability of
6548 the base register of the copied masking operation at the point of
6549 the instruction replacement.
6550 (recombine_stvx_pattern): Likewise.
6552 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6554 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6555 undocumented switches.
6556 (-mpower9-dform-vector): Likewise.
6557 (-mpower9-dform): Likewise.
6558 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6559 comments to delete references to -mpower9-dform* switches.
6560 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6561 Delete reference to -mpower9-dform* switches, test for
6562 -mpower9-vector instead.
6563 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6564 (OTHER_P9_VECTOR_MASKS): Likewise.
6565 (POWERPC_MASKS): Likewise.
6566 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6567 tests against -mpower9-dform* to -mpower9-vector. Delete code
6568 that checked for -mpower9-dform* consistancy with other options.
6569 Add test for -mpower9-misc to enable other power9 switches.
6570 (rs6000_init_hard_regno_mode_ok): Likewise.
6571 (rs6000_option_override_internal): Likewise.
6572 (rs6000_emit_prologue): Likewise.
6573 (rs6000_emit_epilogue): Likewise.
6574 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6575 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6576 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6578 (emit_fusion_p9_store): Likewise.
6579 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6580 resetting these macros if the assembler does not support ISA 3.0
6582 (TARGET_P9_DFORM_VECTOR): Likewise.
6583 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6584 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6586 2017-08-25 Alan Modra <amodra@gmail.com>
6588 PR rtl-optimization/81747
6589 * cse.c (cse_extended_basic_block): Don't attempt to record
6590 equivalences for degenerate conditional branches that branch
6591 to their fall-through.
6593 2017-08-24 Martin Sebor <msebor@redhat.com>
6596 * gimple-fold.c (size_must_be_zero_p): New function.
6597 (gimple_fold_builtin_memory_op): Call it.
6599 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6601 * config/rs6000/mm_malloc.h: New file.
6603 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6605 PR tree-optimization/81913
6606 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6607 analysis when either IVs in condition can wrap.
6609 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6611 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6612 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6614 2017-08-24 Richard Biener <rguenther@suse.de>
6617 * targhooks.c (default_target_can_inline_p): Properly
6618 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6619 is present and always compare.
6620 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6621 infer -mfpmath=sse from TARGET_SSE_P.
6622 (ix86_can_inline_p): Properly use target_option_default_node when
6623 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6625 2017-08-24 Richard Biener <rguenther@suse.de>
6628 * dwarf2out.c (output_die): Handle flag_generate_offload like
6630 (output_comp_unit): Likewise.
6631 (gen_array_type_die): Likewise.
6632 (dwarf2out_early_finish): Likewise.
6633 (note_variable_value_in_expr): Likewise.
6634 (dwarf2out_finish): Likewise. Adjust assert.
6635 * cgraphunit.c (symbol_table::compile): Move setting of
6636 flag_generate_offload earlier ...
6637 (symbol_table::finalize_compilation_unit): ... here, before
6638 early debug finalization.
6640 2017-08-24 Richard Biener <rguenther@suse.de>
6642 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6643 and ipa-fnsummary.h.
6644 (ix86_can_inline_p): When ix86_fpmath flags do not match
6645 check whether the callee uses FP math at all.
6647 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6650 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6651 instead of TYPE_PRECISION.
6653 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6654 Alan Hayward <alan.hayward@arm.com>
6655 David Sherwood <david.sherwood@arm.com>
6657 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6659 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6661 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6663 (ix86_frame::stack_realign_allocate): New field.
6664 (struct machine_frame_state): Modify comments.
6665 (machine_frame_state::sp_realigned_fp_end): New field.
6666 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6668 (sp_valid_at): Add assertion to assure no attempt to access invalid
6669 offset of a realigned stack.
6670 (fp_valid_at): Likewise.
6671 (choose_baseaddr): Modify comments.
6672 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6673 ix86_expand_prologue.
6674 (ix86_expand_prologue): Modify stack realignment and allocation.
6675 (ix86_expand_epilogue): Modify comments.
6676 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6677 avx2_runtime, avx512f, and avx512f_runtime.
6679 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6681 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6682 (mstackrealign): Do not init to -1.
6683 * config/i386/i386.c (ix86_option_override_internal):
6684 Check opts_set, not opts when setting default value of
6685 opts->x_ix86_force_align_arg_pointer.
6687 2017-08-23 Richard Biener <rguenther@suse.de>
6689 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6690 lang_hooks.decl_printable_name.
6691 * print-rtl-function.c (print_rtx_function): Likewise.
6692 * tree-pretty-print.c (dump_function_header): Likewise.
6694 2017-08-23 Richard Biener <rguenther@suse.de>
6697 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6698 -g0 at compile-time.
6700 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6703 * doc/sourcebuild.texi (Other hardware attributes):
6706 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6709 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6710 Fix single-use check.
6712 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6714 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6716 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6718 * doc/install.texi: Modify to add more details on running selected
6721 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6723 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6724 is combined with -mabi=ms.
6725 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6728 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6730 PR tree-optimization/81488
6731 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6732 and cached_basis fields.
6733 (MAX_SPREAD): New constant.
6734 (alloc_cand_and_find_basis): Initialize new fields.
6735 (clear_visited): New function.
6736 (create_phi_basis_1): Rename from create_phi_basis, set visited
6737 and cached_basis fields.
6738 (create_phi_basis): New wrapper function.
6739 (phi_add_costs_1): Rename from phi_add_costs, add spread
6740 parameter, set visited field, short-circuit when limits reached.
6741 (phi_add_costs): New wrapper function.
6742 (record_phi_increments_1): Rename from record_phi_increments, set
6744 (record_phi_increments): New wrapper function.
6745 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6746 (phi_incr_cost): New wrapper function.
6747 (all_phi_incrs_profitable_1): Rename from
6748 all_phi_incrs_profitable, set visited field.
6749 (all_phi_incrs_profitable): New wrapper function.
6751 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6752 Alan Hayward <alan.hayward@arm.com>
6753 David Sherwood <david.sherwood@arm.com>
6755 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6756 that takes the outer and inner modes.
6757 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6758 comparison as the canonical test for a paradoxical subreg.
6759 * combine.c (simplify_set): Use paradoxical_subreg_p.
6760 (make_extraction): Likewise.
6761 (force_to_mode): Likewise.
6762 (rtx_equal_for_field_assignment_p): Likewise.
6763 (gen_lowpart_for_combine): Likewise.
6764 (simplify_comparison): Likewise.
6765 * cse.c (equiv_constant): Likewise.
6766 * expmed.c (store_bit_field_1): Likewise.
6767 * final.c (alter_subreg): Likewise.
6768 * fwprop.c (propagate_rtx): Likewise.
6769 (forward_propagate_subreg): Likewise.
6770 * ira-conflicts.c (ira_build_conflicts): Likewise.
6771 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6772 * lra-constraints.c (curr_insn_transform): Likewise.
6773 (split_reg): Likewise.
6774 * lra-eliminations.c (move_plus_up): Likewise.
6775 (lra_eliminate_regs_1): Likewise.
6776 * recog.c (general_operand): Likewise.
6777 * ree.c (combine_reaching_defs): Likewise.
6778 * reload.c (push_reload): Likewise.
6779 (find_reloads): Likewise.
6780 * reload1.c (elimination_effects): Likewise.
6781 (compute_reload_subreg_offset): Likewise.
6782 (choose_reload_regs): Likewise.
6783 * rtlanal.c (subreg_lsb_1): Likewise.
6784 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6785 (simplify_subreg): Likewise.
6786 * var-tracking.c (track_loc_p): Likewise.
6787 * emit-rtl.c (byte_lowpart_offset): Likewise.
6788 (paradoxical_subreg_p): Delete out-of-line definition.
6790 2017-08-22 Jeff Law <law@redhat.com>
6792 PR tree-optimization/81741
6793 PR tree-optimization/71947
6794 * tree-ssa-dom.c: Include tree-inline.h.
6795 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
6796 equivalences if one is more expensive to compute than the other.
6797 * tree-ssa-scopedtables.h (class const_or_copies): Make
6798 record_const_or_copy_raw method private.
6799 (class avail_exprs_stack): New method simplify_binary_operation.
6800 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
6801 avail_exprs_stack::simplify_binary_operation as needed.
6802 (avail_exprs_stack::simplify_binary_operation): New function.
6804 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
6806 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
6807 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
6808 (DOT_SYMBOLS): Likewise.
6809 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
6810 (RELOCATABLE_NEEDS_FIXUP): Likewise.
6811 (RS6000_ABI_NAME): Likewise.
6812 (TARGET_CMODEL): Likewise.
6813 (TOC_SECTION_ASM_OP): Likewise.
6814 (SET_CMODEL): New macro.
6815 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6817 2017-08-22 Richard Biener <rguenther@suse.de>
6819 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6820 to allow for free-lang-data replacements similar to verify_type_variant.
6822 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6823 Alan Hayward <alan.hayward@arm.com>
6824 David Sherwood <david.sherwood@arm.com>
6826 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6827 VOIDmode for the LABEL_REF.
6829 2017-08-22 Richard Biener <rguenther@suse.de>
6831 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6833 2017-08-22 Richard Biener <rguenther@suse.de>
6835 * common.opt (feliminate-dwarf2-dups): Ignore.
6836 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6837 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6838 same_die_p_wrap, compute_section_prefix,
6839 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6840 (comdat_symbol_id, comdat_symbol_number): Likewise.
6841 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6843 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6844 (output_die): Mark unreachable path unreachable.
6845 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6846 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6847 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6848 (dwarf2out_early_finish): Likewise.
6850 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6852 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6854 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6857 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6858 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6859 (avr_attribute_table) <io, io_low, address>: Initialize
6860 .decl_required with true.
6862 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6864 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6865 undocumented debugging options.
6866 (-mvsx-scalar-double): Likewise.
6867 (-mallow-df-permute): Likewise.
6868 (-mvectorize-builtins): Likewise.
6869 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6870 (rs6000_builtin_vectorized_function): Likewise.
6871 (rs6000_builtin_md_vectorized_function): Likewise.
6872 (rs6000_opt_vars): Likewise.
6874 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6877 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6878 (*btrq_imm): Rename from *btrq.
6879 (*btcq_imm): Rename from *btcq.
6880 (btsc): New code attribute.
6881 (*<btsc><mode>): New insn pattern.
6882 (*btr<mode>): Ditto.
6883 (*<btsc><mode>_mask): New insn_and_split pattern.
6884 (*btr<mode>_mask): Ditto.
6886 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6887 Alan Hayward <alan.hayward@arm.com>
6888 David Sherwood <david.sherwood@arm.com>
6890 * function.c (pad_below): Simplify padding calculation.
6892 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6893 Alan Hayward <alan.hayward@arm.com>
6894 David Sherwood <david.sherwood@arm.com>
6896 * target.def (function_prologue): Remove frame size argument.
6897 (function_epilogue): Likewise.
6898 * doc/tm.texi: Regenerate.
6899 * final.c (final_start_function): Update call to function_prologue.
6900 (final_end_function): Update call to function_epilogue.
6901 (default_function_pro_epilogue): Remove frame size argument.
6902 * output.h (default_function_pro_epilogue): Likewise.
6903 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6904 (arm_output_function_prologue): Likewise.
6905 * config/frv/frv.c (frv_function_prologue): Likewise.
6906 (frv_function_epilogue): Likewise.
6907 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6908 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6909 (ia64_output_function_epilogue): Likewise.
6910 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6911 (m32r_output_function_epilogue): Likewise.
6912 * config/microblaze/microblaze.c (microblaze_function_prologue)
6913 (microblaze_function_epilogue): Likewise.
6914 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6915 (mips_output_function_epilogue): Likewise.
6916 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6917 (mmix_target_asm_function_epilogue): Likewise.
6918 * config/msp430/msp430.c (msp430_start_function): Likewise.
6919 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6920 (nds32_asm_function_epilogue): Likewise.
6921 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6922 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6923 (pa_output_function_epilogue): Likewise.
6924 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6925 (rs6000_output_function_epilogue): Likewise.
6926 * config/rl78/rl78.c (rl78_start_function): Likewise.
6927 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6928 (rs6000_output_function_epilogue): Likewise.
6929 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6930 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6931 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6932 (sparc_asm_function_epilogue): Likewise.
6934 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6936 * tree.h (type_has_mode_precision_p): New function.
6937 * convert.c (convert_to_integer_1): Use it.
6938 * expr.c (expand_expr_real_2): Likewise.
6939 (expand_expr_real_1): Likewise.
6940 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6941 * match.pd: Likewise.
6942 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6943 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6944 * tree-tailcall.c (process_assignment): Likewise.
6945 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6946 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6947 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6948 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6949 (vectorizable_assignment): Likewise.
6950 (vectorizable_shift): Likewise.
6951 (vectorizable_operation): Likewise.
6952 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6954 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6956 * match.pd: Add pow (C, x) simplification.
6958 2017-08-21 Richard Biener <rguenther@suse.de>
6960 PR tree-optimization/81900
6961 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6962 for blocks with abnormal predecessors.
6963 (compute_antic): Do not set visited flag prematurely.
6965 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6968 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6970 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6972 * stor-layout.h (vector_type_mode): Move to...
6973 * tree.h (vector_type_mode): ...here.
6974 * stor-layout.c (vector_type_mode): Move to...
6975 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6977 2017-08-21 Richard Biener <rguenther@suse.de>
6979 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6980 register_external_die hooks.
6981 (debug_false_tree_charstarstar_uhwistar): Declare.
6982 (debug_nothing_tree_charstar_uhwi): Likewise.
6983 * debug.c (do_nothing_debug_hooks): Adjust.
6984 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6985 (debug_nothing_tree_charstar_uhwi): Likewise.
6986 * dbxout.c (dbx_debug_hooks): Adjust.
6987 (xcoff_debug_hooks): Likewise.
6988 * sdbout.c (sdb_debug_hooks): Likewise.
6989 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6990 * dwarf2out.c (macinfo_label_base): New global.
6991 (dwarf2out_register_external_die): New function for the
6992 register_external_die hook.
6993 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6994 (dwarf2_debug_hooks): Use them.
6995 (dwarf2_lineno_debug_hooks): Adjust.
6996 (struct die_struct): Add with_offset flag.
6997 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6998 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6999 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7000 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7001 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7002 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7003 defining section names for the early LTO debug variants.
7004 (reset_indirect_string): New helper.
7005 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7006 (print_dw_val): Add support for offsetted symbol references.
7007 (get_ultimate_context): Split out from is_cxx.
7008 (is_cxx): Use get_ultimate_context.
7009 (is_fortran): Add decl overload.
7010 (compute_comp_unit_symbol): Split out worker from
7011 compute_section_prefix.
7012 (compute_section_prefix): Call compute_comp_unit_symbol and
7013 set comdat_type_p here.
7014 (output_die): Skip DIE symbol output for the LTO added one.
7015 Handle DIE symbol references with offset.
7016 (output_comp_unit): Guard section name mangling properly.
7017 For LTO debug sections emit a symbol at the section beginning
7018 which we use to refer to its DIEs.
7019 (add_abstract_origin_attribute): For DIEs registered via
7020 dwarf2out_register_external_die directly refer to the early
7021 DIE rather than indirectly through the shadow one we created.
7022 Remove obsolete call to dwarf2out_abstract_function for
7023 non-function/block origins.
7024 (gen_array_type_die): When generating early LTO debug do
7025 not emit DW_AT_string_length.
7026 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7027 late when in LTO. As suggested place a gcc_unreachable for
7028 the DECL_ABSTRACT_P case.
7029 (gen_subprogram_die): Avoid another specification DIE
7030 for early built declarations/definitions for the late LTO case.
7031 (gen_variable_die): Add type references for late duplicated VLA dies
7033 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7034 we have the abstract instance already.
7035 (process_scope_var): Adjust decl DIE contexts in LTO which
7036 first puts them in limbo.
7037 (gen_decl_die): Do not generate type DIEs late apart from
7038 types for VLAs or for decls we do not yet have a DIE. Do not
7039 call dwarf2out_abstract_function late.
7040 (dwarf2out_early_global_decl): Make sure to create DIEs
7041 for abstract instances of a decl first.
7042 (dwarf2out_late_global_decl): Adjust comment.
7043 (output_macinfo_op): With multiple macro sections use
7044 macinfo_label_base to distinguish labels.
7045 (output_macinfo): Likewise. Update macinfo_label_base.
7046 Pass in the line info label.
7047 (note_variable_value_in_expr): When generating LTO resolve
7048 all variable values here by generating DIEs as needed.
7049 (init_sections_and_labels): Add early LTO debug flag parameter
7050 and generate different sections and names if set. Add generation
7051 counter for the labels so we can have multiple of them.
7052 (reset_dies): Helper to allow DIEs to be output multiple times.
7053 (dwarf2out_finish): When outputting DIEs to the fat part of an
7054 LTO object first reset DIEs.
7055 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7056 (modified_type_die): Check for decl_ultimate_origin being self
7058 (gen_type_die_with_usage): Likewise.
7059 (gen_typedef_die): Allow decl_ultimate_origin being self.
7060 (set_decl_abstract_flags): Remove.
7061 (set_block_abstract_flags): Likewise.
7062 (dwarf2out_abstract_function): Treat the early generated DIEs
7063 as the abstract copy and only add DW_AT_inline and
7064 DW_AT_artificial here and call set_decl_origin_self.
7065 If the DIE has an abstract origin don't do anything.
7066 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7067 if we have none yet (Go fails to build one, PR78628).
7068 (variably_modified_type_p): Prevent endless recursion for Ada
7069 cyclic pointer types.
7070 * lto-streamer-in.c: Include debug.h.
7071 (dref_queue): New global.
7072 (lto_read_tree_1): Stream in DIE references.
7073 (lto_input_tree): Register DIE references.
7074 (input_function): Stream DECL_DEBUG_ARGS.
7075 * lto-streamer-out.c: Include debug.h.
7076 (lto_write_tree_1): Output DIE references.
7077 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7078 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7079 (output_function): Stream DECL_DEBUG_ARGS.
7080 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7081 Stream DECL_ABSTRACT_ORIGIN.
7082 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7083 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7084 DECL_CONTEXT for file-scope decls.
7085 * lto-streamer.h (struct dref_entry): Declare.
7086 (dref_queue): Likewise.
7087 * cfgexpand.c (pass_expand::execute): Do not call the
7088 outlining_inline_function hook here.
7089 * lto-wrapper.c (debug_obj): New global.
7090 (tool_cleanup): Unlink it if required.
7091 (debug_objcopy): New function.
7092 (run_gcc): Handle early debug sections in the IL files by
7093 extracting them to separate files, partially linkin them and
7094 feeding the result back as result to the linker.
7095 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7096 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7097 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7098 sections into a separate segment.
7099 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7101 (darwin_asm_dwarf_section): Likewise.
7102 (darwin_asm_output_dwarf_offset): Likewise.
7103 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7105 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayward <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * read-md.h (md_reader::record_potential_iterator_use): Replace
7110 pointer argument with an rtx and an index.
7111 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7112 (apply_mode_iterator): Likewise.
7113 (apply_code_iterator): Likewise.
7114 (apply_int_iterator): Likewise.
7115 (apply_subst_iterator): Likewise.
7116 (record_iterator_use): Likewise.
7117 (record_attribute_use): Likewise.
7118 (md_reader::record_potential_iterator_use): Likewise. Update calls
7119 to record_iterator_use and apply_iterator.
7120 (iterator_use): Replace ptr with x and index.
7121 (attribute_use): Likewise.
7122 (apply_attribute_uses): Update calls to apply_iterator.
7123 (apply_iterators): Likewise. Update initialization of iterator_use.
7124 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7125 and record_potential_iterator_use.
7126 (rtx_reader::read_rtx_operand): Likewise.
7128 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7129 Alan Hayward <alan.hayward@arm.com>
7130 David Sherwood <david.sherwood@arm.com>
7132 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7135 2017-08-21 Richard Biener <rguenther@suse.de>
7138 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7139 at struct end conservatively when comparing common bases.
7141 2017-08-21 Richard Biener <rguenther@suse.de>
7143 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7144 (mem_ref_in_stmt): Remove.
7145 (determine_max_movement): Use ref index to get at the reference.
7146 (invariantness_dom_walker::before_dom_children): Deal with
7147 lim data already initialized.
7148 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7150 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7152 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7153 (TARGET_ISA_ROUND): Ditto.
7154 (TARGET_ROUND): Ditto.
7155 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7156 * config/i386/i386.md: Ditto.
7157 * config/i386/sse.md: Ditto.
7158 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7159 with OPTION_MASK_ISA_SSE4_1.
7161 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7164 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7165 __builtin_ia32_lzcnt_u16.
7167 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7170 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7171 (rs6000_set_current_function): Rewrite function to use it.
7173 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7176 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7177 and TYPE_WARN_IF_NOT_ALIGN.
7178 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7179 (handle_warn_if_not_align): New.
7180 (place_union_field): Call handle_warn_if_not_align.
7181 (place_field): Call handle_warn_if_not_align.
7182 Copy TYPE_WARN_IF_NOT_ALIGN.
7183 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7184 (layout_type): Likewise.
7185 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7187 (tree_decl_common): Add warn_if_not_align.
7188 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7189 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7190 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7191 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7192 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7193 * doc/extend.texi: Document warn_if_not_aligned attribute.
7194 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7196 2017-08-17 Martin Liska <mliska@suse.cz>
7199 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7200 (get_data_dependence): Use it as pointer type.
7201 (distribute_loop): Likewise.
7203 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7205 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7206 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7207 (p8_vmrgow_<mode>_direct): New define_insn.
7208 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7209 handle endianness for vmrgew and vmrgow permute patterns.
7211 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7213 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7214 * config/rs6000/rs6000-cpus.def: Remove comment.
7215 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7216 (POWERPC_MASKS): Likewise.
7217 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7218 use of TARGET_VSX_TIMODE.
7219 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7220 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7221 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7222 (rs6000_option_override_internal): Remove dead code.
7223 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7224 (rs6000_legitimize_reload_address): Likewise.
7225 (rs6000_legitimate_address_p): Likewise.
7226 (rs6000_opt_masks): Delete "vsx-timode".
7227 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7228 from function comment.
7229 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7230 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7231 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7233 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7234 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7235 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7238 (define_peephole2): Likewise.
7240 2017-08-17 Martin Sebor <msebor@redhat.com>
7243 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7244 past the end of an array.
7245 (test_pp_format): Add test cases.
7247 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7249 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7250 missing ECF_NOTHROW flags.
7252 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7255 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7256 operands residing in integer registers.
7257 (*vsx_le_perm_load_<mode>): Likewise.
7258 (*vsx_le_perm_store_<mode>): Likewise.
7259 (define_peephole2): Add peepholes to optimize the above.
7261 2017-08-17 Marek Polacek <polacek@redhat.com>
7264 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7265 to mimic what shorten_compare did. Change the return type to bool.
7266 (fold_cond_expr_with_comparison): Update call to
7267 operand_equal_for_comparison_p.
7268 (fold_ternary_loc): Likewise.
7270 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7272 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7274 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7276 2017-08-17 Richard Biener <rguenther@suse.de>
7278 * tree-ssa-structalias.c (solve_graph): When propagating
7279 to successors update the graphs succ edges and avoid duplicate work.
7281 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7284 * config/i386/i386.c (ix86_option_override_internal): Save target
7285 specific options after ix86_stack_protector_guard_reg was changed.
7287 2017-08-17 Richard Biener <rguenther@suse.de>
7289 PR tree-optimization/81827
7290 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7291 (new_var_info): Initialize it conservatively.
7292 (get_call_vi): Mark register vars.
7293 (new_scalar_tmp_constraint_exp): Likewise.
7294 (handle_rhs_call): Likewise.
7295 (handle_const_call): Likewise.
7296 (create_function_info_for): Likewise.
7297 (solve_constraints): Sort varinfos to separate register from
7298 non-register vars to pack points-to solution bitmaps during
7301 2017-08-17 Marek Polacek <polacek@redhat.com>
7303 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7305 2017-08-17 Richard Biener <rguenther@suse.de>
7307 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7308 to true when overflow is undefined and we saturated the result.
7310 2017-08-17 Alan Modra <amodra@gmail.com>
7313 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7314 Don't use store multiple if only one reg needs saving.
7315 (interesting_frame_related_regno): New function.
7316 (rs6000_frame_related): Don't emit frame info for regs that
7318 (rs6000_emit_epilogue): Likewise.
7320 2017-08-16 Nathan Sidwell <nathan@acm.org>
7322 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7323 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7324 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7325 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7326 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7327 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7328 (verify_type): Adjust for TYPE_BINFO move.
7329 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7330 process TYPE_BINFO directly.
7331 (hash_tree): Likewise.
7332 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7334 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7337 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7339 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7341 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7344 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7345 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7346 Add ix86_binary_operator_ok to insn constraint.
7347 (*iordi_1_bts): Ditto.
7348 (*xordi_1_btc): Ditto.
7349 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7350 Update corresponding peephole2 pattern.
7354 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7356 PR tree-optimization/81832
7357 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7358 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7360 2017-08-16 Marek Polacek <polacek@redhat.com>
7363 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7364 regarding size_zero_node.
7366 2017-08-16 Martin Liska <mliska@suse.cz>
7369 * config.gcc: Respect previously set extra_objs in case
7372 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7374 PR tree-optimization/81835
7375 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7376 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7377 not depend on the phi.
7379 2017-08-16 Alan Modra <amodra@gmail.com>
7381 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7384 2017-08-16 Alan Modra <amodra@gmail.com>
7386 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7387 (save_reg_p): ..into this. Update all callers.
7388 (first_reg_to_save): Simplify.
7390 2017-08-16 Alan Modra <amodra@gmail.com>
7392 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7395 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7399 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7400 constant count if that count is less than 32.
7402 2017-08-15 Nathan Sidwell <nathan@acm.org>
7404 * gcc.c (execute): Emit friendlier message if inferior is killed
7405 by an external cause.
7407 2017-08-15 Richard Biener <rguenther@suse.de>
7409 PR tree-optimization/81790
7410 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7411 CONSTRUCTORs from simplifying and VN.
7413 2017-08-14 Martin Sebor <msebor@redhat.com>
7415 * builtin-attrs.def: Add comments.
7417 2017-08-14 Martin Sebor <msebor@redhat.com>
7420 * doc/extend.texi (attribute nonstring): Document new attribute.
7422 2017-08-14 Martin Sebor <msebor@redhat.com>
7425 * tree-diagnostic.c (default_tree_printer): Handle %G.
7426 * gimple-pretty-print.h (percent_G_format): Declare new function.
7427 * gimple-pretty-print.c (percent_G_format): Define.
7428 * tree-pretty-print.c (percent_K_format): Add argument.
7430 2017-08-14 Martin Sebor <msebor@redhat.com>
7432 PR translation/79998
7433 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7434 Remove a stray space.
7436 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7439 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7440 (*iordi_1_bts): Ditto.
7441 (*xordi_1_btc): Ditto.
7443 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7446 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7447 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7449 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7450 quoted strings, and make more translator-friendly.
7451 (darwin_rs6000_override_options): Likewise.
7452 (rs6000_option_override_internal): Likewise.
7453 (rs6000_return_in_memory): Fix overlong line.
7454 (init_cmulative_args): Use quoted strings, and make more
7455 translator-friendly.
7456 (rs6000_pass_by_reference): Fix overlong line.
7457 (def_builtin): Use quoted strings.
7458 (altivec_expand_predicate_builtin): Use quoted strings, and make
7459 more translator-friendly.
7460 (htm_expand_builtin): Use quoted strings.
7461 (cpu_expand_builtin): Use quoted strings, and make more
7462 translator-friendly.
7463 (altivec_expand_builtin): Likewise.
7464 (paired_expand_predicate_builtin): Likewise.
7465 (rs6000_invalid_builtin): Likewise.
7466 (builtin_function_type): Use quoted strings.
7467 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7468 more translator-friendly.
7469 (rs6000_trampoline_init): Likewise.
7470 (rs6000_handle_altivec_attribute): Likewise.
7471 (rs6000_inner_target_options): Use quoted strings.
7472 (rs6000_disable_incompatible_switches): Likewise.
7473 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7474 strings, and make more translator-friendly.
7475 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7477 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7479 PR tree-optimization/81799
7480 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7481 cond_expr to simple gimple operand.
7483 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7486 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7488 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7492 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7493 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7494 TARGET_GASISR_PROLOGUES.
7495 * config/avr/avr.c (avr_option_override): Same.
7496 (avr_pass_pre_proep::execute): Same.
7498 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7501 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7502 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7504 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7506 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7507 define_insn_and_split. Split to a memory load from 0 in
7508 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7509 using PTR mode iterator.
7510 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7511 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7512 (*add_tp_<mode>): Redefine as define_insn_and_split.
7513 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7514 address space. Merge with *add_tp_x32 using PTR mode iterator.
7515 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7516 Split to an add with a memory load from 0 in
7517 DEFAULT_TLS_SEG_REG address space.
7519 2017-08-12 Andrew Pinski <apinski@cavium.com>
7521 * config/aarch64/aarch64-option-extensions.def (rdma):
7522 Fix feature string to what Linux prints out in /proc/cpuinfo.
7524 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7527 * dwarf2out.c (modified_type_die): For C typedef types that have
7528 an ultimate origin, process the ultimate origin instead of the
7530 (gen_typedef_die): Assert that input DECLs have no ultimate
7532 (gen_type_die_with_usage): For typedef variants that have an
7533 ultimate origin, just call gen_decl_die on the original DECL.
7534 (process_scope_var): Avoid creating DIEs for local typedefs and
7535 concrete static variables.
7537 2017-08-12 Alan Modra <amodra@gmail.com>
7541 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7542 match gnu-user.h startfile.
7543 (ENDFILE_LINUX_SPEC): Similarly.
7545 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7548 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7550 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7552 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7553 * config/aarch64/aarch64.md (mov<mode>): Change.
7554 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7555 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7556 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7558 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7560 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7561 for storage order barriers.
7563 2017-08-11 Martin Liska <mliska@suse.cz>
7566 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7567 variables of void type.
7569 2017-08-11 Martin Liska <mliska@suse.cz>
7571 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7572 TARGET_SUPPORTS_ALIASES.
7573 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7574 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7575 (optimize_weakref): Likewise.
7576 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7577 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7578 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7580 2017-08-11 Martin Liska <mliska@suse.cz>
7583 * config/i386/i386.c (make_resolver_func): Do complete
7584 refactoring of the function.
7586 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7589 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7590 * config/i386/i386.c (ix86_stack_protect_guard): Use
7591 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7592 * doc/invoke.texi (x86 Options): Document
7593 -mstack-protector-guard-symbol= option.
7595 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7597 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7598 * config/i386/i386.c (ix86_split_stack_guard): New function.
7599 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7600 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7601 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7602 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7603 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7604 (split_stack_space_check): Call ix86_split_stack_guard.
7606 2017-08-10 Martin Sebor <msebor@redhat.com>
7608 * print-tree.c (print_node): Print location using the established
7610 Replace spaces with colons.
7611 (debug_raw, debug): Ditto.
7613 2017-08-10 Martin Sebor <msebor@redhat.com>
7616 * pretty-print.c (pp_format): Correct the handling of %s precision.
7618 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7621 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7623 (ix86_finalize_stack_frame_flags): This. Also clear
7624 frame_pointer_needed if -fno-omit-frame-pointer is used without
7626 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7627 with ix86_finalize_stack_frame_flags.
7628 (ix86_expand_epilogue): Likewise.
7629 (ix86_expand_split_stack_prologue): Likewise.
7630 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7632 2017-08-10 Martin Liska <mliska@suse.cz>
7635 * c-attribs.c (handle_target_attribute):
7636 Report warning for an empty string argument of target attribute.
7638 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7641 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7643 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7644 or DECL_NONLOCAL labels.
7645 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7646 or DECL_NONLOCAL labels here.
7648 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7651 to indicate when early gimple folding has been disabled.
7652 (rs6000_gimple_fold_builtin): Add debug content.
7653 (rs6000_invalid_builtin): Fix whitespace.
7654 (rs6000_expand_builtin): Fix whitespace.
7655 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7657 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7660 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7661 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7663 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7665 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7666 (qdf24xx): Likewise.
7667 * config/aarch64/aarch64.md: Include falkor.md.
7668 * config/aarch64/falkor.md: New.
7670 2017-08-09 Marek Polacek <polacek@redhat.com>
7673 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7674 * diagnostic.c (emit_diagnostic): Add a comment.
7675 (emit_diagnostic_valist): New function.
7677 2017-08-09 Marek Polacek <polacek@redhat.com>
7680 * input.c (make_location): New overload.
7681 * input.h (make_location): Declare.
7683 2017-08-08 Alan Modra <amodra@gmail.com>
7684 H.J. Lu <hongjiu.lu@intel.com>
7687 * gcc.c (NO_PIE_SPEC): Delete.
7688 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7690 (LINK_PIE_SPEC): ..to here.
7691 (LINK_COMMAND_SPEC): Support -no-pie.
7692 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7693 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7695 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7696 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7697 (ENDFILE_CRTEND_SPEC): Similarly.
7699 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7702 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7703 (mstack-protector-guard-offset=): Ditto.
7704 * config/i386/i386.c (ix86_option_override): Handle
7705 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7707 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7708 ix86_stack_protect_guard_offset variables.
7709 (TARGET_STACK_PROTECT_GUARD): Always define.
7710 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7711 and -mstack-protector-guard-offset= options.
7713 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7715 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7716 boundary case for the last candidate.
7718 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7720 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7721 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7723 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7726 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7727 * config/aarch64/aarch64-builtins.c
7728 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7729 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7730 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7732 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7733 Andrew Pinski <pinskia@gmail.com>
7736 * internal-fn.def (XORSIGN): New.
7737 * optabs.def (xorsign_optab): New.
7738 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7739 (convert_expand_mult_copysign): New.
7740 (pass_optimize_widening_mul::execute): Call
7741 convert_expand_mult_copysign.
7743 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7745 PR tree-optimization/81354
7746 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7747 Insert on edges rather than explicitly creating landing pads.
7748 (analyze_candidates_and_replace): Commit edge inserts.
7750 2017-08-08 Richard Biener <rguenther@suse.de>
7753 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7754 (expand_simple_operations): Also look through ADDR_EXPRs with
7755 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7757 2017-08-08 Richard Biener <rguenther@suse.de>
7759 PR tree-optimization/81723
7760 * tree-vect-slp.c (struct bst_traits): New hash traits.
7761 (bst_fail): New global.
7762 (vect_build_slp_tree_2): New worker, split out from ...
7763 (vect_build_slp_tree): ... this now wrapping it with using
7764 bst_fail set to cache SLP tree build fails. Properly handle
7766 (vect_analyze_slp_instance): Allocate and free bst_fail.
7768 2017-08-08 Martin Liska <mliska@suse.cz>
7771 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7772 LABEL_DECLs that can be from a different function.
7774 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7776 PR tree-optimization/81744
7777 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7778 loop's number of iterations.
7780 2017-08-08 Martin Liska <mliska@suse.cz>
7782 * asan.c: Include header files.
7783 * attribs.c (build_decl_attribute_variant): New function moved
7785 (build_type_attribute_qual_variant): Likewise.
7786 (cmp_attrib_identifiers): Likewise.
7787 (simple_cst_list_equal): Likewise.
7788 (omp_declare_simd_clauses_equal): Likewise.
7789 (attribute_value_equal): Likewise.
7790 (comp_type_attributes): Likewise.
7791 (build_type_attribute_variant): Likewise.
7792 (lookup_ident_attribute): Likewise.
7793 (remove_attribute): Likewise.
7794 (merge_attributes): Likewise.
7795 (merge_type_attributes): Likewise.
7796 (merge_decl_attributes): Likewise.
7797 (merge_dllimport_decl_attributes): Likewise.
7798 (handle_dll_attribute): Likewise.
7799 (attribute_list_equal): Likewise.
7800 (attribute_list_contained): Likewise.
7801 * attribs.h (lookup_attribute): New function moved from tree.[ch].
7802 (lookup_attribute_by_prefix): Likewise.
7803 * bb-reorder.c: Include header files.
7804 * builtins.c: Likewise.
7805 * calls.c: Likewise.
7806 * cfgexpand.c: Likewise.
7807 * cgraph.c: Likewise.
7808 * cgraphunit.c: Likewise.
7809 * convert.c: Likewise.
7810 * dwarf2out.c: Likewise.
7811 * final.c: Likewise.
7812 * fold-const.c: Likewise.
7813 * function.c: Likewise.
7814 * gimple-expr.c: Likewise.
7815 * gimple-fold.c: Likewise.
7816 * gimple-pretty-print.c: Likewise.
7817 * gimple.c: Likewise.
7818 * gimplify.c: Likewise.
7819 * hsa-common.c: Likewise.
7820 * hsa-gen.c: Likewise.
7821 * internal-fn.c: Likewise.
7822 * ipa-chkp.c: Likewise.
7823 * ipa-cp.c: Likewise.
7824 * ipa-devirt.c: Likewise.
7825 * ipa-fnsummary.c: Likewise.
7826 * ipa-inline.c: Likewise.
7827 * ipa-visibility.c: Likewise.
7829 * lto-cgraph.c: Likewise.
7830 * omp-expand.c: Likewise.
7831 * omp-general.c: Likewise.
7832 * omp-low.c: Likewise.
7833 * omp-offload.c: Likewise.
7834 * omp-simd-clone.c: Likewise.
7835 * opts-global.c: Likewise.
7836 * passes.c: Likewise.
7837 * predict.c: Likewise.
7838 * sancov.c: Likewise.
7839 * sanopt.c: Likewise.
7840 * symtab.c: Likewise.
7841 * toplev.c: Likewise.
7842 * trans-mem.c: Likewise.
7843 * tree-chkp.c: Likewise.
7844 * tree-eh.c: Likewise.
7845 * tree-into-ssa.c: Likewise.
7846 * tree-object-size.c: Likewise.
7847 * tree-parloops.c: Likewise.
7848 * tree-profile.c: Likewise.
7849 * tree-ssa-ccp.c: Likewise.
7850 * tree-ssa-live.c: Likewise.
7851 * tree-ssa-loop.c: Likewise.
7852 * tree-ssa-sccvn.c: Likewise.
7853 * tree-ssa-structalias.c: Likewise.
7854 * tree-ssa.c: Likewise.
7855 * tree-streamer-in.c: Likewise.
7856 * tree-vectorizer.c: Likewise.
7857 * tree-vrp.c: Likewise.
7859 * ubsan.c: Likewise.
7860 * varasm.c: Likewise.
7861 * varpool.c: Likewise.
7862 * tree.c: Remove functions moved to attribs.[ch].
7864 * config/aarch64/aarch64.c: Add attrs.h header file.
7865 * config/alpha/alpha.c: Likewise.
7866 * config/arc/arc.c: Likewise.
7867 * config/arm/arm.c: Likewise.
7868 * config/avr/avr.c: Likewise.
7869 * config/bfin/bfin.c: Likewise.
7870 * config/c6x/c6x.c: Likewise.
7871 * config/cr16/cr16.c: Likewise.
7872 * config/cris/cris.c: Likewise.
7873 * config/darwin.c: Likewise.
7874 * config/epiphany/epiphany.c: Likewise.
7875 * config/fr30/fr30.c: Likewise.
7876 * config/frv/frv.c: Likewise.
7877 * config/ft32/ft32.c: Likewise.
7878 * config/h8300/h8300.c: Likewise.
7879 * config/i386/winnt.c: Likewise.
7880 * config/ia64/ia64.c: Likewise.
7881 * config/iq2000/iq2000.c: Likewise.
7882 * config/lm32/lm32.c: Likewise.
7883 * config/m32c/m32c.c: Likewise.
7884 * config/m32r/m32r.c: Likewise.
7885 * config/m68k/m68k.c: Likewise.
7886 * config/mcore/mcore.c: Likewise.
7887 * config/microblaze/microblaze.c: Likewise.
7888 * config/mips/mips.c: Likewise.
7889 * config/mmix/mmix.c: Likewise.
7890 * config/mn10300/mn10300.c: Likewise.
7891 * config/moxie/moxie.c: Likewise.
7892 * config/msp430/msp430.c: Likewise.
7893 * config/nds32/nds32-isr.c: Likewise.
7894 * config/nds32/nds32.c: Likewise.
7895 * config/nios2/nios2.c: Likewise.
7896 * config/nvptx/nvptx.c: Likewise.
7897 * config/pa/pa.c: Likewise.
7898 * config/pdp11/pdp11.c: Likewise.
7899 * config/powerpcspe/powerpcspe.c: Likewise.
7900 * config/riscv/riscv.c: Likewise.
7901 * config/rl78/rl78.c: Likewise.
7902 * config/rx/rx.c: Likewise.
7903 * config/s390/s390.c: Likewise.
7904 * config/sh/sh.c: Likewise.
7905 * config/sol2.c: Likewise.
7906 * config/sparc/sparc.c: Likewise.
7907 * config/spu/spu.c: Likewise.
7908 * config/stormy16/stormy16.c: Likewise.
7909 * config/tilegx/tilegx.c: Likewise.
7910 * config/tilepro/tilepro.c: Likewise.
7911 * config/v850/v850.c: Likewise.
7912 * config/vax/vax.c: Likewise.
7913 * config/visium/visium.c: Likewise.
7914 * config/xtensa/xtensa.c: Likewise.
7916 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7919 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7920 constraints since the -mupper-regs-* switches have been
7922 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7923 into a vector from a double word element that was extracted from
7924 another vector, and eliminate extra XXPERMDI instructions.
7925 (vsx_concat_<mode>_2): Likewise.
7926 (vsx_concat_<mode>_3): Likewise.
7927 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7928 concat to allow optimizing inserts from previous extracts.
7930 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7932 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7933 memory reference to a SSP offset in TLS address space.
7934 (ix86_print_operand) <case '@'>: Remove.
7935 (ix86_print_operand_punct_valid_p): Remove '@' code.
7936 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7938 (stack_tls_protect_set_<mode>): Remove.
7939 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7940 (stack_tls_protect_test_<mode>): Remove.
7941 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7943 2017-08-07 Olivier Hainque <hainque@adacore.com>
7946 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7948 2017-08-07 Douglas Rupp <rupp@adacore.com>
7950 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7951 variable was referenced as multidir in command.
7953 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7956 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7959 2017-08-07 Martin Liska <mliska@suse.cz>
7961 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7962 * config/rl78/rl78.c: Add include of attribs.h.
7963 * config/sh/sh.c: Likewise.
7964 * config/v850/v850.c: Likewise.
7966 2017-08-07 Tom de Vries <tom@codesourcery.com>
7969 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7971 2017-08-07 Martin Liska <mliska@suse.cz>
7973 * config/mips/mips.c: Include attribs.h.
7975 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7978 * doc/invoke.texi: Document change in behvaior for -Ofast for
7981 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7983 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7985 (aarch64_pop_regs): Likewise.
7986 (aarch64_gen_load_pair): Likewise.
7987 (aarch64_save_callee_saves): Likewise.
7988 (aarch64_restore_callee_saves): Likewise.
7990 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7992 * config/i386/i386.c: Revert the last change.
7994 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7997 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7999 (ix86_finalize_stack_frame_flags): This. Also clear
8000 frame_pointer_needed if -fno-omit-frame-pointer is used without
8002 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8003 with ix86_finalize_stack_frame_flags.
8004 (ix86_expand_epilogue): Likewise.
8005 (ix86_expand_split_stack_prologue): Likewise.
8007 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8010 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8011 to P_AES for Westmere.
8013 2017-08-07 Jonathan Yong <10walls@gmail.com>
8015 * config/i386/mingw.opt (fset-stack-executable): Removed.
8016 * config/i386/cygming.opt (fset-stack-executable): Moved
8018 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8020 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8022 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8024 2017-08-07 Marek Polacek <polacek@redhat.com>
8027 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8029 2017-08-07 Martin Liska <mliska@suse.cz>
8031 * attribs.h (canonicalize_attr_name): New function.
8032 (cmp_attribs): Move from c-format.c and adjusted.
8033 (is_attribute_p): Moved from tree.h.
8034 * tree-inline.c: Add new includes.
8035 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8036 (private_is_attribute_p): Remove.
8037 (private_lookup_attribute): Likewise.
8038 (private_lookup_attribute_by_prefix): Simplify.
8039 (remove_attribute): Use is_attribute_p.
8040 * tree.h: Remove removed declarations.
8042 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8045 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8046 instead of computing it in the function. Formatting fix.
8047 (expand_case): Don't rely on default_edge being the first edge,
8048 clear it if removing it, pass default_edge to
8049 emit_case_dispatch_table.
8050 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8053 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8055 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8056 insn in the function, emit NOP after the insn.
8058 2017-08-06 Tom de Vries <tom@codesourcery.com>
8060 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8063 2017-08-06 Tom de Vries <tom@codesourcery.com>
8065 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8068 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8070 PR tree-optimization/57371
8071 * match.pd: New pattern.
8073 2017-08-04 Marek Polacek <polacek@redhat.com>
8076 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8077 perform the computation in offset_int.
8079 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8081 PR tree-optimization/81136
8082 * tree-vectorizer.h: Include tree-hash-traits.h.
8083 (vec_base_alignments): New typedef.
8084 (vec_info): Add a base_alignments field.
8085 (vect_record_base_alignments): Declare.
8086 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8088 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8089 (create_data_ref): Add an is_conditional_in_stmt argument.
8090 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8091 the is_conditional_in_stmt field.
8092 (data_ref_loc): Add an is_conditional_in_stmt field.
8093 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8094 (find_data_references_in_stmt): Update call to create_data_ref.
8095 (graphite_find_data_references_in_stmt): Likewise.
8096 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8097 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8098 (vect_record_base_alignment): New function.
8099 (vect_record_base_alignments): Likewise.
8100 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8101 for nested statements even if we fail to compute a misalignment.
8102 Use pooled base alignments for unconditional references.
8103 (vect_find_same_alignment_drs): Compare base addresses instead
8105 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8106 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8108 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8110 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8111 Add an explicit name for the enum. Use auto_vec for slp_instances
8113 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8115 (_bb_vec_info): Add a constructor and destructor.
8116 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8117 (destroy_loop_vec_info): Delete.
8118 (vect_destroy_datarefs): Likewise.
8119 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8120 (vec_info::vec_info): New function.
8121 (vec_info::~vec_info): Likewise.
8122 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8123 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8124 destroy_loop_vec_info.
8125 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8126 (_loop_vec_info::_loop_vec_info): ...this.
8127 (destroy_loop_vec_info): Replace with...
8128 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8129 the stmt_vec_infos. Leave handling of vec_info information to its
8130 destructor. Remove explicit vector releases.
8131 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8132 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8133 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8134 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8135 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8136 (destroy_bb_vec_info): Replace with...
8137 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8138 information to its destructor.
8139 (vect_slp_analyze_bb_1): Use new and delete instead of
8140 new_bb_vec_info and destroy_bb_vec_info.
8141 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8144 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8146 * tree-data-ref.h (subscript): Add access_fn field.
8147 (data_dependence_relation): Add could_be_independent_p.
8148 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8149 (same_access_functions): Move to tree-data-ref.c.
8150 * tree-data-ref.c (ref_contains_union_access_p): New function.
8151 (access_fn_component_p): Likewise.
8152 (access_fn_components_comparable_p): Likewise.
8153 (dr_analyze_indices): Add a reference to access_fn_component_p.
8154 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8156 (constant_access_functions): Likewise.
8157 (add_other_self_distances): Likewise.
8158 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8159 (initialize_data_dependence_relation): Use XCNEW and remove
8160 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8161 of access functions that have the same type. Allow the
8162 subsequence to end with different bases in some circumstances.
8163 Record the chosen access functions in SUB_ACCESS_FN.
8164 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8165 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8166 (subscript_dependence_tester_1): Likewise dra and drb.
8167 (build_classic_dist_vector): Update calls accordingly.
8168 (subscript_dependence_tester): Likewise.
8169 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8170 DDR_COULD_BE_INDEPENDENT_P.
8171 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8172 comp_alias_ddrs instead of may_alias_ddrs.
8173 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8175 (vect_analyze_data_ref_dependence): Use it if
8176 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8177 distance vectors if that fails.
8178 (dependence_distance_ge_vf): New function.
8179 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8180 LOOP_VINFO_MAY_ALIAS_DDRS.
8182 2017-08-04 Richard Biener <rguenther@suse.de>
8185 * fold-const.c (fold_binary_loc): Properly restrict
8186 minus_var0 && minus_var1 case when associating undefined overflow
8189 2017-08-04 Tom de Vries <tom@codesourcery.com>
8191 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8193 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8195 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8196 Don't start diagnostic messages with a capital letter.
8197 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8199 (rs6000_invalid_builtin): Likewise.
8200 (rs6000_trampoline_init): Likewise.
8202 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8205 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8206 after setting changeable df flags.
8208 2017-08-03 Richard Biener <rguenther@suse.de>
8210 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8211 up if the use is in USE - X.
8213 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8215 * toplev.c (dumpfile.h): New include.
8216 (internal_error_reentered): New static function. Use it...
8217 (internal_error_function): ...here to handle reentered internal_error.
8219 2017-08-03 Richard Biener <rguenther@suse.de>
8222 * fold-const.c (split_tree): Add minus_var and minus_con
8223 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8224 here but always use minus_*.
8225 (associate_trees): Assert we never associate with MINUS_EXPR
8226 and NULL first operand. Do not recurse for PLUS_EXPR operands
8227 when associating as MINUS_EXPR either.
8228 (fold_binary_loc): Track minus_var and minus_con.
8230 2017-08-03 Tom de Vries <tom@codesourcery.com>
8233 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8234 ACCEL_COMPILER, apply finish_options on
8235 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8237 2017-08-03 Tom de Vries <tom@codesourcery.com>
8240 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8241 function_entry_patch_area_size > 0.
8243 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8246 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8247 instead of 10??LU, perform unit multiplication in wide_int,
8248 don't change alloc_object_size_limit if the limit is larger
8251 PR tree-optimization/81655
8252 PR tree-optimization/81588
8253 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8254 the case when ranges[i].low and high are 1 for unsigned type with
8258 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8259 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8261 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8263 * tree-vrp.h: Add include guard.
8265 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8268 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8269 (ud2): New insn pattern.
8270 * config/i386/i386.c (ix86_expand_epilogue):
8271 For naked functions, generate ud2 instead of trap insn.
8273 2017-08-02 Marek Polacek <polacek@redhat.com>
8276 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8278 2017-08-02 Tom de Vries <tom@codesourcery.com>
8279 Cesar Philippidis <cesar@codesourcery.com>
8281 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8282 Add missing edge probabilities.
8284 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8286 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8289 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8292 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8293 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8294 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8295 prologue_seq sequences - if any.
8297 2017-08-02 Richard Biener <rguenther@suse.de>
8299 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8300 via vectors if supported, integer extracts via punning if supported
8301 or otherwise vector extracts.
8303 2017-08-02 Richard Biener <rguenther@suse.de>
8305 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8307 (bitmap_insert_into_set): ... this.
8309 2017-08-02 Richard Biener <rguenther@suse.de>
8311 PR tree-optimization/81633
8313 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8315 PR tree-optimization/71752
8316 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8318 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8320 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8321 (machine_function::call_ms2sysv_pad_out): Remove field.
8322 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8323 (ix86_compute_frame_layout): Likewise.
8325 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8328 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8329 attribute with interrupt attribute.
8331 2017-08-01 Andrew Pinski <apinski@cavium.com>
8333 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8334 BIT_INSERT_EXPR's operand 1
8335 to see if the types precision matches.
8337 2017-08-01 Martin Liska <mliska@suse.cz>
8340 * builtins.c (expand_builtin_memcpy_args): Remove.
8341 (expand_builtin_memcpy): Call newly added function
8342 expand_builtin_memory_copy_args.
8343 (expand_builtin_memcpy_with_bounds): Likewise.
8344 (expand_builtin_mempcpy): Remove last argument.
8345 (expand_builtin_mempcpy_with_bounds): Likewise.
8346 (expand_builtin_memory_copy_args): New function created from
8347 expand_builtin_mempcpy_args with small modifications.
8348 (expand_builtin_mempcpy_args): Remove.
8349 (expand_builtin_stpcpy): Remove unused argument.
8350 (expand_builtin): Likewise.
8351 (expand_builtin_with_bounds): Likewise.
8353 2017-08-01 Martin Liska <mliska@suse.cz>
8356 Make mempcpy more optimal (PR middle-end/70140).
8358 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8362 __builtin_vec_cmpne verify both arguments are compatible vectors
8363 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8364 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8365 move computation of aligned to after checking the argument types.
8369 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8370 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8373 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8374 Doug Rupp <rupp@adacore.com>
8375 Olivier Hainque <hainque@adacore.com>
8377 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8378 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8380 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8381 for TARGET_OS_CPP_BUILTIN.
8382 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8383 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8385 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8386 passing required abi options to the assembler for EABI configurations.
8387 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8388 of .text.hot and .text.unlikely sections for kernel modules when
8389 using ARM style exceptions.
8390 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8391 options. Add EXTRA_CC1_SPEC.
8392 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8394 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8396 (ARM_TARGET2_DWARF_FORMAT): Define.
8397 * config/arm/t-vxworks: Adjust multilib control to removal of the
8398 Diab command line options.
8400 2017-08-01 Martin Liska <mliska@suse.cz>
8402 PR gcov-profile/81561
8403 * gcov.c (unblock): Make unblocking safe as we need to preserve
8404 index correspondence of blocks and block_lists.
8406 2017-08-01 Richard Biener <rguenther@suse.de>
8408 PR tree-optimization/81181
8409 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8410 (compute_antic): ... end of iteration here.
8412 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8414 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8415 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8416 (ftree-slp-vectorize): Likewise.
8417 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8418 can no longer be set independent of flag_tree_loop_vectorize.
8419 * omp-general.c (emp_max_vf): Likewise.
8420 * opts.c (enable_fdo_optimizations): Remove references to
8421 flag_tree_vectorize, these are now implicit.
8422 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8423 and leave it for the options machinery.
8425 2017-08-01 Martin Liska <mliska@suse.cz>
8428 * builtins.c (expand_builtin_memcpy_args): Remove.
8429 (expand_builtin_memcpy): Call newly added function
8430 expand_builtin_memory_copy_args.
8431 (expand_builtin_memcpy_with_bounds): Likewise.
8432 (expand_builtin_mempcpy): Remove last argument.
8433 (expand_builtin_mempcpy_with_bounds): Likewise.
8434 (expand_builtin_memory_copy_args): New function created from
8435 expand_builtin_mempcpy_args with small modifications.
8436 (expand_builtin_mempcpy_args): Remove.
8437 (expand_builtin_stpcpy): Remove unused argument.
8438 (expand_builtin): Likewise.
8439 (expand_builtin_with_bounds): Likewise.
8441 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8444 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8445 print "ds:" only for immediates in generic address space.
8447 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8450 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8451 (ix86_function_ok_for_sibcall): Return false for naked functions.
8453 2017-08-01 Richard Biener <rguenther@suse.de>
8455 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8456 (compute_antic): Seed worklist with exit block predecessors.
8457 * cfganal.c (dfs_find_deadend): For a cycle return the source
8458 of the edge closing it.
8460 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8462 * config/aarch64/aarch64.c
8463 (aarch64_can_const_movi_rtx_p): Move 0 check.
8465 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8467 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8468 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8470 * match.pd: Ditto in address comparison pattern.
8472 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8474 PR tree-optimization/81627
8475 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8476 closed ssa form for store-store chain.
8478 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8480 PR tree-optimization/81620
8481 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8482 for store-store chain.
8484 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8486 PR tree-optimization/81588
8487 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8488 ranges[i].in_p, invert comparison code ccode. For >/>=,
8489 swap rhs1 and rhs2 and comparison code unconditionally,
8490 for </<= don't do that. Don't swap rhs1/rhs2 again if
8491 ranges[i].in_p, instead invert comparison code ccode if
8492 opcode or oe->rank is BIT_IOR_EXPR.
8495 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8496 a direct optab to conversion optab.
8497 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8498 with GET_MODE_INNER as last argument instead of optab_handler.
8499 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8500 vector extraction if possible and optab is available.
8501 * expr.c (store_constructor): Use convert_optab_handler instead
8502 of optab_handler. Use vector initialization from smaller
8503 vectors if possible and optab is available.
8504 * tree-vect-stmts.c (vectorizable_load): Likewise.
8505 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8507 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8508 of vec_init from half-sized vectors with the same element mode.
8509 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8510 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8511 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8512 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8513 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8514 after mode in gen_vec_extract* calls.
8515 (vec_extract<mode>): Renamed to ...
8516 (vec_extract<mode><ssescalarmodelower>): ... this.
8517 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8518 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8519 element mode after mode in gen_vec_init* calls.
8520 (VEC_INIT_HALF_MODE): New mode iterator.
8521 (vec_init<mode>): Renamed to ...
8522 (vec_init<mode><ssescalarmodelower>): ... this.
8523 (vec_init<mode><ssehalfvecmodelower>): New expander.
8524 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8525 (vec_extractv2sfsf): ... this.
8526 (vec_initv2sf): Renamed to ...
8527 (vec_initv2sfsf): ... this.
8528 (vec_extractv2si): Renamed to ...
8529 (vec_extractv2sisi): ... this.
8530 (vec_initv2si): Renamed to ...
8531 (vec_initv2sisi): ... this.
8532 (vec_extractv4hi): Renamed to ...
8533 (vec_extractv4hihi): ... this.
8534 (vec_initv4hi): Renamed to ...
8535 (vec_initv4hihi): ... this.
8536 (vec_extractv8qi): Renamed to ...
8537 (vec_extractv8qiqi): ... this.
8538 (vec_initv8qi): Renamed to ...
8539 (vec_initv8qiqi): ... this.
8540 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8541 (vec_init<mode>): Renamed to ...
8542 (vec_init<mode><VEC_base_l>): ... this.
8543 (vec_extract<mode>): Renamed to ...
8544 (vec_extract<mode><VEC_base_l>): ... this.
8545 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8546 (vec_initv2sfsf): ... this.
8547 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8548 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8549 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8550 element mode after mode in gen_vec_init* calls.
8551 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8552 (vec_init<mode><Vel>): ... this.
8553 (vec_extract<mode>): Renamed to ...
8554 (vec_extract<mode><Vel>): ... this.
8555 * config/aarch64/iterators.md (Vel): New mode attribute.
8556 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8557 Add element mode after mode in gen_vec_extract* calls.
8558 * config/s390/vector.md (non_vec_l): New mode attribute.
8559 (vec_extract<mode>): Renamed to ...
8560 (vec_extract<mode><non_vec_l>): ... this.
8561 (vec_init<mode>): Renamed to ...
8562 (vec_init<mode><non_vec_l>): ... this.
8563 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8564 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8566 * config/arm/iterators.md (V_elem_l): New mode attribute.
8567 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8568 (vec_extract<mode><V_elem_l>): ... this.
8569 (vec_extractv2di): Renamed to ...
8570 (vec_extractv2didi): ... this.
8571 (vec_init<mode>): Renamed to ...
8572 (vec_init<mode><V_elem_l>): ... this.
8573 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8574 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8575 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8576 Add element mode after gen_vec_extract* calls.
8577 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8578 (vec_init<mode><unitmode>): ... this.
8579 (vec_extract<mode>): Renamed to ...
8580 (vec_extract<mode><unitmode>): ... this.
8581 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8582 (vec_init<mode><unitmode>): ... this.
8583 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8584 (vec_initv2sfsf): ... this.
8585 (vec_extractv2sf): Renamed to ...
8586 (vec_extractv2sfsf): ... this.
8587 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8588 Add element mode after gen_vec_extract* calls.
8589 * config/mips/mips.md (unitmode): New mode iterator.
8590 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8591 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8592 * config/spu/spu.md (inner_l): New mode attribute.
8593 (vec_init<mode>): Renamed to ...
8594 (vec_init<mode><inner_l>): ... this.
8595 (vec_extract<mode>): Renamed to ...
8596 (vec_extract<mode><inner_l>): ... this.
8597 * config/sparc/sparc.md (veltmode): New mode iterator.
8598 (vec_init<VMALL:mode>): Renamed to ...
8599 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8600 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8601 (vec_initv2sisi): ... this.
8602 (vec_initv2sf): Renamed to ...
8603 (vec_initv2sfsf): ... this.
8604 (vec_extractv2sf): Renamed to ...
8605 (vec_extractv2sfsf): ... this.
8606 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8607 (vec_init<mode>): Renamed to ...
8608 (vec_init<mode><VEC_base_l>): ... this.
8609 (vec_extract<mode>): Renamed to ...
8610 (vec_extract<mode><VEC_base_l>): ... this.
8611 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8612 (vec_initv2sfsf): ... this.
8613 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8614 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8615 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8616 gen_vec_init* calls.
8618 2017-08-01 Richard Biener <rguenther@suse.de>
8620 PR tree-optimization/81297
8621 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8622 TREE_OVERFLOW from INTEGER_CSTs.
8624 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8626 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8628 2017-07-31 Carl Love <cel@us.ibm.com>
8630 * config/rs6000/rs6000-c: Add support for built-in functions
8631 vector signed char vec_xl_be (signed long long, signed char *);
8632 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8633 vector signed int vec_xl_be (signed long long, signed int *);
8634 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8635 vector signed long long vec_xl_be (signed long long, signed long long *);
8636 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8637 vector signed short vec_xl_be (signed long long, signed short *);
8638 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8639 vector double vec_xl_be (signed long long, double *);
8640 vector float vec_xl_be (signed long long, float *);
8641 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8642 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8643 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8645 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8646 (altivec_expand_builtin): Add switch statement to call
8647 altivec_expand_xl_be for each builtin.
8648 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8649 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8650 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8651 __builtin_vsx_le_be_v16qi.
8652 * doc/extend.texi: Update the built-in documentation file for the
8653 new built-in functions.
8655 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8658 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8660 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8662 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8664 * config.gcc: Add z14.
8665 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8666 CPU model numbers for z13s and z14.
8667 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8669 * config/s390/s390-opts.h (enum processor_type): Rename
8670 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8671 * config/s390/s390.c (processor_table): Add field for CPU name to
8672 be passed to Binutils.
8673 (s390_asm_output_machine_for_arch): Use the new field in
8674 processor_table for Binutils.
8675 (s390_expand_builtin): Replace arch12 with z14.
8676 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8677 (s390_get_sched_attrmask): Likewise.
8678 (s390_get_unit_mask): Likewise.
8679 * config/s390/s390.opt: Add z14 to processor_type enum.
8681 2017-07-31 Martin Jambor <mjambor@suse.cz>
8684 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8685 regardless of optimization level.
8687 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8688 Martin Liska <mliska@suse.cz>
8690 * predict.def: Remove old comment and adjust probability.
8691 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8694 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8697 * config/i386/i386.c (ix86_function_naked): New function.
8698 (ix86_can_use_return_insn_p): Return false for naked functions.
8699 (ix86_expand_prologue): Skip prologue for naked functions.
8700 (ix86_expand_epilogue): Skip epilogue for naked functions
8701 and emit trap instruction.
8702 (ix86_warn_func_return): New function.
8703 (ix86_attribute_table): Add "naked" attribute specification.
8704 (TARGET_WARN_FUNC_RETURN): Define.
8705 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8707 2017-07-31 Martin Liska <mliska@suse.cz>
8709 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8710 (dump_gimple_bb_header): Always dump BB info.
8711 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8713 2017-07-31 Martin Liska <mliska@suse.cz>
8716 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8717 also with current_function_decl non-null equality.
8719 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8722 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8723 change type to the element type, instead add eltype variable and
8724 use it where we are interested in the element type.
8726 PR tree-optimization/81603
8727 * ipa-polymorphic-call.c
8728 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8729 offset arithmetic in offset_int, bail out if the resulting bit offset
8730 doesn't fit into shwi.
8732 2017-07-31 Martin Liska <mliska@suse.cz>
8734 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8735 (gimplify_save_expr): Fix comment.
8737 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8740 * config/i386/i386.c (ix86_function_arg): Update arguments for
8742 (ix86_compute_frame_layout): Set the initial stack offset to
8743 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8744 INCOMING_FRAME_SP_OFFSET.
8745 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8746 stack before exception handler returns.
8747 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8748 the 'ERROR_CODE' for exception handler.
8750 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8752 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8753 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8754 (ASM_OUTPUT_REG_POP): Ditto.
8755 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8756 instead of asm_fprintf to output pure string.
8758 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8760 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8761 to imported_module_or_decl hook.
8762 (debug_nothing_tree_tree_tree_bool): Remove.
8763 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8764 * debug.c (do_nothing_debug_hooks): Use
8765 debug_nothing_tree_tree_tree_bool_bool instead of
8766 debug_nothing_tree_tree_tree_bool.
8767 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8768 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8769 * sdbout.c (sdb_debug_hooks): Likewise.
8770 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8771 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8773 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8774 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8775 attribute, don't add anything.
8777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8779 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8780 (fold_build2_stat_loc): Likewise.
8781 (fold_build3_stat_loc): Likewise.
8782 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8783 (fold_build1_loc): Remove macro.
8784 (fold_build2_loc): Likewise.
8785 (fold_build3_loc): Likewise.
8787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8789 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8790 (gimple_build_debug_bind_source_stat): Likewise.
8791 * gimple.h (gimple_build_debug_bind): Remove macro.
8792 (gimple_build_debug_bind_source): Likewise.
8794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8796 * bitmap.c (bitmap_alloc): Adjust.
8797 (bitmap_gc_alloc): Likewise.
8798 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
8800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8802 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
8803 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
8804 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
8805 (bitmap_gc_alloc_stat): Likewise.
8806 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
8808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8810 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
8811 * rtl.h (shallow_copy_rtx): Remove macro.
8813 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8815 * emit-rtl.c (gen_raw_REG): Adjust.
8816 * gengenrtl.c (gendef): Likewise.
8817 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8818 * rtl.h (rtx_alloc): Remove macro.
8820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8822 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8823 (build_tree_list_stat): Likewise.
8824 * tree.h (build_tree_list): Remove macro.
8825 (build_tree_list_vec): Likewise.
8827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8829 * tree.c (make_vector_stat): Remove _stat from name.
8830 (build_vector_stat): Likewise.
8831 * tree.h (make_vector_stat): Remove macro.
8832 (build_vector_stat): Likewise.
8834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8836 * tree.h (build_var_debug_value): Remove prototype.
8838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8840 * tree.c (tree_cons_stat): Remove _stat from name.
8841 * tree.h (tree_cons): Remove macro.
8843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8845 * tree.c (build_vl_exp_stat): Remove _stat from name.
8846 * tree.h (build_vl_exp): Remove macro.
8848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8850 * tree.c (build_decl_stat): Remove _stat from name.
8851 * tree.h (build_decl): Remove macro.
8853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8855 * gimple.c (gimple_build_with_ops_stat): Adjust.
8856 (gimple_alloc_stat): Remove _stat from name.
8857 * gimple.h (gimple_alloc): Remove macro.
8859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8861 * tree.c (make_tree_vec_stat): Remove _stat from name.
8862 (grow_tree_vec_stat): Likewise.
8863 * tree.h (make_tree_vec_stat): Adjust prototype.
8864 (grow_tree_vec_stat): Likewise.
8865 (make_tree_vec): Remove macro.
8866 (grow_tree_vec): Likewise.
8868 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8870 * fold-const.c (fold_build1_stat_loc): Adjust.
8871 (fold_build2_stat_loc): Likewise.
8872 (fold_build3_stat_loc): Likewise.
8873 * tree.c (build0_stat): Remove _stat from name.
8874 (build1_stat): Likewise.
8875 (build2_stat): Likewise.
8876 (build3_stat): Likewise.
8877 (build4_stat): Likewise.
8878 (build5_stat): Likewise.
8879 * tree.h (build1_loc): Remove macro, and rename _stat function
8881 (build2_loc): Likewise.
8882 (build3_loc): Likewise.
8883 (build4_loc): Likewise.
8884 (build5_loc): Likewise.
8886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8888 * tree.c (make_int_cst_stat): Remove _stat from name.
8889 * tree.h (make_int_cst_stat): Adjust prototype.
8890 (make_int_cst): Remove macro.
8892 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8894 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8895 * tree.h (make_tree_binfo_stat): Adjust prototype.
8896 (make_tree_binfo): Remove.
8898 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8900 * tree.c (copy_node_stat): Rename to copy_node.
8901 (build_distinct_type_copy): Adjust.
8902 * tree.h (copy_node_stat): Adjust prototype.
8903 (copy_node): Remove macro.
8905 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8907 * tree.c (make_node_stat): rename to make_node.
8908 (build_tree_list_stat): Adjust.
8909 (build0_stat): Likewise.
8910 (build2_stat): Likewise.
8911 (build3_stat): Likewise.
8912 (build4_stat): Likewise.
8913 (build5_stat): Likewise.
8914 (build_decl_stat): Likewise.
8915 * tree.h (make_node_stat): Adjust prototype.
8916 (make_node): remove macro.
8918 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8920 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8921 (PPC_FEATURE2_SCV): Likewise.
8922 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8924 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8926 * config/aarch64/aarch64.c
8927 (aarch64_internal_mov_immediate): Add new special pattern.
8928 * config/aarch64/aarch64.md (*movdi_aarch64):
8929 Add reg/32bit const mov case.
8931 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8932 Richard Sandiford <richard.sandiford@linaro.org>
8934 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8935 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8936 Add integer and movi cases.
8937 (movi-split-hf-df-sf split, fp16): New.
8938 (enabled): Added TARGET_FP_F16INST.
8939 * config/aarch64/iterators.md (GPF_HF): New.
8940 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8942 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8944 * config/aarch64/aarch64.c
8945 (aarch64_simd_container_mode): Add prototype.
8946 (aarch64_expand_mov_immediate): Add HI support.
8947 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8948 (aarch64_can_const_movi_rtx_p): New.
8949 (aarch64_preferred_reload_class):
8950 Remove restrictions of using FP registers for certain SIMD operations.
8951 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8952 (aarch64_valid_floating_const): Add integer move validation.
8953 (aarch64_simd_imm_scalar_p): Remove.
8954 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8955 (aarch64_legitimate_constant_p): Expand list of supported cases.
8956 * config/aarch64/aarch64-protos.h
8957 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8958 (aarch64_reinterpret_float_as_int): New.
8959 (aarch64_simd_imm_scalar_p): Remove.
8960 * config/aarch64/constraints.md (Uvi): New.
8961 (Dd): Split into Ds and new Dd.
8962 * config/aarch64/aarch64.md (*movsi_aarch64):
8964 (*movdi_aarch64): Add SIMD mov case.
8966 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8968 * tree-predcom.c: (struct chain): Handle store-store chain in which
8969 stores for elimination only store loop invariant values.
8970 (execute_pred_commoning_chain): Ditto.
8971 (prepare_initializers_chain_store_elim): Ditto.
8972 (prepare_finalizers): Ditto.
8973 (is_inv_store_elimination_chain): New function.
8974 (initialize_root_vars_store_elim_1): New function.
8976 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8978 * tree-predcom.c: Revise general description of the pass.
8979 (enum chain_type): New enum type for store elimination.
8980 (struct chain): New field supporting store elimination.
8981 (struct component): Ditto.
8982 (dump_chain): Dump store-stores chain.
8983 (release_chain): Release resources.
8984 (split_data_refs_to_components): Compute and create component
8985 contains only stores for elimination.
8986 (get_chain_last_ref_at): New function.
8987 (make_invariant_chain): Initialization.
8988 (make_rooted_chain): Specify chain type in parameter and record it.
8989 (add_looparound_copies): Skip for store-stores chain.
8990 (determine_roots_comp): Compute type of chain and pass it to
8992 (initialize_root_vars_store_elim_2): New function.
8993 (finalize_eliminated_stores): New function.
8994 (remove_stmt): Handle store for elimination.
8995 (execute_pred_commoning_chain): Execute predictive commoning on
8997 (determine_unroll_factor): Skip unroll for store-stores chain.
8998 (prepare_initializers_chain_store_elim): New function.
8999 (prepare_initializers_chain): Hanlde store-store chain.
9000 (prepare_finalizers_chain, prepare_finalizers): New function.
9001 (tree_predictive_commoning_loop): Return integer value indicating
9002 if loop is unrolled or lcssa form is corrupted.
9003 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9005 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9007 * tree-predcom.c (initialize_root): Delete.
9008 (execute_pred_commoning_chain): Initialize root vars and replace
9009 reference of non-combined chain directly, rather than call above
9012 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9014 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9015 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9017 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9019 * tree-predcom.c (struct chain): New field init_seq.
9020 (release_chain): Release init_seq.
9021 (prepare_initializers_chain): Record intialization stmts in above
9023 (insert_init_seqs): New function.
9024 (tree_predictive_commoning_loop): Call insert_init_seqs.
9026 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9028 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9030 2017-07-28 Richard Biener <rguenther@suse.de>
9032 * match.pd: Remove superfluous :c.
9033 * genmatch.c (simplify::id): Add member.
9034 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9036 (current_id): New global.
9037 (dt_node::parent): Move from ...
9038 (dt_operand::parent): ... here. Add for_id member.
9039 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9040 (decision_tree::find_node): Relax order requirement when
9041 merging DT_TRUE nodes to ones inbetween the current simplify
9042 and the one we try to merge with. Add diagnostic whenever
9043 we need to enforce pattern order by not merging.
9044 (decision_tree::insert): Set current_id.
9045 (decision_tree::print_node): Dump parent node and for_id.
9046 (parser::last_id): Add member.
9047 (parser::push_simplify): Assign unique id.
9048 (parser::parser): Initialize last_id.
9050 2017-07-28 Martin Liska <mliska@suse.cz>
9053 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9054 gimple_build_debug_bind.
9056 2017-07-28 Richard Biener <rguenther@suse.de>
9058 PR tree-optimization/81502
9059 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9061 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9063 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9064 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9065 for BIT_FIELD_REF args.
9066 * fold-const.c (make_bit_field_ref): Likewise.
9067 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9069 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9072 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9073 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9074 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9075 Or it into SANITIZER_UNDEFINED.
9076 * ubsan.c: Include gimple-fold.h and varasm.h.
9077 (ubsan_expand_ptr_ifn): New function.
9078 (instrument_pointer_overflow): New function.
9079 (maybe_instrument_pointer_overflow): New function.
9080 (instrument_object_size): Formatting fix.
9081 (pass_ubsan::execute): Call instrument_pointer_overflow
9082 and maybe_instrument_pointer_overflow.
9083 * internal-fn.c (expand_UBSAN_PTR): New function.
9084 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9085 * sanitizer.def (__ubsan_handle_pointer_overflow,
9086 __ubsan_handle_pointer_overflow_abort): New builtins.
9087 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9088 * internal-fn.def (UBSAN_PTR): New internal function.
9089 * opts.c (sanitizer_opts): Add pointer-overflow.
9090 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9091 * fold-const.c (build_range_check): Compute pointer range check in
9092 integral type if pointer arithmetics would be needed. Formatting
9095 2017-07-28 Martin Liska <mliska@suse.cz>
9098 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9099 parameters that are of a variable-length.
9101 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9103 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9105 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9106 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9107 (CRT_CALL_STATIC_FUNCTION): Likewise.
9108 (ASM_DEFAULT_SPEC): New define.
9109 (ASM_SPEC32): Likewise.
9110 (ASM_SPEC64): Likewise.
9111 (ASM_SPEC_COMMON): Likewise.
9112 (ASM_SPEC): Likewise.
9113 (INVALID_64BIT): Likewise.
9114 (LINK_OS_DEFAULT_SPEC): Likewise.
9115 (LINK_OS_SPEC32): Likewise.
9116 (LINK_OS_SPEC64): Likewise.
9117 (POWERPC_LINUX): Likewise.
9118 (PTRDIFF_TYPE): Likewise.
9119 (RESTORE_FP_PREFIX): Likewise.
9120 (RESTORE_FP_SUFFIX): Likewise.
9121 (SAVE_FP_PREFIX): Likewise.
9122 (SAVE_FP_SUFFIX): Likewise.
9123 (SIZE_TYPE): Likewise.
9124 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9125 (TARGET_64BIT): Likewise.
9126 (TARGET_64BIT): Likewise.
9127 (TARGET_AIX): Likewise.
9128 (WCHAR_TYPE_SIZE): Likewise.
9129 (WCHAR_TYPE): Undefine.
9130 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9131 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9132 (CPP_OS_RTEMS_SPEC): Delete.
9133 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9134 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9136 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9138 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9140 PR tree-optimization/81578
9141 * tree-parloops.c (build_new_reduction): Bail out if
9142 reduction_code isn't one of the standard OpenMP reductions.
9143 Move the details printing after that decision.
9145 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9147 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9148 related to reload_in_progress.
9149 (splat_input_operand): Likewise.
9150 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9152 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9154 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9155 (TARGET_INSTANTIATE_DECLS): Likewise.
9156 (legitimate_indexed_address_p): Delete reload_in_progress code.
9157 (rs6000_debug_legitimate_address_p): Likewise.
9158 (rs6000_eliminate_indexed_memrefs): Likewise.
9159 (rs6000_emit_le_vsx_store): Likewise.
9160 (rs6000_emit_move_si_sf_subreg): Likewise.
9161 (rs6000_emit_move): Likewise.
9162 (register_to_reg_type): Likewise.
9163 (rs6000_pre_atomic_barrier): Likewise.
9164 (rs6000_machopic_legitimize_pic_address): Likewise.
9165 (rs6000_allocate_stack_temp): Likewise.
9166 (rs6000_address_for_fpconvert): Likewise.
9167 (rs6000_address_for_altivec): Likewise.
9168 (rs6000_secondary_memory_needed_rtx): Delete function.
9169 (rs6000_check_sdmode): Likewise.
9170 (rs6000_alloc_sdmode_stack_slot): Likewise.
9171 (rs6000_instantiate_decls): Likewise.
9172 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9173 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9174 Delete reload_in_progress.
9175 (*vec_reload_and_plus_<mptrsize>): Likewise.
9176 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9177 (vsx_div_v2di): Likewise.
9178 (vsx_udiv_v2di): Likewise.
9180 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9182 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9183 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9184 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9185 (rs6000_debug_reg_global): Delete print of LRA status.
9186 (rs6000_option_override_internal): Delete dead LRA related code.
9187 (rs6000_lra_p): Delete function.
9188 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9190 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9192 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9193 * config/riscv/rtems.h: New file.
9195 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9196 Sudakshina Das <sudi.das@arm.com>
9198 * config/aarch64/aarch64.md
9199 (define_split for and<mode>3nr_compare): Move
9200 non aarch64_logical_operand to a register.
9201 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9202 register immediate operand to a register.
9203 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9205 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9208 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9210 2017-07-27 Richard Biener <rguenther@suse.de>
9212 PR tree-optimization/81573
9213 PR tree-optimization/81494
9214 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9215 multi defuse cycle case.
9217 2017-07-27 Richard Biener <rguenther@suse.de>
9219 PR tree-optimization/81571
9220 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9223 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9225 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9226 earlier and only if MASK_FPU is set. Adjust formatting.
9228 2017-07-27 Martin Liska <mliska@suse.cz>
9230 * opt-functions.awk: Add validation of value of Init.
9231 * optc-gen.awk: Pass new argument.
9233 2017-07-27 Martin Liska <mliska@suse.cz>
9235 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9236 Fix wrong condition.
9238 2017-07-27 Martin Liska <mliska@suse.cz>
9240 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9241 BBs and edges seen by autoFDO.
9243 2017-07-27 Richard Biener <rguenther@suse.de>
9245 PR tree-optimization/81502
9246 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9247 with incompatible but same sized type.
9248 (execute_update_addresses_taken): Likewise.
9250 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9252 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9253 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9255 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9258 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9259 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9260 Change s_operand to memory_operand.
9262 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9264 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9265 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9266 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9267 Emit instructions rather than returning an expression. Handle TFmode
9268 and KFmode by casting to TImode.
9269 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9270 (rs6000_emit_le_vsx_store): Likewise.
9271 * config/rs6000/vsx.md (VSX_TI): New iterator.
9272 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9273 (*vsx_le_undo_permute_<mode>): Likewise.
9274 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9275 emit the split sequence.
9276 (*vsx_le_perm_store_<mode>): Likewise.
9278 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9280 PR tree-optimization/81555
9281 PR tree-optimization/81556
9282 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9283 if true, force CHANGED for the recursive invocation.
9284 (reassociate_bb): Remember original length of ops array, pass
9285 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9287 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9288 attributes for noipa attribute. For naked attribute use
9289 lookup_attribute first before lookup_attribute_spec.
9290 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9292 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9293 for functions with noipa attribute.
9294 (cgraph_externally_visible_p): Return true for functions with noipa
9296 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9297 for functions with noipa attribute.
9298 * doc/extend.texi: Document noipa function attribute.
9299 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9300 also for functions with noipa attribute.
9301 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9303 2017-07-26 Andrew Pinski <apinski@cavium.com>
9305 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9306 vec_unalign_load_cost and vec_unalign_store_cost.
9308 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9310 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9311 -mvsx-small-integer option.
9312 (ISA_3_0_MASKS_IEEE): Likewise.
9313 (OTHER_VSX_VECTOR_MASKS): Likewise.
9314 (POWERPC_MASKS): Likewise.
9315 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9316 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9317 code, only testing for DImode being allowed in non-VSX floating
9319 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9320 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9322 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9323 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9324 TARGET_P8_VECTOR test.
9325 (rs6000_secondary_reload_simple_move): Likewise.
9326 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9327 since TARGET_P9_VECTOR was already tested.
9328 (rs6000_opt_masks): Remove -mvsx-small-integer.
9329 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9330 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9332 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9333 test for TARGET_VEXTRACTUB was used, and that uses
9335 (p9 extract splitter): Likewise.
9336 (vsx_extract_<mode>_di_p9): Likewise.
9337 (vsx_extract_<mode>_store_p9): Likewise.
9338 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9339 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9340 the elimination of TARGET_VSX_SMALL_INTEGER.
9341 (vsx_extract_<mode>_p8): Likewise.
9342 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9343 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9344 (vsx_set_<mode>_p9): Likewise.
9345 (vsx_set_v4sf_p9): Likewise.
9346 (vsx_set_v4sf_p9_zero): Likewise.
9347 (vsx_insert_extract_v4sf_p9): Likewise.
9348 (vsx_insert_extract_v4sf_p9_2): Likewise.
9349 * config/rs6000/rs6000.md (sign extend splitter): Change
9350 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9351 (floatsi<mode>2_lfiwax_mem): Likewise.
9352 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9353 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9354 since a test for TARGET_P9_VECTOR was used.
9355 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9356 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9357 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9358 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9359 TARGET_P8_VECTOR test.
9360 (fix_trunc<mode>si2_stfiwx): Likewise.
9361 (fix_trunc<mode>si2_internal): Likewise.
9362 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9363 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9365 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9366 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9367 TARGET_P8_VECTOR test.
9368 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9369 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9370 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9372 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9373 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9374 since a test for TARGET_P9_VECTOR was used.
9375 (splitter for loading small constants): Likewise.
9377 2017-07-26 Andrew Pinski <apinski@cavium.com>
9379 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9382 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9385 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9386 (fp_valid_at): Likewise.
9388 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9390 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9391 (qdf24xx_addrcost_table): Likewise.
9392 (cortexa57_tunings): Update to use generic_branch_cost.
9393 (cortexa72_tunings): Likewise.
9394 (cortexa73_tunings): Likewise.
9395 (qdf24xx_tunings): Likewise.
9397 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9399 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9400 (thunderx2t99_branch_cost): Likewise.
9401 (cortexa35_tunings): Update to use generic_branch_cost.
9402 (cortexa53_tunings): Likewise.
9403 (cortexa57_tunings): Likewise.
9404 (cortexa72_tunings): Likewise.
9405 (cortexa73_tunings): Likewise.
9406 (thunderx2t99_tunings): Likewise.
9408 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9410 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9411 (sparc_option_override): Honour MASK_FSMULD.
9412 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9413 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9414 * config/sparc/sparc.opt (mfsmuld): New option.
9415 * doc/invoke.texi (mfsmuld): Document option.
9417 2017-07-26 Marek Polacek <polacek@redhat.com>
9420 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9422 2017-07-26 Richard Biener <rguenther@suse.de>
9424 * gimple-match-head.c (do_valueize): Return OP if valueize
9426 (get_def): New helper to get at the def stmt of a SSA name
9428 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9429 do_valueize to get at the def stmt.
9430 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9432 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9435 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9437 2017-07-26 Martin Liska <mliska@suse.cz>
9440 * function.c (expand_function_start): Make expansion of
9441 nonlocal_goto_save_area after parm_birth_insn.
9443 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9445 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9446 from all CPU target flags enable members.
9448 2017-07-26 Richard Biener <rguenther@suse.de>
9450 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9451 (decision_tree::gen): Make 'type' const.
9452 (write_predicate): Likewise.
9454 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9456 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9457 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9458 (rs6000_option_override_internal): Likewise.
9459 (rs6000_expand_vector_set): Likewise.
9460 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9461 (TARGET_UPPER_REGS_SF): Likewise.
9462 (TARGET_UPPER_REGS_DI): Likewise.
9463 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9464 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9465 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9466 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9467 (Splitters for DI constants in Altivec registers): Likewise.
9468 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9469 (vsx_set_v4sf_p9): Likewise.
9470 (vsx_set_v4sf_p9_zero): Likewise.
9471 (vsx_insert_extract_v4sf_p9): Likewise.
9472 (vsx_insert_extract_v4sf_p9_2): Likewise.
9474 2017-07-25 Carl Love <cel@us.ibm.com>
9476 * doc/extend.texi: Update the built-in documentation file for the
9477 existing built-in functions
9478 vector signed char vec_cnttz (vector signed char);
9479 vector unsigned char vec_cnttz (vector unsigned char);
9480 vector signed short vec_cnttz (vector signed short);
9481 vector unsigned short vec_cnttz (vector unsigned short);
9482 vector signed int vec_cnttz (vector signed int);
9483 vector unsigned int vec_cnttz (vector unsigned int);
9484 vector signed long long vec_cnttz (vector signed long long);
9485 vector unsigned long long vec_cnttz (vector unsigned long long);
9487 2017-07-25 Andrew Pinski <apinski@cavium.com>
9489 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9490 accesses where the use is for the first operand of a BIT_INSERT.
9492 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9495 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9496 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9498 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9500 * config/i386/gstabs.h: Delete.
9501 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9503 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9505 * config/i386/i386.c (ix86_decompose_address): Do not check for
9506 register RTX when looking at index_reg or base_reg.
9507 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9509 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9511 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9512 to update EH info here.
9514 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9516 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9518 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9520 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9522 2017-07-25 Torsten Duwe <duwe@suse.de>
9524 * common.opt: Introduce -fpatchable-function-entry
9525 command line option, and its variables function_entry_patch_area_size
9526 and function_entry_patch_area_start.
9527 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9528 including a two-value parser.
9529 * target.def (print_patchable_function_entry): New target hook.
9530 * targhooks.h (default_print_patchable_function_entry): New function.
9531 * targhooks.c (default_print_patchable_function_entry): Likewise.
9532 * toplev.c (process_options): Switch off IPA-RA if
9533 patchable function entries are being generated.
9534 * varasm.c (assemble_start_function): Look at the
9535 patchable-function-entry command line switch and current
9536 function attributes and maybe generate NOP instructions by
9537 calling the print_patchable_function_entry hook.
9538 * doc/extend.texi: Document patchable_function_entry attribute.
9539 * doc/invoke.texi: Document -fpatchable_function_entry
9540 command line option.
9541 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9543 * doc/tm.texi: Re-generate.
9545 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9548 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9549 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9551 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9553 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9555 2017-07-25 Richard Biener <rguenther@suse.de>
9557 PR tree-optimization/81455
9558 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9559 not walk in cycles when looking for guards.
9561 2017-07-25 Richard Biener <rguenther@suse.de>
9563 PR tree-optimization/81529
9564 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9565 when optimizing backedge uses.
9567 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9569 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9571 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9572 to dl_section_ref. On AIX, append an expression to subtract
9573 the size of the section length to dl_section_ref.
9575 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9577 * configure.ac: If any of the config.* scripts fail, exit 1.
9578 * configure: Regenerate.
9580 2017-07-25 Richard Biener <rguenther@suse.de>
9583 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9584 of immediate uses, be more verbose on errors.
9586 2017-07-25 Richard Biener <rguenther@suse.de>
9588 PR tree-optimization/81510
9589 * tree-vect-loop.c (vect_is_simple_reduction): When the
9590 reduction stmt is not inside the loop bail out.
9592 2017-07-25 Richard Biener <rguenther@suse.de>
9594 PR tree-optimization/81303
9595 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9596 profitability check against LOOP_VINFO_NITERSM1.
9598 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9600 * domwalk.c (cmp_bb_postorder): Simplify.
9601 (sort_bbs_postorder): New function. Use it...
9602 (dom_walker::walk): ...here to optimize common cases.
9604 2017-07-25 Martin Liska <mliska@suse.cz>
9607 * ipa-visibility.c (function_and_variable_visibility): Make the
9608 redirection just on target that supports aliasing.
9609 Fix GNU coding style.
9611 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9614 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9616 * config/arm/rtems.h: Likewise.
9617 * config/bfin/rtems.h: Likewise.
9618 * config/i386/rtemself.h: Likewise.
9619 * config/lm32/rtems.h: Likewise.
9620 * config/m32c/rtems.h: Likewise.
9621 * config/m68k/rtemself.h: Likewise.
9622 * config/microblaze/rtems.h: Likewise.
9623 * config/mips/rtems.h: Likewise.
9624 * config/moxie/rtems.h: Likewise.
9625 * config/nios2/rtems.h: Likewise.
9626 * config/powerpcspe/rtems.h: Likewise.
9627 * config/rs6000/rtems.h: Likewise.
9628 * config/rtems.h: Likewise.
9629 * config/sh/rtems.h: Likewise.
9630 * config/sh/rtemself.h: Likewise.
9631 * config/sparc/rtemself.h: Likewise.
9633 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9636 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9637 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9638 * tree-ssa-structalias.c (alias_get_name): Same.
9640 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9643 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9644 instructions if no du chain is found.
9646 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9648 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9650 2017-07-25 Richard Biener <rguenther@suse.de>
9653 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9656 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9658 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9660 (ISA_2_7_MASKS_SERVER): Likewise.
9661 (ISA_3_0_MASKS_IEEE): Likewise.
9662 (OTHER_P8_VECTOR_MASKS): Likewise.
9663 (OTHER_VSX_VECTOR_MASKS): Likewise.
9664 (POWERPC_MASKS): Likewise.
9665 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9666 duplicate list of options.
9667 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9668 explicit -mupper-regs options.
9669 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9670 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9671 alias for -mupper-regs-df.
9672 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9673 (rs6000_init_hard_regno_mode_ok): Likewise.
9674 (rs6000_option_override_internal): Likewise.
9675 (rs6000_opt_masks): Likewise.
9676 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9677 options in terms of whether -mvsx or -mpower8-vector was used.
9678 (TARGET_UPPER_REGS_DI): Likewise.
9679 (TARGET_UPPER_REGS_SF): Likewise.
9680 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9681 -mupper-regs-* options.
9683 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9685 * passes.c (emergency_dump_function): Print some empty lines and a
9686 header before the RTL dump.
9688 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9690 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9692 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9695 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9696 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9698 2017-07-24 Carl Love <cel@us.ibm.com>
9700 * config/rs6000/rs6000-c.c: Add support for built-in functions
9701 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9702 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9703 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9704 vec_extract_fp_from_shortl): Add defines for the two builtins.
9705 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9706 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9708 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9709 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9710 * doc/extend.texi: Update the built-in documentation file for the
9711 new built-in function.
9713 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9716 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9718 * doc/generic.texi: Likewise.
9719 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9720 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9722 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9724 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9725 (aarch64_mls_elt_merge<mode>): Likewise.
9727 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9729 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9730 having __cxa_atexit.
9732 2017-07-23 Michael Collison <michael.collison@arm.com>
9734 * config/arm/arm.c (arm_option_override): Deprecate
9735 use of -mstructure-size-boundary.
9736 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9737 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9739 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9742 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9743 Reduce cost estimate for direct moves.
9745 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9748 * config/i386/i386.c (ix86_option_override_internal): Disable
9749 BMI, BMI2 and TBM instructions for -m16.
9751 2017-07-21 Carl Love <cel@us.ibm.com>
9753 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9754 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9755 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9756 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9757 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9758 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9759 VMULOSW): New enum "unspec" values.
9760 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9761 altivec_vmulosw): New patterns.
9762 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9763 VMULOSW): Add definitions.
9765 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9767 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9768 (qdf24xx): Likewise.
9769 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9770 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9771 (AARCH64_FL_V8_1): Renumber.
9772 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9773 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9774 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9775 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9776 rdma to feature modifiers list.
9778 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9781 * ipa-visibility (function_and_variable_visibility): Convert
9782 recursive PLT call to direct call if appropriate.
9784 2017-07-21 Andrew Pinski <apinski@cavium.com>
9786 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9787 operand 1 to see if the types precision matches.
9788 * fold-const.c (operand_equal_p): Likewise.
9790 2017-07-21 Richard Biener <rguenther@suse.de>
9792 PR tree-optimization/81303
9793 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
9794 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
9795 (vect_peeling_hash_get_lowest_cost): Adjust.
9796 (vect_enhance_data_refs_alignment): Likewise. Use
9797 vect_get_peeling_costs_all_drs to compute the penalty for no
9798 peeling to match up costs.
9800 2017-07-21 Richard Biener <rguenther@suse.de>
9802 PR tree-optimization/81500
9803 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
9804 we didn't identify a reduction path.
9806 2017-07-21 Tom de Vries <tom@codesourcery.com>
9807 Cesar Philippidis <cesar@codesourcery.com>
9809 PR gcov-profile/81442
9810 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
9813 2017-07-21 Tom de Vries <tom@codesourcery.com>
9816 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9818 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9819 nvptx_override_options_after_change.
9821 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9823 * dwarf2out.c (output_file_names): Avoid double testing for
9826 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9828 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9830 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9832 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9834 (try_crossjump_to_edge): Do not punt on partitioned functions.
9836 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9838 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9839 Put all BBs reachable only via paths crossing cold region to cold
9841 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9843 2016-07-21 Richard Biener <rguenther@suse.de>
9845 PR tree-optimization/81303
9846 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9847 into account prologue and epilogue iterations when raising
9848 min_profitable_iters to sth at least covering one vector iteration.
9850 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9852 * config/arm/arm.c (arm_test_cpu_arch_dat):
9855 2017-07-20 Nathan Sidwell <nathan@acm.org>
9857 Remove TYPE_METHODS.
9858 * tree.h (TYPE_METHODS): Delete.
9859 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9860 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9861 (dbxout_type_methods): Scan TYPE_FIELDS.
9862 (dbxout_type): Don't check TYPE_METHODS here.
9863 * function.c (use_register_for_decl): Always ignore register for
9864 class types when not optimizing.
9865 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9866 * tree.c (free_lang_data_in_type): Stitch out member functions and
9867 templates from TYPE_FIELDS.
9868 (build_distinct_type_copy, verify_type_variant,
9869 verify_type): Member fns are on TYPE_FIELDS.
9870 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9871 * tree-pretty-print.c (dump_generic_node): Likewise.
9873 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9876 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9877 V2TImode and V4TImode.
9878 (ix86_expand_vector_extract): Likewise.
9879 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9880 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9881 (ssescalarmode): Handle V4TImode and V2TImode.
9882 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9883 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9884 (VEXTRACTI128_MODE): New mode iterator.
9885 (splitter for *vec_extractv?ti first element): New.
9886 (VEC_INIT_MODE): New mode iterator.
9887 (vec_init<mode>): Consolidate 3 expanders into one using
9888 VEC_INIT_MODE mode iterator.
9890 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9892 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9893 non_spilled_static_chain_regno_p.
9895 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9897 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9899 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9901 * bb-reorder.c (connect_traces): Allow copying of blocks within
9904 2017-07-20 Richard Biener <rguenther@suse.de>
9906 * gimple.h (gimple_phi_result): Add gphi * overload.
9907 (gimple_phi_result_ptr): Likewise.
9908 (gimple_phi_arg): Likewise. Adjust index assert to only
9909 allow actual argument accesses rather than all slots available
9911 (gimple_phi_arg_def): Add gphi * overload.
9912 * tree-phinodes.c (make_phi_node): Initialize only actual
9914 (resize_phi_node): Clear memory not covered by old node,
9915 do not initialize excess argument slots.
9916 (reserve_phi_args_for_new_edge): Initialize new argument slot
9919 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9921 PR tree-optimization/81388
9923 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9925 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9926 by removing computation of may_be_zero.
9928 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9929 Tom de Vries <tom@codesourcery.com>
9932 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9933 when gimple level profile disagrees with what RTL expander did.
9935 2017-07-20 Richard Biener <rguenther@suse.de>
9937 PR tree-optimization/61171
9938 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9939 (vect_analyze_stmt): Add slp instance parameter.
9940 (vectorizable_reduction): Likewise.
9941 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9942 (vect_is_simple_reduction): Deal with chains not detected
9943 as SLP reduction chain, specifically not properly associated
9944 chains containing a mix of plus/minus.
9945 (get_reduction_op): Remove.
9946 (get_initial_defs_for_reduction): Simplify, pass in whether
9947 this is a reduction chain, pass in the SLP node for the PHIs.
9948 (vect_create_epilog_for_reduction): Get the SLP instance as
9950 (vectorizable_reduction): Get the SLP instance as arg.
9951 During analysis remember the SLP node with the PHIs in the
9952 instance. Simplify getting at the vectorized reduction PHIs.
9953 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9954 through SLP instance.
9955 (vect_slp_analyze_operations): Likewise.
9956 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9957 (vect_transform_stmt): Likewise.
9959 2017-07-20 Tom de Vries <tom@codesourcery.com>
9961 PR tree-optimization/81489
9962 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9963 read of phi arg location to before loop that modifies phi.
9965 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9967 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9970 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9973 * except.c (execute): Fix ordering issue.
9975 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9977 PR rtl-optimization/81423
9978 * combine.c (make_compound_operation_int): Don't try to optimize
9979 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9981 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9983 PR rtl-optimization/81423
9984 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9985 with a constant that is -1 in the truncated to mode.
9987 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9989 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9990 (determine_unlikely_bbs): ... here.
9991 * predict.h (propagate_unlikely_bbs_forward): Declare.
9992 * cfgexpand.c (pass_expand::execute): Use it.
9993 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9995 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9996 propagate_unlikely_bbs_forward.
9998 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10000 PR middle-end/81331
10001 * except.c (maybe_add_nop_after_section_switch): New function.
10004 2017-07-19 Tom de Vries <tom@codesourcery.com>
10006 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10008 2017-07-19 Tom de Vries <tom@codesourcery.com>
10010 * gimple.h (gimple_phi_arg): Make assert more strict.
10012 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10014 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10015 * config/rs6000/mmintrin.h: New file.
10016 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10018 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10020 PR tree-optimization/81346
10021 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10023 2017-07-19 Tom de Vries <tom@codesourcery.com>
10025 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10027 2017-07-19 Tom de Vries <tom@codesourcery.com>
10029 * config/nvptx/nvptx-modes.def: Add V2DImode.
10030 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10031 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10032 (nvptx_output_mov_insn): Handle lack of mov.b128.
10033 (nvptx_print_operand): Handle 'H' and 'L' codes.
10034 (nvptx_vector_mode_supported): Allow V2DImode.
10035 (nvptx_preferred_simd_mode): New function.
10036 (nvptx_data_alignment): New function.
10037 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10038 nvptx_preferred_simd_mode.
10039 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10041 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10043 2017-07-19 Tom de Vries <tom@codesourcery.com>
10045 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10046 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10047 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10048 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10049 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10050 (mov<VECIM>_insn): New define_insn.
10051 (define_expand "mov<VECIM>): New define_expand.
10053 2017-07-19 Tom de Vries <tom@codesourcery.com>
10055 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10057 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10059 PR tree-optimization/81346
10060 * fold-const.h (fold_div_compare, range_check_type): Declare.
10061 * fold-const.c (range_check_type): New function.
10062 (build_range_check): Use range_check_type.
10063 (fold_div_compare): No longer static, rewritten into
10064 a match.pd helper function.
10065 (fold_comparison): Don't call fold_div_compare here.
10066 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10067 as helper function.
10069 2017-07-19 Nathan Sidwell <nathan@acm.org>
10071 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10072 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10073 * tree.c (find_decls_types_r, verify_type): Use
10074 TYPE_{MIN,MAX}_VALUE_RAW.
10075 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10076 (hash_tree): Likewise.
10077 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10079 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10082 2017-07-18 Tom de Vries <tom@codesourcery.com>
10084 PR middle-end/81464
10085 * omp-expand.c (expand_omp_for_static_chunk): Handle
10086 equal-argument loop exit phi.
10088 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10091 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10092 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10093 operand 2 predicate.
10094 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10095 operand 2 predicate.
10096 (ror,rol -> rorx splitters): Use const_int_operand as
10097 operand 2 predicate.
10099 2017-06-18 Richard Biener <rguenther@suse.de>
10101 PR tree-optimization/81410
10102 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10103 the gap in the ! slp_perm SLP case after each group.
10105 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10107 PR middle-end/81463
10108 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10111 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10113 PR middle-end/81462
10114 * predict.c (set_even_probabilities): Cleanup; do not affect
10115 probabilities that are already known.
10116 (combine_predictions_for_bb): Call even when count is set.
10118 2017-07-18 Nathan Sidwell <nathan@acm.org>
10120 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10123 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10126 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10127 optimization for loop niter analysis.
10129 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10132 * config/avr/avr.c (avr_optimize_casesi): Don't use
10133 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10135 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10137 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10138 body_cost_vec from _vect_peel_extended_info.
10139 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10140 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10143 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10145 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10147 2017-07-18 Richard Biener <rguenther@suse.de>
10149 PR tree-optimization/80620
10150 PR tree-optimization/81403
10151 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10152 info when re-using a VN table entry.
10154 2017-07-18 Richard Biener <rguenther@suse.de>
10156 PR tree-optimization/81418
10157 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10158 vectype_in. Verify that with lane-reducing reduction operations
10159 we have a single def-use cycle.
10161 2017-07-17 Carl Love <cel@us.ibm.com>
10163 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10165 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10166 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10167 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10169 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10170 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10171 VMULOSW): New enum "unspec" values.
10172 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10173 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10174 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10175 altivec_vmulosw): New patterns.
10176 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10177 VMULOSW): Add definitions.
10178 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10180 * config/alpha/alpha.c: Include predict.h.
10182 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10184 * tree-vrp.c (compare_assert_loc): Fix comparison function
10185 to return predictable results.
10187 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10189 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10191 (subdi3): Likewise.
10192 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10193 * doc/invoke.texi (mexpand-adddi): Update text.
10195 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10197 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10198 that also clobbers the CC register. The old expand code is moved
10200 (*arc_clzsi2): ... here.
10201 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10202 the CC register. The old expand code is moved to ...
10203 (arc_ctzsi2): ... here.
10205 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10207 * config/arc/arc.opt (mindexed-loads): Use initial value
10208 TARGET_INDEXED_LOADS_DEFAULT.
10209 (mauto-modify-reg): Use initial value
10210 TARGET_AUTO_MODIFY_REG_DEFAULT.
10211 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10212 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10213 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10214 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10216 2017-07-17 Martin Liska <mliska@suse.cz>
10219 * opts.c (finish_options): Do not allow -fgnu-tm
10220 w/ -fsanitize={kernel-,}address. Say sorry.
10222 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10225 * tree-loop-distribution.c (classify_partition): Only assert on
10226 numer of iterations.
10227 (merge_dep_scc_partitions): Delete prameter. Update function call.
10228 (distribute_loop): Remove code handling loop with unknown niters.
10229 (pass_loop_distribution::execute): Skip loop with unknown niters.
10231 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10234 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10235 function sort_partitions_by_post_order.
10237 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10239 PR tree-optimization/81374
10240 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10241 the max index of basic blocks, rather than number of basic blocks.
10243 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10245 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10247 (arc_legitimate_pic_operand_p): Likewise.
10248 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10250 (arc_needs_pcl_p): Likewise.
10251 (arc_legitimate_pc_offset_p): Likewise.
10252 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10253 function is also used in constrains.md.
10254 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10255 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10256 PLUS. Only return true/false in known cases, otherwise assert.
10257 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10258 is already called in arc_legitimate_constant_p.
10259 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10261 (LEGITIMATE_PIC_OPERAND_P): Use
10262 arc_raw_symbolic_reference_mentioned_p function.
10263 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10268 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10269 Andrew Burgess <andrew.burgess@embecosm.com>
10271 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10272 (arc_return_address_register): New function.
10273 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10274 (arc_handle_fndecl_attribute): Add naked attribute.
10275 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10276 (TARGET_WARN_FUNC_RETURN): Likewise.
10277 (arc_allocate_stack_slots_for_args): New function.
10278 (arc_warn_func_return): Likewise.
10279 (machine_function): Change type fn_type.
10280 (arc_compute_function_type): Consider new naked function type,
10281 change function return type.
10282 (arc_must_save_register): Adapt to handle new
10283 arc_compute_function_type's return type.
10284 (arc_expand_prologue): Likewise.
10285 (arc_expand_epilogue): Likewise.
10286 (arc_return_address_regs): Delete.
10287 (arc_return_address_register): New function.
10288 (arc_epilogue_uses): Use above function.
10289 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10290 (arc_function_type): Change encoding, add naked type.
10291 (ARC_INTERRUPT_P): Change to handle the new encoding.
10292 (ARC_FAST_INTERRUPT_P): Likewise.
10293 (ARC_NORMAL_P): Define.
10294 (ARC_NAKED_P): Likewise.
10295 (arc_compute_function_type): Delete prototype.
10296 * config/arc/arc.md (in_ret_delay_slot): Use
10297 arc_return_address_register function.
10298 (simple_return): Likewise.
10299 (p_return_i): Likewise.
10301 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10303 PR tree-optimization/81428
10304 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10305 can't be built for those types.
10307 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10309 Remove stuff dead since r239246.
10311 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10312 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10313 (avr_inform_devices): Remove dead stuff.
10315 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10317 * config/arm/arm_neon.h: Fix softp typo.
10319 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10321 PR tree-optimization/81365
10322 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10323 aggregate moves onto bb predecessor edges, make sure there are no
10324 loads that could alias the lhs in between the start of bb and the
10327 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10330 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10331 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10332 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10334 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10336 PR tree-optimization/81396
10337 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10338 (init_symbolic_number): Initialize it to 1.
10339 (perform_symbolic_merge): Add n_ops from both operands into the new
10341 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10342 without base_addr as useless if they need more than one operation.
10343 (bswap_replace): Handle !bswap case for NULL base_addr.
10345 2017-07-17 Tom de Vries <tom@codesourcery.com>
10348 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10351 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10353 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10354 conditional builtin define __FIX_LEON3FT_B2BST.
10356 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10358 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10359 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10362 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10364 PR rtl-optimization/81424
10365 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10366 to remove potential trapping from operands if -fnon-call-exceptions.
10368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10370 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10371 profile_proability for scalling.
10372 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10374 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10376 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10378 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10380 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10381 fixpoint arithmetics.
10383 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10385 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10386 fixpoint arithmetics.
10388 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10390 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10391 fixpoint arithmetics.
10393 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10395 * profile-count.h (profile_probability::from_reg_br_prob_note,
10396 profile_probability::to_reg_br_prob_note): New functions.
10397 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10398 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10399 * predict.c (probability_reliable_p): Update.
10400 (edge_probability_reliable_p): Update.
10401 (br_prob_note_reliable_p): Update.
10402 (invert_br_probabilities): Update.
10403 (add_reg_br_prob_note): New function.
10404 (combine_predictions_for_insn): Update.
10405 * asan.c (asan_clear_shadow): Update.
10406 * cfgbuild.c (compute_outgoing_frequencies): Update.
10407 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10408 (update_br_prob_note): Update.
10409 (rtl_verify_edges): Update.
10410 (purge_dead_edges): Update.
10411 (fixup_reorder_chain): Update.
10412 * emit-rtl.c (try_split): Update.
10413 * ifcvt.c (cond_exec_process_insns): Update.
10414 (cond_exec_process_if_block): Update.
10415 (dead_or_predicable): Update.
10416 * internal-fn.c (expand_addsub_overflow): Update.
10417 (expand_neg_overflow): Update.
10418 (expand_mul_overflow): Update.
10419 * loop-doloop.c (doloop_modify): Update.
10420 * loop-unroll.c (compare_and_jump_seq): Update.
10421 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10422 * predict.h: Update.
10423 * reorg.c (mostly_true_jump): Update.
10425 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10426 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10427 * config/arc/arc.c: (emit_unlikely_jump): Update.
10428 * config/arm/arm.c: (emit_unlikely_jump): Update.
10429 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10430 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10431 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10432 (ix86_print_operand): Update.
10433 (ix86_split_fp_branch): Update.
10434 (predict_jump): Update.
10435 * config/ia64/ia64.c (ia64_print_operand): Update.
10436 * config/mmix/mmix.c (mmix_print_operand): Update.
10437 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10438 (rs6000_expand_split_stack_prologue): Update.
10439 * config/rs6000/rs6000.c: Update.
10440 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10441 (s390_expand_vec_movstr): Update.
10442 (s390_expand_cs_tdsi): Update.
10443 (s390_expand_split_stack_prologue): Update.
10444 * config/sh/sh.c (sh_print_operand): Update.
10445 (expand_cbranchsi4): Update.
10446 (expand_cbranchdi4): Update.
10447 * config/sparc/sparc.c (output_v9branch): Update.
10448 * config/spu/spu.c (get_branch_target): Update.
10449 (ea_load_store_inline): Update.
10450 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10451 * config/tilepro/tilepro.c: Update.
10453 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10455 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10456 (gimplify_save_expr): Likewise.
10458 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10460 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10462 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10464 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10466 * timevar.def (TV_IPA_FNSUMMARY): Define.
10468 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10470 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10471 to back store errata sensitive sequence from being generated.
10472 (sqrtdf2_fix): Likewise.
10474 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10476 * tree-ssa-threadupdate.c (compute_path_counts,
10477 update_joiner_offpath_counts): Use profile_probability.
10479 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10482 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10484 * config/arm/arm-c.c (arm_cpu_builtins): Define
10485 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10487 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10489 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10490 array entries to represent __ieee128 versions of the
10491 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10492 scalar_extract_sig, and scalar_insert_exp built-in functions.
10493 (altivec_resolve_overloaded_builtin): Add special case handling
10494 for the __builtin_scalar_insert_exp function, as represented by
10495 the P9V_BUILTIN_VEC_VSIEDP constant.
10496 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10497 exponent support for __ieee128 argument.
10498 (VSESQP): Add scalar extract signature support for __ieee128
10500 (VSTDCNQP): Add scalar test negative support for __ieee128
10502 (VSIEQP): Add scalar insert exponent support for __int128 argument
10503 with __ieee128 result.
10504 (VSIEQPF): Add scalar insert exponent support for __ieee128
10505 argument with __ieee128 result.
10506 (VSTDCQP): Add scalar test data class support for __ieee128
10508 (VSTDCNQP): Add overload support for scalar test negative with
10509 __ieee128 argument.
10510 (VSTDCQP): Add overload support for scalar test data class
10511 __ieee128 argument.
10512 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10513 UNSPEC_VSX_SXSIGDP.
10514 (UNSPEC_VSX_SIEXPQP): New constant.
10515 (xsxexpqp): New insn for VSX scalar extract exponent quad
10517 (xsxsigqp): New insn for VSX scalar extract significand quad
10519 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10520 precision with floating point argument.
10521 (xststdcqp): New expand for VSX scalar test data class quad
10523 (xststdcnegqp): New expand for VSX scalar test negative quad
10525 (xststdcqp): New insn to match expansions for VSX scalar test data
10526 class quad precision and VSX scalar test negative quad precision.
10527 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10528 special case operand checking to enforce that second operand of
10529 VSX scalar test data class with quad precision argument is a 7-bit
10531 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10532 prototypes and descriptions of __ieee128 versions of
10533 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10534 scalar_test_data_class, and scalar_test_neg built-in functions.
10536 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10538 PR tree-optimization/81162
10539 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10540 replace a negate with an add.
10542 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10544 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10546 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10548 * config/arm/arm-c.c (arm_cpu_builtins): Define
10549 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10551 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10553 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10554 (armv8-r): Set ARM Cortex-R52 as default CPU.
10555 * config/arm/arm-tables.opt: Regenerate.
10556 * config/arm/arm-tune.md: Regenerate.
10557 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10559 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10560 extension for -mcpu=cortex-r52.
10562 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10564 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10565 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10566 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10567 (fp-armv8): Define it as FP_ARMv8 only.
10568 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10569 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10571 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10572 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10573 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10574 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10575 than TARGET_FPU_ARMV8.
10576 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10577 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10578 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10580 * config/arm/neon.md (neon_vrint): Likewise.
10581 (neon_vcvt): Likewise.
10582 (neon_<fmaxmin_op><mode>): Likewise.
10583 (<fmaxmin><mode>3): Likewise.
10584 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10585 * config/arm/predicates.md (arm_cond_move_operator): Check against
10586 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10588 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10590 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10591 to top of function.
10593 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10595 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10596 loop in comment with memset.
10598 2017-07-14 Martin Liska <mliska@suse.cz>
10600 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10601 * dwarf2out.c (is_java): Remove the function.
10602 (output_pubname): Remove usage of the function.
10603 (lower_bound_default): Remove usage of DW_LANG_Java.
10604 (gen_compile_unit_die): Likewise.
10605 * gcc.c: Remove compiler defaults for .java and .zip files.
10606 * gimple-expr.c (remove_suffix): Change as there's no longer
10607 extension than 4-letter one.
10608 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10609 (gimplify_save_expr): Likewise.
10610 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10611 as it's possible even for other languages than Java.
10612 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10613 * lto-opts.c (lto_write_options): Remove reference to Java.
10614 * opts.c (strip_off_ending): Update file extension handling.
10615 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10616 * tree-eh.c (lower_resx): Likewise.
10617 * tree.c (free_lang_data_in_type): Remove dead code.
10618 (find_decls_types_r): Likewise.
10619 (build_common_builtin_nodes): Remove Java from a comment.
10620 (verify_type): Remove dead code.
10621 * varasm.c (assemble_external): Remove Java from a comment.
10623 2017-07-14 Martin Liska <mliska@suse.cz>
10625 * opts.c (finish_options): Add quotes.
10626 (common_handle_option): Likewise.
10628 2017-07-14 Martin Liska <mliska@suse.cz>
10630 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10631 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10632 Remove N_SO_PASCAL.
10633 * dwarf2out.c (lower_bound_default): Do not handle
10635 (gen_compile_unit_die): Likewise.
10636 * gcc.c: Remove default extension binding for GNU Pascal.
10637 * stmt.c: Remove Pascal language from a comment.
10638 * xcoffout.c: Likewise.
10640 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10643 * diagnostic-show-locus.c (fixit_cmp): New function.
10644 (layout::layout): Sort m_fixit_hints.
10645 (column_range::column_range): Assert that the values are valid.
10646 (struct char_span): New struct.
10647 (correction::overwrite): New method.
10648 (struct source_line): New struct.
10649 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10650 calls in terms of classes source_line and char_span, and
10651 correction::overwrite.
10652 (selftest::test_overlapped_fixit_printing_2): New function.
10653 (selftest::diagnostic_show_locus_c_tests): Call it.
10655 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10657 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10658 early if there is no lhs.
10660 2017-07-13 Martin Liska <mliska@suse.cz>
10662 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10663 (gen_reference_type_die): Likewise.
10664 * stor-layout.c: Remove Pascal-related comment.
10666 2017-07-13 Martin Liska <mliska@suse.cz>
10668 * opts.c (finish_options): Add quotes to error messages.
10669 (parse_sanitizer_options): Likewise.
10671 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10673 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10675 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10677 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10679 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10681 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10683 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10685 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10688 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10689 provides the hardware capability bits, define the macro
10690 __BUILTIN_CPU_SUPPORTS__.
10691 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10692 if GLIBC does not provide the hardware capability bits. Add a
10693 gcc_unreachable call if the built-in cpu function is neither
10694 __builtin_cpu_is nor __builtin_cpu_supports.
10695 (rs6000_get_function_versions_dispatcher): Change the warning
10696 that an old GLIBC is used which does not export the capability
10697 bits to be an error.
10698 * doc/extend.texi (target_clones attribute): Document the
10699 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10700 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10701 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10702 the macros defined by GCC if the newer GLIBC is available.
10704 2017-07-12 Jeff Law <law@redhat.com>
10706 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10707 remaining includes slightly.
10708 * config/riscv/riscv-builtins.c: Include profile-count.h.
10710 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10713 * config/avr/avr.c (avr_set_current_function): In diagnostic
10714 messages: Quote keywords and (parts of) identifiers.
10715 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10718 2017-07-12 Carl Love <cel@us.ibm.com>
10720 * config/rs6000/rs6000-c.c: Add support for built-in functions
10721 vector bool char vec_revb (vector bool char);
10722 vector bool short vec_revb (vector short char);
10723 vector bool int vec_revb (vector bool int);
10724 vector bool long long vec_revb (vector bool long long);
10725 * doc/extend.texi: Update the built-in documentation file for the
10726 new built-in functions.
10728 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10730 * config/s390/s390.md: Remove movcc splitter.
10732 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10734 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10735 load/store on condition.
10737 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10740 * config/avr/avr.c (avr_encode_section_info)
10741 [progmem && !TREE_READONLY]: Error if progmem object needs
10744 2017-07-11 Michael Collison <michael.collison@arm.com>
10746 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10749 2017-07-11 Carl Love <cel@us.ibm.com>
10751 * config/rs6000/rs6000-c.c: Add support for builtins
10752 vector unsigned int vec_parity_lsbb (vector signed int);
10753 vector unsigned int vec_parity_lsbb (vector unsigned int);
10754 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10755 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10756 vector unsigned long long vec_parity_lsbb (vector signed long long);
10757 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10758 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10759 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10760 * doc/extend.texi: Update the built-in documentation file for the
10761 new built-in functions.
10763 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10765 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10766 (layout::m_primary_loc): New field.
10767 (layout::layout): Initialize new field. Move location filtering
10768 logic from here to...
10769 (layout::maybe_add_location_range): ...this new method. Add
10770 support for filtering to just the lines already specified by other
10772 (layout::will_show_line_p): New method.
10773 (gcc_rich_location::add_location_if_nearby): New method.
10774 (selftest::test_add_location_if_nearby): New test function.
10775 (selftest::diagnostic_show_locus_c_tests): Call it.
10776 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10779 2017-07-11 Tom de Vries <tom@codesourcery.com>
10781 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10782 (bb_first_real_insn): New function.
10783 (nvptx_single): Add extra initialization of broadcasted condition
10786 2017-07-11 Nathan Sidwell <nathan@acm.org>
10788 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10790 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10792 * doc/extend.texi (AVR Function Attributes): Remove weblink to
10793 Binutils doc as TEXI will mess them up.
10794 * doc/invoke.texi (AVR Options): Same here.
10796 2017-07-11 Daniel Cederman <cederman@gaisler.com>
10798 * config/sparc/sparc.opt (mfix-ut700): New option.
10799 (mfix-gr712rc): Likewise.
10800 (sparc_fix_b2bst): New variable.
10801 * doc/invoke.texi (SPARC options): Document them.
10802 (ARM options): Fix warnings.
10803 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
10804 instructions to prevent sequences that can trigger the store-store
10805 errata for certain LEON3FT processors.
10806 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
10807 (sparc_option_override): Set sparc_fix_b2bst appropriately.
10808 * config/sparc/sparc.md (fix_b2bst): New attribute.
10809 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
10811 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
10814 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
10816 (*rsqrtsf2_sse): Ditto.
10818 (div<mode>3): Macroize insn from divdf3 and divsf3
10819 using MODEF mode iterator.
10821 2017-07-10 Martin Sebor <msebor@redhat.com>
10823 PR tree-optimization/80397
10824 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10825 instead of testing for equality to INTEGER_TYPE.
10827 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10829 * config.gcc: Remove uclibc from arc target spec.
10831 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10833 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10835 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10838 * lto-wrapper.c (remove_option): New function.
10839 (merge_and_complain): Merge PIC/PIE options more realistically.
10841 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10843 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10847 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10848 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10849 * config.in: Regenerate.
10850 * configure: Regenerate.
10851 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10852 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10853 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10854 (TARGET_GASISR_PROLOGUES): ...target mask.
10855 * common/config/avr/avr-common.c
10856 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10857 Set -mgas-isr-prologues.
10858 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10859 INSERT_PASS_BEFORE for it.
10860 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10861 * config/avr/avr.c (avr_option_override)
10862 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10863 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10864 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10865 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10866 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10868 (make_avr_pass_pre_proep): New function.
10869 (emit_push_sfr) <treg>: Add argument to function and use it
10870 instead of TMP_REG.
10871 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10872 and set machine->gasisr.yes.
10873 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10874 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10875 __gcc_isr.n_pushed to .L__stack_usage.
10876 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10877 (avr_asm_final_postscan_insn): ...this new static function.
10878 * config/avr/avr.h (machine_function)
10879 <is_no_gccisr, use_L__stack_usage>: New fields.
10880 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10881 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10882 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10883 (gasisr, *gasisr): New expander and insn.
10884 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10885 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10886 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10888 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10890 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10893 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10895 Move jump-tables out of .text again.
10898 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10899 (ASM_OUTPUT_ADDR_VEC): New function.
10900 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10901 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10902 INSN_ADDRESSes as asm comment.
10903 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10904 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10905 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10906 * config/avr/avr.md (*tablejump): Adjust comment.
10907 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10908 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10910 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10911 (avr_output_addr_vec): New proto.
10912 (avr_log_t) <insn_addresses>: New field.
10914 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10917 * config/i386/i386.c (ix86_function_arg_advance): Set
10918 outgoing_args_on_stack to true if there are outgoing arguments
10920 (ix86_function_arg): Likewise.
10921 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10922 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10923 * config/i386/i386.h (machine_function): Add
10924 outgoing_args_on_stack.
10926 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10928 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10929 supporting pthreds.
10930 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10932 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10934 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10935 (REAL_H): Remove $(MACHMODE_H).
10936 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10938 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10939 $(MACHMODE_H) and double-int.h.
10940 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10942 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10945 2017-07-07 Andrew Pinski <apinski@cavium.com>
10947 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10948 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10950 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10952 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10953 Add warning if GCC was not configured to link against a GLIBC that
10954 exports the hardware capability bits.
10955 (make_resolver_func): Make resolver function private and not a
10956 COMDAT function. Create the name with clone_function_name instead
10957 of make_unique_name.
10960 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10961 correct operand in doing the split.
10963 2017-07-07 Carl Love <cel@us.ibm.com>
10965 * config/rs6000/rs6000-c: Add support for built-in function
10966 vector unsigned short vec_pack_to_short_fp32 (vector float,
10968 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10969 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10970 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10971 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10972 (convert_4f32_8i16): Add define_expand.
10973 * doc/extend.texi: Update the built-in documentation file for the
10974 new built-in function.
10976 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10978 * config/sparc/m8.md: New file.
10979 * config/sparc/sparc.md: Include m8.md.
10981 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10983 * config/sparc/sparc.opt: New option -mvis4b.
10984 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10985 (sparc_option_override): Handle VIS4B.
10986 (enum sparc_builtins): Define
10987 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10988 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10989 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10990 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10991 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10992 (check_constant_argument): New function.
10993 (sparc_vis_init_builtins): Define builtins
10994 __builtin_vis_dictunpack{8,16,32},
10995 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10996 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10997 __builtin_vis_fpcmpde{8,16,32}shl and
10998 __builtin_vis_fpcmpur{8,16,32}shl.
10999 (sparc_expand_builtin): Check that the constant operands to
11000 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11001 constant and in range.
11002 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11004 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11005 (SPARC_IMM5_P): Likewise.
11006 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11007 (enabled): Handle vis4b.
11008 (UNSPEC_DICTUNPACK): New unspec.
11009 (UNSPEC_FPCMPSHL): Likewise.
11010 (UNSPEC_FPUCMPSHL): Likewise.
11011 (UNSPEC_FPCMPDESHL): Likewise.
11012 (UNSPEC_FPCMPURSHL): Likewise.
11013 (cpu_feature): New CPU feature `vis4b'.
11014 (dictunpack{8,16,32}): New insns.
11015 (FPCSMODE): New mode iterator.
11016 (fpcscond): New code iterator.
11017 (fpcsucond): Likewise.
11018 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11019 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11020 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11021 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11022 * config/sparc/constraints.md: Define constraints `q' for unsigned
11023 2-bit integer constants and `t' for unsigned 5-bit integer
11025 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11027 (imm5_operand_dictunpack16): Likewise.
11028 (imm5_operand_dictunpack32): Likewise.
11029 (imm2_operand): Likewise.
11030 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11031 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11032 ditunpack* and fpcmp*shl builtins.
11034 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11036 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11037 * config.in: Add HAVE_AS_SPARC6 define.
11038 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11040 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11042 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11043 (CPP_CPU_SPEC): Handle m8.
11044 (ASM_CPU_SPEC): Likewise.
11045 * config/sparc/sparc-opts.h (enum processor_type): Add
11047 * config/sparc/sparc.c (m8_costs): New struct.
11048 (sparc_option_override): Handle TARGET_CPU_m8.
11049 (sparc32_initialize_trampoline): Likewise.
11050 (sparc64_initialize_trampoline): Likewise.
11051 (sparc_issue_rate): Likewise.
11052 (sparc_register_move_cost): Likewise.
11053 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11054 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11055 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11056 (CPP_CPU_SPEC): Handle M8.
11057 (ASM_CPU_SPEC): Likewise.
11058 (AS_M8_FLAG): Define.
11059 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11060 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11061 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11063 * configure: Regenerate.
11064 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11067 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11069 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11071 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11072 ("*movdi_insn_sp32"): Do not set v3pipe.
11073 ("*movsi_insn"): Likewise.
11074 ("*movdi_insn_sp64"): Likewise.
11075 ("*movsf_insn"): Likewise.
11076 ("*movdf_insn_sp32"): Likewise.
11077 ("*movdf_insn_sp64"): Likewise.
11078 ("*zero_extendsidi2_insn_sp64"): Likewise.
11079 ("*sign_extendsidi2_insn"): Likewise.
11080 ("*mov<VM32:mode>_insn"): Likewise.
11081 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11082 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11083 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11084 ("<vlop:code><VL:mode>3"): Likewise.
11085 ("*not_<vlop:code><VL:mode>3"): Likewise.
11086 ("*nand<VL:mode>_vis"): Likewise.
11087 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11088 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11089 ("one_cmpl<VL:mode>2"): Likewise.
11090 ("faligndata<VM64:mode>_vis"): Likewise.
11091 ("alignaddrsi_vis"): Likewise.
11092 ("alignaddrdi_vis"): Likweise.
11093 ("alignaddrlsi_vis"): Likewise.
11094 ("alignaddrldi_vis"): Likewise.
11095 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11096 ("bmaskdi_vis"): Likewise.
11097 ("bmasksi_vis"): Likewise.
11098 ("bshuffle<VM64:mode>_vis"): Likewise.
11099 ("cmask8<P:mode>_vis"): Likewise.
11100 ("cmask16<P:mode>_vis"): Likewise.
11101 ("cmask32<P:mode>_vis"): Likewise.
11102 ("pdistn<P:mode>_vis"): Likewise.
11103 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11105 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11107 * config/sparc/sparc.md ("subtype"): New insn attribute.
11108 ("*wrgsr_sp64"): Set insn subtype.
11109 ("*rdgsr_sp64"): Likewise.
11110 ("alignaddrsi_vis"): Likewise.
11111 ("alignaddrdi_vis"): Likewise.
11112 ("alignaddrlsi_vis"): Likewise.
11113 ("alignaddrldi_vis"): Likewise.
11114 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11115 ("fexpand_vis"): Likewise.
11116 ("fpmerge_vis"): Likewise.
11117 ("faligndata<VM64:mode>_vis"): Likewise.
11118 ("bshuffle<VM64:mode>_vis"): Likewise.
11119 ("cmask8<P:mode>_vis"): Likewise.
11120 ("cmask16<P:mode>_vis"): Likewise.
11121 ("cmask32<P:mode>_vis"): Likewise.
11122 ("fchksm16_vis"): Likewise.
11123 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11124 ("fmean16_vis"): Likewise.
11125 ("fp<plusminus_insn>64_vis"): Likewise.
11126 ("<plusminus_insn>v8qi3"): Likewise.
11127 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11128 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11129 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11130 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11131 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11132 ("*movqi_insn"): Likewise.
11133 ("*movhi_insn"): Likewise.
11134 ("*movsi_insn"): Likewise.
11135 ("movsi_pic_gotdata_op"): Likewise.
11136 ("*movdi_insn_sp32"): Likewise.
11137 ("*movdi_insn_sp64"): Likewise.
11138 ("movdi_pic_gotdata_op"): Likewise.
11139 ("*movsf_insn"): Likewise.
11140 ("*movdf_insn_sp32"): Likewise.
11141 ("*movdf_insn_sp64"): Likewise.
11142 ("*zero_extendhisi2_insn"): Likewise.
11143 ("*zero_extendqihi2_insn"): Likewise.
11144 ("*zero_extendqisi2_insn"): Likewise.
11145 ("*zero_extendqidi2_insn"): Likewise.
11146 ("*zero_extendhidi2_insn"): Likewise.
11147 ("*zero_extendsidi2_insn_sp64"): Likewise.
11148 ("ldfsr"): Likewise.
11149 ("prefetch_64"): Likewise.
11150 ("prefetch_32"): Likewise.
11151 ("tie_ld32"): Likewise.
11152 ("tie_ld64"): Likewise.
11153 ("*tldo_ldub_sp32"): Likewise.
11154 ("*tldo_ldub1_sp32"): Likewise.
11155 ("*tldo_ldub2_sp32"): Likewise.
11156 ("*tldo_ldub_sp64"): Likewise.
11157 ("*tldo_ldub1_sp64"): Likewise.
11158 ("*tldo_ldub2_sp64"): Likewise.
11159 ("*tldo_ldub3_sp64"): Likewise.
11160 ("*tldo_lduh_sp32"): Likewise.
11161 ("*tldo_lduh1_sp32"): Likewise.
11162 ("*tldo_lduh_sp64"): Likewise.
11163 ("*tldo_lduh1_sp64"): Likewise.
11164 ("*tldo_lduh2_sp64"): Likewise.
11165 ("*tldo_lduw_sp32"): Likewise.
11166 ("*tldo_lduw_sp64"): Likewise.
11167 ("*tldo_lduw1_sp64"): Likewise.
11168 ("*tldo_ldx_sp64"): Likewise.
11169 ("*mov<VM32:mode>_insn"): Likewise.
11170 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11171 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11173 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11175 * config/sparc/sparc.md ("type"): New insn type viscmp.
11176 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11178 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11179 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11180 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11181 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11183 ("n7_vis_logical_11cycle"): Likewise.
11184 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11185 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11186 * config/sparc/niagara.md ("niag_vis"): Likewise.
11187 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11188 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11190 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11192 * config/sparc/sparc.md: New instruction type `bmask'.
11193 (bmaskdi_vis): Use the `bmask' type.
11194 (bmasksi_vis): Likewise.
11195 * config/sparc/ultra3.md (us3_array): Likewise.
11196 * config/sparc/niagara7.md (n7_array): Likewise.
11197 * config/sparc/niagara4.md (n4_array): Likewise.
11198 * config/sparc/niagara2.md (niag2_vis): Likewise.
11199 (niag3_vis): Likewise.
11200 * config/sparc/niagara.md (niag_vis): Likewise.
11202 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11204 * ipa-comdats.c: Remove optimize check from gate.
11205 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11206 for functions not optimized.
11207 (ipa_fn_summary_read): Skip optimize check.
11208 (ipa_fn_summary_write): Likewise.
11209 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11211 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11213 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11214 (check_callers): Give up on caller which is not optimized.
11215 (inline_small_functions): Likewise.
11216 (ipa_inline): Do not give up when not optimizing.
11217 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11218 away unoptimizes cdtors.
11219 (whole_program_function_and_variable_visibility): Do
11220 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11221 * ipa.c (process_references): Do not check optimize.
11222 (symbol_table::remove_unreachable_nodes): Update optimize check.
11223 (set_writeonly_bit): Update optimize check.
11224 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11225 (pass_ipa_single_use::gate): Remove.
11227 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11229 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11230 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11231 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11232 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11233 permute_load, permute_store, adjust_extract, adjust_splat,
11234 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11235 replace_swap_with_copy, dump_swap_insn_table,
11236 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11237 recombine_lvx_pattern, recombine_stvx_pattern,
11238 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11239 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11240 to file rs6000-p8swap.c.
11241 * config/rs6000/rs6000-p8swap.c: New file.
11242 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11243 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11244 and rs6000*-*-* targets.
11246 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11248 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11250 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11252 * lto-wrapper.c (merge_and_complain): Do not merge
11253 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11254 fsigned_zeros, ftrapping_math, fwrapv.
11255 (append_compiler_options): Do not track these options.
11256 (append_linker_options): Likewie
11258 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11260 * cgraphunit.c (cgraph_node::finalize_function): When
11261 !flag_toplevel_reorde set no_reorder flag.
11262 (varpool_node::finalize_decl): Likewise.
11263 (symbol_table::compile): Drop no toplevel reorder path.
11265 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11267 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11268 edges; zero probability is not better than uninitialized.
11270 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11272 * asan.h (asan_sanitize_allocas_p): Declare.
11273 * asan.c (asan_sanitize_allocas_p): New function.
11274 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11275 (handle_builtin_alloca): Likewise.
11276 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11277 if !asan_sanitize_allocas_p.
11278 * params.def (asan-instrument-allocas): Add new option.
11279 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11280 * opts.c (common_handle_option): Disable allocas sanitization for
11283 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11285 * asan.c: Include gimple-fold.h.
11286 (get_last_alloca_addr): New function.
11287 (handle_builtin_stackrestore): Likewise.
11288 (handle_builtin_alloca): Likewise.
11289 (asan_emit_allocas_unpoison): Likewise.
11290 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11291 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11292 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11293 (instrument_builtin_call): Pass gimple iterator to
11294 get_mem_refs_of_builtin_call.
11295 (last_alloca_addr): New global.
11296 * asan.h (asan_emit_allocas_unpoison): Declare.
11297 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11298 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11299 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11300 if function calls alloca.
11301 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11302 * gimple-fold.h (replace_call_with_value): Declare.
11303 * internal-fn.c: Include asan.h.
11304 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11305 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11307 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11309 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11310 (C_SELFTEST_FLAGS): New.
11311 (CPP_SELFTEST_FLAGS): New.
11312 (SELFTEST_DEPS): New, from deps of s-selftest.
11313 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11314 (CPP_SELFTEST_DEPS): New.
11315 (selftest): Add dependency on s-selftest-c++.
11316 (s-selftest): Rename to...
11317 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11318 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11319 than SELFTEST_FLAGS.
11320 (selftest-gdb): Rename to...
11321 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11323 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11324 (selftest-valgrind): Rename to...
11325 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11327 (selftest-valgrind): Reintroduce as an alias for
11328 selftest-c-valgrind.
11329 (s-selftest-c++): New.
11330 (selftest-c++-gdb): New.
11331 (selftest-c++-valgrind): New.
11333 2017-07-06 Olivier Hainque <hainque@adacore.com>
11335 * gcc.c (process_command): When deciding if undefined variables
11336 should be ignored when processing specs, accept "gcc -v" as well.
11338 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11340 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11341 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11343 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11345 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11346 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11347 * config/arm/arm-tables.opt: Regenerate.
11348 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11350 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11352 2017-07-06 Carl Love <cel@us.ibm.com>
11354 * ChangeLog: Clean up from mid air collision
11356 2017-07-06 Carl Love <cel@us.ibm.com>
11358 * config/rs6000/rs6000-c.c: Add support for built-in functions
11359 vector signed int vec_subc (vector signed int, vector signed int);
11360 vector signed __int128 vec_subc (vector signed __int128,
11361 vector signed __int128);
11362 vector unsigned __int128 vec_subc (vector unsigned __int128,
11363 vector unsigned __int128);
11364 vector signed int vec_sube (vector signed int, vector signed int,
11365 vector signed int);
11366 vector unsigned int vec_sube (vector unsigned int,
11367 vector unsigned int,
11368 vector unsigned int);
11369 vector signed __int128 vec_sube (vector signed __int128,
11370 vector signed __int128,
11371 vector signed__int128);
11372 vector unsigned __int128 vec_sube (vector unsigned __int128,
11373 vector unsigned __int128,
11374 vector unsigned __int128);
11375 vector signed int vec_subec (vector signed int, vector signed int,
11376 vector signed int);
11377 vector unsigned int vec_subec (vector unsigned int,
11378 vector unsigned int,
11379 vector unsigned int);
11380 vector signed __int128 vec_subec (vector signed __int128,
11381 vector signed __int128,
11382 vector signed__int128);
11383 vector unsigned __int128 vec_subec (vector unsigned __int128,
11384 vector unsigned __int128,
11385 vector unsigned __int128);
11386 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11387 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11388 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11389 BU_ALTIVEC_OVERLOAD_X definitions.
11390 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11391 * doc/extend.texi: Update the built-in documentation file for the new
11392 built-in functions.
11394 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11397 * diagnostic-show-locus.c (layout::layout): Use start and finish
11398 spelling location for the start and finish of each range.
11399 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11400 Add unused aspect param.
11401 * input.c (expand_location_1): Add "aspect" param, and use it
11402 to access the correct part of the location.
11403 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11405 (expand_location_to_spelling_point): Likewise.
11406 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11407 param, and pass it to expand_location_1.
11409 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11411 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11412 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11413 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11414 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11415 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11416 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11417 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11418 _mm_maskz_getmant_ss): New intrinsics.
11419 (__builtin_ia32_getexpss128_mask): Changed to ...
11420 __builtin_ia32_getexpss128_round ... this.
11421 (__builtin_ia32_getexpsd128_mask): Changed to ...
11422 __builtin_ia32_getexpsd128_round ... this.
11423 * config/i386/i386-builtin-types.def
11424 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11425 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11426 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11427 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11428 __builtin_ia32_getmantss_mask_round): New builtins.
11429 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11430 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11431 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11432 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11433 * config/i386/sse.md
11434 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11435 avx512f_sgetexp<mode><mask_scalar_name>
11436 <round_saeonly_scalar_name> ... this.
11437 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11438 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11439 vgetexp<ssescalarmodesuffix>
11440 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11441 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11442 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11443 avx512f_vgetmant<mode><mask_scalar_name>
11444 <round_saeonly_scalar_name> ... this.
11445 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11446 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11447 vgetmant<ssescalarmodesuffix>
11448 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11449 %0<mask_scalar_operand4>, %1, %2
11450 <round_saeonly_scalar_mask_op4>, %3} ... this.
11451 * config/i386/subst.md (mask_scalar_operand4,
11452 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11453 round_saeonly_scalar_nimm_predicate): New subst attributes.
11455 2017-07-06 Julia Koval <julia.koval@intel.com>
11457 * config/i386/i386.c (ix86_erase_embedded_rounding):
11458 Remove code for old rounding pattern.
11460 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11462 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11464 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11466 * doc/sourcebuild.texi (Test Directives, Variants of
11467 dg-require-support): Add documentation for dg-require-stack-check.
11469 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11471 * config/i386/subst.md (mask_scalar, round_scalar,
11472 round_saeonly_scalar): New meta-templates.
11473 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11474 round_scalar_mask_operand3, round_scalar_mask_op3,
11475 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11476 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11477 round_saeonly_scalar_constraint,
11478 round_saeonly_scalar_prefix): New subst attribute.
11479 * config/i386/sse.md
11480 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11481 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11482 <round_scalar_name> ... this.
11483 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11484 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11485 <round_scalar_name> ... this.
11486 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11487 <sse>_vm<code><mode>3<mask_scalar_name>
11488 <round_saeonly_scalar_name> ... this.
11489 (v<plusminus_mnemonic><ssescalarmodesuffix>
11490 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11491 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11492 v<plusminus_mnemonic><ssescalarmodesuffix>
11493 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11494 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11495 (v<multdiv_mnemonic><ssescalarmodesuffix>
11496 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11497 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11498 v<multdiv_mnemonic><ssescalarmodesuffix>
11499 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11500 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11501 (v<maxmin_float><ssescalarmodesuffix>
11502 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11503 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11504 v<maxmin_float><ssescalarmodesuffix>
11505 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11506 %0<mask_scalar_operand3>, %1, %<iptr>2
11507 <round_saeonly_scalar_mask_op3>} ... this.
11509 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11511 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11512 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11514 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11515 Alan Hayward <alan.hayward@arm.com>
11516 David Sherwood <david.sherwood@arm.com>
11518 * combine.c (simplify_if_then_else): Remove "enum" before
11520 * compare-elim.c (can_eliminate_compare): Likewise.
11521 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11523 (aarch64_lookup_simd_builtin_type): Likewise.
11524 (aarch64_simd_builtin_type): Likewise.
11525 (aarch64_init_simd_builtin_types): Likewise.
11526 (aarch64_simd_expand_args): Likewise.
11527 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11529 (aarch64_reverse_mask): Likewise.
11530 (aarch64_simd_emit_reg_reg_move): Likewise.
11531 (aarch64_gen_adjusted_ldpstp): Likewise.
11532 (aarch64_ccmp_mode_to_code): Likewise.
11533 (aarch64_operands_ok_for_ldpstp): Likewise.
11534 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11535 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11537 (aarch64_min_divisions_for_recip_mul): Likewise.
11538 (aarch64_reassociation_width): Likewise.
11539 (aarch64_get_condition_code_1): Likewise.
11540 (aarch64_simd_emit_reg_reg_move): Likewise.
11541 (aarch64_simd_attr_length_rglist): Likewise.
11542 (aarch64_reverse_mask): Likewise.
11543 (aarch64_operands_ok_for_ldpstp): Likewise.
11544 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11545 (aarch64_gen_adjusted_ldpstp): Likewise.
11546 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11548 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11549 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11550 (arm_lookup_simd_builtin_type): Likewise.
11551 (arm_simd_builtin_type): Likewise.
11552 (arm_init_simd_builtin_types): Likewise.
11553 (arm_expand_builtin_args): Likewise.
11554 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11555 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11556 (ft32_setup_incoming_varargs): Likewise.
11557 (ft32_function_arg): Likewise.
11558 (ft32_function_arg_advance): Likewise.
11559 (ft32_pass_by_reference): Likewise.
11560 (ft32_arg_partial_bytes): Likewise.
11561 (ft32_valid_pointer_mode): Likewise.
11562 (ft32_addr_space_pointer_mode): Likewise.
11563 (ft32_addr_space_legitimate_address_p): Likewise.
11564 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11566 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11567 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11568 (iamcu_alignment): Likewise.
11569 (canonicalize_vector_int_perm): Likewise.
11570 (ix86_noce_conversion_profitable_p): Likewise.
11571 (ix86_mpx_bound_mode): Likewise.
11572 (ix86_operands_ok_for_move_multiple): Likewise.
11573 * config/microblaze/microblaze-protos.h
11574 (microblaze_expand_conditional_branch_reg): Likewise.
11575 * config/microblaze/microblaze.c
11576 (microblaze_expand_conditional_branch_reg): Likewise.
11577 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11579 (rs6000_reassociation_width): Likewise.
11580 (rs6000_invalid_binary_op): Likewise.
11581 (fusion_p9_p): Likewise.
11582 (emit_fusion_p9_load): Likewise.
11583 (emit_fusion_p9_store): Likewise.
11584 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11586 (riscv_hard_regno_mode_ok_p): Likewise.
11587 (riscv_address_insns): Likewise.
11588 (riscv_split_symbol): Likewise.
11589 (riscv_legitimize_move): Likewise.
11590 (riscv_function_value): Likewise.
11591 (riscv_hard_regno_nregs): Likewise.
11592 (riscv_expand_builtin): Likewise.
11593 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11594 (riscv_build_integer): Likewise.
11595 (riscv_split_integer): Likewise.
11596 (riscv_legitimate_constant_p): Likewise.
11597 (riscv_cannot_force_const_mem): Likewise.
11598 (riscv_regno_mode_ok_for_base_p): Likewise.
11599 (riscv_valid_base_register_p): Likewise.
11600 (riscv_valid_offset_p): Likewise.
11601 (riscv_valid_lo_sum_p): Likewise.
11602 (riscv_classify_address): Likewise.
11603 (riscv_legitimate_address_p): Likewise.
11604 (riscv_address_insns): Likewise.
11605 (riscv_load_store_insns): Likewise.
11606 (riscv_force_binary): Likewise.
11607 (riscv_split_symbol): Likewise.
11608 (riscv_force_address): Likewise.
11609 (riscv_legitimize_address): Likewise.
11610 (riscv_move_integer): Likewise.
11611 (riscv_legitimize_const_move): Likewise.
11612 (riscv_legitimize_move): Likewise.
11613 (riscv_address_cost): Likewise.
11614 (riscv_subword): Likewise.
11615 (riscv_output_move): Likewise.
11616 (riscv_canonicalize_int_order_test): Likewise.
11617 (riscv_emit_int_order_test): Likewise.
11618 (riscv_function_arg_boundary): Likewise.
11619 (riscv_pass_mode_in_fpr_p): Likewise.
11620 (riscv_pass_fpr_single): Likewise.
11621 (riscv_pass_fpr_pair): Likewise.
11622 (riscv_get_arg_info): Likewise.
11623 (riscv_function_arg): Likewise.
11624 (riscv_function_arg_advance): Likewise.
11625 (riscv_arg_partial_bytes): Likewise.
11626 (riscv_function_value): Likewise.
11627 (riscv_pass_by_reference): Likewise.
11628 (riscv_setup_incoming_varargs): Likewise.
11629 (riscv_print_operand): Likewise.
11630 (riscv_elf_select_rtx_section): Likewise.
11631 (riscv_save_restore_reg): Likewise.
11632 (riscv_for_each_saved_reg): Likewise.
11633 (riscv_register_move_cost): Likewise.
11634 (riscv_hard_regno_mode_ok_p): Likewise.
11635 (riscv_hard_regno_nregs): Likewise.
11636 (riscv_class_max_nregs): Likewise.
11637 (riscv_memory_move_cost): Likewise.
11638 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11639 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11640 (rl78_addr_space_address_mode): Likewise.
11641 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11643 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11644 (rs6000_reassociation_width): Likewise.
11645 (rs6000_invalid_binary_op): Likewise.
11646 (fusion_p9_p): Likewise.
11647 (emit_fusion_p9_load): Likewise.
11648 (emit_fusion_p9_store): Likewise.
11649 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11650 (ok_for_simple_move_operands): Likewise.
11651 (ok_for_simple_move_strict_operands): Likewise.
11652 (ok_for_simple_arith_logic_operands): Likewise.
11653 (visium_legitimize_reload_address): Likewise.
11654 (visium_select_cc_mode): Likewise.
11655 (output_cbranch): Likewise.
11656 (visium_split_double_move): Likewise.
11657 (visium_expand_copysign): Likewise.
11658 (visium_expand_int_cstore): Likewise.
11659 (visium_expand_fp_cstore): Likewise.
11660 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11661 (visium_function_arg): Likewise.
11662 (visium_function_arg_advance): Likewise.
11663 (visium_libcall_value): Likewise.
11664 (visium_setup_incoming_varargs): Likewise.
11665 (visium_legitimate_constant_p): Likewise.
11666 (visium_legitimate_address_p): Likewise.
11667 (visium_legitimize_address): Likewise.
11668 (visium_secondary_reload): Likewise.
11669 (visium_register_move_cost): Likewise.
11670 (visium_memory_move_cost): Likewise.
11671 (prepare_move_operands): Likewise.
11672 (ok_for_simple_move_operands): Likewise.
11673 (ok_for_simple_move_strict_operands): Likewise.
11674 (ok_for_simple_arith_logic_operands): Likewise.
11675 (visium_function_value_1): Likewise.
11676 (rtx_ok_for_offset_p): Likewise.
11677 (visium_legitimize_reload_address): Likewise.
11678 (visium_split_double_move): Likewise.
11679 (visium_expand_copysign): Likewise.
11680 (visium_expand_int_cstore): Likewise.
11681 (visium_expand_fp_cstore): Likewise.
11682 (visium_split_cstore): Likewise.
11683 (visium_select_cc_mode): Likewise.
11684 (visium_split_cbranch): Likewise.
11685 (output_cbranch): Likewise.
11686 (visium_print_operand_address): Likewise.
11687 * expmed.c (flip_storage_order): Likewise.
11688 * expmed.h (emit_cstore): Likewise.
11689 (flip_storage_order): Likewise.
11690 * genrecog.c (validate_pattern): Likewise.
11691 * hsa-gen.c (gen_hsa_addr): Likewise.
11692 * internal-fn.c (expand_arith_overflow): Likewise.
11693 * ira-color.c (allocno_copy_cost_saving): Likewise.
11694 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11695 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11696 (process_invariant_for_inheritance): Likewise.
11697 * lra-eliminations.c (move_plus_up): Likewise.
11698 * omp-low.c (lower_oacc_reductions): Likewise.
11699 * simplify-rtx.c (simplify_subreg): Likewise.
11700 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11701 (TARGET_CHKP_BOUND_MODE): Likewise..
11702 * targhooks.c (default_chkp_bound_mode): Likewise.
11703 (default_setup_incoming_vararg_bounds): Likewise.
11704 * targhooks.h (default_chkp_bound_mode): Likewise.
11705 (default_setup_incoming_vararg_bounds): Likewise.
11706 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11707 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11708 (have_whole_vector_shift): Likewise.
11709 * tree-vect-stmts.c (vectorizable_load): Likewise.
11710 * doc/tm.texi: Regenerate.
11712 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11714 Graceful degrade if Binutils PR21472 is not available.
11717 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11718 .rodata in flash test fails.
11719 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11720 * confgure: Regenerate.
11721 * config.in: Regenerate.
11722 * config/avr/avr.c (avr_asm_named_section)
11723 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11724 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11725 (avr_asm_init_sections): Same.
11727 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11729 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11730 (fma<VH:mode>4_intrinsic): Likewise.
11731 (*fmsub<VCVTF:mode>4): Likewise.
11732 (*fmsub<VH:mode>4_intrinsic): Likewise.
11734 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11737 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11738 Don't depend on "optimize > 0".
11739 (out_movhi_r_mr, out_movqi_mr_r): Same.
11740 (out_movhi_mr_r, out_movqi_r_mr): Same.
11741 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11742 io_address_operand on "optimize > 0".
11744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11746 * tree-loop-distribution.c: Add general explanantion on the pass.
11747 (generate_loops_for_partition): Mark distributed loop.
11748 (pg_add_dependence_edges): New parameter. Handle alias data
11749 dependence specially and record it in the parameter if asked.
11750 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11751 (init_partition_graph_vertices, add_partition_graph_edge): New.
11752 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11753 (free_partition_graph_vdata, build_partition_graph): New.
11754 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11755 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11756 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11757 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11758 (version_for_distribution_p, finalize_partitions): New.
11759 (distribute_loop): Handle alias data dependence specially. Factor
11760 out loop fusion code as functions and call these functions.
11762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11764 * tree-loop-distribution.c (classify_partition): New parameter and
11765 better handle reduction statement.
11766 (rdg_build_partitions): Revise comment.
11767 (distribute_loop): Compute statements in all partitions and pass it
11768 to classify_partition.
11770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11772 * tree-loop-distribution.c (enum partition_type): New.
11773 (struct partition): New field type.
11774 (partition_merge_into): Add parameter. Update partition type.
11775 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11776 (build_rdg_partition_for_vertex): Compute partition type.
11777 (rdg_build_partitions): Dump partition type.
11778 (distribute_loop): Update calls to partition_merge_into.
11780 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11782 * tree-loop-distribution.c (struct ddr_hasher): New.
11783 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11785 (classify_partition): Call get_data_dependence.
11786 (pg_add_dependence_edges): Ditto.
11787 (distribute_loop): Release data dependence hash table.
11789 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11791 * tree-loop-distribution.c (ref_base_address): Delete.
11792 (similar_memory_accesses): Rename ...
11793 (share_memory_accesses): ... to this. Check if partitions access
11794 the same memory reference.
11795 (distribute_loop): Call share_memory_accesses.
11797 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11799 * tree-loop-distribution.c (struct partition): New field recording
11800 its data reference.
11801 (partition_alloc, partition_free): Init and release data refs.
11802 (partition_merge_into): Merge data refs.
11803 (build_rdg_partition_for_vertex): Collect data refs for partition.
11804 (pg_add_dependence_edges): Change parameters from vector to bitmap.
11806 (distribute_loop): Remve data refs from vertice data of partition
11809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11811 * tree-loop-distribution.c (params.h): Include header file.
11812 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
11813 (datarefs_vec): New global var.
11814 (create_rdg_vertices): Use datarefs_vec directly.
11815 (free_rdg): Don't free data references.
11816 (build_rdg): Update use. Don't free data references.
11817 (distribute_loop): Compute global variable for data references.
11818 Bail out if there are too many data references.
11820 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11822 * tree-loop-distribution.c (loop_nest): New global var.
11823 (build_rdg): Use loop directly, rather than loop nest.
11824 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11826 (distribute_loop): Compute global variable loop nest. Update use.
11828 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11830 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11831 (partition_merge_into): New parameter. Dump reason for fusion.
11832 (distribute_loop): Update use of partition_merge_into.
11834 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11836 * tree-loop-distribution.c (bb_top_order_index): New.
11837 (bb_top_order_index_size, bb_top_order_cmp): New.
11838 (stmts_from_loop): Use topological order.
11839 (pass_loop_distribution::execute): Compute and release topological
11840 order for basic blocks.
11842 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11844 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11849 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11850 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11851 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11852 * internal-fn.def (LOOP_DIST_ALIAS): New.
11853 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11854 (fold_loop_internal_call): ... this.
11855 (vect_loop_dist_alias_call): New function.
11856 (set_uid_loop_bbs): Call fold_loop_internal_call.
11857 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11860 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11863 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11864 Require dead FLAGS_REG at the beginning of a peephole.
11866 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11869 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11870 arguments in the call to __builtin_ia32_sbb_u32.
11871 (_subborrow_u64): Swap _X and _Y arguments in the call to
11872 __builtin_ia32_sbb_u64.
11874 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11877 * tree-vrp.c (compare_assert_loc): Turn into a function template
11878 with stable template parameter. Only test if a->e is NULL,
11879 !a->e == !b->e has been verified already. Use e == NULL or
11880 e != NULL instead of e or ! e tests. If stable is true, don't use
11881 iterative_hash_expr, on the other side allow a or b or both NULL
11882 and sort the NULLs last.
11883 (process_assert_insertions): Sort using compare_assert_loc<false>
11884 instead of compare_assert_loc, later sort using
11885 compare_assert_loc<true> before calling process_assert_insertions_for
11886 in a loop. Use break instead of continue once seen NULL pointer.
11888 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11890 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11891 Cortex-R7 and Cortex-R8 processors.
11893 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11895 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11896 uninitialized while src is not.
11898 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11900 * common/config/arm/arm-common.c: Adjust include path for
11902 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11903 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11904 (arm-cpu-data.h): Likewise.
11905 (arm-cpu-cdata.h): Likewise.
11906 * config/arm/arm-cpu.h: Delete.
11907 * config/arm/arm-cpu-cdata.h: Delete.
11908 * config/arm/arm-cpu-data.h: Delete.
11910 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11912 * config/arm/arm-cpus.in (cortex-a55): New.
11913 (cortex-a75): Likewise.
11914 (cortex-a75.cortex-a55): Likewise.
11915 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11917 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11918 * config/arm/arm-cpu-cdata.h: Regenerate.
11919 * config/arm/arm-cpu-data.h: Regenerate.
11920 * config/arm/arm-cpu.h: Regenerate.
11921 * config/arm/arm-tables.opt: Regenerate.
11922 * config/arm/arm-tune.md: Regenerate.
11924 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11926 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11928 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11930 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11933 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11935 PR tree-optimization/81292
11936 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11937 full_string_p, also call adjust_related_strinfos if the adjustment
11938 is simple, otherwise invalidate related strinfos.
11940 2017-07-04 Martin Liska <mliska@suse.cz>
11943 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11944 newly created variable as DECL_IGNORED_P.
11946 2017-07-04 Martin Liska <mliska@suse.cz>
11949 * ipa-inline.c (inline_small_functions):
11950 Use xstrdup_for_dump.
11952 2017-07-04 Tom de Vries <tom@codesourcery.com>
11954 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11956 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11959 * config/darwin.c (darwin_function_switched_text_sections):
11962 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11964 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11966 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11968 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11970 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11972 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11973 min_profitable_iters, and th as inclusive lower bounds.
11974 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11975 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11976 for min_profitable_iters and min_profitable_estimate.
11977 (vect_transform_loop): Treat th as an inclusive lower bound.
11978 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11980 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11983 * config/darwin.c (darwin_function_switched_text_sections):
11984 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11985 in two pieces, and suppress the use of buf.
11987 2017-07-03 Nathan Sidwell <nathan@acm.org>
11989 * hash-table.h (hash_table_mod1): Fix indentation.
11991 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11993 PR middle-end/81290
11994 * predict.c (force_edge_cold): Be more careful about propagation
11996 * profile-count.h (profile_probability::guessed,
11997 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11999 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12001 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12003 * doc/invoke.texi (rcpc architecture extension): Document it.
12005 2017-07-03 Richard Biener <rguenther@suse.de>
12007 PR tree-optimization/60510
12008 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12009 the scalar reduction PHI and use it.
12010 (vectorizable_reduction): Properly guard the single_defuse_cycle
12011 path for non-SLP reduction chains where we cannot use it.
12012 Rework reduc_def/index and vector type deduction. Rework
12013 vector operand gathering during reduction op code-gen.
12014 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12015 chains dissolve the chain and leave it to non-SLP reduction
12018 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12020 * tree-data-ref.h (dr_alignment): Declare.
12021 * tree-data-ref.c (dr_alignment): New function.
12022 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12023 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12025 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12027 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12029 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12030 and base_misalignment fields.
12031 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12032 * tree-data-ref.c: Include builtins.h.
12033 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12034 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12035 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12036 * tree-vect-data-refs.c: Include tree-cfg.h.
12037 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12038 fields instead of calculating an alignment here.
12039 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12040 innermost_loop_behavior fields.
12042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12044 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12046 (DR_STEP_ALIGNMENT): New macro.
12047 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12048 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12049 (create_data_ref): Print it.
12050 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12051 to tell whether the step preserves vector (mis)alignment.
12052 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12053 Move the check for an integer step and generalise to all INTEGER_CST.
12054 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12055 Print the outer step alignment.
12057 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12059 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12060 with offset_alignment.
12061 (DR_ALIGNED_TO): Delete.
12062 (DR_OFFSET_ALIGNMENT): New macro.
12063 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12064 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12065 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12066 (create_data_ref): Likewise.
12067 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12068 (vect_analyze_data_refs): Likewise.
12069 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12070 creating dummy innermost behavior.
12072 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12074 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12075 with a "innermost_loop_behavior *" and refeence tree.
12076 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12077 (create_data_ref): Update call accordingly.
12078 * tree-predcom.c (find_looparound_phi): Likewise.
12080 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12082 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12083 fields with dr_wrt_vec_loop.
12084 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12085 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12086 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12087 (vect_dr_behavior): New function.
12088 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12089 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12090 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12091 track whether the step preserves the misalignment.
12092 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12093 Use vect_dr_behavior.
12094 (vect_setup_realignment): Update call accordingly.
12095 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12096 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12097 call to vect_create_addr_base_for_vector_ref.
12098 (vect_create_cond_for_align_checks): Likewise.
12099 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12100 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12101 (vect_recog_mask_conversion_pattern): Likewise.
12102 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12103 (new_stmt_vec_info): Remove redundant zeroing.
12105 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12107 * common/config/arm/arm-common.c (arm_be8_option): New function.
12108 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12109 (ISA_ARMv6): Add isa_bit_be8.
12110 * config/arm/arm.h (arm_be8_option): Add prototype.
12111 (BE8_SPEC_FUNCTION): New define.
12112 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12113 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12114 (mlittle-endian): Similarly.
12115 (mbe8, mbe32): New options.
12116 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12117 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12119 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12121 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12123 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12125 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12126 (cleanup_tree_cfg_bb): Use it.
12127 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12129 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12131 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12134 * loop-doloop.c (add_test): Update profile.
12136 2017-07-03 Martin Liska <mliska@suse.cz>
12139 * sanopt.c (rewrite_usage_of_param): New function.
12140 (sanitize_rewrite_addressable_params): Likewise.
12141 (pass_sanopt::execute): Call rewrite_usage_of_param.
12143 2017-07-03 Richard Biener <rguenther@suse.de>
12145 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12146 back to using VIEW_CONVERT_EXPR.
12148 2017-07-03 Martin Liska <mliska@suse.cz>
12151 * doc/extend.texi: Document when a resolver function is
12152 generated for target_clones.
12154 2017-07-03 Martin Liska <mliska@suse.cz>
12156 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12157 and shadow memory of auto variables which are subject of
12158 use-after-scope sanitization.
12159 (asan_expand_mark_ifn): Add do set only when is_poison.
12161 2016-07-03 Richard Biener <rguenther@suse.de>
12163 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12165 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12166 (vectorizable_reduction): Perform reduction PHI creation when
12167 visiting a reduction PHI and adjust and simplify code generation
12168 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12169 (vect_transform_loop): Visit reduction PHIs.
12170 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12171 defs into the SLP tree.
12172 (vect_build_slp_tree): Reduction defs terminate the recursion.
12173 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12175 (vect_get_vec_defs_for_stmt_copy): Export.
12176 (vect_get_vec_defs): Likewise.
12177 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12179 (vect_get_vec_defs_for_stmt_copy): Declare.
12180 (vect_get_vec_defs): Likewise.
12182 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12184 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12185 parameter with a "loop" parameter and use it instead of the
12186 loop containing DR_STMT. Don't check simple_iv when doing
12187 BB analysis. Describe the two analysis modes in the comment.
12189 2017-07-03 Tom de Vries <tom@codesourcery.com>
12191 PR tree-optimization/69468
12192 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12193 (find_same_succ_bb): Handle ignore_edge_flags.
12195 2017-07-03 Tom de Vries <tom@codesourcery.com>
12197 PR tree-optimization/81192
12198 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12200 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12202 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12204 2017-07-03 Tom de Vries <tom@codesourcery.com>
12206 PR tree-optimization/81192
12207 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12208 BB_SAME_SUCC (bb) == NULL.
12210 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12212 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12215 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12217 * dumpfile.c: Include profile-count.h
12218 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12220 (insert_cond_bb): Update profile.
12221 * tree-cfg.h (insert_cond_bb): Update prototype.
12222 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12223 * tree-dump.c: Do not include tree-cfg.
12225 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12227 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12229 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12231 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12236 * tree-complex.c (expand_complex_div_wide): update profile.
12238 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12239 Alan Hayward <alan.hayward@arm.com>
12240 David Sherwood <david.sherwood@arm.com>
12242 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12243 (CORETYPES_H): New define.
12244 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12245 (insn-modes-inline.h, s-modes-inline-h): New rules.
12246 (generated_files): Add insn-modes-inline.h.
12247 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12248 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12249 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12250 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12251 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12252 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12253 (build/gencodes.o, build/genconditions.o): Likewise.
12254 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12255 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12256 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12257 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12258 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12259 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12260 * coretypes.h: Include everything up to real.h for generators.
12261 Include insn-modes.h first. Include wide-int-print.h after
12262 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12263 * machmode.h: Don't include insn-modes.h here.
12264 * function-tests.c: Remove includes of signop.h, machmode.h,
12265 double-int.h and wide-int.h.
12267 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12269 * optc-save-gen.awk: Likewise.
12270 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12271 * godump.c: Remove include of wide-int-print.h.
12272 * pretty-print.h: Likewise.
12273 * wide-int-print.cc: Likewise.
12274 * wide-int.cc: Likewise.
12275 * hash-map-tests.c: Remove include of signop.h.
12276 * hash-set-tests.c: Likewise.
12277 * rtl-tests.c: Likewise.
12278 * mkconfig.sh: Remove include of machmode.h.
12279 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12281 (emit_insn_modes_inline_h): ...this new function. Emit the code
12282 into an insn-modes-inline.h header file, adding appropriate
12283 include guards and end comments.
12284 (emit_insn_modes_c_header): Remove include of machmode.h.
12285 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12287 (main): Handle -i flag and call emit_insn_modes_inline_h when
12290 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12292 * tree-ssa-strlen.c (strinfo): Rename the length field to
12293 nonzero_chars. Add a full_string_p field.
12294 (compare_nonzero_chars, zero_length_string_p): New functions.
12295 (get_addr_stridx): Add an offset_out parameter.
12296 Use compare_nonzero_chars.
12297 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12298 (new_strinfo): Update after above changes to strinfo.
12299 (set_endptr_and_length): Set full_string_p.
12300 (get_string_length): Update after above changes to strinfo.
12301 (unshare_strinfo): Update call to new_strinfo.
12302 (maybe_invalidate): Likewise.
12303 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12304 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12305 as a uhwi instead of an shwi. Update after above changes to
12306 strinfo and new_strinfo.
12307 (zero_length_string): Assert that chainsi contains full strings.
12308 Use zero_length_string_p. Update call to new_strinfo.
12309 (adjust_related_strinfos): Update after above changes to strinfo.
12310 Copy full_string_p from origsi.
12311 (adjust_last_stmt): Use zero_length_string_p.
12312 (handle_builtin_strlen): Update after above changes to strinfo and
12313 new_strinfo. Install the lhs as the string length if the previous
12314 entry didn't describe a full string.
12315 (handle_builtin_strchr): Update after above changes to strinfo
12317 (handle_builtin_strcpy): Likewise.
12318 (handle_builtin_strcat): Likewise.
12319 (handle_builtin_malloc): Likewise.
12320 (handle_pointer_plus): Likewise.
12321 (handle_builtin_memcpy): Likewise. Track nonzero characters
12322 that aren't necessarily followed by a nul terminator.
12323 (handle_char_store): Likewise.
12325 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12327 PR tree-optimization/80769
12328 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12329 for malloc and calloc. Document the new invariant that all related
12330 strinfos have delayed lengths or none do.
12331 (verify_related_strinfos): Move earlier in file.
12332 (set_endptr_and_length): New function, split out from...
12333 (get_string_length): ...here. Also set the lengths of related
12335 (zero_length_string): Assert that chainsi has known (rather than
12337 (adjust_related_strinfos): Likewise.
12339 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12341 PR tree-optimization/81136
12342 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12343 assert that two references with the same misalignment have the same
12344 compile-time misalignment if those compile-time misalignments
12347 2017-07-01 Andi Kleen <ak@linux.intel.com>
12349 * print-tree.c (print_node): Print all attributes.
12351 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12353 * cfg.c (scale_bbs_frequencies): New function.
12354 * cfg.h (scale_bbs_frequencies): Declare it.
12355 * cfgloopanal.c (single_likely_exit): Cleanup.
12356 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12358 (scale_loop_profile): Likewise.
12359 (loop_version): Likewise.
12360 (create_empty_loop_on_edge): Update.
12361 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12362 scale_loop_frequencies, scale_loop_profile, loopify,
12363 loop_version): Update prototypes.
12364 * modulo-sched.c (sms_schedule): Update.
12365 * predict.c (unlikely_executed_edge_p): Also check probability.
12366 (probably_never_executed_edge_p): Fix typo.
12367 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12368 * tree-parloops.c (gen_parallel_loop): Update.
12369 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12370 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12371 * tree-ssa-loop-split.c (split_loop): Update.
12372 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12373 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12374 (vect_loop_versioning): Update.
12375 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12377 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12379 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12381 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12383 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12384 to keep profile consistent.
12386 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12388 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12389 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12390 * profile-count.h (max_safe_multiplier): Make unsigned.
12391 (profile_count::guessed_zero): New.
12393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12395 * bb-reorder.c (fix_up_crossing_landing_pad,
12396 fix_crossing_conditional_branches): Use make_single_succ_edge
12397 to keep profile consistent.
12399 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12401 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12404 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12407 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12408 the right scopes, make sure cond_jump isn't preserved between multiple
12409 iterations. Search for fallthru edge whenever there are 3+ edges and
12410 use find_fallthru_edge for it.
12412 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12414 Patch by Alexander Monakov <amonakov@ispras.ru>
12415 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12416 probabilities consistently.
12418 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12420 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12421 emit_cmp_and_jump_insns.
12423 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12426 * tree-inline.c (expand_call_inline): Combine profile statuses.
12428 2017-06-30 Andrew Pinski <apinski@cavium.com>
12430 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12431 fold_stmt returned true.
12433 2017-06-30 Nathan Sidwell <nathan@acm.org>
12435 * ggc.h (empty_string): Delete.
12436 * cfgexpand.c (expand_asm_stmt): Use plain "".
12437 * optabs.c (expand_asm_memory_barrier): Likewise.
12438 * stringpool.c (empty_string): Delete.
12439 (digit_vector, digit_string): Delete.
12440 (ggc_alloc_string): Use plain "", don't optimize single digit
12441 strings. Use ggc_alloc_atomic.
12443 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12445 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12446 comparison set and one other set, use the cost of the non-comparison
12449 2017-06-30 Nathan Sidwell <nathan@acm.org>
12451 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12454 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12456 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12457 loops. Remove now unneeded calls to gimple_switch_set_label() that
12458 just set removed labels to NULL_TREE.
12460 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12462 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12463 (set_range_info): ...here. Only call set_range_info_raw if domain
12465 (set_nonzero_bits): Call set_range_info_raw.
12466 * tree-ssanames.h (set_range_info_raw): New.
12468 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12471 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12472 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12473 of nonimmediate_operand and <store_mask_constraint> instead of m
12474 for the input operand. For V8FI iterator, always split if input
12475 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12476 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12477 <store_mask_predicate> instead of register_operand and
12478 <store_mask_constraint> instead of v for the input operand. Make
12479 sure both operands aren't MEMs for if not <mask_applied>.
12481 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12483 * lto-wrapper.c (copy_file) Close both file descriptors before
12486 2017-06-30 Martin Liska <mliska@suse.cz>
12489 * multiple_target.c (create_dispatcher_calls): Make ifunc
12490 also for function that don't have calls or are not referenced.
12492 2017-06-30 Richard Biener <rguenther@suse.de>
12494 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12495 analyze the first scalar stmt. Move vector type computation
12496 for the BB case here from ...
12497 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12498 live operation processing in the SLP case properly.
12500 2017-06-30 Richard Biener <rguenther@suse.de>
12502 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12504 2017-06-30 Martin Liska <mliska@suse.cz>
12507 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12508 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12510 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12512 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12515 2017-06-30 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/81249
12518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12519 condition reduction result to original scalar type.
12521 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12523 * profile-count.h (enum profile_quality): Fix typos and whitespace
12526 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12528 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12529 type for branch probabilities.
12531 2017-06-29 Julian Brown <julian@codesourcery.com>
12532 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12534 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12535 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12536 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12537 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12539 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12541 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12542 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12543 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12544 CC usage from generic code to here.
12545 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12546 CC usage into the target macros.
12548 2017-06-29 Maya Rashish <coypu@sdf.org>
12550 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12553 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12555 * arm/arm-builtins.c: Include profile-count.h
12556 * except.c (sjlj_emit_function_enter): Use
12557 profile_probability::unlikely.
12559 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12561 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12562 and tocrel_offset be pointer args rather than implicitly using
12564 (legitimate_constant_pool_address_p, rs6000_emit_move,
12565 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12566 tocrel_offset and use in toc_relative_expr_p call.
12567 (print_operand, print_operand_address): Use static tocrel_base_oac
12568 and tocrel_offset_oac.
12569 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12572 2017-06-29 Maya Rashish <coypu@sdf.org>
12574 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12576 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12578 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12579 objects, take into account only the alignment of 'op0' and 'mode1' if
12582 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12584 * ccmp.c (ccmp_tree_comparison_p): New function.
12585 (ccmp_candidate_p): Update to use above function.
12586 (get_compare_parts): New function.
12587 (expand_ccmp_next): Update to use new functions.
12588 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12590 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12591 take mode as argument.
12592 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12593 * expr.c (expand_expr_real_1): Pass mode as argument.
12595 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12597 * combine.c (combine_instructions): Print insns to dump_file, together
12600 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12602 * asan.c (asan_emit_stack_protection): Update.
12603 (create_cond_insert_point): Update.
12604 * auto-profile.c (afdo_propagate_circuit): Update.
12605 * basic-block.h (struct edge_def): Turn probability to
12606 profile_probability.
12607 (EDGE_FREQUENCY): Update.
12608 * bb-reorder.c (find_traces_1_round): Update.
12609 (better_edge_p): Update.
12610 (sanitize_hot_paths): Update.
12611 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12612 (make_single_succ_edge): Update.
12613 (check_bb_profile): Update.
12614 (dump_edge_info): Update.
12615 (update_bb_profile_for_threading): Update.
12616 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12617 probabilitycount to 0.
12618 * cfgbuild.c (compute_outgoing_frequencies): Update.
12619 * cfgcleanup.c (try_forward_edges): Update.
12620 (outgoing_edges_match): Update.
12621 (try_crossjump_to_edge): Update.
12622 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12623 (expand_gimple_tailcall): Update.
12624 (construct_init_block): Use make_single_succ_edge.
12625 (construct_exit_block): Use make_single_succ_edge.
12626 * cfghooks.c (verify_flow_info): Update.
12627 (redirect_edge_succ_nodup): Update.
12628 (split_edge): Update.
12629 (account_profile_record): Update.
12630 * cfgloopanal.c (single_likely_exit): Update.
12631 * cfgloopmanip.c (scale_loop_profile): Update.
12632 (set_zero_probability): Remove.
12633 (duplicate_loop_to_header_edge): Update.
12634 * cfgloopmanip.h (loop_version): Update prototype.
12635 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12636 (force_nonfallthru_and_redirect): Update.
12637 (update_br_prob_note): Update.
12638 (rtl_verify_edges): Update.
12639 (purge_dead_edges): Update.
12640 (rtl_lv_add_condition_to_bb): Update.
12641 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12642 * cgraphunit.c (init_lowered_empty_function): Update.
12643 (cgraph_node::expand_thunk): Update.
12644 * cilk-common.c: Include profile-count.h
12645 * dojump.c (inv): Remove.
12646 (jumpifnot): Update.
12647 (jumpifnot_1): Update.
12648 (do_jump_1): Update.
12650 (do_jump_by_parts_greater_rtx): Update.
12651 (do_compare_rtx_and_jump): Update.
12652 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12653 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12654 * dwarf2cfi.c: Include profile-count.h
12655 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12656 (sjlj_emit_dispatch_table): Likewise.
12657 * explow.c: Include profile-count.h
12658 * expmed.c (emit_store_flag_force): Update.
12659 (do_cmp_and_jump): Update.
12660 * expr.c (compare_by_pieces_d::generate): Update.
12661 (compare_by_pieces_d::finish_mode): Update.
12662 (emit_block_move_via_loop): Update.
12663 (store_expr_with_bounds): Update.
12664 (store_constructor): Update.
12665 (expand_expr_real_2): Update.
12666 (expand_expr_real_1): Update.
12667 * expr.h (try_casesi, try_tablejump): Update prototypes.
12668 * gimple-pretty-print.c (dump_probability): Update.
12669 (dump_profile): New.
12670 (dump_gimple_label): Update.
12671 (dump_gimple_bb_header): Update.
12672 * graph.c (draw_cfg_node_succ_edges): Update.
12673 * hsa-gen.c (convert_switch_statements): Update.
12674 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12675 (find_if_case_1): Update.
12676 (find_if_case_2): Update.
12677 * internal-fn.c (expand_arith_overflow_result_store): Update.
12678 (expand_addsub_overflow): Update.
12679 (expand_neg_overflow): Update.
12680 (expand_mul_overflow): Update.
12681 (expand_vector_ubsan_overflow): Update.
12682 * ipa-cp.c (good_cloning_opportunity_p): Update.
12683 * ipa-split.c (split_function): Use make_single_succ_edge.
12684 * ipa-utils.c (ipa_merge_profiles): Update.
12685 * loop-doloop.c (add_test): Update.
12686 (doloop_modify): Update.
12687 * loop-unroll.c (compare_and_jump_seq): Update.
12688 (unroll_loop_runtime_iterations): Update.
12689 * lra-constraints.c (lra_inheritance): Update.
12690 * lto-streamer-in.c (input_cfg): Update.
12691 * lto-streamer-out.c (output_cfg): Update.
12692 * mcf.c (adjust_cfg_counts): Update.
12693 * modulo-sched.c (sms_schedule): Update.
12694 * omp-expand.c (expand_omp_for_init_counts): Update.
12695 (extract_omp_for_update_vars): Update.
12696 (expand_omp_ordered_sink): Update.
12697 (expand_omp_for_ordered_loops): Update.
12698 (expand_omp_for_generic): Update.
12699 (expand_omp_for_static_nochunk): Update.
12700 (expand_omp_for_static_chunk): Update.
12701 (expand_cilk_for): Update.
12702 (expand_omp_simd): Update.
12703 (expand_omp_taskloop_for_outer): Update.
12704 (expand_omp_taskloop_for_inner): Update.
12705 * omp-simd-clone.c (simd_clone_adjust): Update.
12706 * optabs.c (expand_doubleword_shift): Update.
12707 (expand_abs): Update.
12708 (emit_cmp_and_jump_insn_1): Update.
12709 (expand_compare_and_swap_loop): Update.
12710 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12711 * predict.c (predictable_edge_p): Update.
12712 (edge_probability_reliable_p): Update.
12713 (set_even_probabilities): Update.
12714 (combine_predictions_for_insn): Update.
12715 (combine_predictions_for_bb): Update.
12716 (propagate_freq): Update.
12717 (estimate_bb_frequencies): Update.
12718 (force_edge_cold): Update.
12719 * profile-count.c (profile_count::dump): Add missing space into dump.
12720 (profile_count::debug): Add newline.
12721 (profile_count::differs_from_p): Explicitly convert to unsigned.
12722 (profile_count::stream_in): Update.
12723 (profile_probability::dump): New member function.
12724 (profile_probability::debug): New member function.
12725 (profile_probability::differs_from_p): New member function.
12726 (profile_probability::differs_lot_from_p): New member function.
12727 (profile_probability::stream_in): New member function.
12728 (profile_probability::stream_out): New member function.
12729 * profile-count.h (profile_count_quality): Rename to ...
12730 (profile_quality): ... this one.
12731 (profile_probability): New.
12732 (profile_count): Update.
12733 * profile.c (compute_branch_probabilities): Update.
12734 * recog.c (peep2_attempt): Update.
12735 * sched-ebb.c (schedule_ebbs): Update.
12736 * sched-rgn.c (find_single_block_region): Update.
12737 (compute_dom_prob_ps): Update.
12738 (schedule_region): Update.
12739 * sel-sched-ir.c (compute_succs_info): Update.
12740 * stmt.c (struct case_node): Update.
12741 (do_jump_if_equal): Update.
12742 (get_outgoing_edge_probs): Update.
12743 (conditional_probability): Update.
12744 (emit_case_dispatch_table): Update.
12745 (expand_case): Update.
12746 (expand_sjlj_dispatch_table): Update.
12747 (emit_case_nodes): Update.
12748 * targhooks.c: Update.
12749 * tracer.c (better_p): Update.
12750 (find_best_successor): Update.
12751 * trans-mem.c (expand_transaction): Update.
12752 * tree-call-cdce.c: Update.
12753 * tree-cfg.c (gimple_split_edge): Upate.
12754 (move_sese_region_to_fn): Upate.
12755 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12756 * tree-eh.c (lower_resx): Upate.
12757 (cleanup_empty_eh_move_lp): Upate.
12758 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12759 * tree-inline.c (copy_edges_for_bb): Update.
12760 (copy_cfg_body): Update.
12761 * tree-parloops.c (gen_parallel_loop): Update.
12762 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12763 (gimple_gen_time_profiler): Update.
12764 * tree-ssa-dce.c (remove_dead_stmt): Update.
12765 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12766 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12767 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12768 (unloop_loops): Update.
12769 (try_peel_loop): Update.
12770 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12771 * tree-ssa-loop-split.c (connect_loops): Update.
12772 (split_loop): Update.
12773 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12774 (hoist_guard): Update.
12775 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12776 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12777 (value_replacement): Update.
12778 * tree-ssa-reassoc.c (branch_fixup): Update.
12779 * tree-ssa-tail-merge.c (replace_block_by): Update.
12780 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12781 (create_edge_and_update_destination_phis): Update.
12782 (compute_path_counts): Update.
12783 (recompute_probabilities): Update.
12784 (update_joiner_offpath_counts): Update.
12785 (freqs_to_counts_path): Update.
12786 (duplicate_thread_path): Update.
12787 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12788 (struct switch_conv_info): Update.
12789 (gen_inbound_check): Update.
12790 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12791 (vect_do_peeling): Update.
12792 (vect_loop_versioning): Update.
12793 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12794 (optimize_mask_stores): Update.
12795 * ubsan.c (ubsan_expand_null_ifn): Update.
12796 * value-prof.c (gimple_divmod_fixed_value): Update.
12797 (gimple_divmod_fixed_value_transform): Update.
12798 (gimple_mod_pow2): Update.
12799 (gimple_mod_pow2_value_transform): Update.
12800 (gimple_mod_subtract): Update.
12801 (gimple_mod_subtract_transform): Update.
12802 (gimple_ic): Update.
12803 (gimple_stringop_fixed_value): Update.
12804 (gimple_stringops_transform): Update.
12805 * value-prof.h: Update.
12807 2017-06-29 Carl Love <cel@us.ibm.com>
12809 * config/rs6000/rs6000-c.c: Add support for built-in functions
12810 vector signed int vec_signed (vector float);
12811 vector signed long long vec_signed (vector double);
12812 vector signed int vec_signed2 (vector double, vector double);
12813 vector signed int vec_signede (vector double);
12814 vector signed int vec_signedo (vector double);
12815 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
12816 instruction generator.
12817 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12818 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12819 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12821 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12822 vunsignede_v2df): Add define_expands.
12823 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12824 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12825 VEC_UNSIGNEDO): Add definitions.
12826 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12827 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12828 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12829 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12830 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12831 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12832 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12833 vec_unsignede, vec_unsignedo): Add builtin defines.
12834 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12836 * doc/extend.texi: Update the built-in documentation file for the
12837 new built-in functions.
12839 2017-06-29 Richard Biener <rguenther@suse.de>
12841 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12842 reduction chains to LOOP_VINFO_REDUCTIONS.
12843 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12844 SLP reductions after processing reduction chains.
12846 2017-06-29 Nathan Sidwell <nathan@acm.org>
12848 * builtins.c (fold_builtin_FUNCTION): Use
12849 lang_hooks.decl_printable_name.
12851 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12853 PR middle-end/81194
12854 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12855 with only one label.
12856 * stmt.c (expand_case): Assert NCASES is greater than one.
12858 2017-06-29 Richard Biener <rguenther@suse.de>
12860 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12862 (group_case_labels): Likewise.
12863 (find_taken_edge): Push sanity checking on val to workers...
12864 (find_taken_edge_cond_expr): ... here
12865 (find_taken_edge_switch_expr): ... and here, handle cases
12866 with just a default label.
12867 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12868 (group_case_labels): Likewise.
12869 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12870 group_case_labels does anything cleanup the CFG again.
12872 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12874 PR tree-optimization/81196
12875 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12876 exit condition comparing two IVs.
12878 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12880 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12881 profile to the dummy entry at the end of the list of architectures.
12882 * config/arm/arm-cpu-cdata.h: Regenerated.
12884 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12885 Michael Collison <michael.collison@arm.com>
12888 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12890 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12891 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12892 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12893 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12894 of shift when the shift amount is masked with constant equal to
12895 the size of the mode.
12896 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12899 2017-06-29 Martin Liska <mliska@suse.cz>
12901 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12903 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12906 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12909 (edge_growth_cache_entry::edge_growth_cache_entry): New
12911 (reset_edge_growth_cache): Update to use constructor.
12913 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12915 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12916 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12917 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12919 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12921 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12922 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12924 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12926 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12927 (*-linux-uclibc*): Add t-uclibc tmake_file.
12928 * config/t-musl: New.
12929 * config/t-uclibc: New.
12931 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12933 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12935 (gen_comm_data): Emit architectural setting of arch_prof.
12936 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12938 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12939 (armv8-m.base, armv8-m.main): Likewise.
12940 * arm-protos.h (arm_build_target): Add profile field.
12941 (arch_option): Likewise.
12942 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12944 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12945 arm_active_target.profile.
12947 2017-06-28 Richard Biener <rguenther@suse.de>
12949 PR middle-end/81227
12950 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12951 TYPE_OVERFLOW_WRAPS.
12952 * match.pd (negate_expr_p): Likewise.
12953 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12954 fold_build2, not fold_binary.
12956 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12958 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12959 Convert memory address to Pmode.
12960 (aarch64_print_operand): Assert MEM operands are always Pmode.
12962 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12965 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12966 Remove redundant if.
12967 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12968 * config/arm/aarch-common-protos.h
12969 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12970 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12972 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12975 * multiple_target.c (create_dispatcher_calls): Set the default
12976 clone to be static, not public.
12978 2017-06-28 Richard Biener <rguenther@suse.de>
12980 * tree-vect-loop.c (vectorizable_reduction): Move special
12981 cond reduction IV var creation ...
12982 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12983 parameter. Use STMT_VINFO_VECTYPE.
12984 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12987 2017-06-28 Martin Liska <mliska@suse.cz>
12990 * ipa-visibility.c (non_local_p): Handle visibility.
12992 2017-06-28 Martin Liska <mliska@suse.cz>
12995 * common.opt: Add IntegerRange to various options.
12996 * opt-functions.awk (integer_range_info): New function.
12997 * optc-gen.awk: Add integer_range_info to cl_options struct.
12998 * opts-common.c (decode_cmdline_option): Handle
12999 CL_ERR_INT_RANGE_ARG.
13000 (cmdline_handle_error): Likewise.
13001 * opts.c (print_filtered_help): Show valid interval in
13002 when --help is provided.
13003 * opts.h (struct cl_option): Add range_min and range_max fields.
13004 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13006 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13008 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13009 (x * C EQ/NE y * C): New transformation.
13011 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13013 * genmultilib (combination_space): Accept '+' in option names.
13015 2017-06-28 Martin Liska <mliska@suse.cz>
13018 * asan.c (instrument_derefs): Bail out inner references
13019 that are hard register variables.
13021 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13024 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13025 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13027 2017-06-28 Richard Biener <rguenther@suse.de>
13029 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13030 (vect_get_slp_defs): Adjust.
13031 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13033 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13035 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13036 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13037 (vectorizable_reduction): Adjust.
13038 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13040 (vect_get_slp_defs): Likewise.
13041 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13042 (vectorizable_bswap): Adjust.
13043 (vectorizable_call): Likewise.
13044 (vectorizable_conversion): Likewise.
13045 (vectorizable_assignment): Likewise.
13046 (vectorizable_shift): Likewise.
13047 (vectorizable_operation): Likewise.
13048 (vectorizable_store): Likewise.
13049 (vectorizable_condition): Likewise.
13050 (vectorizable_comparison): Likewise.
13052 2017-06-28 Michael Collison <michael.collison@arm.com>
13055 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13057 (arm_gen_movmemqi): Removed unused variable 'i'.
13058 Convert 'for' loop into 'while' loop.
13059 (arm_expand_prologue): Remove last unnecessary set of insn.
13060 (thumb_pop): Remove unused variable 'pushed_words'.
13061 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13063 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13065 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13066 * config/s390/s390-protos.h: Add prototype of
13067 s390_rel_address_ok_p.
13068 * config/s390/s390.c (s390_got_symbol): New function.
13069 (s390_rel_address_ok_p): New function.
13070 (legitimize_pic_address): Use s390_rel_address_ok_p.
13071 (s390_load_got): Use s390_got_symbol.
13072 (s390_option_override): Issue error if
13073 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13074 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13076 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13078 2017-06-27 Andrew Pinski <apinski@cavium.com>
13080 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13081 (X * copysign (1.0, X)): New pattern.
13082 (X * copysign (1.0, -X)): New pattern.
13083 (copysign (-1.0, CST)): New pattern.
13085 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13087 * genmultilib (combination_space): Remove variable.
13088 Validate reuse rules against regular expression for any sequence
13089 of multilib options in any order.
13091 2017-06-27 Michael Collison <michael.collison@arm.com>
13093 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13094 call aarch64_split_simd_combine.
13095 * (aarch64_combine_internal<mode>): Delete pattern.
13096 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13097 Allow register and subreg operands.
13099 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13101 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13102 specific need, just fallback on defaults.
13103 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13105 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13106 Olivier Hainque <hainque@adacore.com>
13108 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13110 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13111 targets. Pick a default if no particular attempt applied.
13112 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13115 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13117 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13118 (x86_64-wrs-vxworks7): Likewise.
13120 2017-06-27 Marek Polacek <polacek@redhat.com>
13123 * ubsan.c (instrument_null): Check get_base_address's result for null.
13125 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13127 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13129 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13131 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13132 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13133 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13134 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13135 New function types.
13136 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13137 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13138 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13139 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13140 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13141 BUILT_IN_FEUPDATEENV): New builtins.
13142 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13143 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13144 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13145 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13147 (builtin_structptr_types): Adjust size.
13148 * tree.c (builtin_structptr_types): Add four entries.
13150 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13151 Olivier Hainque <hainque@adacore.com>
13153 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13154 (TLS_SYM): New local macro, forcing reference to __tls__ on
13155 link command lines for VxWorks 7 RTPs, triggering initialization
13157 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13158 OS features TLS support, true for RTPs on VxWorks 7.
13159 * config/vxworks.c (vxworks_override_options): Setup emutls
13162 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13164 * predict.c (test_prediction_value_range): Use -1U instead of -1
13165 to avoid narrowing conversion warning.
13166 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13167 to avoid narrowing conversion warning.
13168 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13170 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13172 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13174 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13175 64bit configurations.
13176 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13177 (SIZE_TYPE): Likewise.
13178 * config/vxworks.c (vxworks_emutls_var_fields): Use
13179 long_unsigned_type_node instead of unsigned_type_node as the offset
13180 field type, which is "pointer" mode in emutls.c.
13182 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13185 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13187 PR middle-end/81207
13188 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13189 gimple_vuse copying separately from gimple_vdef copying.
13191 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13193 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13196 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13197 Olivier Hainque <hainque@adacore.com>
13199 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13200 for all vxworks7 targets.
13201 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13202 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13203 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13204 variations for VX6/VX7 and 32/64bits later on in ...
13205 (VXWORKS_LIB_SPEC): Leverage new macros.
13206 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13207 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13209 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13211 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13212 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13214 2017-06-26 Carl Love <cel@us.ibm.com>
13216 * config/rs6000/rs6000-c.c: Add support for built-in functions
13217 vector bool char vec_reve (vector bool char);
13218 vector signed char vec_reve (vector signed char);
13219 vector unsigned char vec_reve (vector unsigned char);
13220 vector bool int vec_reve (vector bool int);
13221 vector signed int vec_reve (vector signed int);
13222 vector unsigned int vec_reve (vector unsigned int);
13223 vector bool long long vec_reve (vector bool long long);
13224 vector signed long long vec_reve (vector signed long long);
13225 vector unsigned long long vec_reve (vector unsigned long long);
13226 vector bool short vec_reve (vector bool short);
13227 vector signed short vec_reve (vector signed short);
13228 vector double vec_reve (vector double);
13229 vector float vec_reve (vector float);
13230 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13231 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13232 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13233 (altivec_vreve): New pattern.
13234 * config/rs6000/altivec.h (vec_reve): New define.
13235 * doc/extend.texi (vec_rev): Update the built-in documentation file
13236 for the new built-in functions.
13238 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13240 PR tree-optimization/71815
13241 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13243 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13245 (slsr_process_phi): Likewise.
13246 (replace_uncond_cands_and_profitable_phis): Don't replace a
13247 multiply candidate with a stride of 1 (copy or cast).
13248 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13250 (lowest_cost_path): Likewise.
13251 (total_savings): Likewise.
13253 2017-06-26 Richard Biener <rguenther@suse.de>
13256 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13257 Use def_builtin_pure for all gather builtins.
13259 2017-06-26 Richard Biener <rguenther@suse.de>
13261 PR tree-optimization/81203
13262 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13263 non-dominating BBs.
13265 2017-06-26 Marek Polacek <polacek@redhat.com>
13268 * doc/invoke.texi: Document -Wmultistatement-macros.
13270 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13272 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13273 arm_neon_ok_no_float_abi effective target.
13275 2017-06-26 Richard Biener <rguenther@suse.de>
13277 PR tree-optimization/80928
13278 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13279 (copy_bbs): Set BB_DUPLICATED flag early.
13280 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13282 (execute_on_shrinking_pred): Likewise.
13283 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13284 BB_DUPLICATED blocks.
13285 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13286 iterate over all PHIs considering removal of *gsi.
13288 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13290 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13293 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13295 * config/rs6000/rs6000-string.c: (expand_block_clear,
13296 do_load_for_compare, select_block_compare_mode,
13297 compute_current_alignment, expand_block_compare,
13298 expand_strncmp_align_check, expand_strn_compare,
13299 expand_block_move, rs6000_output_load_multiple)
13300 Move functions related to string/block move/compare
13301 to a separate file.
13302 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13303 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13304 for this function which is now used in two files.
13305 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13306 * config.gcc: Add rs6000-string.o to extra_objs for
13307 targets powerpc*-*-* and rs6000*-*-*.
13309 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13312 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13313 32-bit, since indexed is not valid for DImode.
13314 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13315 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13316 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13317 (define_peephole2 for Altivec d-form store): Likewise.
13320 * multiple_target.c (create_dispatcher_calls): Only create the
13321 dispatcher call if the function is the default clone of a
13322 versioned function.
13324 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13326 PR middle-end/80902
13327 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13328 a call, force the call to not be a tail call.
13330 2017-06-23 Jeff Law <law@redhat.com>
13332 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13335 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13337 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13338 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13339 handling for early expansion of vector shifts (sl,sr,sra,rl).
13340 (builtin_function_type): Add vector shift right instructions
13341 to the unsigned argument list.
13343 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13345 rtl-optimizatoin/79286
13346 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13347 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13348 trap. PIC register plus a const unspec without offset can never trap.
13350 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13352 * tree.h (builtin_structptr_type): New type.
13353 (builtin_structptr_types): Declare new array.
13354 * tree.c (builtin_structptr_types): New array.
13355 (free_lang_data, build_common_tree_nodes): Use it.
13357 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13360 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13363 2017-06-22 Matt Turner <mattst88@gmail.com>
13365 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13366 Lake models to skylake case. Assume skylake for unknown
13367 models with clflushopt.
13369 2017-06-22 Jeff Law <law@redhat.com>
13371 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13372 frame sizes that do not satisfy aarch64_uimm12_shift.
13374 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13376 * profile-count.h (apply_probability,
13377 apply_scale, probability_in): Fix checks for zero.
13379 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13381 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13382 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13384 2016-06-22 Richard Biener <rguenther@suse.de>
13386 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13387 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13388 REDUC_MAX_EXPR support.
13389 (vectorizable_reduction): Likewise.
13390 (vect_create_epilog_for_reduction): Likewise.
13392 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13394 * match.pd (A / (1 << B) -> A >> B): New.
13395 * generic-match-head.c: Include optabs-tree.h.
13396 * gimple-match-head.c: Likewise.
13397 * optabs-tree.h (target_supports_op_p): New.
13398 * optabs-tree.c (target_supports_op_p): New.
13400 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13402 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13403 $gcc_cv_ld --help output.
13404 (gcc_cv_ld_demangle): Likewise.
13405 (gcc_cv_ld_eh_frame_hdr): Likewise.
13406 (gcc_cv_ld_pie): Likewise.
13407 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13408 (gcc_cv_ld_buildid): Likewise.
13409 (gcc_cv_ld_sysroot): Likewise.
13410 (ld_bndplt_support): Likewise.
13411 (ld_pushpopstate_support): Likewise.
13412 * configure: Regenerate.
13413 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13415 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13418 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13419 operands indexes to avoid gap between operands and match_dups.
13421 2017-06-21 Andrew Pinski <apinski@cavium.com>
13423 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13424 Increment Arith_shift and Arith_shift_reg by 1.
13425 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13427 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13428 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13429 (aarch64_strip_extend): Add new argument and test for it.
13430 (aarch64_cheap_mult_shift_p): New function.
13431 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13432 add a cost if it is true.
13433 Update calls to aarch64_strip_extend.
13434 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13436 2017-06-21 Andrew Pinski <apinski@cavium.com>
13438 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13440 (thunderxt88): Likewise.
13441 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13442 (thunderx_prefetch_tune): New variable.
13443 (thunderx2t99_prefetch_tune): Update for the correct values.
13444 (thunderxt88_tunings): New variable.
13445 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13446 generic_prefetch_tune.
13447 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13449 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13451 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13452 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13453 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13454 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13455 (aarch64_atomic_cas<mode>, GPI): Likewise.
13457 2017-06-21 Martin Liska <mliska@suse.cz>
13459 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13460 statements on cold and hot labels.
13461 * predict.c (tree_estimate_probability_bb): Remove the
13462 prediction from this place.
13464 2017-06-21 Martin Liska <mliska@suse.cz>
13466 PR tree-optimization/79489
13467 * gimplify.c (maybe_add_early_return_predict_stmt): New
13469 (gimplify_return_expr): Call the function.
13470 * predict.c (tree_estimate_probability_bb): Remove handling
13472 * predict.def: Update comment about early return predictor.
13473 * gimple-predict.h (is_gimple_predict): New function.
13474 * predict.def: Change default value of early return to 66.
13475 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13477 * passes.def: Put pass_strip_predict_hints to the beginning of
13480 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13482 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13483 FUNCTION_DECL declarations.
13484 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13486 (dwaf2out_decl): Likewise.
13487 * godump.c (go_early_global_decl): Skip call to the real debug hook
13488 for FUNCTION_DECL declarations.
13489 * passes.c (rest_of_decl_compilation): Skip call to the
13490 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13491 -fdump-go-spec is passed.
13493 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13495 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13496 Reorder for compactness.
13497 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13498 (def_builtin_pure, def_builtin_pure2): New functions.
13499 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13501 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13503 * match.pd (nop_convert): New predicate.
13504 ((A +- CST1) +- CST2): Allow some NOP conversions.
13506 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13509 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13510 with ctors/dtors if GOVD_SHARED is set.
13512 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13514 * config/aarch64/aarch64.md (movti_aarch64):
13515 Emit mov rather than orr.
13516 (movtf_aarch64): Likewise.
13517 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13518 Emit mov rather than orr.
13520 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13522 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13523 Swap alternatives, make integer dup more expensive.
13525 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13527 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13528 Return true for non-tls symbols.
13530 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13532 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13533 (cortex-a75): Likewise.
13534 (cortex-a75.cortex-a55): Likewise.
13535 * config/aarch64/aarch64-tune.md: Regenerate.
13536 * doc/invoke.texi (-mtune): Document new values for -mtune.
13538 2017-06-21 Tom de Vries <tom@codesourcery.com>
13540 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13541 stack_size feature.
13542 (Effective-Target Keywords, Other attributes): Suggest using
13543 dg-add-options stack_size feature to get stack limit in stack_size
13544 effective target documentation.
13546 2017-06-21 Julian Brown <julian@codesourcery.com>
13547 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13549 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13550 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13551 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13553 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13554 attribute type list for neon_multiply.
13555 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13556 attribute type list for neon_multiply.
13557 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13558 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13559 attribute type list for neon_multiply.
13560 * config/arm/types.md (crypto_pmull): Add.
13561 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13562 attribute type list.
13564 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13566 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13569 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13571 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13572 to make sure not to dereference a NULL cost_classes_ptr pointer.
13574 2017-06-20 Carl Love <cel@us.ibm.com>
13576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13577 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13578 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13580 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13581 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13582 VMULOSW): New enum "unspec" values.
13583 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13584 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13585 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13586 altivec_vmulosw): New patterns.
13587 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13588 VMULOSW): Add definitions.
13590 2017-06-20 Julia Koval <julia.koval@intel.com>
13592 * config/i386/i386.c: Fix rounding expand for new pattern.
13593 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13595 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13597 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13598 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13600 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13602 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13605 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13607 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13608 architecture, then by implementer ID.
13609 * config/aarch64/aarch64-tune.md: Regenerate.
13611 2017-06-20 Richard Biener <rguenther@suse.de>
13613 PR middle-end/81097
13614 * fold-const.c (split_tree): Fold to type before negating.
13616 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13618 * diagnostic-show-locus.c
13619 (selftest::test_fixit_deletion_affecting_newline): New function.
13620 (selftest::diagnostic_show_locus_c_tests): Call it.
13622 2017-06-20 Andreas Schwab <schwab@suse.de>
13625 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13628 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13630 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13631 __ARM_FEATURE_COPROC according to support.
13633 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13635 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13636 Rewritten to avoid overflow for > 32-bit pointers.
13639 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13640 by removing enum keyword.
13641 (ubsan_type_descriptor): Likewise. Formatting fix.
13644 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13645 splitter): Require TARGET_SSE2 in the condition.
13647 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13650 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13651 for doing vector set of SFmode on ISA 3.0.
13652 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13653 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13655 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13656 SFmode value into a V4SF variable that was extracted from another
13657 V4SF variable without converting the element to double precision
13658 and back to single precision vector format.
13659 (vsx_insert_extract_v4sf_p9_2): Likewise.
13661 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13663 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13664 in UWHI to avoid undefined overflow.
13667 * ubsan.h (enum ubsan_encode_value_phase): New.
13668 (ubsan_encode_value): Change second argument to
13669 enum ubsan_encode_value_phase with default value of
13670 UBSAN_ENCODE_VALUE_GENERIC.
13671 * ubsan.c (ubsan_encode_value): Change second argument to
13672 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13673 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13674 create_tmp_var_raw instead of create_tmp_var and use a
13676 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13677 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13678 ubsan_encode_value callers.
13681 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13682 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13683 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13685 2017-06-19 Richard Biener <rguenther@suse.de>
13687 PR middle-end/81118
13688 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13689 estimates if we changed anything.
13691 2017-06-19 Richard Biener <rguenther@suse.de>
13693 PR tree-optimization/80887
13694 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13695 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13696 simplified lookups, then reset mprts_hook.
13697 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13699 (try_to_simplify): Likewise.
13701 2017-06-19 Martin Liska <mliska@suse.cz>
13704 * gimplify.c (gimplify_switch_expr):
13705 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13707 2017-06-19 Martin Liska <mliska@suse.cz>
13709 * doc/install.texi: Document that PGO runs in 4 stages.
13711 2017-06-19 Martin Liska <mliska@suse.cz>
13714 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13715 to dispatcher function name.
13716 * multiple_target.c (replace_function_decl): New function.
13717 (create_dispatcher_calls): Redirect both edges and references.
13719 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13721 * profile-count.c (profile_count::dump): Dump quality.
13722 (profile_count::differs_from_p): Update for unsigned val.
13723 * profile-count.h (profile_count_quality): New enum.
13724 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13726 2017-06-19 Richard Biener <rguenther@suse.de>
13728 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13729 struct function as arg.
13730 (estimate_numbers_of_iterations): Export overload with loop arg.
13731 (free_numbers_of_iterations_estimates_loop): Use an overload of
13732 free_numbers_of_iterations_estimates instead.
13733 * tree-cfg.c (remove_bb): Adjust.
13734 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13735 * tree-parloops.c (gen_parallel_loop): Likewise.
13736 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13738 (tree_unroll_loops_completely): Likewise.
13739 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13740 Use an overload instead and export.
13741 (estimated_loop_iterations): Adjust.
13742 (max_loop_iterations): Likewise.
13743 (likely_max_loop_iterations): Likewise.
13744 (estimate_numbers_of_iterations): Take struct function as arg
13746 (loop_exits_before_overflow): Adjust.
13747 (free_numbers_of_iterations_estimates_loop): Use an overload.
13748 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13749 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13751 2017-06-19 Richard Biener <rguenther@suse.de>
13754 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13755 RANGE_EXPR conservatively.
13757 2017-06-16 Carl Love <cel@us.ibm.com>
13759 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13760 definitions for vec_float, vec_float2, vec_floato,
13761 vec_floate built-ins.
13762 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13763 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13765 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13766 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13767 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13768 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13769 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13770 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13771 vec_floato): Add builtin defines.
13772 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13773 Update the built-in documentation file for the new built-in
13776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13778 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13779 (mthumb): Mark as the negative of -marm.
13781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13783 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13785 (ARM Options, -mtune): Document that this accepts the same
13786 extension options as -mcpu.
13787 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13791 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13792 permitted extensions.
13794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13796 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
13797 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
13798 (armv8-m.main): Add option +nodsp.
13799 * config/arm/arm-cpu-cdata.h: Regenerated.
13801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13803 * config/arm/t-fuchsia: New file.
13804 * config.gcc (arm*-*-fuchsia*): Use it.
13806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13808 * config/arm/t-symbian: Rewrite for new option infrastructure.
13810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13812 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
13813 (MULTILIB_REQUIRED): Likewise.
13815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13817 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13818 (MULTILIB_RESUE): Likewise.
13819 (MULTILIB_MATCHES): Likewise.
13820 (MULTLIB_REQUIRED): Likewise.
13822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13824 * config/arm/t-rtems: Rewrite for new option framework.
13826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13828 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13829 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13830 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13831 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13832 * config/arm/t-multilib: ... here.
13833 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13834 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13835 armv7-a and armv8*-a when A-profile libraries have not been built.
13836 * config/arm/t-rmprofile: Rewrite.
13838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13840 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13841 with a backslash. Remove the backslash after substituting unescaped
13843 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13847 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13848 the driver to pass through the default setting of -mfloat-abi.
13849 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13851 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13852 (all_feat_combs): New rule.
13853 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13855 * config/arm/t-aprofile: Rewrite.
13857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13859 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13860 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13861 fpu is not specified by the user/command-line.
13862 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13863 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13864 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13865 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13866 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13867 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13871 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13872 * config/arm/t-arm-elf: Rewritten.
13874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13876 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13877 have some floating-point instructions.
13878 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13879 (TARGET_MAYBE_HARD_FLOAT): New macro.
13880 * config/arm/arm-builtins.c (arm_init_builtins): Use
13881 TARGET_MAYBE_HARD_FLOAT.
13882 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13886 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13887 (configargs.h): Include it.
13888 (arm_print_hint_for_fpu_option): New function.
13889 (arm_parse_fpu_option): New function.
13890 (candidate_extension): New class.
13891 (arm_canon_for_multilib): New function.
13892 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13893 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13894 (ARCH_CANONICAL_SPECS): New macro.
13895 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13899 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13900 are set after handling multilib fragments. Set target_cpu_default2
13903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13905 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13907 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13908 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13910 (check_arch): Likewise.
13911 * config/arm/arm.c (arm_configure_build_target): Handle
13912 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13913 options in the default.
13915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13917 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13918 when an option is an alias of another.
13919 * config/arm/parsecpu.awk (optalias): New parser token.
13920 (gen_comm_data): Mark non-alias options as such. Emit entries
13921 for extension aliases.
13922 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13923 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13924 (armv6kz, armv6zk, armv6t2): Likewise.
13925 (armv7): Make vfpv3-d16 an alias.
13926 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13928 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13929 Sort in canonical order.
13930 (armv8-a): Sort in canonical order.
13931 (armv8.1-a, armv8.2-a): Likewise.
13932 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13934 (cortex-a9): Sort in canonical order.
13935 * config/arm/arm.c (selftests.h): Include it.
13936 (arm_test_cpu_arch_data): New function.
13937 (arm_run_self_tests): New function.
13938 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13939 (targetm): Move declaration to the end of the file.
13940 * arm-cpu-cdata.h: Regenerated.
13942 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13944 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13945 call to target_mode_check describing the type of option passed.
13946 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13947 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13948 arm_parse_cpu_option_name to match parameters against list of
13950 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13951 arm_arch_core_flags data structure.
13952 * config/arm/arm-cpu_cdata.h: Regenerated.
13954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13956 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13958 (arm_print_hint_for_cpu_option): Likewise.
13959 (arm_print_hint_for_arch_option): Likewise.
13960 (arm_parse_cpu_option_name): Likewise.
13961 (arm_parse_arch_option_name): Likewise.
13962 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13963 of entries in the all_fpus list.
13964 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13965 (arm_parse_cpu_option_name): Declare.
13966 (arm_parse_arch_option_name): Declare.
13967 (arm_parse_option_features): Declare.
13968 (arm_intialize_isa): Declare.
13969 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13971 (gen_comm_data): ... here. Make definitions non-static.
13972 * config/arm/arm-cpu-data.h: Regenerated.
13973 * config/arm/arm-cpu-cdata.h: Regenerated.
13975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13977 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13978 (cpu_arch_extension): New structure.
13979 (cpu_arch_option, arch_option, cpu_option): New structures.
13980 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13981 architecture types.
13982 (gen_data): Generate new format data tables.
13983 * config/arm/arm.c (cpu_tune): New structure.
13984 (cpu_option, processors): Delete.
13985 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13986 (arm_print_hint_for_cpu_option): ... this and ...
13987 (arm_print_hint_for_arch_option): ... this.
13988 (arm_parse_arch_cpu_name): Delete. Replace with ...
13989 (arm_parse_cpu_option_name): ... this and ...
13990 (arm_parse_arch_option_name): ... this.
13991 (arm_unrecognized_feature): Change type of target parameter to
13993 (arm_parse_arch_cpu_features): Delete. Replace with ...
13994 (arm_parse_option_features): ... this.
13995 (arm_configure_build_target): Rework to use new configuration data
13997 (arm_print_tune_info): Rework for new configuration data tables.
13998 * config/arm/arm-cpu-data.h: Regenerated.
13999 * config/arm/arm-cpu.h: Regenerated.
14001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14003 * Makefile.in (OBJS): Move sbitmap.o from here ...
14004 (OBJS-libcommon): ... to here.
14006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14008 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14009 (ISA_ALL_CRYPTO): New macro.
14010 (ISA_ALL_SIMD): New macro
14011 (ISA_ALL_FP): New macro.
14012 * config/arm/arm.c (fpu_bitlist): Update initializer.
14013 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14015 (arm9e): Add fpu. Add option for nofp
14016 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14017 (arm926ej-s, arm1026ej-s): Likewise.
14018 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14019 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14020 neon-fp16, neon-vfpv4, nofp and nosimd.
14021 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14022 (cortex-a8): Add fpu. Add option for nofp.
14023 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14024 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14025 (cortex-r4f): Add fpu.
14026 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14027 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14029 (cortex-r8): Likewise.
14030 (cortex-m4): Add fpu. Add option for nofp.
14031 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14032 (cortex-a17.cortex-a7): Likewise.
14033 (cortex-a32): Add fpu. Add options for crypto and nofp.
14034 (cortex-a35, cortex-a53): Likewise.
14035 (cortex-a57): Add fpu. Add option for crypto.
14036 (cortex-a72, cortex-a73): Likewise.
14037 (exynos-m1): Likewise.
14038 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14039 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14040 (cortex-m33): Add fpu. Add option for nofp.
14041 * config/arm/arm-cpu-cdata.h: Regenerated
14042 * config/arm/arm-cpu-data.h: Regenerated.
14044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14046 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14047 (armv5te, armv5tej): Likewise.
14048 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14049 (armv7): Add options fp and vfpv3-d16.
14050 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14051 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14053 (armv7ve): Likewise.
14054 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14055 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14056 (armv8-a): Add nocrypto option.
14057 (armv8.1-a, armv8.2-a): Likewise.
14058 (armv8-m.main): add options fp, fp.dp and nofp.
14060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14062 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14064 (armv8-a+crc): Delete.
14065 (armv8.1-a): Add options simd, crypto and nofp.
14066 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14067 (armv8.2-a+fp16): Delete.
14068 (armv8-m.main): Add option dsp.
14069 (armv8-m.main+dsp): Delete.
14070 (cortex-a8): Add fpu. Add nofp option.
14071 (cortex-a9): Add fpu. Add nofp and nosimd options.
14072 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14073 link to main cpu and architecture data structures.
14074 (gen_comm_data): Only put isa attributes from the main architecture
14076 (option): New statement for architecture and CPU entries.
14077 * arm.c (struct cpu_option): New structure.
14078 (struct processors): Add entry for options.
14079 (arm_unrecognized_feature): New function.
14080 (arm_parse_arch_cpu_name): Ignore any characters after the first
14082 (arm_parse_arch_cpu_feature): New function.
14083 (arm_configure_build_target): Separate out any CPU and architecture
14084 features and parse separately. Don't error out if -mfpu=auto is
14085 used with only an architecture string.
14086 (arm_print_asm_arch_directives): New function.
14087 (arm_file_start): Call it.
14088 * config/arm/arm-cpu-cdata.h: Regenerated.
14089 * config/arm/arm-cpu-data.h: Likewise.
14090 * config/arm/arm-tables.opt: Likewise.
14092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14094 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14095 assembler when it is not -mfpu=auto.
14097 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14099 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14100 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14101 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14102 (ASM_CPU_SPEC): Rewrite.
14103 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14104 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14105 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14106 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14107 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14108 copied string is NUL-terminated. Also strip any characters prefixed
14110 (arm_rewrite_selected_arch): New function.
14111 (arm_rewrite_march): New function.
14113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14115 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14116 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14117 (march, mcpu, mtune): Convert to string-based options.
14118 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14119 (arm_parse_arch_cpu_name): New function.
14120 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14121 identify selected architecture or CPU.
14122 (arm_option_save): New function.
14123 (TARGET_OPTION_SAVE): Redefine.
14124 (arm_option_restore): Restore string options.
14125 (arm_option_print): Print string options.
14127 2017-06-16 Martin Sebor <msebor@redhat.com>
14129 PR tree-optimization/80933
14130 PR tree-optimization/80934
14131 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14132 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14133 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14134 (gimple_fold_builtin): Call them.
14136 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14138 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14139 as unlikely; update profile.
14141 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14143 * predict.c (force_edge_cold): Handle declaring edges impossible
14146 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14148 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14150 (try_unroll_loop_completely): Fix reporting.
14152 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14154 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14156 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14159 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14160 if given a non-constant argument for an intrinsic which requires a
14163 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14165 * profile.c (compare_freqs): New function.
14166 (branch_prob): Sort edge list.
14167 (find_spanning_tree): Assume that the list is priority sorted.
14169 2017-06-16 Richard Biener <rguenther@suse.de>
14171 PR tree-optimization/81090
14172 * passes.def (pass_record_bounds): Remove.
14173 * tree-pass.h (make_pass_record_bounds): Likewise.
14174 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14175 make_pass_record_bounds): Likewise.
14176 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14177 not free niter estimates at the beginning but at the end.
14178 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14180 2017-06-16 Richard Biener <rguenther@suse.de>
14182 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14183 initializer to workaround ICE in host GCC 4.8.
14185 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14187 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14189 (clone_inlined_nodes): Update.
14191 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14193 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14194 prefetch settings, and enable prefetching by default at -O3.
14196 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14198 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14199 Set flag_prefetch_loop_arrays according to tuning data.
14201 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14203 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14204 New tune structure.
14205 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14206 [Unrelated to main purpose of the patch] Place the pointer field last
14207 to enable type checking errors when tune structure are wrongly merged.
14208 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14209 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14210 (thunderx2t99_prefetch_tune): New tune constants.
14211 (tune_params *_tunings): Update all tunings (no functional change).
14212 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14213 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14214 from tunings structures.
14216 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14219 * ubsan.c (instrument_null): Add T argument, use it instead
14220 of computing it based on IS_LHS.
14221 (instrument_object_size): Likewise.
14222 (pass_ubsan::execute): Adjust instrument_null and
14223 instrument_object_size callers to pass gimple_get_lhs or
14224 gimple_assign_rhs1 result to it. Use instrument_null instead of
14225 calling get_base_address and instrument_mem_ref. Handle
14226 aggregate call arguments for object-size sanitization.
14228 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14230 PR tree-optimization/81089
14231 * tree-vrp.c (is_masked_range_test): Validate operands of
14234 2017-06-15 Martin Sebor <msebor@redhat.com>
14237 * dumpfile.c (dump_register): Avoid calling memset to initialize
14238 a class with a default ctor.
14239 * gcc.c (struct compiler): Remove const qualification.
14240 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14241 * hash-table.h: Ditto.
14242 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14244 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14245 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14247 * params.h (struct param_info): Make struct members non-const.
14248 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14249 with default initialization.
14250 * vec.h (vec_copy_construct, vec_default_construct): New helper
14252 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14253 with vec_copy_construct.
14254 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14255 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14256 * doc/invoke.texi (-Wclass-memaccess): Document.
14258 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14260 * emit-rtl.h (is_leaf): Update comment about local
14261 register allocator.
14263 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14266 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14267 for a variable to have a section before checking if the section has a
14269 Set section to.persistent if persistent attribute is set.
14270 Warn if .persistent attribute is used on an automatic variable.
14272 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14274 PR rtl-optimization/80474
14275 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14277 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14279 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14282 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14284 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14285 (casesi): Emit bounds checking as RTL.
14286 (casesi_internal_mips16_<mode>): Remove bounds checking.
14288 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14290 * config/xtensa/xtensa.c (xtensa_option_override): Append
14291 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14292 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14293 xtensa_doloop_hooks): Define unconditionally.
14294 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14296 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14297 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14298 for it in xtensa_option_override.
14299 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14300 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14302 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14304 * doc/cppopts.texi: Document '-' special value to -MF.
14306 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14308 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14309 (cortex_a53_fconst): Likewise.
14310 (cortex_a53_fpmul): Likewise.
14311 (cortex_a53_f_load_64): Likewise.
14312 (cortex_a53_f_load_many): Likewise.
14313 (cortex_a53_advsimd_alu): Likewise.
14314 (cortex_a53_advsimd_alu_q): Likewise.
14315 (cortex_a53_advsimd_mul): Likewise.
14316 (cortex_a53_advsimd_mul_q): Likewise.
14317 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14318 Add missing fmul, r2f_cvt and fconst cases.
14320 2017-06-14 Richard Biener <rguenther@suse.de>
14322 PR middle-end/81088
14323 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14325 (fold_binary_loc): When associating do not treat pre-existing
14326 TREE_OVERFLOW on literal constants as a reason to allow
14327 TREE_OVERFLOW on associated literal constants.
14329 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14331 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14332 (MASK_FEATURES): New macro.
14333 * config/sparc/sparc.c (sparc_option_override): Remove the special
14334 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14336 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14338 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14339 a division of 0 if non-call exceptions are enabled.
14341 2017-06-14 Andrew Pinski <apinski@cavium.com>
14342 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14345 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14346 Improve vector initialization code gen for only variable case.
14348 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14350 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14352 2017-06-14 Richard Biener <rguenther@suse.de>
14354 PR tree-optimization/81083
14355 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14358 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14360 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14361 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14362 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14363 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14364 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14365 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14367 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14369 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14370 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14372 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14374 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14376 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14378 * config/rs6000/t-rtems: Don't handle SPE.
14380 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14382 * config/rs6000/t-linux: Don't handle SPE.
14384 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14386 * config/rs6000/eabispe.h: Delete file.
14388 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14390 * config/rs6000/t-spe: Delete file.
14392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14394 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14395 (rs6000_legitimate_offset_address_p): Return false for anything in
14396 V2SImode or V2SFmode.
14398 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14400 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14401 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14402 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14404 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14405 (rs6000_legitimate_offset_address_p): Ditto.
14406 (rs6000_emit_move): Ditto.
14407 (rs6000_init_builtins): Remove V4HI_type_node.
14409 2017-06-13 Martin Liska <mliska@suse.cz>
14412 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14413 (gate_asan): Likewise.
14414 * asan.h (asan_no_sanitize_address_p): Remove the function.
14415 (sanitize_flags_p): New function.
14416 * builtins.def: Fix coding style.
14417 * common.opt: Use renamed enum value.
14418 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14419 * doc/extend.texi: Document no_sanitize attribute.
14420 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14421 to SANITIZE_UNDEFINED_NONDEFAULT.
14422 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14423 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14424 Use sanitize_flags_p.
14425 * gimplify.c (gimplify_function_tree): Likewise.
14426 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14427 * opts.c (parse_no_sanitize_attribute): New function.
14428 (common_handle_option): Use renamed enum value.
14429 * opts.h (parse_no_sanitize_attribute): Declare.
14430 * tree.c (sanitize_flags_p): New function.
14431 * tree.h: Declared here.
14432 * tsan.c: Use sanitize_flags_p.
14433 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14434 (instrument_mem_ref): Likewise.
14435 (instrument_bool_enum_load): Likewise.
14436 (do_ubsan_in_current_function): Remove the function.
14437 (pass_ubsan::execute): Use sanitize_flags_p.
14438 * ubsan.h: Remove do_ubsan_in_current_function
14439 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14440 (dump_function_to_file): Use it here.
14442 2017-06-13 Martin Jambor <mjambor@suse.cz>
14444 PR tree-optimization/80803
14445 PR tree-optimization/81063
14446 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14447 (propagate_subaccesses_across_link): Enqueue subtree whenever
14448 necessary instead of relying on the caller.
14450 2017-06-13 Martin Jambor <mjambor@suse.cz>
14452 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14453 that have a first_link.
14454 (sort_and_splice_var_accesses): Do not check first_link before
14456 (subtree_mark_written_and_enqueue): Likewise.
14457 (propagate_all_subaccesses): Likewise and do not stop at first
14458 parent with a first_link.
14460 2017-06-13 Martin Jambor <mjambor@suse.cz>
14462 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14465 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14467 * match.pd: New pattern.
14469 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14471 * tree-vrp.c (is_masked_range_test): New function.
14472 (register_edge_assert_for): Determine ranges for
14475 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14477 PR tree-optimization/67328
14478 * fold-const.c (maskable_range_p): New function.
14479 (build_range_check): Generate bittests if possible.
14481 2017-06-13 Martin Liska <mliska@suse.cz>
14483 * gimple-pretty-print.c (dump_probability): Add new argument.
14484 (dump_edge_probability): Dump both probability and count.
14485 (dump_gimple_label): Likewise.
14486 (dump_gimple_bb_header): Likewise.
14488 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14491 * config/avr/avr-devices.c: Fix indentation.
14492 * config/avr/gen-avr-mmcu-specs.c: Dito.
14494 2017-06-13 Richard Biener <rguenther@suse.de>
14496 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14497 instead get vector type from stmt_info.
14498 (vectorizable_reduction): Adjust. Remove dead code.
14500 2017-06-13 Richard Biener <rguenther@suse.de>
14502 PR middle-end/81065
14503 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14504 case of C * (x * C2 + C3).
14505 (fold_addr_of_array_ref_difference): Properly fold index difference.
14507 2017-06-12 David S. Miller <davem@davemloft.net>
14510 * config/sparc/sparc.md (return expander): Emit frame blockage if
14511 function uses alloca.
14513 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14515 * combine.c (make_field_assignment): Check len rather than the mode
14516 precision when calling force_to_mode.
14518 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14520 Support multilibs and devices that see flash in RAM address range.
14523 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14524 (avr_mcu_t) <flash_pm_offset>: New field.
14525 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14526 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14527 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14528 (AVR_TINY_PM_OFFSET): Remove macro.
14529 * config/avr/avr.opt (-mshort-calls): New option.
14530 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14531 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14532 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14533 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14534 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14535 instead of avr_arch->have_jmp_call.
14536 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14537 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14538 avr_arch->flash_pm_offset to define.
14539 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14540 new field flash_pm_offset. Add entry for avrxmega3.
14541 (avr_texinfo): Add entry for avrxmega3.
14542 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14543 attiny212, attiny214,
14544 attiny412, attiny414, attiny416, attiny417,
14545 attiny814, attiny816, attiny817,
14546 attiny1614, attiny1616, attiny1617,
14547 attiny3214, attiny3216, attiny3217.
14548 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14549 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14550 (avr_print_operand_address) [AVR_TINY]: Same.
14551 (avr_asm_init_sections) <readonly_data_section>: Only patch
14552 callback if avr_arch->flash_pm_offset = 0.
14553 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14554 for rodata if avr_arch->flash_pm_offset != 0.
14555 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14556 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14557 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14558 (m_options): Append opt_rcall.
14559 (m_dirnames): Append dir_rcall.
14560 * config/avr/t-multilib: Regenerate.
14562 * configure.ac [target=avr]: Check whether avrxmega3 default
14563 linker description file works as needed.
14564 * configure: Regenerate.
14565 * doc/avr-mmcu.texi: Regenerate.
14566 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14567 <__AVR_ARCH__>: Document avrxmega3 and 103.
14568 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14569 <__AVR_SHORT_CALLS__>: Document it.
14570 <__AVR_PM_BASE_ADDRESS__>: Document it.
14571 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14572 (AVR Variable Attributes) <progmem>: Document this is
14573 not needed for avrxmega3.
14574 (AVR Named Address Spaces) <__flash>: Dito.
14576 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14578 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14580 2017-06-12 Doug Rupp <rupp@adacore.com>
14582 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14583 Append vxworks-stdint.h to the tm_file list.
14584 * config/vxworks-stdint.h: New file.
14586 2017-06-12 Martin Liska <mliska@suse.cz>
14588 PR tree-optimization/81041
14589 * tree-profile.c (gimple_gen_ic_func_profiler):
14590 Create an extra BB in profile-generate
14591 (gimple_gen_time_profiler): Likewise.
14593 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14595 PR tree-optimization/81003
14596 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14597 (update_range_test): Use it instead of force_gimple_operand_gsi.
14599 2017-06-12 Richard Biener <rguenther@suse.de>
14601 PR tree-optimization/81053
14602 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14603 with backedge value not defined in loop. Simplify def stmt
14606 2017-06-11 Tom de Vries <tom@codesourcery.com>
14609 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14611 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14612 nvptx_cannot_force_const_mem.
14614 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14616 * opts.c (finish_options): Move test for flag_split_stack after
14617 it has been initialized.
14619 2017-06-11 Jason Merrill <jason@redhat.com>
14621 * tree.h (id_equal): New.
14622 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14623 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14624 instead of strcmp of IDENTIFIER_POINTER.
14626 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14628 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14629 (mark_all_inlined_calls_cdtor): Fix formating.
14630 (inline_transform): Rescale profile before inlining.
14632 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14634 * cgraph.h (cgraph_edge::clone): Update prototype.
14635 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14636 (cgraph_node::create_clone): Update.
14637 (cgraph_node::create_version_clone): Update.
14638 * tree-inline.c (copy_bb): Update.
14639 (expand_call_inline): Update.
14641 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14643 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14644 factored out from ...
14645 (rs6000_emit_prologue): ... here.
14647 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14649 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14650 factored out from ...
14651 (rs6000_emit_prologue): ... here.
14653 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14655 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14657 (handle_missing_profiles): Fix computation of tp_first_run.
14658 (counts_to_freqs): Do not touch freqs when count is 0.
14660 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14662 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14665 2017-06-10 Tom de Vries <tom@codesourcery.com>
14667 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14668 attributes): Document signal effective target.
14670 2017-06-10 Tom de Vries <tom@codesourcery.com>
14672 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14673 Document effective target stack_size.
14675 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14677 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14678 to the edit_context if they can be auto-applied.
14680 2017-06-9 Ian Lance Taylor <iant@golang.org>
14682 * opts.c (finish_options): If -fsplit-stack, disable implicit
14683 -forder-blocks-and-partition.
14684 * doc/invoke.texi (Optimize Options): Document that when using
14685 -fsplit-stack -forder-blocks-and-partition is not implicitly
14688 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14690 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14691 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14692 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14693 * builtins.def (abort, trap, unreachable): Declare cold.
14694 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14695 * tree-core.h (ECF_COLD): New.
14696 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14697 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14699 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14701 * predict.c (unlikely_executed_stmt_p): Cleanup.
14703 2017-06-09 Richard Biener <rguenther@suse.de>
14705 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14706 model if the ref is always written to.
14708 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14710 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14712 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14714 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14717 2017-06-09 Tom de Vries <tom@codesourcery.com>
14720 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14721 "target cannot support label values" when encountering LABEL_REF.
14723 2017-06-09 Martin Liska <mliska@suse.cz>
14725 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14726 (gimple_gen_ic_func_profiler): Emit direct comparison
14727 of __gcov_indirect_call_callee with NULL.
14728 (gimple_gen_time_profiler): Change probability from
14729 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14731 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14733 * profile.c (edge_gcov_counts): Turn to pointer.
14734 (compute_branch_probabilities, compute_branch_probabilities): Update.
14735 (branch_prob): Do not clear edge_gcov_count.
14736 * profile.h (edge_gcov_counts): Turn to pointer.
14737 (edge_gcov_count): Update.
14739 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14741 * gimple.h (gimple_check_failed): Mark cold.
14743 2017-06-09 Richard Biener <rguenther@suse.de>
14745 PR tree-optimization/66623
14746 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14747 refactor check_reduction into two parts, properly computing
14748 whether we have to check reduction validity for outer loop
14751 2017-06-09 Richard Biener <rguenther@suse.de>
14753 PR tree-optimization/79483
14754 * graphite-scop-detection.c (order): New global.
14755 (get_order): Compute bb to order mapping that satisfies code
14756 generation constraints.
14757 (cmp_pbbs): New helper.
14758 (build_scops): Start domwalk at entry block, sort generated
14761 2017-06-09 Richard Biener <rguenther@suse.de>
14763 PR middle-end/81007
14764 * ipa-polymorphic-call.c
14765 (ipa_polymorphic_call_context::restrict_to_inner_class):
14766 Skip FIELD_DECLs with error_mark_node type.
14767 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14770 2017-06-09 Martin Liska <mliska@suse.cz>
14772 * predict.c (struct branch_predictor): New struct.
14773 (test_prediction_value_range): New test.
14774 (predict_c_tests): New function.
14775 * selftest-run-tests.c (selftest::run_tests): Run the function.
14776 * selftest.h: Declare new tests.
14778 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14781 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14782 gen_add3_insn did not fail.
14783 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14784 r0, construct that number in a temporary reg and add that reg to r0.
14785 If asked to put the result in r0 as well, fail.
14787 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14790 for early expansion of vec_eqv.
14792 2017-06-08 Jakub Jelinek <jakub@redhat.com>
14794 PR middle-end/81005
14795 * ubsan.c (instrument_null): Avoid pointless code temporary.
14796 (pass_ubsan::execute): Instrument aggregate arguments of calls.
14798 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
14802 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
14805 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
14808 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14810 * predict.c (unlikely_executed_edge_p): Move ahead.
14811 (probably_never_executed_edge_p): Use it.
14813 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14815 PR middle-end/79988
14816 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14817 gimple_call_builtin_p call.
14819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14821 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14822 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14823 rtl_check_failed_type2, rtl_check_failed_code1,
14824 rtl_check_failed_code2, rtl_check_failed_code_mode,
14825 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14826 rtvec_check_failed_bounds, rtl_check_failed_flag,
14827 _fatal_insn_not_found, _fatal_insn): Likewise.
14828 * tree.h (tree_contains_struct_check_failed,
14829 tree_check_failed, tree_not_check_failed,
14830 tree_class_check_failed, tree_range_check_failed,
14831 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14832 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14833 tree_operand_check_failed, omp_clause_check_failed,
14834 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14837 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14839 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14840 flag_branch_probabilities.
14841 * ipa-inline.c (edge_badness): Likewise.
14842 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14843 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14844 * predict.c (maybe_hot_frequency_p): Likewise.
14845 (probably_never_executed): Likewise.
14846 * sched-ebb.c (schedule_ebbs): Likewise.
14847 * sched-rgn.c (find_single_block_region): Likewise.
14848 * tracer.c (tail_duplicate): Likewise.
14850 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14852 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14853 longer requires x_flag_profile_use.
14855 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14857 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14858 instead of flag_reorder_blocks_and_partition.
14859 * dbxout.c (dbxout_function_end): Likewise.
14860 * dwarf2out.c (gen_subprogram_die): Likewise.
14861 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14862 * hw-doloop.c (reorg_loops): Likewise.
14863 * varasm.c (assemble_start_function,
14864 assemble_end_function): Likewise.
14865 (decide_function_section): Do not check for
14866 flag_reorder_blocks_and_partition.
14868 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14870 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14872 (chkp_get_hard_register_fake_addr_expr): Ditto.
14873 (chkp_build_addr_expr): Add check for hard reg case.
14874 (chkp_parse_array_and_component_ref): Ditto.
14875 (chkp_find_bounds_1): Ditto.
14876 (chkp_process_stmt): Don't generate bounds store for
14879 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14881 * predict.c (maybe_hot_bb_p): Do not check profile status.
14882 (maybe_hot_edge_p): Likewise.
14883 (probably_never_executed): Check for zero counts even if profile
14885 (unlikely_executed_edge_p): New function.
14886 (unlikely_executed_stmt_p): New function.
14887 (unlikely_executed_bb_p): New function.
14888 (set_even_probabilities): Use unlikely predicates.
14889 (combine_predictions_for_bb): Likewise.
14890 (predict_paths_for_bb): Likewise.
14891 (predict_paths_leading_to_edge): Likewise.
14892 (determine_unlikely_bbs): New function.
14893 (estimate_bb_frequencies): Use it.
14894 (compute_function_frequency): Use zero counts even if profile is
14896 * profile-count.h: Fix typo.
14898 2017-08-08 Julia Koval <julia.koval@intel.com>
14900 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14901 _mm512_mask_cvtsepi16_storeu_epi8,
14902 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14903 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14904 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14905 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14906 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14907 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14908 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14909 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14910 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14911 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14912 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14913 __builtin_ia32_pmovuswb256mem_mask,
14914 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14915 __builtin_ia32_pmovwb512mem_mask): New builtins.
14917 2017-08-08 Julia Koval <julia.koval@intel.com>
14919 PR target/73350,80862
14920 * config/i386/subst.md (round): Fix round pattern.
14921 * config/i386/i386.c (ix86_erase_embedded_rounding):
14922 Fix erasing rounding for the fixed pattern.
14924 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14926 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14928 2017-06-08 Martin Liska <mliska@suse.cz>
14930 PR gcov-profile/80911
14931 * gcov.c (block_info::block_info): New constructor.
14933 2017-06-07 Carl Love <cel@us.ibm.com>
14935 * config/rs6000/rs6000-c: The return type of the following
14936 built-in functions was implemented as int not long long. Fix sign
14937 of return value for the unsigned version of vec_mulo and vec_mule.
14938 vector unsigned long long vec_bperm (vector unsigned long long,
14939 vector unsigned char)
14940 vector signed long long vec_mule (vector signed int,
14942 vector unsigned long long vec_mule (vector unsigned int,
14943 vector unsigned int)
14944 vector signed long long vec_mulo (vector signed int,
14946 vector unsigned long long vec_mulo (vector unsigned int,
14947 vector unsigned int)
14948 * doc/extend.texi: Fix the documentation for the built-in
14951 2017-06-07 Carl Love <cel@us.ibm.com>
14954 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14957 2017-06-07 Carl Love <cel@us.ibm.com>
14959 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14960 support, Generate doublehv for signed int/float for BE case only.
14962 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14964 * doc/invoke.texi (mcx16): Rewrite.
14966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14968 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14969 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14970 *mov<mode>_softfloat, and an anonymous splitter): Use
14971 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14975 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14977 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14978 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14979 (rs6000_debug_reg_global): Adjust.
14980 (rs6000_init_hard_regno_mode_ok): Adjust.
14981 (rs6000_dbx_register_number): Adjust.
14982 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14983 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14984 Remove SPE_ACC and SPEFSCR.
14985 (REG_ALLOC_ORDER): Ditto.
14986 (FRAME_POINTER_REGNUM): Change to 111.
14987 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14988 (REG_CLASS_NAMES): Ditto.
14989 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14990 (REGISTER_NAMES): Ditto.
14991 (ADDITIONAL_REG_NAMES): Ditto.
14992 (rs6000_reg_names): Ditto.
14993 * config/rs6000/rs6000.md: Renumber some register number
14996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14998 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15000 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15001 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15003 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15004 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15005 Delete the SPE high registers.
15006 (REG_ALLOC_ORDER): Ditto.
15007 (enum reg_class): Remove SPE_HIGH_REGS.
15008 (REG_CLASS_NAMES): Ditto.
15009 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15010 (REGISTER_NAMES): Ditto.
15011 (rs6000_reg_names): Ditto.
15012 * doc/tm.texi.in: Remove SPE as example.
15013 * doc/tm.texi: Regenerate.
15015 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15017 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15018 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15019 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15020 * config/rs6000/rs6000.md (type): Remove "brinc".
15022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15024 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15025 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15026 * config/rs6000/linuxspe.h: Delete file.
15027 * config/rs6000/rs6000.md: Don't include spe.md.
15028 * config/rs6000/spe.h: Delete file.
15029 * config/rs6000/spe.md: Delete file.
15030 * config/rs6000/t-rs6000: Remove spe.md.
15032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15034 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15035 (reg_or_none500mem_operand): Delete.
15036 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15037 instead of reg_or_none500mem_operand.
15039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15042 handling of SPE flags.
15043 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15045 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15047 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15049 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15050 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15051 paired_divv2sf3): Similar.
15052 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15053 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15054 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15056 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15057 Rename the paired_* instruction patterns.
15058 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15060 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15061 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15062 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15063 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15064 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15065 PAIRED_VECTOR_MODE.
15066 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15067 (spe_func_has_64bit_regs_p): Delete.
15068 (spe_expand_predicate_builtin): Delete.
15069 (spe_expand_evsel_builtin): Delete.
15070 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15071 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15072 (invalid_e500_subreg): Delete.
15073 (rs6000_legitimize_address): Always force_reg op2 as well, for
15074 paired single memory accesses.
15075 (rs6000_member_type_forces_blk): Delete.
15076 (rs6000_spe_function_arg): Delete.
15077 (rs6000_expand_unop_builtin): Delete SPE handling.
15078 (rs6000_expand_binop_builtin): Ditto.
15079 (spe_expand_stv_builtin): Delete.
15080 (bdesc_2arg_spe): Delete.
15081 (spe_expand_builtin): Delete.
15082 (spe_expand_predicate_builtin): Delete.
15083 (spe_expand_evsel_builtin): Delete.
15084 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15085 (spe_init_builtins): Delete.
15086 (spe_func_has_64bit_regs_p): Delete.
15087 (savres_routine_name): Delete "info" parameter. Adjust callers.
15088 (rs6000_emit_stack_reset): Ditto.
15089 (rs6000_dwarf_register_span): Delete.
15090 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15091 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15092 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15094 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15096 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15097 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15098 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15099 mulv2sf3, divv2sf3): Delete expanders.
15101 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15103 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15107 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15108 * config/rs6000/rs6000.c: Ditto.
15110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15112 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15113 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15114 comparison_operator.
15116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15118 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15119 * config/rs6000/rs6000.opt: Ditto.
15120 * config/rs6000/t-rtems: Ditto.
15122 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15124 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15125 TARGET_E500_SINGLE by 0, simplify.
15126 * config/rs6000/rs6000.c: Ditto.
15127 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15128 (spe_build_register_parallel): Delete.
15129 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15130 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15131 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15132 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15133 (E500_CONVERT): Delete.
15134 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15138 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15139 * config/rs6000/dfp.md: Ditto.
15140 (negdd2, *negdd2_fpr): Merge.
15141 (absdd2, *absdd2_fpr): Merge.
15142 (negtd2, *negtd2_fpr): Merge.
15143 (abstd2, *abstd2_fpr): Merge.
15144 * config/rs6000/e500.h: Delete file.
15145 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15146 TARGET_FPRS by 1 and simplify.
15147 * config/rs6000/rs6000-c.c: Ditto.
15148 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15149 TARGET_DF_SPE by 0.
15150 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15152 * config/rs6000/rs6000.md: Ditto.
15153 (floatdidf2, *floatdidf2_fpr): Merge.
15154 (move_from_CR_gt_bit): Delete.
15155 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15156 (E500_CR_IOR_COMPARE): Delete.
15157 (All patterns that require !TARGET_FPRS): Delete.
15158 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15162 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15164 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15166 * graphds.c (add_edge): Intitialize edge's attached data.
15167 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15168 pointer parameter. Call pointed function on each edge during
15169 graph traversing. Skip traversing the edge when the function
15171 (graphds_dfs, graphds_scc): Ditto.
15172 (for_each_edge): New parameter. Pass the new parameter to callback
15174 * graphds.h (skip_edge_callback): New function pointer type.
15175 (graphds_dfs, graphds_scc): New function pointer parameter.
15176 (graphds_edge_callback, for_each_edge): New parameter.
15178 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15180 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15181 out code checking if runtime alias check is possible to below ...
15182 Call the new function.
15183 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15184 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15186 2017-06-07 Marek Polacek <polacek@redhat.com>
15189 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15190 TYPE_OVERFLOW_WRAPS check.
15192 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15194 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15195 if versioning is required.
15196 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15197 peeling with the check for versioning.
15199 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15201 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15202 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15203 Set true to new parameter if new ssa variable is defined.
15204 (vect_gen_vector_loop_niters): Refactor. Set range information
15205 for the new vector loop bound variable.
15206 (vect_do_peeling): Ditto.
15208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15210 * tree-affine.c (ssa.h): Include header file.
15211 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15212 has wrapping overflow behavior.
15214 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15216 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15218 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15220 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15221 (tree_to_aff_combination): ... here.
15223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15225 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15226 reg_pressure model function.
15227 (ivopts_global_cost_for_size): Delete.
15228 (determine_set_costs, iv_ca_recount_cost): Call new model function
15229 ivopts_estimate_reg_pressure.
15231 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15233 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15234 expensive than udiv. Remove floating point cases from mod.
15236 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15238 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15239 Increase idiv cost.
15241 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15243 * config/aarch64/aarch64.md
15244 (copysignsf3): Fix mask generation.
15246 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15248 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15250 (class dump_manager): Add register_dumps method.
15251 * dumpfile.c: Include langhooks.h.
15252 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15253 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15254 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15255 (dump_manager::dump_register): Start with 512 entries instead of 32.
15256 (dump_manager::register_dumps): New method.
15257 * toplev.c (general_init): Instead of invoking register_dumps
15258 langhook, invoke register_dumps method on the dump manager.
15259 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15262 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15264 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15265 Say that # requires an associated define_split to exist, and that
15266 the define_split must be suitable for use after register allocation.
15268 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15270 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15271 (compute_outgoing_frequencies): Also initialize zero counts.
15272 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15273 around loops; preserve more of profile when nothing changes.
15275 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15277 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15279 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15280 * config/arm/arm-cpu-cdata.h: Regenerate.
15281 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15282 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15283 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15284 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15285 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15287 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15289 * config/arm/t-rmprofile: Likewise.
15290 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15292 2017-06-06 David S. Miller <davem@davemloft.net>
15295 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15296 blockage if function uses alloca.
15298 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15300 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15301 New "uid" fields to hold pretty-print IDs of group and ref.
15302 Memory references are now identified as <group_id>:<ref_id>
15303 instead of using [random] addresses.
15304 (dump_mem_details): Simplify, no functional change.
15305 (dump_mem_ref): Simplify and make output more concise.
15306 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15307 (find_or_create_group): Initialize group uid.
15308 (record_ref): Initialize ref uid. Improve debug output.
15309 (prune_group_by_reuse, should_issue_prefetch_p,)
15310 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15311 (mark_nontemporal_store, determine_loop_nest_reuse):
15312 Improve debug output.
15314 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15316 * dbgcnt.def (prefetch): New debug counter.
15317 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15318 (schedule_prefetches): Stop issueing prefetches if debug counter
15321 2017-06-06 Tom de Vries <tom@codesourcery.com>
15323 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15324 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15326 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15328 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15329 Use aarch64_reg_or_zero predicate for operand 4.
15330 (aarch64_compare_and_swap<mode> define_insn_and_split):
15331 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15332 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15334 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15336 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15337 (arm_compute_save_core_reg_mask): This.
15338 (thumb1_compute_save_reg_mask): Rename into ...
15339 (thumb1_compute_save_core_reg_mask): This.
15340 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15341 (arm_compute_frame_layout): Likewise.
15343 2017-06-06 Richard Biener <rguenther@suse.de>
15345 PR tree-optimization/80974
15346 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15347 keep or clear leaders SSA info.
15349 2017-06-06 Tom de Vries <tom@codesourcery.com>
15351 * config/nvptx/nvptx.c (split_mode_p): New function.
15352 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15354 2017-06-06 Tom de Vries <tom@codesourcery.com>
15356 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15358 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15361 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15364 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15366 * shrink-wrap.c (handle_simple_exit): Update profile.
15367 (try_shrink_wrapping): Upate profile.
15369 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15371 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15372 (tree_guess_outgoing_edge_probabilities): New.
15373 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15374 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15376 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15378 * ipa-split.c (split_function): Initialize return bb profile.
15380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15382 * profile.c (compute_branch_probabilities): Also initialize
15383 EXIT_BLOCK profile.
15385 2017-06-06 Richard Biener <rguenther@suse.de>
15387 PR tree-optimization/80928
15388 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15389 (vect_analyze_loop_operations): Properly guard analysis for
15391 (vect_transform_loop): Likewise.
15392 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15393 (vect_model_induction_cost): Do not cost for pure SLP.
15394 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15395 of induction in inner loop vectorization.
15396 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15397 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15398 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15400 (vect_analyze_slp_cost_1): Cost induction.
15401 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15402 (vect_get_slp_vect_defs): Likewise.
15403 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15404 (vect_transform_stmt): Handle SLP reductions.
15405 * tree-vectorizer.h (vectorizable_induction): Adjust.
15407 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15409 * config/rs6000/rs6000.c (make_resolver_func): Update
15410 init_lowered_empty_function call.
15412 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15414 * doc/invoke.texi: Document the -fprofile-abs-path option.
15415 * common.opt (fprofile-abs-path): New option.
15416 * gcov-io.h (gcov_write_filename): Declare.
15417 * gcov-io.c (gcov_write_filename): New function.
15418 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15419 * profile.c (output_location): Likewise.
15421 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15423 * shring-wrap.c: Revert accidental commit.
15425 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15427 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15429 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15431 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15433 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15434 profile in callgraph edge.
15435 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15436 (apply_scale): Likewise.
15437 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15438 Also scale profile when inlining function with zero profile.
15439 (initialize_cfun): Update exit block profile even when it is zero.
15440 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15441 when profile is read.
15443 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15445 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15446 (CLONE_*): New constants to define the processors we can generate
15447 code for with the target_clone attribute.
15448 (rs6000_clone_map): New array to identify which clone processors
15449 the current program is running on.
15450 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15451 target_clone attribute.
15452 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15453 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15454 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15455 (cpu_expand_builtin): Add support for target_clone attribute.
15456 (rs6000_valid_attribute_p): Allow "default" attribute.
15457 (get_decl_name): New debug function to simplify printing the
15458 current function name in debugging statements.
15459 (rs6000_clone_priority): New functions to support the target_clone
15460 attribute, and be able to generate code to switch between ISA 2.05
15461 through ISA 3.0 (power6 through power9).
15462 (rs6000_compare_version_priority): Likewise.
15463 (rs6000_get_function_versions_dispatcher): Likewise.
15464 (make_resolver_func): Likewise.
15465 (add_condition_to_bb): Likewise.
15466 (dispatch_function_versions): Likewise.
15467 (rs6000_generate_version_dispatcher_body): Likewise.
15468 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15469 (fusion_gpr_load_p): Fix a spacing issue.
15470 * doc/extend.texi (Common Function Attributes): Document that the
15471 PowerPC supports the target_clone attribute.
15473 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15475 * config/arm/arm.h: explain F symbol found in description of ARM
15476 register allocation in its legend.
15478 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15480 * config/mips/frame-header-opt.c: Include profile-count.h.
15481 * config/riscv/riscv.c: Include profile-count.h
15483 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15485 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15487 (sm_set_flag_if_changed): Add bbs field.
15488 (execute_sm_if_changed_flag_set): Pass BBS.
15489 (execute_sm): Update.
15491 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15493 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15496 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15498 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15499 (peephole2): New peephole2 to emit the above.
15500 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15502 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15504 * config/aarch64/aarch64.c (define_peephole2 above
15505 *sub_<shift>_<mode>): New peephole.
15507 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15509 * config/i386/i386.c (make_resolver_func): Update.
15510 * Makefile.in: Add profile-count.h and profile-count.o
15511 * auto-profile.c (afdo_indirect_call): Update to new API.
15512 (afdo_set_bb_count): Update.
15513 (afdo_propagate_edge): Update.
15514 (afdo_propagate_circuit): Update.
15515 (afdo_calculate_branch_prob): Update.
15516 (afdo_annotate_cfg): Update.
15517 * basic-block.h: Include profile-count.h
15518 (struct edge_def): Turn count to profile_count.
15519 (struct basic_block_def): Likewie.
15520 (REG_BR_PROB_BASE): Move to profile-count.h
15521 (RDIV): Move to profile-count.h
15522 * bb-reorder.c (max_entry_count): Turn to profile_count.
15523 (find_traces): Update.
15524 (rotate_loop):Update.
15525 (connect_traces):Update.
15526 (sanitize_hot_paths):Update.
15527 * bt-load.c (migrate_btr_defs): Update.
15528 * cfg.c (RDIV): Remove.
15529 (init_flow): Use alloc_block.
15530 (alloc_block): Uninitialize count.
15531 (unchecked_make_edge): Uninitialize count.
15532 (check_bb_profile): Update.
15533 (dump_edge_info): Update.
15534 (dump_bb_info): Update.
15535 (update_bb_profile_for_threading): Update.
15536 (scale_bbs_frequencies_int): Update.
15537 (scale_bbs_frequencies_gcov_type): Update.
15538 (scale_bbs_frequencies_profile_count): New.
15539 * cfg.h (update_bb_profile_for_threading): Update.
15540 (scale_bbs_frequencies_profile_count): Declare.
15541 * cfgbuild.c (compute_outgoing_frequencies): Update.
15542 (find_many_sub_basic_blocks): Update.
15543 * cfgcleanup.c (try_forward_edges): Update.
15544 (try_crossjump_to_edge): Update.
15545 * cfgexpand.c (expand_gimple_tailcall): Update.
15546 (construct_exit_block): Update.
15547 * cfghooks.c (verify_flow_info): Update.
15548 (dump_bb_for_graph): Update.
15549 (split_edge): Update.
15550 (make_forwarder_block): Update.
15551 (duplicate_block): Update.
15552 (account_profile_record): Update.
15553 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15554 (get_estimated_loop_iterations): Update.
15555 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15556 (single_likely_exit): Update.
15557 * cfgloopmanip.c (scale_loop_profile): Update.
15559 (set_zero_probability): Update.
15560 (lv_adjust_loop_entry_edge): Update.
15561 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15562 (purge_dead_edges): Update.
15563 (rtl_account_profile_record): Update.
15564 * cgraph.c (cgraph_node::create): Uninitialize count.
15565 (symbol_table::create_edge): Uninitialize count.
15566 (cgraph_update_edges_for_call_stmt_node): Update.
15567 (cgraph_edge::dump_edge_flags): Update.
15568 (cgraph_node::dump): Update.
15569 (cgraph_edge::maybe_hot_p): Update.
15570 * cgraph.h: Include profile-count.h
15571 (create_clone), create_edge, create_indirect_edge): Update.
15572 (cgraph_node): Turn count to profile_count.
15573 (cgraph_edge0: Likewise.
15574 (make_speculative, clone): Update.
15575 (create_edge): Update.
15576 (init_lowered_empty_function): Update.
15577 * cgraphclones.c (cgraph_edge::clone): Update.
15578 (duplicate_thunk_for_node): Update.
15579 (cgraph_node::create_clone): Update.
15580 * cgraphunit.c (cgraph_node::analyze): Update.
15581 (cgraph_node::expand_thunk): Update.
15582 * final.c (dump_basic_block_info): Update.
15583 * gimple-streamer-in.c (input_bb): Update.
15584 * gimple-streamer-out.c (output_bb): Update.
15585 * graphite.c (print_global_statistics): Update.
15586 (print_graphite_scop_statistics): Update.
15587 * hsa-brig.c: Include basic-block.h.
15588 * hsa-dump.c: Include basic-block.h.
15589 * hsa-gen.c (T sum_slice): Update.
15590 (convert_switch_statements):Update.
15591 * hsa-regalloc.c: Include basic-block.h.
15592 * ipa-chkp.c (chkp_produce_thunks): Update.
15593 * ipa-cp.c (struct caller_statistics): Update.
15594 (init_caller_stats): Update.
15595 (gather_caller_stats): Update.
15596 (ipcp_cloning_candidate_p): Update.
15597 (good_cloning_opportunity_p): Update.
15598 (get_info_about_necessary_edges): Update.
15599 (dump_profile_updates): Update.
15600 (update_profiling_info): Update.
15601 (update_specialized_profile): Update.
15602 (perhaps_add_new_callers): Update.
15603 (decide_about_value): Update.
15604 (ipa_cp_c_finalize): Update.
15605 * ipa-devirt.c (struct odr_type_warn_count): Update.
15606 (struct decl_warn_count): Update.
15607 (struct final_warning_record): Update.
15608 (possible_polymorphic_call_targets): Update.
15609 (ipa_devirt): Update.
15610 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15611 * ipa-icf.c (sem_function::merge): Update.
15612 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15613 * ipa-inline.c (compute_uninlined_call_time): Update.
15614 (compute_inlined_call_time): Update.
15615 (want_inline_small_function_p): Update.
15616 (want_inline_self_recursive_call_p): Update.
15617 (edge_badness): Update.
15618 (lookup_recursive_calls): Update.
15619 (recursive_inlining): Update.
15620 (inline_small_functions): Update.
15621 (dump_overall_stats): Update.
15622 (dump_inline_stats): Update.
15623 * ipa-profile.c (ipa_profile_generate_summary): Update.
15624 (ipa_propagate_frequency): Update.
15625 (ipa_profile): Update.
15626 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15627 * ipa-utils.c (ipa_merge_profiles): Update.
15628 * loop-doloop.c (doloop_modify): Update.
15629 * loop-unroll.c (report_unroll): Update.
15630 (unroll_loop_runtime_iterations): Update.
15631 * lto-cgraph.c (lto_output_edge): Update.
15632 (lto_output_node): Update.
15633 (input_node): Update.
15634 (input_edge): Update.
15635 (merge_profile_summaries): Update.
15636 * lto-streamer-in.c (input_cfg): Update.
15637 * lto-streamer-out.c (output_cfg): Update.
15638 * mcf.c (create_fixup_graph): Update.
15639 (adjust_cfg_counts): Update.
15640 (sum_edge_counts): Update.
15641 * modulo-sched.c (sms_schedule): Update.
15642 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15643 * predict.c (maybe_hot_count_p): Update.
15644 (probably_never_executed): Update.
15645 (dump_prediction): Update.
15646 (combine_predictions_for_bb): Update.
15647 (propagate_freq): Update.
15648 (handle_missing_profiles): Update.
15649 (counts_to_freqs): Update.
15650 (rebuild_frequencies): Update.
15651 (force_edge_cold): Update.
15652 * predict.h: Include profile-count.h
15653 (maybe_hot_count_p, counts_to_freqs): UPdate.
15654 * print-rtl-function.c: Do not include cfg.h
15655 * print-rtl.c: Include basic-block.h
15656 * profile-count.c: New file.
15657 * profile-count.h: New file.
15658 * profile.c (is_edge_inconsistent): Update.
15659 (correct_negative_edge_counts): Update.
15660 (is_inconsistent): Update.
15661 (set_bb_counts): Update.
15662 (read_profile_edge_counts): Update.
15663 (compute_frequency_overlap): Update.
15664 (compute_branch_probabilities): Update; Initialize and deinitialize
15666 (branch_prob): Update.
15667 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15668 (edge_gcov_count): New.
15669 (bb_gcov_count): New.
15670 * shrink-wrap.c (try_shrink_wrapping): Update.
15671 * tracer.c (better_p): Update.
15672 * trans-mem.c (expand_transaction): Update.
15673 (ipa_tm_insert_irr_call): Update.
15674 (ipa_tm_insert_gettmclone_call): Update.
15675 * tree-call-cdce.c: Update.
15676 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15677 (gimple_duplicate_sese_tail): Update.
15678 (gimple_account_profile_record): Update.
15679 (execute_fixup_cfg): Update.
15680 * tree-inline.c (copy_bb): Update.
15681 (copy_edges_for_bb): Update.
15682 (initialize_cfun): Update.
15683 (freqs_to_counts): Update.
15684 (copy_cfg_body): Update.
15685 (expand_call_inline): Update.
15686 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15687 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15688 (try_unroll_loop_completely): Update.
15689 (try_peel_loop): Update.
15690 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15691 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15692 * tree-ssa-loop-split.c (connect_loops): Update.
15693 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15694 * tree-ssa-reassoc.c (branch_fixup): Update.
15695 * tree-ssa-tail-merge.c (replace_block_by): Update.
15696 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15697 (compute_path_counts): Update.
15698 (update_profile): Update.
15699 (recompute_probabilities): Update.
15700 (update_joiner_offpath_counts): Update.
15701 (estimated_freqs_path): Update.
15702 (freqs_to_counts_path): Update.
15703 (clear_counts_path): Update.
15704 (ssa_fix_duplicate_block_edges): Update.
15705 (duplicate_thread_path): Update.
15706 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15707 (struct switch_conv_info): Update.
15708 * tree-tailcall.c (decrease_profile): Update.
15709 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15710 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15711 * value-prof.c (check_counter): Update.
15712 (gimple_divmod_fixed_value): Update.
15713 (gimple_mod_pow2): Update.
15714 (gimple_mod_subtract): Update.
15715 (gimple_ic_transform): Update.
15716 (gimple_stringop_fixed_value): Update.
15717 * value-prof.h (gimple_ic): Update.
15719 2017-06-02 Carl Love <cel@us.ibm.com>
15721 * config/rs6000/rs6000-c: Add support for built-in functions
15722 vector double vec_doublee (vector signed int);
15723 vector double vec_doublee (vector unsigned int);
15724 vector double vec_doublee (vector float);
15725 vector double vec_doubleh (vector signed int);
15726 vector double vec_doubleh (vector unsigned int);
15727 vector double vec_doubleh (vector float);
15728 vector double vec_doublel (vector signed int);
15729 vector double vec_doublel (vector unsigned int);
15730 vector double vec_doublel (vector float);
15731 vector double vec_doubleo (vector signed int);
15732 vector double vec_doubleo (vector unsigned int);
15733 vector double vec_doubleo (vector float);.
15734 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15735 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15737 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15738 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15739 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15741 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15742 vec_doublel, vec_doubleh.
15743 * doc/extend.texi: Update the built-in documentation file for the
15744 new built-in functions.
15746 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15749 * ipa-inline-analysis.c (free_growth_caches): Set
15750 edge_removal_hook_holder to NULL after removing it.
15752 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15754 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15755 comparision with zero.
15757 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15758 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15759 for early expansion of vec_min and vec_max builtins.
15760 (builtin_function_type): Add min/max unsigned variants to those
15761 identified as having unsigned arguments.
15763 2017-06-02 Olivier Hainque <hainque@adacore.com>
15765 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15767 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15769 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15770 Use VALL_F16 iterator rather than VALL.
15772 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15774 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15775 Emit CBNZ inside loop when doing a strong exchange and comparing
15776 against zero. Generate the CC flags after the loop.
15778 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15780 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15781 (dl_section_ref): New.
15782 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15783 On AIX, append an expression to subtract the size of the
15784 section length to dl_section_ref.
15786 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15788 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15789 for early expansion of vector absolute builtins.
15791 2017-06-02 Richard Biener <rguenther@suse.de>
15793 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
15794 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
15796 2017-06-02 Richard Biener <rguenther@suse.de>
15798 PR tree-optimization/80948
15799 * tree-tailcall.c (find_tail_calls): Track stmts to move in
15800 stmt order as well.
15802 2017-06-02 Richard Biener <rguenther@suse.de>
15804 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
15806 * tree-vect-stmts.c (process_use): Do not mark backedge defs
15807 for inductions as relevant.
15809 2017-06-02 Richard Biener <rguenther@suse.de>
15811 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
15812 (vectorizable_induction): ... this. Remove dead code.
15814 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
15816 * builtins. (expand_builtin_alloca): Remove second parameter and
15817 infer its value from the first parameter instead.
15818 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15820 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15822 PR rtl-optimization/80903
15823 * loop-doloop.c (add_test): Unshare sequence.
15825 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15827 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15829 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15831 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15833 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15834 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15835 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15836 (xlogue_layout::get_stub_rtx): Make static.
15837 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15838 (xlogue_layout::compute_stub_managed_regs): Rename to...
15839 (xlogue_layout::count_stub_managed_regs): ...this.
15840 (xlogue_layout::is_stub_managed_reg): New function.
15841 (xlogue_layout::m_stub_names): Rename to...
15842 (xlogue_layout::s_stub_names): ...this, make static.
15843 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15844 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15845 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15846 xlogue_layout::s_stub_names): Instantiate statics.
15847 (stub_managed_regs): Remove.
15848 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15849 (disable_call_ms2sysv_xlogues): Rename to...
15850 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15851 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15853 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15854 change after reload_completed.
15855 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15857 (ix86_expand_prologue): Likewise.
15858 (ix86_expand_epilogue): Likewise.
15859 (ix86_expand_split_stack_prologue): Likewise.
15860 (ix86_compute_frame_layout): Remove frame parameter ...
15861 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15862 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15864 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15865 (ix86_frame): Move from here ...
15866 * config/i386/i386.h (ix86_frame): ... to here.
15867 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15868 complete ix86_frame data structure instead. Remove some_ld_name.
15870 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15872 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15873 symbols that hold a DECL_VALUE_EXPR.
15875 2017-06-01 Martin Jambor <mjambor@suse.cz>
15877 PR tree-optimization/80898
15878 * tree-sra.c (process_subtree_disqualification): Removed.
15879 (disqualify_candidate): Do not acll
15880 process_subtree_disqualification.
15881 (subtree_mark_written_and_enqueue): New function.
15882 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15883 RHS has been disqualified and re-queue LHS if necessary. Apart
15884 from that, ignore disqualified RHS.
15886 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15888 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15889 address fetch for z10 or later.
15891 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15893 * config/arc/arc.md (tst_movb): Add guard when splitting.
15895 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15897 * config/arc/arc.c (arc_can_eliminate): Test against
15898 arc_frame_pointer_needed.
15900 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15902 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15903 to prevent store reordering.
15904 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15905 (type): Add block type.
15906 (stack_tie): Define special instruction to be used in
15909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15911 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15912 constraint. It is not valid for the pattern.
15913 (noncommutative_binary_comparison): Likewise.
15915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15917 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15920 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15922 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15923 be used by the reg-alloc.
15925 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15927 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15928 reg-alloc when having mul64 or mul32x16 instructions.
15929 (mulsidi3): Likewise.
15930 (umulsidi3): Likewise.
15931 (mulsi32x16): New pattern.
15932 (mulsi64): Likewise.
15933 (mulsidi64): Likewise.
15934 (umulsidi64): Likewise.
15935 (MUL32x16_REG): Define.
15936 (mul64_600): Use MUL32x16_REG.
15937 (mac64_600): Likewise.
15938 (umul64_600): Likewise.
15939 (umac64_600): Likewise.
15941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15943 * config/arc/arc.md (mulsi3_700): Make it commutative.
15945 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15947 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15949 (*sign_extendsidi2_insn): Likewise for movstosw.
15951 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15953 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15954 the type of the input discriminant value. Convert the
15955 discriminant value of signedness vary.
15957 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15959 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15960 Add to -Wall section.
15962 2017-06-01 Richard Biener <rguenther@suse.de>
15964 PR middle-end/66313
15965 * fold-const.c (fold_plusminus_mult_expr): If the factored
15966 factor may be zero use a wrapping type for the inner operation.
15967 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15968 and handle moved defs.
15969 (process_assignment): Properly guard the unary op case. Return a
15970 tri-state indicating that moving the stmt before the call may allow
15971 to continue. Pass through to_move.
15972 (find_tail_calls): Handle moving unrelated defs before
15975 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15978 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15979 splitter result in the canonical way.
15981 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15983 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15984 also for 32bit target. Update insn attributes.
15985 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15987 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15989 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15990 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15991 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15992 (_mm_maskz_min_ss): New intrinsics.
15994 2017-05-31 Martin Liska <mliska@suse.cz>
15996 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15997 Change comment style to one we normally use.
15998 (vectorizable_reduction): Likewise.
15999 (vectorizable_induction): Likewise.
16000 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16001 (vectorizable_call): Likewise.
16002 (vectorizable_simd_clone_call): Likewise.
16003 (vectorizable_conversion): Likewise.
16004 (vectorizable_assignment): Likewise.
16005 (vectorizable_shift): Likewise.
16006 (vectorizable_operation): Likewise.
16007 (vectorizable_store): Likewise.
16008 (vectorizable_load): Likewise.
16009 * tree-vectorizer.h: Likewise.
16011 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16013 * passes.c (emergency_dump_function): New.
16014 * tree-pass.h (emergency_dump_function): Declare.
16015 * plugin.c (plugins_internal_error_function): Remove.
16016 * plugin.h (plugins_internal_error_function): Remove declaration.
16017 * toplev.c (internal_error_function): New static function. Use it...
16018 (general_init): ...here.
16020 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16022 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16023 (arc_rtx_costs): Add costs for new patterns.
16024 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16025 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16027 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16029 * tree-ssa-strlen.c (get_next_strinfo): New function.
16030 (get_stridx_plus_constant): Use it.
16031 (zero_length_string): Likewise.
16032 (adjust_related_strinfos): Likewise.
16033 (adjust_last_stmt): Likewise.
16035 2017-05-31 Richard Biener <rguenther@suse.de>
16038 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16039 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16041 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16043 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16044 loop_vinfo argument and use of dependence distance vectors.
16045 Check instead whether the two references differ only in their
16046 initial value and assume that they have the same alignment if the
16047 difference is a multiple of the vector alignment.
16048 (vect_analyze_data_refs_alignment): Update call accordingly.
16050 2017-05-31 Martin Liska <mliska@suse.cz>
16053 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16055 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16057 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16058 (create_intersect_range_checks): Move from ...
16059 * tree-data-ref.c (create_intersect_range_checks_index)
16060 (create_intersect_range_checks): ... to here.
16061 (create_runtime_alias_checks): New function factored from ...
16062 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16063 here. Call above function.
16064 * tree-data-ref.h (create_runtime_alias_checks): New function.
16066 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16068 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16069 segment length for dr_b and compute it in wide_int.
16071 2017-05-31 Richard Biener <rguenther@suse.de>
16073 PR tree-optimization/80906
16074 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16075 and pass through iv_map.
16076 (copy_bb_and_scalar_dependences): Adjust.
16077 (translate_pending_phi_nodes): Likewise.
16078 (copy_loop_close_phi_args): Handle code-generating IVs instead
16081 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16083 * diagnostic-color.c (color_dict): Add "type-diff".
16084 (parse_gcc_colors): Update comment.
16085 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16086 -fdiagnostics-show-template-tree and -fno-elide-type.
16087 (GCC_COLORS): Add type-diff to example.
16089 (-fdiagnostics-show-template-tree): New.
16090 (-fno-elide-type): New.
16091 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16092 the pp_format_decoder callback. Call any m_format_postprocessor's
16094 (pretty_printer::pretty_printer): Initialize
16095 m_format_postprocessor.
16096 (pretty_printer::~pretty_printer): Delete any
16097 m_format_postprocessor.
16098 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16099 (class format_postprocessor): New class.
16100 (struct pretty_printer::format_decoder): Document the new parameters.
16101 (struct pretty_printer::m_format_postprocessor): New field.
16102 * tree-diagnostic.c (default_tree_printer): Update for new
16103 bool and const char ** params.
16104 * tree-diagnostic.h (default_tree_printer): Likewise.
16106 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16108 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16109 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16111 rs6000_gen_cell_microcode code.
16112 (rs6000_final_prescan_insn): Delete.
16113 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16114 "warn-cell-microcode" entries.
16115 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16116 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16117 throughout. Change cc_reg_not_micro_cr0_operand to
16118 cc_reg_not_cr0_operand throughout.
16119 (*extendhi<mode>2_noload): Delete.
16120 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16121 (mwarn-cell-microcode): Delete.
16122 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16123 -mgen-cell-microcode and -mwarn-cell-microcode.
16125 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16128 * config/i386/constraints.md (Yd): New constraint.
16130 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16131 and (?Yd, r) alternatives. Update insn attributes.
16132 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16133 and (?*Yd, r) alternatives. Update insn attributes.
16134 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16136 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16138 * gimplify.c (gimplify_modify_expr): Don't create a
16139 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16142 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16144 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16146 2017-05-30 Richard Biener <rguenther@suse.de>
16148 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16149 and reduc_def fields.
16150 (STMT_VINFO_REDUC_TYPE): New define.
16151 (STMT_VINFO_REDUC_DEF): Likewise.
16152 (vect_force_simple_reduction): Adjust prototype.
16153 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16154 (vect_is_simple_reduction): Remove check_reduction argument.
16155 (vect_force_simple_reduction): Adjust and set
16156 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16157 (vectorizable_reduction): Do not re-do reduction analysis
16158 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16159 * tree-parloops.c (gather_scalar_reductions): Adjust.
16161 2017-05-30 Richard Biener <rguenther@suse.de>
16163 PR middle-end/80901
16164 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16167 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16169 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16170 Introduce unknown_misalignment parameter and remove vf.
16171 (vect_peeling_hash_get_lowest_cost):
16172 Pass unknown_misalignment parameter.
16173 (vect_enhance_data_refs_alignment):
16174 Fix unsupportable data ref treatment.
16176 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16178 * tree-vect-data-refs.c (vect_get_data_access_cost):
16179 Workaround for SLP handling.
16180 (vect_enhance_data_refs_alignment):
16181 Compute costs for doing no peeling at all, compare to the best
16182 peeling costs so far and avoid peeling if cheaper.
16184 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16186 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16187 Return peeling info and set costs to zero for unlimited cost
16189 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16190 with unknown misalignment. Compute and costs for unknown
16191 misalignment, compare them to the costs for known misalignment
16192 and choose the cheapest for peeling.
16194 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16196 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16197 (vect_get_peeling_costs_all_drs): Create function.
16198 (vect_peeling_hash_get_lowest_cost):
16199 Use vect_get_peeling_costs_all_drs.
16200 (vect_peeling_supportable): Create function.
16201 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16203 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16205 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16206 DR_HAS_NEGATIVE_STEP.
16207 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16208 (vect_enhance_data_refs_alignment): Use.
16209 (vect_duplicate_ssa_name_ptr_info): Use.
16210 * tree-vectorizer.h (dr_misalignment): Use.
16211 (known_alignment_for_access_p): Use.
16213 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16216 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16218 (has_section_name): New function.
16220 2017-05-30 Martin Liska <mliska@suse.cz>
16223 * auto-profile.c (get_function_decl_from_block): Fix
16226 2017-05-30 Richard Biener <rguenther@suse.de>
16228 PR middle-end/80876
16229 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16231 2017-05-30 Martin Liska <mliska@suse.cz>
16233 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16234 * dumpfile.h (struct dump_file_info): Remove ctors.
16236 2017-05-30 Martin Liska <mliska@suse.cz>
16238 * predict.def: Fix GNU coding style.
16240 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16242 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16243 Mark 'to' argument with ATTRIBUTE_UNUSED.
16245 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16247 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16248 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16249 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16252 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16254 * doc/install.texi (Options specification): Restore entry of
16255 --enable-sjlj-exceptions.
16257 2017-05-27 Michael Eager <eager@eagercon.com>
16260 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16262 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16264 * config/microblaze/microblaze.h
16265 (FIXED_REGISTERS): Update in macro.
16266 (CALL_USED_REGISTERS): Update in macro.
16268 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16270 * doc/install.texi: Add links to macOS binary distributions.
16272 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16276 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16278 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16280 2017-05-26 Martin Liska <mliska@suse.cz>
16282 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16284 2017-05-26 Martin Liska <mliska@suse.cz>
16286 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16287 always leading ';; '.
16288 (dump_bb_info): Likewise.
16289 (brief_dump_cfg): Likewise.
16290 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16291 * dumpfile.c: Remove usage of TDF_VERBOSE.
16292 * dumpfile.h (enum dump_kind): Likewise.
16293 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16294 * print-tree.c (debug_verbose): Remove.
16295 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16296 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16297 * tree-diagnostic.c (default_tree_printer): Replace
16298 TDF_DIAGNOSTIC with TDF_SLIM.
16300 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16302 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16303 in parameter loop, rather than loop_vinfo.
16304 (create_intersect_range_checks): Ditto.
16305 (vect_create_cond_for_alias_checks): Update call to above functions.
16307 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16309 PR tree-optimization/80815
16310 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16311 for merging runtime alias checks. Handle negative DR_STEPs.
16313 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16315 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16317 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16318 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16319 out code pruning runtime alias checks.
16320 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16321 factored out from above.
16322 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16324 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16326 (prune_runtime_alias_test_list): New decalaration.
16328 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16330 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16331 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16332 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16333 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16334 (operator==, comp_dr_with_seg_len_pair): Ditto.
16335 (vect_prune_runtime_alias_test_list): Ditto.
16337 2017-05-26 Martin Liska <mliska@suse.cz>
16340 * params.def: Bound partial-inlining-entry-probability param.
16342 2017-05-26 Marek Polacek <polacek@redhat.com>
16345 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16348 2017-05-26 Richard Biener <rguenther@suse.de>
16350 PR tree-optimization/80842
16351 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16354 2017-05-26 Richard Biener <rguenther@suse.de>
16356 PR tree-optimization/80844
16357 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16359 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16361 * doc/md.texi (Machine Constraints): Update x86 family
16362 machine constraints section to match 'config/i386/constraints.md'.
16364 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16366 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16368 2017-05-25 Nathan Sidwell <nathan@acm.org>
16370 * doc/invoke.texi (--enable-languages): Update documentation.
16372 2017-05-25 Martin Liska <mliska@suse.cz>
16374 * dumpfile.c: Add TDF_FOLDING.
16375 * dumpfile.h (enum dump_kind): Likewise.
16376 * genmatch.c (dt_simplify::gen_1): Use it.
16378 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16380 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16382 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16384 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16385 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16387 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16389 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16390 * match.pd (X == C): Rewrite it here.
16391 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16392 with_certain_nonzero_bits2): New predicates.
16393 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16395 2017-05-24 Nathan Sidwell <nathan@acm.org>
16397 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16400 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16403 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16405 * config/powerpcspe: New port. Files are copied from the rs6000
16406 port, with "rs6000" in filenames replaced by "powerpcspe".
16408 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16410 PR rtl-optimization/80754
16411 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16413 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16415 * config/sparc/sparc.md (length): Return the correct value for -mflat
16416 sibcalls to match output_sibcall.
16418 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16422 * config/rs6000/rs6000.c (struct machine_function): Add new field
16424 (rs6000_get_separate_components): Init that field, use it.
16425 (rs6000_components_for_bb): Use the field.
16427 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16429 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16431 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16433 PR middle-end/80823
16434 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16436 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16439 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16440 address against address_operand predicate.
16441 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16443 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16445 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16446 parameters passed indirectly.
16448 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16450 * config/i386/i386.md (*movdi_internal): Remove SSE4
16451 alternative 18 (?r, *v). Update insn attributes.
16452 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16453 Update insn attributes.
16454 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16455 Update insn attributes.
16456 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16457 alternative 1 (r, v). Remove isa attribute.
16458 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16459 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16460 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16462 2017-05-23 Tom de Vries <tom@codesourcery.com>
16464 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16467 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16469 * cgraphunit.c (symbol_table::process_new_functions): Update.
16470 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16471 (inline_generate_summary): Rename to ...
16472 (ipa_fn_summary_generate): ... this one.
16473 (inline_read_summary): Rename to ...
16474 (ipa_fn_summary_read): ... this one.
16475 (inline_write_summary): Rename to ...
16476 (ipa_fn_summary_write): ... this one.
16477 (inline_free_summary): Rename to ...
16478 (ipa_free_fn_summary): ... this one.
16479 (pass_data_local_fn_summary, pass_local_fn_summary,
16480 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16481 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16482 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16483 make_pass_ipa_fn_summary): New.
16484 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16485 inline_write_summary, inline_free_summary): Remove.
16486 (ipa_free_fn_summary) : New.
16487 * ipa-inline.c (ipa_inline): Update.
16488 (pass_ipa_inline): Do not generate summaries.
16489 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16491 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16492 and add pass_ipa_fn_summary.
16493 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16495 (make_pass_inline_parameters): Remove.
16497 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16499 * omp-low.c (struct omp_context): Remove "default_kind" member.
16502 * omp-offload.c (execute_oacc_device_lower): Remove the
16503 parallelism dimensions function attributes for unparallelized
16504 OpenACC kernels constructs.
16506 2017-05-23 Martin Liska <mliska@suse.cz>
16508 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16510 (cgraph_edge::make_speculative): Likewise.
16511 (cgraph_edge::resolve_speculation): Likewise.
16512 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16513 (cgraph_node::dump): Likewise.
16514 * cgraph.h: Likewise.
16515 * cgraphunit.c (analyze_functions): Likewise.
16516 (symbol_table::compile): Likewise.
16517 * ipa-cp.c (print_all_lattices): Likewise.
16518 (determine_versionability): Likewise.
16519 (initialize_node_lattices): Likewise.
16520 (ipcp_verify_propagated_values): Likewise.
16521 (estimate_local_effects): Likewise.
16522 (update_profiling_info): Likewise.
16523 (create_specialized_node): Likewise.
16524 (perhaps_add_new_callers): Likewise.
16525 (decide_about_value): Likewise.
16526 (decide_whether_version_node): Likewise.
16527 (identify_dead_nodes): Likewise.
16528 (ipcp_store_bits_results): Likewise.
16529 * ipa-devirt.c (dump_targets): Likewise.
16530 (ipa_devirt): Likewise.
16531 * ipa-icf.c (sem_item::dump): Likewise.
16532 (sem_function::equals): Likewise.
16533 (sem_variable::equals): Likewise.
16534 (sem_item_optimizer::read_section): Likewise.
16535 (sem_item_optimizer::execute): Likewise.
16536 (congruence_class::dump): Likewise.
16537 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16538 (dump_inline_summary): Likewise.
16539 (estimate_node_size_and_time): Likewise.
16540 (inline_analyze_function): Likewise.
16541 * ipa-inline-transform.c (inline_call): Likewise.
16542 * ipa-inline.c (report_inline_failed_reason): Likewise.
16543 (want_early_inline_function_p): Likewise.
16544 (edge_badness): Likewise.
16545 (update_edge_key): Likewise.
16546 (inline_small_functions): Likewise.
16547 * ipa-profile.c (ipa_profile): Likewise.
16548 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16549 (ipa_make_edge_direct_to_target): Likewise.
16550 (remove_described_reference): Likewise.
16551 (ipa_impossible_devirt_target): Likewise.
16552 (propagate_controlled_uses): Likewise.
16553 (ipa_print_node_params): Likewise.
16554 (ipcp_transform_function): Likewise.
16555 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16556 (propagate_pure_const): Likewise.
16557 * ipa-reference.c (generate_summary): Likewise.
16558 (read_write_all_from_decl): Likewise.
16559 (propagate): Likewise.
16560 (ipa_reference_read_optimization_summary): Likewise.
16561 * ipa-utils.c (ipa_merge_profiles): Likewise.
16562 * ipa.c (walk_polymorphic_call_targets): Likewise.
16563 (symbol_table::remove_unreachable_nodes): Likewise.
16564 (ipa_single_use): Likewise.
16565 * passes.c (execute_todo): Likewise.
16566 * predict.c (drop_profile): Likewise.
16567 * symtab.c (symtab_node::get_dump_name): New function.
16568 (symtab_node::dump_name): Likewise.
16569 (symtab_node::dump_asm_name): Likewise.
16570 (symtab_node::dump_references): Likewise.
16571 (symtab_node::dump_referring): Likewise.
16572 (symtab_node::dump_base): Likewise.
16573 (symtab_node::debug_symtab): Likewise.
16574 * tree-sra.c (convert_callers_for_node): Likewise.
16575 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16576 * value-prof.c (init_node_map): Likewise.
16578 2017-05-23 Martin Liska <mliska@suse.cz>
16580 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16581 and symtab_node::debug_symtab to symbol_table::debug.
16582 * cgraphunit.c (analyze_functions): Use the renamed function.
16583 (symbol_table::compile): Likewise.
16584 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16585 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16586 * passes.c (execute_todo): Likewise.
16587 * symtab.c (symbol_table::dump): New function.
16588 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16590 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16592 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16593 that nonconst implies exec.
16595 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16597 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16598 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16599 (inline_edge_summary_vec): Turn into ...
16600 (ipa_call_summaries): ... this one.
16601 (redirect_to_unreachable, edge_set_predicate,
16602 evaluate_properties_for_edge, inline_summary_alloc,
16603 reset_ipa_call_summary, reset_inline_summary,
16604 inline_summary_t::duplicate): Update.
16605 (inline_edge_duplication_hook): Turn to ...
16606 (ipa_call_summary_t::duplicate): ... this one.
16607 (inline_edge_removal_hook): Turn to ...
16608 (ipa_call_summary_t::remove): ... this one.
16609 (dump_inline_edge_summary): Turn to ...
16610 (dump_ipa_call_summary): ... this one.
16611 (estimate_function_body_sizes): Update.
16612 (inline_update_callee_summaries): Update.
16613 (remap_edge_change_prob): Update.
16614 (remap_edge_summaries): Update.
16615 (inline_merge_summary): Update.
16616 (do_estimate_edge_time): Update.
16617 (inline_generate_summary): Update.
16618 (inline_read_section): Update.
16619 (inline_read_summary): Update.
16620 (inline_free_summary): Update.
16621 * ipa-inline.c (can_inline_edge_p): Update.
16622 (compute_inlined_call_time): Update.
16623 (want_inline_small_function_p): Update.
16624 (edge_badness): Update.
16625 (early_inliner): Update.
16626 * ipa-inline.h (inline_edge_summary): Turn to ...
16627 (ipa_call_summary): ... this one.
16628 (ipa_call_summary_t): New class.
16629 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16630 (ipa_call_summaries): New.
16631 (inline_edge_summary): Remove.
16632 (estimate_edge_growth): Update.
16633 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16634 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16635 * ipa-split.c (execute_split_functions): Update.
16636 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16638 2017-05-23 Tom de Vries <tom@codesourcery.com>
16640 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16641 attributes): Document rdrand effective target.
16643 2017-05-23 Tom de Vries <tom@codesourcery.com>
16645 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16646 attributes): Sort alphabetically.
16648 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16650 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16652 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16655 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16656 V2DF/V2DI splat into two separate patterns, one that handles
16657 registers, and the other that only handles memory. Drop support
16658 for splatting from a GPR on ISA 2.07 and then splitting the
16659 splat into direct move and splat.
16660 (vsx_splat_<mode>_reg): Likewise.
16661 (vsx_splat_<mode>_mem): Likewise.
16663 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16665 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16667 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16669 PR middle-end/80809
16670 * omp-low.c (finish_taskreg_remap): New function.
16671 (finish_taskreg_scan): If unit size of ctx->record_type
16672 is non-constant, unshare the size expression and replace
16673 decls in it with possible outer var refs.
16675 PR middle-end/80809
16676 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16677 GOVD_SHARED rather than GOVD_PRIVATE with it.
16678 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16679 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16681 PR middle-end/80853
16682 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16683 as last argument to build_outer_var_ref for pointer bases of array
16684 section reductions.
16686 2017-05-19 Martin Sebor <msebor@redhat.com>
16688 * print-tree.c (print_node): Print DECL_READ_P flag.
16690 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16692 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16693 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16694 * cgraph.c: Likewise.
16695 * cgraphunit.c: Likewise.
16696 * gengtype.c: Likewise.
16697 * ipa-cp.c: Likewise.
16698 * ipa-devirt.c: Likewise.
16699 * ipa-icf.c: Likewise.
16700 * ipa-predicate.c: Likewise.
16701 * ipa-profile.c: Likewise.
16702 * ipa-prop.c: Likewise.
16703 * ipa-split.c: Likewise.
16705 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16706 edge_predicate_pool, dump_inline_hints,
16707 inline_summary::account_size_time, redirect_to_unreachable,
16708 edge_set_predicate, set_hint_predicate,
16709 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16710 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16711 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16712 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16713 ipa_call_summary_t::remove, initialize_growth_caches,
16714 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16715 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16716 mark_modified, unmodified_parm_1, unmodified_parm,
16717 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16718 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16719 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16720 will_be_nonconstant_predicate, record_modified_bb_info,
16721 get_minimal_bb, record_modified, param_change_prob,
16722 phi_result_unknown_predicate, predicate_for_phi_result,
16723 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16724 estimate_function_body_sizes, compute_inline_parameters,
16725 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16726 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16727 inline_update_callee_summaries, remap_edge_change_prob,
16728 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16729 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16730 inline_analyze_function, inline_summary_t::insert,
16731 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16732 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16733 inline_free_summary): Move to ipa-fnsummary.h
16734 (predicate_t): Remove.
16735 * ipa-fnsummary.c: New file.
16736 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16737 (enum inline_hints_vals, inline_hints, agg_position_info,
16738 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16739 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16740 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16741 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16742 inline_read_summary, inline_write_summary, inline_free_summary,
16743 inline_analyze_function, initialize_inline_failed,
16744 inline_merge_summary, inline_update_overall_summary,
16745 compute_inline_parameters): Move to ipa-fnsummary.h
16746 * ipa-fnsummary.h: New file.
16747 * ipa-inline-transform.h: Include ipa-inline.h.
16748 * ipa-inline.c: LIkewise.
16750 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16752 * ipa-inline.c (edge_badness): Use inlined_time instead of
16753 inline_summaries->get.
16755 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16757 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16759 2017-05-22 Nathan Sidwell <nathan@acm.org>
16761 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16762 (fdump-lang): Document 'raw' option.
16763 * dumpfile.h (TDI_tu): Delete.
16764 * dumpfile.c (dump_files): Remove translation-unit.
16765 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16767 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16769 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16770 command option from $(AWK) call.
16771 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16772 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16773 [FORMAT]: Remove handling of variable.
16774 * config/avr/t-multilib: Regenerate.
16776 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16778 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16780 (dump_inline_summary): Do not print self_time.
16781 (estimate_function_body_sizes): Do not set self_time.
16782 (compute_inline_parameters): Likewise.
16783 (inline_read_section, inline_write_summary): Do not stream self_time.
16784 * ipa-inline.h (inline_summary): Drop self_time.
16786 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16788 * ipa-inline-analysis.c (account_size_time): Rename to ...
16789 (inline_summary::account_size_time): ... this one.
16790 (reset_ipa_call_summary): Turn to ...
16791 (ipa_call_summary::reset): ... this one.
16792 (reset_inline_summary): Turn to ...
16793 (inline_summary::reset): ... this one.
16794 (inline_summary_t::remove): Update.
16795 (inline_summary_t::duplicate): Update.
16796 (ipa_call_summary_t::remove): Update.
16797 (dump_inline_summary): Update.
16798 (estimate_function_body_sizes): Update.
16799 (compute_inline_parameters): Update.
16800 (estimate_node_size_and_time): Update.
16801 (inline_merge_summary): Update.
16802 (inline_update_overall_summary): Update.
16803 (inline_read_section): Update.
16804 (inline_write_summary): Update.
16805 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
16806 add account_size_time and reset member functions.
16807 (ipa_call_summary): Add reset function.
16808 * ipa-predicate.h (predicate::operator &): Constify.
16810 2017-05-22 Richard Biener <rguenther@suse.de>
16812 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
16814 2017-05-19 Jason Merrill <jason@redhat.com>
16816 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16818 2017-05-19 Marek Polacek <polacek@redhat.com>
16821 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16822 TYPE_OVERFLOW_WRAPS checks.
16824 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16826 * tree-core.h (enum omp_clause_default_kind): Add
16827 "OMP_CLAUSE_DEFAULT_PRESENT".
16828 * tree-pretty-print.c (dump_omp_clause): Handle it.
16829 * gimplify.c (enum gimplify_omp_var_data): Add
16830 "GOVD_MAP_FORCE_PRESENT".
16831 (gimplify_adjust_omp_clauses_1): Map it to
16832 "GOMP_MAP_FORCE_PRESENT".
16833 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16835 * gimplify.c (oacc_default_clause): Clarify.
16837 2017-05-19 Nathan Sidwell <nathan@acm.org>
16839 LANG_HOOK_REGISTER_DUMPS
16840 * toplev.c (general_init): Call register dump lang hook.
16841 * doc/invoke.texi: Document -fdump-lang option family.
16842 * dumpfile.c (dump_files): Remove class dump here.
16843 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16844 * dumpfile.h (tree_dump_index): Remove TDI_class.
16845 * langhooks-def.h (lhd_register_dumps): Declare.
16846 (LANG_HOOKS_REGISTER_DUMPS): Define.
16847 (LANG_HOOKS_INITIALIZER): Add it.
16848 * langhooks.c (lhd_register_dumps): Define.
16849 * langhooks.h (struct lang_hooks): Add register_dumps.
16851 2017-05-19 Nathan Sidwell <nathan@acm.org>
16853 * context.h (context::set_passes): New.
16854 * context.c (context::context): Do not create pass manager.
16855 * toplev.c (general_init): Create pass manager here.
16857 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16859 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16860 use this splitter if two add or or instructions would also work for
16861 the constant we want to generate.
16863 2017-05-19 Richard Biener <rguenther@suse.de>
16866 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16867 predicate evaluation.
16869 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16871 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16873 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16876 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16878 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16879 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16880 (inline_edge_summary_vec): Turn into ...
16881 (ipa_call_summaries): ... this one.
16882 (redirect_to_unreachable, edge_set_predicate,
16883 evaluate_properties_for_edge, inline_summary_alloc,
16884 reset_ipa_call_summary, reset_inline_summary,
16885 inline_summary_t::duplicate): Update.
16886 (inline_edge_duplication_hook): Turn to ...
16887 (ipa_call_summary_t::duplicate): ... this one.
16888 (inline_edge_removal_hook): Turn to ...
16889 (ipa_call_summary_t::remove): ... this one.
16890 (dump_inline_edge_summary): Turn to ...
16891 (dump_ipa_call_summary): ... this one.
16892 (estimate_function_body_sizes): Update.
16893 (inline_update_callee_summaries): Update.
16894 (remap_edge_change_prob): Update.
16895 (remap_edge_summaries): Update.
16896 (inline_merge_summary): Update.
16897 (do_estimate_edge_time): Update.
16898 (inline_generate_summary): Update.
16899 (inline_read_section): Update.
16900 (inline_read_summary): Update.
16901 (inline_free_summary): Update.
16902 * ipa-inline.c (can_inline_edge_p): Update.
16903 (compute_inlined_call_time): Update.
16904 (want_inline_small_function_p): Update.
16905 (edge_badness): Update.
16906 (early_inliner): Update.
16907 * ipa-inline.h (inline_edge_summary): Turn to ...
16908 (ipa_call_summary): ... this one.
16909 (ipa_call_summary_t): New class.
16910 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16911 (ipa_call_summaries): New.
16912 (inline_edge_summary): Remove.
16913 (estimate_edge_growth): Update.
16914 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16915 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16916 * ipa-split.c (execute_split_functions): Update.
16917 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16919 2017-05-19 Richard Biener <rguenther@suse.de>
16921 PR middle-end/80764
16922 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16924 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16926 * config/rs6000/rs6000.c (struct machine_function): Add field
16927 fpr_is_wrapped_separately.
16928 (rs6000_get_separate_components): Use 64 components. Handle the
16929 new FPR components.
16930 (rs6000_components_for_bb): Handle the FPR components.
16931 (rs6000_emit_prologue_components): Handle the FPR components.
16932 (rs6000_emit_epilogue_components): Handle the FPR components.
16933 (rs6000_set_handled_components): Handle the FPR components.
16934 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16935 that are already separately shrink-wrapped.
16936 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16937 that are already separately shrink-wrapped.
16939 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16942 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16945 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16946 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16947 cases where the register allocator uses a move and an offsettable
16948 memory operation to/from a FPR register on ISA 2.06/2.07.
16949 (define_peephole2 for Altivec d-form store): Likewise.
16951 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16954 * config/i386/mmx.md (*mov<mode>_internal): Enable
16955 alternatives 11, 12, 13 and 14 also for 32bit targets.
16956 Remove alternatives 15, 16, 17 and 18.
16957 * config/i386/sse.md (vec_concatv2di): Change
16958 alternative (!x, *y) to (x, ?!*Yn).
16960 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16962 * dumpfile.h (enum dump_kind): Remove stray comma.
16964 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16966 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16967 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16968 predicate::num_conditions
16969 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16970 (CHANGED): turn into predicate::changed.
16971 (agg_position_info): Move to ipa-predicate.h
16972 (add_condition, predicate::add_clause, predicate::operator &=,
16973 predicate::or_with, predicate::evaluate, predicate::probability,
16974 dump_condition, dump_clause, predicate::dump,
16975 predicate::remap_after_duplication, predicate::remap_after_inlining,
16976 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16977 (evaluate_conditions_for_known_args): Update.
16978 (set_cond_stmt_execution_predicate): Update.
16979 * ipa-inline.h: Include ipa-predicate.h
16980 (condition, inline_param_summary, conditions, agg_position_info,
16981 predicate): Move to ipa-predicate.h
16982 * ipa-predicate.c: New file.
16983 * ipa-predicate.h: New file.
16985 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16987 * final.c (leaf_function_p): Check we are not in a sequence.
16989 2017-05-18 Martin Liska <mliska@suse.cz>
16991 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16992 * dumpfile.c (dump_register): Use new enum dump_kind.
16993 (get_dump_file_name): Likewise.
16994 (dump_enable_all): Likewise.
16995 (dump_switch_p_1): Likewise.
16996 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16997 * dumpfile.h (enum dump_kind): New enum type.
16998 (struct dump_file_info): Create constructor and
16999 format fields and comments.
17000 * passes.c (pass_manager::register_one_dump_file):
17002 * statistics.c (statistics_early_init): Likewise.
17003 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17004 TDF_TREE with TDF_SLIM.
17005 (gather_memory_references_ref): Likewise.
17007 2017-05-18 Martin Liska <mliska@suse.cz>
17009 * vec.h (struct vnull): Use it.
17011 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17013 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17014 (true_predicate, false_predicate, true_predicate_p,
17015 false_predicate_p): Remove.
17016 (single_cond_predicate, not_inlined_predicate): Turn to member function
17018 (add_condition): Update.
17019 (add_clause): Turn to...
17020 (predicate::add_clause): ... this one; update; allow passing NULL
17022 (and_predicates): Turn to ...
17023 (predicate::operator &=): ... this one.
17024 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17025 (or_predicates): Turn to ...
17026 (predicate::or_with): ... this one.
17027 (evaluate_predicate): Turn to ...
17028 (predicate::evaluate): ... this one.
17029 (predicate_probability): Turn to ...
17030 (predicate::probability): ... this one.
17031 (dump_condition): Update.
17032 (dump_predicate): Turn to ...
17033 (predicate::dump): ... this one.
17034 (account_size_time): Update.
17035 (edge_set_predicate): Update.
17036 (set_hint_predicate): UPdate.
17037 (evaluate_conditions_for_known_args): Update.
17038 (evaluate_properties_for_edge): Update.
17039 (remap_predicate_after_duplication): Turn to...
17040 (predicate::remap_after_duplication): ... this one.
17041 (remap_hint_predicate_after_duplication): Update.
17042 (inline_summary_t::duplicate): UPdate.
17043 (dump_inline_edge_summary): Update.
17044 (dump_inline_summary): Update.
17045 (set_cond_stmt_execution_predicate): Update.
17046 (set_switch_stmt_execution_predicate): Update.
17047 (compute_bb_predicates): Update.
17048 (will_be_nonconstant_expr_predicate): Update.
17049 (will_be_nonconstant_predicate): Update.
17050 (phi_result_unknown_predicate): Update.
17051 (predicate_for_phi_result): Update.
17052 (array_index_predicate): Update.
17053 (estimate_function_body_sizes): Update.
17054 (estimate_node_size_and_time): Update.
17055 (estimate_ipcp_clone_size_and_time): Update.
17056 (remap_predicate): Rename to ...
17057 (predicate::remap_after_inlining): ... this one.
17058 (remap_hint_predicate): Update.
17059 (inline_merge_summary): Update.
17060 (inline_update_overall_summary): Update.
17061 (estimate_size_after_inlining): Update.
17062 (read_predicate): Rename to ...
17063 (predicate::stream_in): ... this one.
17064 (read_inline_edge_summary): Update.
17065 (write_predicate): Rename to ...
17066 (predicate::stream_out): ... this one.
17067 (write_inline_edge_summary): Update.
17068 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17069 (clause_t): Turn to uint32_t
17070 (predicate): Turn to class; implement constructor and operators
17072 (size_time_entry): Update.
17073 (inline_summary): Update.
17074 (inline_edge_summary): Update.
17076 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17078 * fold-const.c (fold_binary_loc): Move transformation...
17079 * match.pd (C - X CMP X): ... here.
17081 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17083 * config/sparc/sparc.c (sparc_option_override): Set function
17084 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17085 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17087 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17089 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17091 2017-05-18 Marek Polacek <polacek@redhat.com>
17094 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17095 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17096 gimple_assign_load_p.
17098 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17100 PR middle-end/80692
17101 * real.c (do_compare): Give decimal_do_compare preference over
17102 comparing just the signs.
17104 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17106 * doc/md.texi (Canonicalization of Instructions): Describe the
17107 canonical form of instructions that inherently set a condition
17110 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17112 PR middle-end/80775
17113 * tree-cfg.c: Move deletion of unreachable case statements to after
17114 the merging of consecutive case labels.
17116 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17118 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17119 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17120 restoring of callee-saved registers.
17122 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17124 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17125 * config/visium/visium.c (single_set_and_flags): Likewise.
17126 * config/visium/visium.md (Substitutions): Likewise.
17128 2017-05-17 Martin Liska <mliska@suse.cz>
17130 * cfg.c: Introduce dump_flags_t type and
17131 use it instead of int type.
17133 * cfghooks.c: Likewise.
17134 * cfghooks.h (struct cfg_hooks): Likewise.
17135 * cfgrtl.c: Likewise.
17136 * cfgrtl.h: Likewise.
17137 * cgraph.c (cgraph_node::get_body): Likewise.
17138 * coretypes.h: Likewise.
17139 * domwalk.c: Likewise.
17140 * domwalk.h: Likewise.
17141 * dumpfile.c (struct dump_option_value_info): Likewise.
17142 (dump_enable_all): Likewise.
17143 (dump_switch_p_1): Likewise.
17144 (opt_info_switch_p): Likewise.
17145 * dumpfile.h (enum tree_dump_index): Likewise.
17146 (struct dump_file_info): Likewise.
17147 * genemit.c: Likewise.
17148 * generic-match-head.c: Likewise.
17149 * gengtype.c (open_base_files): Likewise.
17150 * gimple-pretty-print.c: Likewise.
17151 * gimple-pretty-print.h: Likewise.
17152 * graph.c (print_graph_cfg): Likewise.
17153 * graphite-scop-detection.c (dot_all_sese): Likewise.
17154 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17155 * loop-unroll.c (report_unroll): Likewise.
17156 * passes.c (pass_manager::register_one_dump_file): Likewise.
17157 * print-tree.c: Likewise.
17158 * statistics.c: Likewise.
17159 * tree-cfg.c: Likewise.
17160 * tree-cfg.h: Likewise.
17161 * tree-dfa.c: Likewise.
17162 * tree-dfa.h: Likewise.
17163 * tree-dump.c (dump_function): Likewise.
17164 * tree-dump.h (struct dump_info): Likewise.
17165 * tree-pretty-print.c: Likewise.
17166 * tree-pretty-print.h: Likewise.
17167 * tree-ssa-live.c: Likewise.
17168 * tree-ssa-live.h: Likewise.
17169 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17170 * tree-vect-loop.c: Likewise.
17171 * tree-vect-slp.c: Likewise.
17173 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17174 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17176 PR tree-optimization/80457
17177 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17178 of all arguments to a statement as scalar_to_vec operations.
17179 (vectorizable_call): Adjust call to vect_model_simple_cost for
17181 (vectorizable_conversion): Likewise.
17182 (vectorizable_assignment): Likewise.
17183 (vectorizable_shift): Likewise.
17184 (vectorizable_operation): Likewise.
17185 (vectorizable_comparison): Likewise.
17186 (vect_is_simple_cond): Record the def types for operands.
17187 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17188 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17189 for statement argument count.
17191 2017-05-16 Carl Love <cel@us.ibm.com>
17193 * config/rs6000/rs6000-c: Add support for built-in functions
17194 vector unsigned long long vec_bperm (vector unsigned long long,
17195 vector unsigned char)
17196 vector signed long long vec_mule (vector signed int,
17198 vector unsigned long long vec_mule (vector unsigned int,
17199 vector unsigned int)
17200 vector signed long long vec_mulo (vector signed int,
17202 vector unsigned long long vec_mulo (vector unsigned int,
17203 vector unsigned int)
17204 vector signed char vec_sldw (vector signed char,
17205 vector signed char,
17207 vector unsigned char vec_sldw (vector unsigned char,
17208 vector unsigned char,
17210 vector signed short vec_sldw (vector signed short,
17211 vector signed short,
17213 vector unsigned short vec_sldw (vector unsigned short,
17214 vector unsigned short,
17216 vector signed int vec_sldw (vector signed int,
17219 vector unsigned int vec_sldw (vector unsigned int,
17220 vector unsigned int,
17222 vector signed long long vec_sldw (vector signed long long,
17223 vector signed long long,
17225 vector unsigned long long vec_sldw (vector unsigned long long,
17226 vector unsigned long long,
17228 * config/rs6000/rs6000-c: Add support for built-in functions
17229 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17230 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17231 * doc/extend.texi: Update the built-in documentation for the
17232 new built-in functions.
17234 2017-05-16 Marek Polacek <polacek@redhat.com>
17238 * tree.c (save_expr): Don't fold the expression.
17240 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17242 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17243 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17244 and (?*y,m). Update insn attributes.
17246 2017-05-16 Martin Liska <mliska@suse.cz>
17248 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17249 flags argument of print_gimple_stmt, print_gimple_expr,
17250 print_generic_stmt and print_generic_expr.
17251 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17252 * coretypes.h: Likewise.
17253 * except.c (dump_eh_tree): Likewise.
17254 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17255 * gimple-pretty-print.h: Likewise.
17256 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17257 (backprop::push_to_worklist): Likewise.
17258 (backprop::pop_from_worklist): Likewise.
17259 (backprop::process_use): Likewise.
17260 (backprop::intersect_uses): Likewise.
17261 (note_replacement): Likewise.
17262 * gimple-ssa-store-merging.c
17263 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17264 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17265 (pass_store_merging::execute): Likewise.
17266 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17267 (ssa_base_cand_dump_callback): Likewise.
17268 (dump_incr_vec): Likewise.
17269 (replace_refs): Likewise.
17270 (replace_mult_candidate): Likewise.
17271 (create_add_on_incoming_edge): Likewise.
17272 (create_phi_basis): Likewise.
17273 (insert_initializers): Likewise.
17274 (all_phi_incrs_profitable): Likewise.
17275 (introduce_cast_before_cand): Likewise.
17276 (replace_one_candidate): Likewise.
17277 * gimplify.c (gimplify_expr): Likewise.
17278 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17279 (set_rename): Likewise.
17280 (rename_uses): Likewise.
17281 (copy_loop_phi_nodes): Likewise.
17282 (add_close_phis_to_merge_points): Likewise.
17283 (copy_loop_close_phi_args): Likewise.
17284 (copy_cond_phi_args): Likewise.
17285 (graphite_copy_stmts_from_block): Likewise.
17286 (translate_pending_phi_nodes): Likewise.
17287 * graphite-poly.c (print_pdr): Likewise.
17288 (dump_gbb_cases): Likewise.
17289 (dump_gbb_conditions): Likewise.
17290 (print_scop_params): Likewise.
17291 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17292 (build_cross_bb_scalars_use): Likewise.
17293 (gather_bbs::before_dom_children): Likewise.
17294 * hsa-dump.c (dump_hsa_immed): Likewise.
17295 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17296 (get_replacement_map): Likewise.
17297 * ipa-inline-analysis.c (dump_condition): Likewise.
17298 (estimate_function_body_sizes): Likewise.
17299 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17300 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17301 * ipa-prop.c (ipa_dump_param): Likewise.
17302 (ipa_print_node_jump_functions_for_edge): Likewise.
17303 (ipa_modify_call_arguments): Likewise.
17304 (ipa_modify_expr): Likewise.
17305 (ipa_dump_param_adjustments): Likewise.
17306 (ipa_dump_agg_replacement_values): Likewise.
17307 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17308 * ipa-pure-const.c (check_stmt): Likewise.
17309 (pass_nothrow::execute): Likewise.
17310 * ipa-split.c (execute_split_functions): Likewise.
17311 * omp-offload.c (dump_oacc_loop_part): Likewise.
17312 (dump_oacc_loop): Likewise.
17313 * trans-mem.c (tm_log_emit): Likewise.
17314 (tm_memopt_accumulate_memops): Likewise.
17315 (dump_tm_memopt_set): Likewise.
17316 (dump_tm_memopt_transform): Likewise.
17317 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17318 (print_loop): Likewise.
17319 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17320 (chkp_gather_checks_info): Likewise.
17321 (chkp_get_check_result): Likewise.
17322 (chkp_remove_check_if_pass): Likewise.
17323 (chkp_use_outer_bounds_if_possible): Likewise.
17324 (chkp_reduce_bounds_lifetime): Likewise.
17325 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17326 (chkp_mark_completed_bounds): Likewise.
17327 (chkp_register_incomplete_bounds): Likewise.
17328 (chkp_mark_invalid_bounds): Likewise.
17329 (chkp_maybe_copy_and_register_bounds): Likewise.
17330 (chkp_build_returned_bound): Likewise.
17331 (chkp_get_bound_for_parm): Likewise.
17332 (chkp_build_bndldx): Likewise.
17333 (chkp_get_bounds_by_definition): Likewise.
17334 (chkp_generate_extern_var_bounds): Likewise.
17335 (chkp_get_bounds_for_decl_addr): Likewise.
17336 * tree-chrec.c (chrec_apply): Likewise.
17337 * tree-data-ref.c (dump_data_reference): Likewise.
17338 (dump_subscript): Likewise.
17339 (dump_data_dependence_relation): Likewise.
17340 (analyze_overlapping_iterations): Likewise.
17341 * tree-inline.c (expand_call_inline): Likewise.
17342 (tree_function_versioning): Likewise.
17343 * tree-into-ssa.c (dump_defs_stack): Likewise.
17344 (dump_currdefs): Likewise.
17345 (dump_names_replaced_by): Likewise.
17346 (dump_update_ssa): Likewise.
17347 (update_ssa): Likewise.
17348 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17349 * tree-parloops.c (build_new_reduction): Likewise.
17350 (try_create_reduction_list): Likewise.
17351 (ref_conflicts_with_region): Likewise.
17352 (oacc_entry_exit_ok_1): Likewise.
17353 (oacc_entry_exit_single_gang): Likewise.
17354 * tree-pretty-print.h: Likewise.
17355 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17356 (get_scalar_evolution): Likewise.
17357 (add_to_evolution): Likewise.
17358 (get_loop_exit_condition): Likewise.
17359 (analyze_evolution_in_loop): Likewise.
17360 (analyze_initial_condition): Likewise.
17361 (analyze_scalar_evolution): Likewise.
17362 (instantiate_scev): Likewise.
17363 (number_of_latch_executions): Likewise.
17364 (gather_chrec_stats): Likewise.
17365 (final_value_replacement_loop): Likewise.
17366 (scev_const_prop): Likewise.
17367 * tree-sra.c (dump_access): Likewise.
17368 (disqualify_candidate): Likewise.
17369 (create_access): Likewise.
17370 (reject): Likewise.
17371 (maybe_add_sra_candidate): Likewise.
17372 (create_access_replacement): Likewise.
17373 (analyze_access_subtree): Likewise.
17374 (analyze_all_variable_accesses): Likewise.
17375 (sra_modify_assign): Likewise.
17376 (initialize_constant_pool_replacements): Likewise.
17377 (find_param_candidates): Likewise.
17378 (decide_one_param_reduction): Likewise.
17379 (replace_removed_params_ssa_names): Likewise.
17380 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17381 * tree-ssa-copy.c (dump_copy_of): Likewise.
17382 (copy_prop_visit_cond_stmt): Likewise.
17383 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17384 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17385 (record_equivalences_from_stmt): Likewise.
17386 * tree-ssa-dse.c (compute_trims): Likewise.
17387 (delete_dead_call): Likewise.
17388 (delete_dead_assignment): Likewise.
17389 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17390 (forward_propagate_into_cond): Likewise.
17391 (pass_forwprop::execute): Likewise.
17392 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17393 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17395 (move_computations_worker): Likewise.
17396 (execute_sm): Likewise.
17397 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17398 (remove_exits_and_undefined_stmts): Likewise.
17399 (remove_redundant_iv_tests): Likewise.
17400 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17401 (adjust_iv_update_pos): Likewise.
17402 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17403 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17404 (value_replacement): Likewise.
17405 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17406 * tree-ssa-pre.c (print_pre_expr): Likewise.
17407 (get_representative_for): Likewise.
17408 (create_expression_by_pieces): Likewise.
17409 (insert_into_preds_of_block): Likewise.
17410 (eliminate_insert): Likewise.
17411 (eliminate_dom_walker::before_dom_children): Likewise.
17412 (eliminate): Likewise.
17413 (remove_dead_inserted_code): Likewise.
17414 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17415 * tree-ssa-reassoc.c (get_rank): Likewise.
17416 (eliminate_duplicate_pair): Likewise.
17417 (eliminate_plus_minus_pair): Likewise.
17418 (eliminate_not_pairs): Likewise.
17419 (undistribute_ops_list): Likewise.
17420 (eliminate_redundant_comparison): Likewise.
17421 (update_range_test): Likewise.
17422 (optimize_range_tests_var_bound): Likewise.
17423 (optimize_vec_cond_expr): Likewise.
17424 (rewrite_expr_tree): Likewise.
17425 (rewrite_expr_tree_parallel): Likewise.
17426 (linearize_expr): Likewise.
17427 (break_up_subtract): Likewise.
17428 (linearize_expr_tree): Likewise.
17429 (attempt_builtin_powi): Likewise.
17430 (attempt_builtin_copysign): Likewise.
17431 (transform_stmt_to_copy): Likewise.
17432 (transform_stmt_to_multiply): Likewise.
17433 (dump_ops_vector): Likewise.
17434 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17435 (print_scc): Likewise.
17436 (set_ssa_val_to): Likewise.
17437 (visit_reference_op_store): Likewise.
17438 (visit_use): Likewise.
17439 (sccvn_dom_walker::before_dom_children): Likewise.
17440 (run_scc_vn): Likewise.
17441 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17443 (expr_hash_elt::print): Likewise.
17444 (const_and_copies::pop_to_marker): Likewise.
17445 (const_and_copies::record_const_or_copy_raw): Likewise.
17446 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17447 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17448 (dump_predicates): Likewise.
17449 (find_uninit_use): Likewise.
17450 (warn_uninitialized_phi): Likewise.
17451 (pass_late_warn_uninitialized::execute): Likewise.
17452 * tree-ssa.c (verify_vssa): Likewise.
17453 (verify_ssa): Likewise.
17454 (maybe_optimize_var): Likewise.
17455 * tree-vrp.c (dump_value_range): Likewise.
17456 (dump_all_value_ranges): Likewise.
17457 (dump_asserts_for): Likewise.
17458 (register_edge_assert_for_2): Likewise.
17459 (vrp_visit_cond_stmt): Likewise.
17460 (vrp_visit_switch_stmt): Likewise.
17461 (vrp_visit_stmt): Likewise.
17462 (vrp_visit_phi_node): Likewise.
17463 (simplify_cond_using_ranges_1): Likewise.
17464 (fold_predicate_in): Likewise.
17465 (evrp_dom_walker::before_dom_children): Likewise.
17466 (evrp_dom_walker::push_value_range): Likewise.
17467 (evrp_dom_walker::pop_value_range): Likewise.
17468 (execute_early_vrp): Likewise.
17470 2017-05-16 Richard Biener <rguenther@suse.de>
17472 * dwarf2out.c (loc_list_from_tree_1): Do not create
17473 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17475 2017-05-16 Richard Biener <rguenther@suse.de>
17477 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17479 (note_variable_value_in_expr): If we resolved the decl ref
17480 do not push to the stack.
17482 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17484 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17485 operations in fast-math mode.
17486 (vaddq_f16): Likewise.
17487 (vmul_f16): Likewise.
17488 (vmulq_f16): Likewise.
17489 (vsub_f16): Likewise.
17490 (vsubq_f16): Likewise.
17491 * config/arm/neon.md (add<mode>3): New.
17493 (fma:<VH:mode>3): New. Also remove outdated comment.
17496 2017-05-16 Martin Liska <mliska@suse.cz>
17500 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17501 (odr_types_equivalent_p): Likewise.
17503 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17505 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17507 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17510 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17511 non-interunit SSE move alternatives with '?'.
17512 (zero-extendsidi peephole2): New peephole to skip intermediate
17513 general register in SSE zero-extend sequence.
17515 2017-05-15 Jeff Law <law@redhat.com>
17517 * reorg.c (relax_delay_slots): Create a new variable to hold
17518 the temporary target rather than clobbering TARGET_LABEL.
17520 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17521 missing argument to extract_bit_field call.
17522 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17524 2017-05-15 Martin Liska <mliska@suse.cz>
17527 * gcc.c (process_command): Do not allow empty argument of -o option.
17529 2017-05-15 Renlin Li <renlin.li@arm.com>
17531 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17532 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17533 * config/aarch64/constraints.md (Usf): Add long call check.
17534 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17535 (call_value): Likewise.
17536 (sibcall): Likewise.
17537 (sibcall_value): Likewise.
17539 (call_value_insn): New.
17540 (sibcall_insn): Update rtx pattern.
17541 (sibcall_value_insn): Likewise.
17542 (call_internal): Remove.
17543 (call_value_internal): Likewise.
17544 (sibcall_internal): Likewise.
17545 (sibcall_value_internal): Likewise.
17546 (call_reg): Likewise.
17547 (call_symbol): Likewise.
17548 (call_value_reg): Likewise.
17549 (call_value_symbol): Likewise.
17551 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17554 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17556 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17558 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17559 compatible with CCGOCmode and with CCZmode.
17561 2017-05-14 Martin Sebor <msebor@redhat.com>
17563 PR middle-end/77671
17564 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17565 (gimple_fold_builtin_snprintf): Same.
17566 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17567 (gimple_fold_builtin_snprintf): Same.
17568 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17569 of character types.
17570 (is_call_safe): New function.
17571 (try_substitute_return_value): Call it.
17572 (try_simplify_call): New function.
17573 (pass_sprintf_length::handle_gimple_call): Call it.
17575 2017-05-14 Martin Sebor <msebor@redhat.com>
17577 PR middle-end/80669
17578 * builtins.c (expand_builtin_stpncpy): Simplify.
17580 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17582 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17583 * config/i386/i386.h
17584 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17585 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17586 (struct machine_function): Add new members call_ms2sysv,
17587 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17588 (struct machine_frame_state): New fields sp_realigned and
17589 sp_realigned_offset.
17590 * config/i386/i386.c
17591 (enum xlogue_stub): New enum.
17592 (enum xlogue_stub_sets): New enum.
17593 (class xlogue_layout): New class.
17594 (struct ix86_frame): New fields stack_realign_allocate_offset,
17595 stack_realign_offset and outlined_save_offset. Modify comments to
17596 detail stack layout when using out-of-line stubs.
17597 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17598 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17599 -mcall-ms2sysv-xlogues.
17600 (stub_managed_regs): New static variable.
17601 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17602 registers managed by out-of-line stub.
17603 (disable_call_ms2sysv_xlogues): New function.
17604 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17605 m->call_ms2sysv when appropriate and compute frame layout for
17607 (sp_valid_at, fp_valid_at): New inline functions.
17608 (choose_basereg): New function.
17609 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17611 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17612 Use align parameter of choose_baseaddr to generated aligned SSE movs
17614 (pro_epilogue_adjust_stack): Modify to track
17615 machine_frame_state::sp_realigned.
17616 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17617 (ix86_nsaved_sseregs): Likewise.
17618 (ix86_emit_save_regs): Likewise.
17619 (ix86_emit_save_regs_using_mov): Likewise.
17620 (ix86_emit_save_sse_regs_using_mov): Likewise.
17621 (get_scratch_register_on_entry): Likewise.
17622 (gen_frame_set): New function.
17623 (gen_frame_load): Likewise.
17624 (gen_frame_store): Likewise.
17625 (emit_outlined_ms2sysv_save): Likewise.
17626 (emit_outlined_ms2sysv_restore): Likewise.
17627 (ix86_expand_prologue): Modify stack re-alignment code and call
17628 emit_outlined_ms2sysv_save when appropriate.
17629 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17630 parameter rtx_insn *insn, which allows the function to be used to only
17631 generate the notes.
17632 (ix86_expand_epilogue): Modify validity checks of frame and stack
17633 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17634 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17635 * config/i386/predicates.md
17636 (save_multiple): New predicate.
17637 (restore_multiple): Likewise.
17638 * config/i386/sse.md
17639 (save_multiple<mode>): New pattern.
17640 (save_multiple_realign<mode>): Likewise.
17641 (restore_multiple<mode>): Likewise.
17642 (restore_multiple_and_return<mode>): Likewise.
17643 (restore_multiple_leave_return<mode>): Likewise.
17644 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17646 2017-05-14 Julia Koval <julia.koval@intel.com>
17648 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17649 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17650 (__builtin_ia32_xsetbv): New builtins.
17651 * config/i386/i386.c (ix86_expand_special_args_builtin):
17653 (ix86_expand_builtin): Special expand for new intrinsics.
17654 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17655 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17656 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17658 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17660 * cfganal.c (inverted_post_order_compute): Change argument type
17662 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17663 * df-core.c (rest_of_handle_df_initialize): Adjust.
17664 (rest_of_handle_df_finish): Likewise.
17665 (df_analyze_1): Likewise.
17666 (df_analyze): Likewise.
17667 (loop_inverted_post_order_compute): Change argument to be a vec *.
17668 (df_analyze_loop): Adjust.
17669 (df_get_n_blocks): Likewise.
17670 (df_get_postorder): Likewise.
17671 * df.h (struct df_d): Change field to be a vec.
17672 * lcm.c (compute_laterin): Adjust.
17673 (compute_available): Likewise.
17674 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17675 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17676 * tree-ssa-pre.c (compute_antic): Likewise.
17678 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17680 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17681 (depth_first_search::depth_first_search): Change structure init
17682 function to this constructor.
17683 (depth_first_search::add_bb): Rename function to this member.
17684 (depth_first_search::execute): Likewise.
17685 (flow_dfs_compute_reverse_finish): Adjust.
17687 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17689 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17690 (longest_simple_path): Likewise.
17691 * shrink-wrap.c (spread_components): Likewise.
17692 (disqualify_problematic_components): Likewise.
17693 (emit_common_heads_for_components): Likewise.
17694 (emit_common_tails_for_components): Likewise.
17695 (insert_prologue_epilogue_for_components): Likewise.
17697 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17699 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17702 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17704 * df-core.c (df_set_blocks): Start using auto_bitmap.
17705 (df_compact_blocks): Likewise.
17706 * df-problems.c (df_rd_confluence_n): Likewise.
17707 * df-scan.c (df_insn_rescan_all): Likewise.
17708 (df_process_deferred_rescans): Likewise.
17709 (df_update_entry_block_defs): Likewise.
17710 (df_update_exit_block_uses): Likewise.
17711 (df_entry_block_bitmap_verify): Likewise.
17712 (df_exit_block_bitmap_verify): Likewise.
17713 (df_scan_verify): Likewise.
17714 * lra-constraints.c (lra_constraints): Likewise.
17715 (undo_optional_reloads): Likewise.
17716 (lra_undo_inheritance): Likewise.
17717 * lra-remat.c (calculate_gen_cands): Likewise.
17718 (do_remat): Likewise.
17719 * lra-spills.c (assign_spill_hard_regs): Likewise.
17720 (spill_pseudos): Likewise.
17721 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17722 (bitmap_set_subtract_values): Likewise.
17724 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17726 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17727 management with auto_bitmap.
17728 (fix_inter_tick): Likewise.
17729 (fix_recovery_deps): Likewise.
17730 * ira.c (add_store_equivs): Likewise.
17731 (find_moveable_pseudos): Likewise.
17732 (split_live_ranges_for_shrink_wrap): Likewise.
17733 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17734 (rtx_reuse_manager::seen_def_p): Likewise.
17735 (rtx_reuse_manager::set_seen_def): Likewise.
17736 * print-rtl.h (class rtx_reuse_manager): Likewise.
17738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17740 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17742 (migrate_btr_def): Likewise.
17743 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17744 * df-core.c (loop_post_order_compute): Likewise.
17745 (loop_inverted_post_order_compute): Likewise.
17746 * hsa-common.h: Likewise.
17747 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17748 * init-regs.c (initialize_uninitialized_regs): Likewise.
17749 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17750 (inline_small_functions): Likewise.
17751 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17752 * ira.c (combine_and_move_insns): Likewise.
17753 (build_insn_chain): Likewise.
17754 * loop-invariant.c (find_invariants): Likewise.
17755 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17756 * predict.c (tree_predict_by_opcode): Likewise.
17757 (predict_paths_leading_to): Likewise.
17758 (predict_paths_leading_to_edge): Likewise.
17759 (estimate_loops_at_level): Likewise.
17760 (estimate_loops): Likewise.
17761 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17762 (spread_components): Likewise.
17763 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17764 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17765 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17766 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17767 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17768 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17769 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17770 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17771 (mark_threaded_blocks): Likewise.
17772 (thread_through_all_blocks): Likewise.
17773 * tree-ssa.c (verify_ssa): Likewise.
17774 (execute_update_addresses_taken): Likewise.
17775 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17777 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17779 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17781 (post_order_compute): Likewise.
17782 (inverted_post_order_compute): Likewise.
17783 (pre_and_rev_post_order_compute_fn): Likewise.
17785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17787 * genrecog.c (int_set::int_set): Explicitly construct our
17788 auto_vec base class.
17789 * vec.h (auto_vec::auto_vec): New constructor.
17791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17793 * bitmap.h (class auto_bitmap): New constructor taking
17794 bitmap_obstack * argument.
17796 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17798 * bitmap.h (class auto_bitmap): Change type of m_bits to
17799 bitmap_head, and adjust ctor / dtor and member operators.
17801 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
17803 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
17804 when returned register mode doesn't match original mode.
17806 2017-05-12 Jeff Law <law@redhat.com>
17807 Jakub Jelinek <jakub@redhat.com>
17809 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
17810 we look for cc setter after the compare-elim changes.
17811 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
17812 within the vector to match what compare-elim now expects.
17813 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
17814 (xorsi3_flags, one_cmplsi2_flags): Likewise.
17816 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17817 after the compare-elim changes.
17818 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17819 the vector to match what compare-elim now expects.
17820 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17821 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17822 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17823 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17824 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17826 * config/visium/visium.c (single_set_and_flags): Fix where
17827 we look for cc setter after the compare-elim changes.
17828 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17829 with the vector to match what compare-elim now expects.
17830 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17831 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17832 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17833 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17834 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17835 (neg<mode>2_insn_set_overflow): Likewise.
17837 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17839 PR middle-end/79794
17840 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17841 maybe_expand_insn call, set ops[0].target. If still set after call,
17842 set alt_rtl. Add extra arg to recursive calls.
17843 (extract_bit_field): Add alt_rtl argument. Pass to
17845 * expmed.h (extract_bit_field): Fix prototype.
17846 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17847 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17848 to extract_bit_field_calls.
17849 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17850 Pass alt_rtl to extract_bit_field calls.
17851 * calls.c (store_unaligned_arguments_into_psuedos)
17852 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17853 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17855 * optabs.h (struct expand_operand): Add target bitfield.
17857 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17859 * compare-elim.c (try_eliminate_compare): Canonicalize
17860 operation with embedded compare to
17861 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17862 (set (reg) (operation)].
17864 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17866 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17869 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17870 cost of adding a carry flag for ADC instruction.
17871 [case MINUS]: Ignore the cost of subtracting a carry flag
17872 for SBB instruction.
17874 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17876 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17878 * config/rs6000/bmiintrin.h: New file.
17879 * config/rs6000/bmi2intrin.h: New file.
17880 * config/rs6000/x86intrin.h: New file.
17882 2017-05-12 Jeff Law <law@redhat.com>
17884 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17887 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17889 PR middle-end/80707
17890 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17892 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17894 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17895 early expansion of vector divide builtins.
17896 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17897 builtins identified as having unsigned arguments.
17899 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17901 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17902 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17903 expansion of vector logical operations (and, andc, or, xor,
17906 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17908 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17909 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17911 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17914 early GIMPLE expansion of vector multiplies.
17916 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17918 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17919 TARGET_HAVE_MOVT conditional.
17920 (movt splitter): Likewise.
17922 2017-05-12 Richard Biener <rguenther@suse.de>
17924 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17925 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17926 Fold all stmts not inplace.
17928 2017-05-12 Richard Biener <rguenther@suse.de>
17930 PR tree-optimization/80713
17931 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17932 inserted_exprs bit for not removed stmts.
17934 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17936 PR middle-end/69921
17937 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17938 parallelized" attribute for parallelized OpenACC kernels.
17939 * omp-offload.c (execute_oacc_device_lower): Use it.
17941 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17942 Set "oacc kernels" attribute.
17943 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17944 parameter. Adjust all users.
17945 (oacc_fn_attrib_kernels_p): Remove function.
17946 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17947 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17948 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17949 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17950 assert "oacc kernels" attribute is set.
17952 2017-05-11 Carl Love <cel@us.ibm.com>
17954 * config/rs6000/rs6000-c: Add support for built-in functions
17955 vector unsigned char vec_popcnt (vector signed char)
17956 vector unsigned char vec_popcnt (vector unsigned char)
17957 vector unsigned short vec_popcnt (vector signed short)
17958 vector unsigned short vec_popcnt (vector unsigned short)
17959 vector unsigned int vec_popcnt (vector signed int)
17960 vector unsigned int vec_popcnt (vector unsigned int)
17961 vector unsigned long long vec_popcnt (vector signed long long)
17962 vector unsigned long long vec_popcnt (vector unsigned long long)
17963 vector signed long long vec_slo (vector signed long long,
17964 vector signed char)
17965 vector signed long long vec_slo (vector signed long long,
17966 vector unsigned char)
17967 vector unsigned long long vec_slo (vector unsigned long long,
17968 vector signed char)
17969 vector unsigned long long vec_slo (vector unsigned long long,
17970 vector unsigned char)
17971 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17972 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17973 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17974 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17975 * doc/extend.texi: Update the built-in documentation file for the
17976 new built-in functions.
17978 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17980 * attribs.h (sorted_attr_string): Move machine independent
17981 functions for target clone support from the i386 port to common
17982 code. Rename ix86_function_versions to common_function_versions.
17983 Rename make_name to make_unique_name.
17984 (common_function_versions): Likewise.
17985 (make_unique_name): Likewise.
17986 (make_dispatcher_decl): Likewise.
17987 (is_function_default_version): Likewise.
17988 * attribs.c (attr_strcmp): Likewise.
17989 (sorted_attr_string): Likewise.
17990 (common_function_versions): Likewise.
17991 (make_unique_name): Likewise.
17992 (make_dispatcher_decl): Likewise.
17993 (is_function_default_version): Likewise.
17994 * config/i386/i386.c (attr_strcmp): Likewise.
17995 (sorted_attr_string): Likewise.
17996 (ix86_function_versions): Likewise.
17997 (make_name): Likewise.
17998 (make_dispatcher_decl): Likewise.
17999 (is_function_default_version): Likewise.
18000 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18002 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18005 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18006 Account for direct move costs for vec_construct of integer
18009 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18012 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18013 (UNSPEC_STX_ATOMIC): Ditto.
18014 (loaddi_via_sse): New insn.
18015 (storedi_via_sse): Ditto.
18016 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18017 Update corresponding peephole2 patterns.
18018 (atomic_storedi_fpu): Ditto.
18020 2017-05-11 Julia Koval <julia.koval@intel.com>
18022 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18023 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18025 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18026 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18027 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18029 2017-05-11 Nathan Sidwell <nathan@acm.org>
18031 * graphite-poly.c: Include dumpfile.h.
18033 * dumpfle.h (dump_function): Declare here ...
18034 * tree-dump.h (dump_function): ... not here.
18035 * dumpfile.c: #include tree-cfg.h.
18036 (dump_function): Move here from ...
18037 * tree-dump.c (dump_function): ... here.
18038 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18039 * graphite-poly.c: Don't include tree-dump.h.
18040 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18041 * print-tree.c: Likewise.
18042 * stor-layout.c: Likewise.
18043 * tree-nested.c: Likewise.
18045 * dumpfile.c (dump_start): Use TDF_FLAGS.
18046 (dump_enable_all): Fix TDF_KIND check thinko.
18048 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18051 array entries to represent two legal parameterizations of the
18052 overloaded __builtin_cmpb function, as represented by the
18053 P6_OV_BUILTIN_CMPB constant.
18054 (altivec_resolve_overloaded_builtin): Add special case handling
18055 for the __builtin_cmpb function, as represented by the
18056 P6_OV_BUILTIN_CMPB constant.
18057 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18058 (BU_P6_64BIT_2): New macro.
18059 (BU_P6_OVERLOAD_2): New macro
18060 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18061 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18062 (CMPB): Add overload support to represent both 32-bit and 64-bit
18063 compare-bytes function.
18064 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18065 support for TARGET_CMPB.
18066 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18067 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18068 documentation of the __builtin_cmpb overloaded built-in function.
18070 2017-05-11 Richard Biener <rguenther@suse.de>
18072 PR tree-optimization/80705
18073 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18074 bases are not vectorizable.
18076 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18078 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18079 when counting register pressure.
18081 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18083 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18084 (iv_ca_more_deps): Renamed to ...
18085 (iv_ca_compare_deps): ... this.
18086 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18090 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18092 (determine_group_iv_costs): ... here.
18093 (find_inv_vars_cb): Record inv var if it's not recorded before.
18095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18097 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18098 (get_shiftadd_cost): Ditto.
18100 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18102 * tree-ssa-address.c: Include header file.
18103 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18105 (add_to_parts): Refactor.
18106 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18107 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18110 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18112 PR tree-optimization/53090
18113 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18115 (extract_cond_operands): Detect condition with IV on both sides
18116 and return COMP_IV_EXPR_2.
18117 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18118 (rewrite_use_compare): Simplify by removing call to function
18119 extract_cond_operands.
18121 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18123 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18124 (extract_cond_operands): Detect condition comparing against non-
18125 invariant bound and return appropriate enum value.
18126 (find_interesting_uses_cond): Update use of extract_cond_operands.
18127 Handle its return value accordingly.
18128 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18132 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18133 nonlinear iv_use computation in loop invariant sensitive way.
18135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18137 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18138 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18140 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18142 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18143 (dump_cand): Support iv_cand.inv_exprs.
18144 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18146 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18151 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18153 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18154 as local function. Include necessary header files.
18155 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18157 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18159 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18163 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18164 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18165 RSHIFT_EXPR and BIT_NOT_EXPR.
18167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18169 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18170 (adjust_setup_cost): New parameter supporting round up adjustment.
18171 (struct address_cost_data): Delete.
18172 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18173 (split_address_cost, ptr_difference_cost): Delete.
18174 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18175 (struct ainc_cost_data): New struct.
18176 (get_address_cost_ainc): New function.
18177 (get_address_cost, get_computation_cost): Reimplement.
18178 (determine_group_iv_cost_address): Record inv_expr for all uses of
18180 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18181 (iv_ca_has_deps): Reimplemented to ...
18182 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18184 (iv_ca_extend): Call iv_ca_more_deps.
18186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18188 * tree-ssa-address.c (struct mem_address): Move to header file.
18189 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18190 * tree-ssa-address.h (struct mem_address): Move from C file.
18191 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18195 * tree-affine.h (aff_combination_type): New interface.
18196 (aff_combination_zero_p): Remove static.
18197 (aff_combination_const_p): New interface.
18198 (aff_combination_singleton_var_p): New interfaces.
18200 2017-05-11 Richard Biener <rguenther@suse.de>
18202 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18203 Skip unreachable blocks and destinations.
18204 (eliminate): Move stmt removal and fixup ...
18205 (fini_eliminate): ... here. Skip inserted exprs.
18206 (pass_pre::execute): Move fini_pre after fini_eliminate.
18207 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18208 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18209 PRE to get rid of dead code that has invalid SSA form and
18210 split critical edges again.
18212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18214 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18216 2017-05-11 Richard Biener <rguenther@suse.de>
18218 * passes.c (execute_function_todo): Verify loops if they are
18219 said to be up-to-date.
18220 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18221 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18223 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18226 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18227 handle calling assemble_external ourself.
18230 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18231 modes with zero size. Enhance comment.
18233 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18235 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18236 built-ins for vec_xl and vec_xst with short and char pointer
18239 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18241 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18242 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18243 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18244 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18245 (_mm_maskz_min_round_ss): New intrinsics.
18246 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18247 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18248 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18249 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18250 (__builtin_ia32_minss_mask_round): New builtins.
18251 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18252 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18253 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18255 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18256 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18258 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18261 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18263 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18264 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18265 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18266 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18267 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18268 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18269 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18270 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18271 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18272 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18273 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18274 (__builtin_ia32_mulss_mask_round): New builtins.
18275 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18276 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18277 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18279 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18280 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18282 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18285 2017-05-10 Julia Koval <julia.koval@intel.com>
18287 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18288 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18289 (_mm256_setr_m128i): New intrinsics.
18291 2017-05-10 Julia Koval <julia.koval@intel.com>
18293 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18294 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18295 (_mm_maskz_rcp14_ss): New intrinsics.
18296 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18297 (__builtin_ia32_rcp14ss_mask): New builtins.
18298 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18300 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18302 PR tree-optimization/51513
18303 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18304 (assert_unreachable_fallthru_edge_p): Use it.
18305 (group_case_labels_stmt): Likewise.
18306 * tree-cfg.h: Prototype it.
18307 * stmt.c: Include cfghooks.h and tree-cfg.h.
18308 (emit_case_dispatch_table) <gap_label>: New local variable.
18309 Use it to fill dispatch table gaps.
18310 Test for default_label before updating probabilities.
18311 (expand_case) <default_label>: Remove unneeded initialization.
18312 Test for unreachable default case statement and remove its edge.
18313 Set default_label accordingly.
18314 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18316 2017-05-10 Carl Love <cel@us.ibm.com>
18318 * config/rs6000/rs6000-c: Add support for built-in functions
18319 vector signed char vec_neg (vector signed char)
18320 vector signed short int vec_neg (vector short int)
18321 vector signed int vec_neg (vector signed int)
18322 vector signed long long vec_neg (vector signed long long)
18323 vector float vec_neg (vector float)
18324 vector double vec_neg (vector double)
18325 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18327 * config/rs6000/altivec.h: Add define for vec_neg
18328 * doc/extend.texi: Update the built-in documentation for the
18329 new built-in functions.
18331 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18333 PR tree-optimization/77644
18334 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18336 2017-05-10 Nathan Sidwell <nathan@acm.org>
18338 * dumpfile.h (TDI_lang_all): New.
18339 (TDF_KIND): New. Renumber others
18340 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18342 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18344 (get_dump_file_name): Adjust suffix generation.
18345 (dump_enable_all): Use TDF_KIND.
18346 * doc/invoke.texi (-fdump-lang-all): Document.
18348 * dumpfile.h: Tabify.
18350 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18353 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18354 Move member access before delete.
18356 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18358 * tree-inline.c (expand_call_inline): Split block at stmt
18361 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18364 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18365 are now unused after splitting mov{sf,sd}_hardfloat.
18366 (f32_lr2): Likewise.
18367 (f32_lm): Likewise.
18368 (f32_lm2): Likewise.
18369 (f32_li): Likewise.
18370 (f32_li2): Likewise.
18371 (f32_lv): Likewise.
18372 (f32_sr): Likewise.
18373 (f32_sr2): Likewise.
18374 (f32_sm): Likewise.
18375 (f32_sm2): Likewise.
18376 (f32_si): Likewise.
18377 (f32_si2): Likewise.
18378 (f32_sv): Likewise.
18379 (f32_dm): Likewise.
18380 (f32_vsx): Likewise.
18381 (f32_av): Likewise.
18382 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18383 For movsf, order stores so the VSX stores occur before the GPR
18384 store which encourages the register allocator to use a traditional
18385 FPR instead of a GPR. For movsd, order the stores so that the GPR
18386 store comes before the VSX stores to allow the power6 to work.
18387 This is due to the power6 not having a 32-bit integer store
18388 instruction from a FPR.
18389 (movsf_hardfloat): Likewise.
18390 (movsd_hardfloat): Likewise.
18392 2017-05-09 Martin Sebor <msebor@redhat.com>
18394 PR translation/80280
18395 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18398 PR translation/80280
18399 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18400 data member added in r247778.
18401 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18403 2017-05-09 Nathan Sidwell <nathan@acm.org>
18405 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18407 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18410 2017-05-09 Marek Polacek <polacek@redhat.com>
18412 * doc/invoke.texi: Fix typo.
18414 2017-05-09 Richard Biener <rguenther@suse.de>
18416 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18417 (vrp_val_is_min): Likewise.
18418 (set_value_range_to_value): Likewise.
18419 (set_value_range_to_nonnegative): Likewise.
18420 (gimple_assign_nonzero_p): Likewise.
18421 (gimple_stmt_nonzero_p): Likewise.
18422 (vrp_int_const_binop): Likewise. Remove unreachable case.
18423 (adjust_range_with_scev): Adjust comments.
18424 (compare_range_with_value): Likewise.
18425 (extract_range_from_phi_node): Likewise.
18426 (test_for_singularity): Likewise.
18428 2017-05-09 Richard Biener <rguenther@suse.de>
18430 * tree-vrp.c (get_single_symbol): Add assert that we don't
18431 get overflowed constants as invariant part.
18432 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18433 checks. Use wi::cmp instead of recursing for integer constants.
18434 (compare_values): Just ignore whether we assumed undefined
18435 overflow instead of failing the compare.
18436 (extract_range_for_var_from_comparison_expr): Add comment before the
18437 TREE_NO_WARNING sets.
18438 (test_for_singularity): Likewise.
18439 (extract_range_from_comparison): Do not disable optimization
18440 when we assumed undefined overflow.
18441 (extract_range_basic): Remove init of unused var.
18443 2017-05-09 Richard Biener <rguenther@suse.de>
18445 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18446 (extract_range_from_multiplicative_op_1): Adjust.
18447 (extract_range_from_binary_expr_1): Use int_const_binop.
18449 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18452 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18453 rs6000_store_data_bypass_p in seven define_bypass directives and
18454 in several comments.
18455 * config/rs6000/rs6000-protos.h: Add prototype for
18456 rs6000_store_data_bypass_p function.
18457 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18458 function implements slightly different (rs6000-specific) semantics
18459 than store_data_bypass_p, returning false rather than aborting
18460 with assertion error when arguments do not satisfy the
18461 requirements of store data bypass.
18462 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18463 rs6000_store_data_bypass_p.
18465 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18467 * config/xtensa/xtensa-protos.h
18468 (xtensa_initial_elimination_offset): New declaration.
18469 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18470 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18471 macro definition, add case for FRAME_POINTER_REGNUM when
18472 FRAME_GROWS_DOWNWARD.
18473 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18474 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18475 xtensa_initial_elimination_offset.
18477 2017-05-08 Nathan Sidwell <nathan@acm.org>
18479 * doc/invoke.texi: Alphabetize -fdump options.
18481 2017-05-08 Martin Sebor <msebor@redhat.com>
18483 PR translation/80280
18484 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18486 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18488 * target.def (compute_frame_layout): New optional target hook.
18489 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18490 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18491 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18493 * reload1.c (verify_initial_elim_offsets): Likewise.
18494 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18495 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18496 (arm_get_frame_offsets): Split up into this ...
18497 (arm_compute_frame_layout): ... and this function.
18499 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18501 * config/aarch64/constraints.md (Usa): New constraint.
18502 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18504 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18506 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18507 with_multilib_list after it has been checked.
18509 2017-05-08 Richard Biener <rguenther@suse.de>
18511 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18512 (bitmap_set_subtract_values): Likewise.
18514 2017-05-08 Richard Biener <rguenther@suse.de>
18516 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18517 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18519 (gimple_stmt_nonzero_warnv_p): Rename to ...
18520 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18522 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18523 (extract_range_basic): Adjust, do not disable propagation on
18524 strict overflow sensitive simplification.
18525 (vrp_visit_cond_stmt): Likewise.
18527 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18529 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18530 body size unconditionally.
18532 2017-05-07 Jeff Law <law@redhat.com>
18535 2017-05-06 Jeff Law <law@redhat.com>
18536 PR tree-optimization/78496
18537 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18540 PR tree-optimization/78496
18541 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18542 (simplify_stmt_using_ranges): Call it.
18543 (vrp_dom_walker::before_dom_children): Extract equivalences
18544 from an ASSERT_EXPR with an equality comparison against a
18547 2017-05-06 Jeff Law <law@redhat.com>
18549 PR tree-optimization/78496
18550 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18553 PR tree-optimization/78496
18554 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18555 (simplify_stmt_using_ranges): Call it.
18556 (vrp_dom_walker::before_dom_children): Extract equivalences
18557 from an ASSERT_EXPR with an equality comparison against a
18560 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18562 * lra-constraints.c (lra_copy_reg_equiv): New function.
18563 (split_reg): Use it to copy equivalence information from the
18564 original register to the spill register.
18566 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18568 PR rtl-optimization/75964
18569 * simplify-rtx.c (simplify_const_relational_operation): Remove
18570 invalid handling of comparisons of integer ABS.
18572 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18574 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18575 initialize to zero.
18576 (init_regs): Remove declaration.
18577 (function_arg_advance_32): Initialize error_p as boolean variable.
18579 2017-05-05 Nathan Sidwell <nathan@acm.org>
18581 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18582 lines. Use for (;;).
18584 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18586 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18587 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18588 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18589 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18590 VF=2 that require versioning.
18592 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18594 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18597 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18599 * diagnostic.h (diagnostic_override_option_index): Convert from
18600 macro to inline function.
18602 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18604 * diagnostic.c (last_module_changed_p): New function.
18605 (set_last_module): New function.
18606 (diagnostic_report_current_module): Convert macro usage to
18607 the above functions.
18608 * diagnostic.h (diagnostic_context::last_module): Strengthen
18609 from const line_map * to const line_map_ordinary *.
18610 (diagnostic_last_module_changed): Delete macro.
18611 (diagnostic_set_last_module): Delete macro.
18613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18615 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18616 with diagnostic_report_diagnostic.
18617 (diagnostic_n_impl_richloc): Likewise.
18618 * diagnostic.h (report_diagnostic): Delete macro.
18619 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18620 with diagnostic_report_diagnostic.
18621 * substring-locations.c (format_warning_va): Likewise.
18623 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18625 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18626 save/restor of format_spec. Move option-printing code to...
18627 (print_option_information): ...this new function, and
18628 reimplement by simply printing to the pretty_printer,
18629 rather than appending to the format string.
18631 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18633 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18634 handling logic into...
18635 (update_effective_level_from_pragmas): ...this new function.
18637 2017-05-04 Andrew Waterman <andrew@sifive.com>
18639 * config/riscv/riscv.opt (mstrict-align): New option.
18640 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18641 (SLOW_UNALIGNED_ACCESS): Define.
18642 (riscv_slow_unaligned_access): Declare.
18643 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18645 (riscv_slow_unaligned_access): New variable.
18646 (rocket_tune_info): Set slow_unaligned_access to true.
18647 (optimize_size_tune_info): Set slow_unaligned_access to false.
18648 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18649 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18650 (riscv_option_override): Set riscv_slow_unaligned_access.
18651 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18653 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18655 * config/riscv/riscv.md: Unify indentation.
18657 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18662 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18664 (extendsi<mode>2): Add support for doing sign extension via
18665 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18666 don't have ISA 3.0 instructions.
18667 (extendsi<mode>2 splitter): Likewise.
18668 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18669 generate the normal insns since SImode can now go in vector
18670 registers. Disallow the special UNSPECs needed for previous
18671 machines to hide SImode being used. Add new insns
18672 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18673 (fix_trunc<mode>si2_stfiwx): Likewise.
18674 (fix_trunc<mode>si2_internal): Likewise.
18675 (fixuns_trunc<mode>si2): Likewise.
18676 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18677 (fctiw<u>z_<mode>_smallint): Likewise.
18678 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18679 of floating point to 32-bit integer from doing a direct move to
18680 the GPR registers to do a store.
18681 (fctiwz_<mode>): Break long line.
18683 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18685 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18686 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18687 (addr_list, addr_offset_valid_p): New.
18688 (split_address_groups): Check offset validity with above function.
18689 (gt-tree-ssa-loop-ivopts.h): Include header file.
18691 2017-05-05 Nathan Sidwell <nathan@acm.org>
18693 * config.gcc (arm*-*-*): Add missing 'fi'.
18695 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18697 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18698 included in -fopt-info does not matter.
18699 * doc/optinfo.texi (-fopt-info): Fix description of default
18700 behavour. Explicitly say order of options included in -fopt-info
18703 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18705 * config.gcc: Allow combinations of aprofile and rmprofile values for
18706 --with-multilib-list.
18707 * config/arm/t-multilib: New file.
18708 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18709 variables. Remove setting of ISA and floating-point ABI in
18710 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18711 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18712 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18713 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18715 * config/arm/t-rmprofile: Likewise except for the matches changes.
18716 * doc/install.texi (--with-multilib-list): Document the combination of
18717 aprofile and rmprofile values and warn about pitfalls in doing that.
18719 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18721 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18722 (movdi_aarch64): Likewise.
18724 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18726 PR tree-optimization/80632
18727 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18729 (build_arrays): Initialize it for virtual phis.
18730 (fix_phi_nodes): Use it for virtual phis.
18732 PR tree-optimization/80558
18733 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18734 [x, y] op z into [x op, y op z] for op & or | if conditions
18737 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18738 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18741 * config/arm/arm.md (use_literal_pool): Remove.
18742 (64-bit immediate split): No longer takes cost into consideration
18743 if arm_disable_literal_pool is enabled.
18744 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18745 used when arm_disable_literal_pool is enabled.
18746 (arm_max_const_double_inline_cost): Remove use of
18747 arm_disable_literal_pool.
18748 (push_minipool_fix): Add assert.
18749 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18750 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18751 (no_literal_pool_sf_immediate): New.
18753 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18755 PR tree-optimization/80613
18756 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18757 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18759 2017-05-05 Richard Biener <rguenther@suse.de>
18761 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18763 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18765 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18766 of this flag from insn conditions due to removal from r247495.
18768 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18770 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18772 (arm_early_store_addr_dep_ptr): Likewise.
18773 * config/arm/aarch-common-protos.h
18774 (arm_early_load_addr_dep_ptr): Add prototype.
18775 (arm_early_store_addr_dep_ptr): Likewise.
18776 * config/arm/cortex-a53.md: Add new bypasses.
18778 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18780 * tree.c (next_type_uid): Change type to unsigned.
18781 (type_hash_canon): Decrement back next_type_uid if
18782 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18783 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18786 2017-05-04 Martin Sebor <msebor@redhat.com>
18788 * builtins.c: Fix a trivial typo in a comment.
18790 PR middle-end/79234
18791 * builtins.c (check_sizes): Adjust to handle reading past the end.
18792 Avoid printing excessive upper bound of ranges. Use %E to print
18793 tree nodes instead of converting them to %wu.
18794 (expand_builtin_memchr): New function.
18795 (compute_dest_size): Rename...
18796 (compute_objsize): ...to this.
18797 (expand_builtin_memcpy): Adjust.
18798 (expand_builtin_mempcpy): Adjust.
18799 (expand_builtin_strcat): Adjust.
18800 (expand_builtin_strcpy): Adjust.
18801 (check_strncat_sizes): Adjust.
18802 (expand_builtin_strncat): Adjust.
18803 (expand_builtin_strncpy): Adjust and simplify.
18804 (expand_builtin_memset): Adjust.
18805 (expand_builtin_bzero): Adjust.
18806 (expand_builtin_memcmp): Adjust.
18807 (expand_builtin): Handle memcmp.
18808 (maybe_emit_chk_warning): Check strncat just once.
18810 2017-05-04 Martin Sebor <msebor@redhat.com>
18812 PR preprocessor/79214
18813 PR middle-end/79222
18814 PR middle-end/79223
18815 * builtins.c (check_sizes): Add inlining context and issue
18816 warnings even when -Wno-system-headers is set.
18817 (check_strncat_sizes): Same.
18818 (expand_builtin_strncat): Same.
18819 (expand_builtin_memmove): New function.
18820 (expand_builtin_stpncpy): Same.
18821 (expand_builtin): Handle memmove and stpncpy.
18823 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18825 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18826 which is not used any more.
18828 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18830 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18832 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18834 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18835 (cortexa53_tunings): Likewise.
18836 (cortexa57_tunings): Likewise.
18837 (cortexa72_tunings): Likewise.
18838 (cortexa73_tunings): Likewise.
18840 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18842 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18843 Set loop alignment to 8.
18845 2017-05-04 Martin Sebor <msebor@redhat.com>
18847 PR translation/80280
18848 * builtins.c (expand_builtin_object_size): Add missing quoting to
18849 %D and like directives.
18850 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18851 (hsa_type_for_tree_type): Same.
18852 (verify_function_arguments): Same.
18853 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18854 * varasm.c (get_section): Same.
18857 2017-05-04 Martin Sebor <msebor@redhat.com>
18859 PR translation/80280
18860 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18862 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18864 * config/aarch64/aarch64.c (generic_addrcost_table):
18865 Change HI/TI mode setting.
18867 2017-05-04 Martin Jambor <mjambor@suse.cz>
18869 PR tree-optimization/80622
18870 * tree-sra.c (comes_initialized_p): New function.
18871 (build_accesses_from_assign): Only set write lazily when
18872 comes_initialized_p is false.
18873 (analyze_access_subtree): Use comes_initialized_p.
18874 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18875 instead of testing for PARM_DECL.
18877 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18879 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18880 constraint on operand 0 to allow more general addressing modes.
18881 Adjust output template.
18882 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18884 * config/aarch64/aarch64-protos.h
18885 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18886 * config/aarch64/constraints.md (Dp): New address constraint.
18887 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18890 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18892 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18893 update use of estimate_ipcp_clone_size_and_time.
18894 (estimate_local_effects): Update use of
18895 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18896 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18897 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18898 Return nonspecialized time.
18900 2017-05-04 Richard Biener <rguenther@suse.de>
18902 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18903 for the last VUSE which def dominates the PHI. Directly call
18905 (get_continuation_for_phi_1): Remove.
18907 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18909 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18910 to explain the use of truncating division. Cap the number of
18911 iterations to the maximum given by nb_iterations_upper_bound,
18914 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18916 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18917 * configure: Regenerate.
18918 * config.in: Regenerate.
18919 * config/i386/driver-mingw32.c: new file.
18920 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18921 * config.host: Link driver-mingw32.o on MinGW host.
18922 * doc/install.texi: Document new --enable-mingw-wildcard configure
18925 2017-05-04 Marek Polacek <polacek@redhat.com>
18927 PR tree-optimization/80612
18928 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18930 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18931 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18933 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18934 (movt splitter): Likewise.
18935 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18936 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18937 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18938 block for Thumb-1 with MOVT.
18939 (thumb2_legitimate_address_p): Move code block ...
18940 (can_avoid_literal_pool_for_label_p): ... into this new function.
18941 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18943 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18944 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18945 "M-profile targets with the MOVT instruction".
18947 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18949 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18950 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18951 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18953 2017-05-04 Martin Liska <mliska@suse.cz>
18955 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18956 variable cond_code.
18958 2017-05-04 Richard Biener <rguenther@suse.de>
18960 * tree.c (array_at_struct_end_p): Handle arrays at struct
18961 end with flexarrays more conservatively. Refactor and treat
18962 arrays of arrays or aggregates more strict. Fix
18963 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18964 * tree.c (array_at_struct_end_p): Adjust prototype.
18965 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18966 * gimple-fold.c (get_range_strlen): Likewise.
18967 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18969 2017-05-04 Richard Biener <rguenther@suse.de>
18971 PR tree-optimization/31130
18972 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18974 (supports_overflow_infinity): Likewise.
18975 (is_negative_overflow_infinity): Likewise.
18976 (is_positive_overflow_infinity): Likewise.
18977 (is_overflow_infinity): Likewise.
18978 (stmt_overflow_infinity): Likewise.
18979 (overflow_infinity_range_p): Likewise.
18980 (usable_range_p): Remove as always returning true.
18981 (make_overflow_infinity): Remove.
18982 (negative_overflow_infinity): Likewise.
18983 (positive_overflow_infinity): Likewise.
18984 (avoid_overflow_infinity): Likewise.
18985 (set_value_range): Adjust accordingly.
18986 (set_value_range_to_nonnegative): Likewise, remove now unused
18987 overflow_infinity arg.
18988 (vrp_operand_equal_p): Adjust.
18989 (update_value_range): Likewise.
18990 (range_int_cst_singleton_p): Likewise.
18991 (operand_less_p): Likewise.
18992 (compare_values_warnv): Likewise.
18993 (extract_range_for_var_from_comparison_expr): Likewise.
18994 (vrp_int_const_binop): Likewise.
18995 (zero_nonzero_bits_from_vr): Likewise.
18996 (extract_range_from_multiplicative_op_1): Likewise.
18997 (extract_range_from_binary_expr_1): Likewise.
18998 (extract_range_from_unary_expr): Likewise.
18999 (extract_range_from_comparison): Likewise.
19000 (extract_range_basic): Likewise.
19001 (adjust_range_with_scev): Likewise.
19002 (compare_ranges): Likewise.
19003 (compare_range_with_value): Likewise.
19004 (dump_value_range): Likewise.
19005 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19007 (simplify_cond_using_ranges): Adjust.
19009 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19011 * brig-builtins.def: Added a builtin for class_f64.
19012 * builtin-types.def: Added a builtin type needed by class_f64.
19014 2017-05-03 Jason Merrill <jason@redhat.com>
19016 * timevar.def: Add TV_CONSTEXPR.
19018 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19020 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19022 2017-05-03 Martin Jambor <mjambor@suse.cz>
19024 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19025 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19026 * ipa-cp.c (ipcp_propagate_stage): Do not call
19027 ipa_update_after_lto_read.
19028 * ipa-inline.c (ipa_inline): Likewise.
19030 2017-05-03 Martin Jambor <mjambor@suse.cz>
19032 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19033 tag. Added a default constructor and a destructor.
19034 (ipa_edge_args_sum_t): New class;
19035 (ipa_edge_args_sum): Declare.
19036 (ipa_edge_args_vector): Remove declaration.
19037 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19038 (ipa_free_edge_args_substructures): Remove declaration.
19039 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19040 (ipa_edge_args_info_available_for_edge_p): Likewise.
19041 * ipa-prop.c (ipa_edge_args_vector): Removed.
19042 (edge_removal_hook_holder): Likewise.
19043 (edge_duplication_hook_holder): Likewise.
19044 (ipa_edge_args_sum): New variable.
19045 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19046 ipa_edge_args_vector.
19047 (ipa_free_edge_args_substructures): Likewise.
19048 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19049 ipa_edge_args_vector.
19050 (ipa_edge_removal_hook): Turned into method
19051 ipa_edge_args_sum_t::remove.
19052 (ipa_edge_duplication_hook): Turned into method
19053 ipa_edge_args_sum_t::duplicate.
19054 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19055 registering edge hooks.
19056 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19057 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19058 ipa_edge_args_sum instead of ipa_edge_args_vector.
19059 * ipa-profile.c (ipa_profile): Likewise.
19061 2017-05-03 Martin Jambor <mjambor@suse.cz>
19063 * symbol-summary.h (function_summary): New method exists.
19064 (function_summary::symtab_removal): Deallocate through release.
19065 (call_summary): New class.
19066 (gt_ggc_mx): New overload.
19067 (gt_pch_nx): Likewise.
19068 (gt_pch_nx): Likewise.
19070 2017-05-03 Jeff Law <law@redhat.com>
19072 PR tree-optimization/78496
19073 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19074 from simplify_cond_using_ranges. Split off code to walk
19075 backwards through casts into ...
19076 (simplify_cond_using_ranges_2): New function.
19077 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19078 (execute_vrp): After identifying jump threads, call
19079 simplify_cond_using_ranges_2.
19081 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19084 * ipa-inline.h (inline_summary): Add ctor.
19085 (create_ggc): Do not use ggc_cleared_alloc.
19087 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19088 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19090 * gcc.c (handle_braces): Support escaping in switch matching
19092 * doc/invoke.texi (Spec Files): Document it.
19093 Remove superfluous @code markup in items.
19095 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19097 * diagnostic-show-locus.c (struct column_range): New struct.
19098 (get_affected_columns): New function.
19099 (get_printed_columns): New function.
19100 (struct correction): New struct.
19101 (correction::ensure_capacity): New function.
19102 (correction::ensure_terminated): New function.
19103 (struct line_corrections): New struct.
19104 (line_corrections::~line_corrections): New dtor.
19105 (line_corrections::add_hint): New function.
19106 (layout::print_trailing_fixits): Reimplement in terms of the new
19108 (selftest::test_overlapped_fixit_printing): New function.
19109 (selftest::diagnostic_show_locus_c_tests): Call it.
19111 2017-05-03 Nathan Sidwell <nathan@acm.org>
19113 Canonicalize canonical type hashing
19114 * tree.h (type_hash_canon_hash): Declare.
19115 * tree.c (type_hash_list, attribute_hash_list): Move into
19116 type_hash_canon_hash.
19117 (build_type_attribute_qual_variant): Break out hash code calc into
19118 type_hash_canon_hash.
19119 (type_hash_canon_hash): New. Generic type hash computation.
19120 (build_range_type_1, build_array_type_1, build_function_type,
19121 build_method_type_directly, build_offset_type, build_complex_type,
19122 make_vector_type): Call it.
19124 2017-05-03 Richard Biener <rguenther@suse.de>
19126 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19127 When all DRs have unknown misaligned do not always peel
19128 when there is a store but apply the same costing model as if
19129 there were only loads.
19131 2017-05-03 Richard Biener <rguenther@suse.de>
19134 PR tree-optimization/80492
19135 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19136 compare_base_decls returning dont-know properly.
19138 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19140 * config/arm/iterators.md (CCSI): New mode iterator.
19141 (arch): New mode attribute.
19142 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19143 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19144 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19145 code iterator for success result mode.
19146 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19147 the corresponding new insn generators.
19149 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19153 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19155 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19157 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19158 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19159 (DDR_A): Wrap DDR argument in brackets.
19160 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19161 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19162 (DDR_REVERSED_P): Likewise.
19164 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19166 PR tree-optimization/79472
19167 * tree-switch-conversion.c (struct switch_conv_info): Add
19168 contiguous_range and default_case_nonstandard fields.
19169 (collect_switch_conv_info): Compute contiguous_range and
19170 default_case_nonstandard fields, don't clear final_bb if
19171 contiguous_range and only the default case doesn't have the required
19173 (check_all_empty_except_final): Set default_case_nonstandard instead
19174 of failing if contiguous_range and the default case doesn't have empty
19176 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19177 and only the default case doesn't have the required constants. Skip
19179 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19180 if default_case_nonstandard.
19181 (build_constructors): Build constant 1 just once. Assert that default
19182 values aren't inserted in between cases if contiguous_range. Skip
19184 (build_arrays): Skip virtual phis.
19185 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19186 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19187 Handle virtual phis.
19188 (gen_inbound_check): Handle default_case_nonstandard case.
19189 (process_switch): Adjust check_final_bb caller. Call
19190 gather_default_values with the first non-default case instead of
19191 default case if default_case_nonstandard.
19193 2017-05-02 Nathan Sidwell <nathan@acm.org>
19195 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19196 check. Fix formatting.
19198 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19200 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19201 errors when comparing specialized and unspecialized times.
19203 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19205 * diagnostic-show-locus.c
19206 (layout::should_print_annotation_line_p): Make private.
19207 (layout::print_annotation_line): Make private.
19208 (layout::annotation_line_showed_range_p): Make private.
19209 (layout::show_ruler): Make private.
19210 (layout::print_source_line): Make private. Pass in line and
19211 line_width, rather than calling location_get_source_line. Drop
19213 (layout::print_leading_fixits): New method.
19214 (layout::print_any_fixits): Rename to...
19215 (layout::print_trailing_fixits): ...this, and make private.
19216 Don't print newline fixits.
19217 (diagnostic_show_locus): Move logic for printing one row into...
19218 (layout::print_line): ...this new function. Move the
19219 location_get_source_line call and error-handling from
19220 print_source_line to here. Call print_leading_fixits, and rename
19221 print_any_fixits to print_trailing_fixits.
19222 (selftest::test_fixit_insert_containing_newline): Update now that
19223 newlines are partially supported.
19224 (selftest::test_fixit_insert_containing_newline_2): New test.
19225 (selftest::test_fixit_replace_containing_newline): Update comments.
19226 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19227 * edit-context.c (class added_line): New class.
19228 (class edited_line): Describe newline handling in comment.
19229 (edited_line::actually_edited_p): New method.
19230 (edited_line::print_content): Delete redundant decl.
19231 (edited_line::m_predecessors): New field.
19232 (edited_file::print_content): Call edited_line::print_content.
19233 (edited_file::print_diff): Update to support newlines.
19234 (edited_file::print_diff_hunk): Likewise.
19235 (edited_file::print_run_of_changed_lines): New function.
19236 (edited_file::print_diff_line): Convert to...
19237 (print_diff_line): ...this.
19238 (edited_file::get_effective_line_count): New function.
19239 (edited_line::edited_line): Initialize new field m_predecessors.
19240 (edited_line::~edited_line): Clean up m_predecessors.
19241 (edited_line::apply_fixit): Handle newlines.
19242 (edited_line::get_effective_line_count): New function.
19243 (edited_line::print_content): New function.
19244 (edited_line::print_diff_lines): New function.
19245 (selftest::test_applying_fixits_insert_containing_newline): New
19247 (selftest::test_applying_fixits_replace_containing_newline): New
19249 (selftest::insert_line): New function.
19250 (selftest::test_applying_fixits_multiple_lines): Add example of
19252 (selftest::edit_context_c_tests): Call the new tests.
19254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19256 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19257 parameter cand. Update dump information.
19258 (get_computation_cost): Update uses.
19260 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19262 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19263 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19264 (get_computation_at, rewrite_use_address): Update use of
19265 get_computation_aff.
19267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19269 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19270 (get_computation): Delete.
19271 (get_computation_cost): Implement like get_computation_cost_at.
19272 Use get_computation_at.
19273 (get_computation_cost_at): Delete.
19274 (rewrite_use_nonlinear_expr): Use get_computation_at.
19275 (rewrite_use_compare, remove_unused_ivs): Ditto.
19277 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19279 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19281 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19283 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19284 (ivopts_global_cost_for_size): Rename parameter and update uses.
19285 (iv_ca_recount_cost): Update uses.
19286 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19287 candidates seperately in n_invs and n_cands.
19288 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19290 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19292 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19293 (find_inv_vars_cb): New.
19294 (find_depends): Renamed to ...
19295 (find_inv_vars): ... this.
19296 (add_candidate_1, force_var_cost): Call find_inv_vars.
19297 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19299 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19301 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19302 inv_vars. Add inv_exprs.
19303 (struct iv_cand): Rename depends_on to inv_vars.
19304 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19305 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19306 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19307 (dump_cand): Dump inv_vars.
19308 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19309 (record_invariant, find_depends, add_candidate_1): Ditto.
19310 (set_group_iv_cost, force_var_cost): Ditto.
19311 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19312 (get_computation_cost_at, get_computation_cost): Ditto.
19313 (determine_group_iv_cost_generic): Ditto.
19314 (determine_group_iv_cost_address): Ditto.
19315 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19316 (determine_group_iv_costs): Ditto.
19317 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19318 (iv_ca_set_remove_invariants): Renamed to ...
19319 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19320 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19321 (iv_ca_set_add_invariants): Renamed to ...
19322 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19323 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19324 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19325 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19326 (create_new_ivs): Remove useless dump.
19328 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19330 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19332 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19333 (iv_ca_set_no_cp, create_new_iv): Ditto.
19335 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19337 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19339 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19341 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19342 function tree_check2.
19344 2017-05-02 Martin Liska <mliska@suse.cz>
19346 * doc/gcov.texi: Add missing preposition.
19347 * gcov.c (function_info::function_info): Properly fill up
19348 all member variables.
19350 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19352 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19354 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19356 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19358 2017-05-02 Martin Liska <mliska@suse.cz>
19361 * lto-streamer-in.c (lto_read_tree_1): Remove
19362 LTO_STREAMER_DEBUG.
19363 * lto-streamer.c (struct tree_hash_entry): Likewise.
19364 (struct tree_entry_hasher): Likewise.
19365 (tree_entry_hasher::hash): Likewise.
19366 (tree_entry_hasher::equal): Likewise.
19367 (lto_streamer_init): Likewise.
19368 (lto_orig_address_map): Likewise.
19369 (lto_orig_address_get): Likewise.
19370 (lto_orig_address_remove): Likewise.
19371 * lto-streamer.h: Likewise.
19372 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19373 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19375 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19377 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19378 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19379 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19380 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19381 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19382 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19383 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19384 (mm_maskz_sub_ss): New intrinsics.
19385 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19386 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19387 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19388 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19389 (__builtin_ia32_subss_mask_round): New builtins.
19390 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19391 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19392 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19394 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19395 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19397 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19400 2017-05-02 Martin Jambor <mjambor@suse.cz>
19402 PR tree-optimization/78687
19403 * tree-sra.c (access): New field parent.
19404 (process_subtree_disqualification): New function.
19405 (disqualify_candidate): Call it.
19406 (build_accesses_from_assign): Reset write flag if creating an
19408 (build_access_subtree): Fill in parent field and also prpagate
19409 down grp_write flag.
19410 (create_artificial_child_access): New parameter set_grp_write, set
19411 grp_write to its value.
19412 (propagate_subaccesses_across_link): Also propagate grp_write flag
19414 (propagate_all_subaccesses): Push the closest parent back to work
19415 queue if add_access_to_work_queue returned true.
19417 2017-05-02 Richard Biener <rguenther@suse.de>
19419 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19420 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19421 -fstrict-overflow documentation.
19422 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19423 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19424 flag_strict_overflow.
19425 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19426 * lto-opts.c (lto_write_options): Do not stream it.
19427 * lto-wrapper.c (merge_and_complain): Do not handle it.
19428 * opts.c (default_options_table): Do not set -fstrict-overflow.
19429 (finish_options): Likewise do not clear it when sanitizing.
19430 * simplify-rtx.c (simplify_const_relational_operation): Do not
19431 test flag_strict_overflow.
19433 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19435 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19436 using enabled attribute.
19437 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19438 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19439 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19440 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19441 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19442 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19443 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19444 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19445 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19446 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19448 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19450 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19452 2017-05-02 Richard Biener <rguenther@suse.de>
19454 PR tree-optimization/80591
19456 2017-04-10 Richard Biener <rguenther@suse.de>
19458 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19461 2017-05-02 Richard Biener <rguenther@suse.de>
19463 PR tree-optimization/80549
19464 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19465 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19466 headers if they do not have a preheader.
19468 2017-05-02 Martin Liska <mliska@suse.cz>
19471 * common.opt: Fix typo.
19472 * doc/invoke.texi: Likewise.
19474 2017-05-01 Jan Beulich <jbeulich@suse.com>
19476 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19477 swapping, add (x,x,m,x,n) alternative.
19479 2017-05-01 Nathan Sidwell <nathan@acm.org>
19481 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19482 unnecessary unadjusted_alignment check.
19484 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19487 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19489 * gimplify.c (gimplify_cilk_detach): New function.
19490 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19491 * tree-core.h: Document EXPR_CILK_SPAWN.
19492 * tree.h (EXPR_CILK_SPAWN): Define.
19494 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19496 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19497 to use new fixit_hint representation, using the "replace" logic.
19498 (get_line_span_for_fixit_hint): Likewise.
19499 (layout::print_any_fixits): Likewise.
19500 (selftest::test_one_liner_many_fixits): Rename to...
19501 (selftest::test_one_liner_many_fixits_1): ...this, and update
19502 comment and expected output to reflect that the multiple fix-it
19503 hints are now consolidated into one insertion.
19504 (selftest::test_one_liner_many_fixits_2): New test.
19505 (selftest::test_diagnostic_show_locus_one_liner): Update for
19507 (selftest::test_fixit_consolidation): Update for fix-it API
19509 * diagnostic.c (print_parseable_fixits): Likewise.
19510 * edit-context.c (edited_line::m_line_events): Convert from
19511 auto_vec <line_event *> to auto_vec <line_event>.
19512 (class line_event): Convert from abstract base class to a concrete
19513 class, taking over the role of replace_event.
19514 (class insert_event): Delete.
19515 (class replace_event): Rename to class line_event. Convert to
19517 (edit_context::add_fixits): Reimplement.
19518 (edit_context::apply_insert): Delete.
19519 (edit_context::apply_replace): Rename to...
19520 (edit_context::apply_fixit): ...this. Convert to half-open range.
19521 (edited_file::apply_insert): Delete.
19522 (edited_file::apply_replace): Rename to...
19523 (edited_file::apply_fixit): ...this.
19524 (edited_line::~edited_line): Drop deletion of events.
19525 (edited_line::apply_insert): Delete.
19526 (edited_line::apply_replace): Rename to...
19527 (edited_line::apply_fixit): ...this. Convert to half-open range.
19528 Update for change to type of m_line_events.
19529 * edit-context.h (edit_context::apply_insert): Delete.
19530 (edit_context::apply_replace): Rename to...
19531 (edit_context::apply_fixit): ...this.
19533 2017-05-01 Martin Sebor <msebor@redhat.com>
19535 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19538 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19541 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19542 __get_cpuid_max returns 0.
19543 (__get_cpuid_count): Ditto.
19545 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19547 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19548 replacement expression is another instance of one of its arguments.
19550 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19553 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19554 check for stack push/pop autoinc.
19555 * config/i386/i386.c (ix86_agi_dependent): Return false
19556 if the only reason why modified_in_p returned true is that
19557 addr is SP based and set_insn is a push or pop.
19559 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19561 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19564 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19567 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19568 (account_size_time): Use two predicates - exec_pred and
19570 (evaluate_conditions_for_known_args): Compute both clause and
19572 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19573 (inline_summary_t::duplicate): Update.
19574 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19576 (compute_inline_parameters): Likewise.
19577 (estimate_edge_size_and_time): Update caluclation of time.
19578 (estimate_node_size_and_time): Compute both time and nonspecialized
19580 (estimate_ipcp_clone_size_and_time): Update.
19581 (inline_merge_summary): Update.
19582 (do_estimate_edge_time): Update.
19583 (do_estimate_edge_size): Update.
19584 (do_estimate_edge_hints): Update.
19585 (inline_read_section, inline_write_summary): Stream both new predicates.
19586 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19588 (compute_inlined_call_time): Cleanup.
19589 (big_speedup_p): Update.
19590 (edge_badness): Update.
19591 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19592 (size_time_entry): Replace predicate by exec_predicate and
19593 nonconst_predicate.
19594 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19595 (estimate_edge_time): Return also nonspec_time.
19596 (reset_edge_growth_cache): Update.
19598 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19600 PR rtl-optimization/80491
19601 * ifcvt.c (noce_process_if_block): When looking for x setter
19602 with missing else_bb, don't check only the insn right before
19603 cond_earliest, but look for the last insn that x is modified in
19604 within the same bb.
19606 PR rtl-optimization/80491
19607 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19609 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19611 PR tree-optimization/80487
19612 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19614 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19616 PR tree-optimization/79697
19617 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19618 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19619 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19621 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19622 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19624 2017-04-28 Martin Sebor <msebor@redhat.com>
19626 PR tree-optimization/80523
19627 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19628 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19630 (maybe_warn, format_directive, parse_directive): Use new functions.
19631 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19633 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19635 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19637 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19639 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19640 target_header_dir): Set correctly.
19641 * configure: Regenerated.
19642 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19643 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19644 instead of SYSTEM_HEADER_DIR.
19646 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19648 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19649 (estimate_local_effects): Likewise.
19650 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19651 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19652 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19653 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19654 do_estimate_edge_time, estimate_edge_time): Likewise.
19655 * ipa-inline-analysis.c (estimate_node_size_and_time,
19656 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19657 (estimate_time_after_inlining): Remove.
19659 2017-04-28 Martin Liska <mliska@suse.cz>
19661 * doc/gcov.texi: Enhance documentation of gcov.
19663 2017-04-28 Martin Liska <mliska@suse.cz>
19665 * doc/gcov.texi: Sort options in alphabetic order.
19666 * doc/gcov-dump.texi: Likewise.
19667 * doc/gcov-tool.texi: Likewise.
19668 * gcov.c (print_usage): Likewise.
19669 * gcov-dump.c (print_usage): Likewise.
19670 * gcov-tool.c (print_merge_usage_message): Likewise.
19671 (print_rewrite_usage_message): Likewise.
19672 (print_overlap_usage_message): Likewise.
19674 2017-04-28 Martin Liska <mliska@suse.cz>
19676 PR gcov-profile/53915
19677 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19679 2017-04-28 Martin Liska <mliska@suse.cz>
19681 PR gcov-profile/79891
19682 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19683 is marked by compiler as living on a line.
19684 (get_cycles_count): Remove usage of the union.
19685 (output_intermediate_file): Likewise.
19686 (find_source): Fix GNU coding style.
19687 (accumulate_line_counts): Remove old non-all block mode.
19688 (output_lines): Remove usage of the union.
19689 * profile.c (output_location): Include all BBs, even if
19690 belonging to a same line (and file) as a previous BB.
19692 2017-04-28 Martin Liska <mliska@suse.cz>
19694 * gcov.c (process_args): Handle new argument 'w'.
19695 (read_graph_file): Assign ID to BBs.
19696 (output_branch_count): Display BB # if verbose flag is set.
19697 (output_lines): Likewise for arcs.
19698 (print_usage): Add '--verbose' option help.
19699 * doc/gcov.texi: Document --verbose (-w) option.
19701 2017-04-28 Martin Liska <mliska@suse.cz>
19703 * gcov.c (struct block_location_info): New struct.
19704 (process_file): Fill up the new structure.
19705 (read_graph_file): Replace usage of encoding by the newly added
19707 (add_line_counts): Likewise.
19708 (accumulate_line_counts): Remove usage of the union.
19709 (function_info::function_info): New function.
19710 (function_info::~function_info): Likewise.
19711 (process_file): Call delete instead of release_function.
19712 (release_function): Release the function.
19713 (release_structures): Call delete instead of release_function.
19714 (solve_flow_graph): Replace usage of num_blocks.
19715 (find_exception_blocks): Likewise.
19716 (output_lines): Fix GNU coding style.
19718 2017-04-28 Martin Liska <mliska@suse.cz>
19721 * coverage.c (coverage_remove_note_file): New function.
19722 * coverage.h: Declare the function.
19723 * toplev.c (finalize): Clean if an error has been seen.
19725 2017-04-28 Martin Liska <mliska@suse.cz>
19727 PR gcov-profile/80031
19728 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19729 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19730 * gcov.c (read_graph_file): Read just number of blocks.
19731 * profile.c (branch_prob): Do not stream 0 flags per a basic
19734 2017-04-28 Martin Liska <mliska@suse.cz>
19736 * gcov-dump.c (tag_*): Add new argument to declarations.
19737 (dump_gcov_file): Likewise.
19738 (tag_blocks): Add and use new argument depth.
19739 (tag_arcs): Likewise.
19740 (tag_lines): Likewise.
19741 (tag_counters): Likewise.
19742 (tag_summary): Likewise.
19743 (dump_working_sets): Use depth to do a proper indentation.
19745 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19748 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19749 * symtab.c (symtab_node::debug_symtab): Move definition here.
19751 2017-04-28 Richard Biener <rguenther@suse.de>
19753 * lto-streamer.h (LTO_major_version): Bump to 7.
19755 2017-04-28 Richard Biener <rguenther@suse.de>
19757 * tree-vrp.c (assert_info): New struct.
19758 (add_assert_info): New helper.
19759 (register_edge_assert_for_2): Refactor to add asserts to a vector
19761 (register_edge_assert_for_1): Likewise.
19762 (register_edge_assert_for): Likewise.
19763 (finish_register_edge_assert_for): New helper actually registering
19764 asserts where live on edge.
19765 (find_conditional_asserts): Adjust.
19766 (find_switch_asserts): Likewise.
19767 (evrp_dom_walker::try_find_new_range): Generalize.
19768 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19770 2017-04-27 Marek Polacek <polacek@redhat.com>
19773 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19774 arg10 and arg11 to itype.
19776 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19778 * doc/extend.texi (Object Size Checking): Improve grammar.
19780 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19783 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19784 that the logic for permitting reciprocal estimates matches that
19787 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19790 * tree.c (type_cache_hasher::equal): Only compare
19791 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19792 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
19793 non-aggregate element types.
19794 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
19795 about the flag on ARRAY_TYPEs in the comment, formatting fix.
19797 2017-04-27 Richard Biener <rguenther@suse.de>
19799 PR middle-end/80533
19800 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
19801 stripping ARRAY_REFs from MEM_EXPR make sure we're not
19802 keeping a reference to a trailing array.
19804 2017-04-27 Richard Biener <rguenther@suse.de>
19806 PR middle-end/80539
19807 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
19808 being in loop-closed SSA form conservatively.
19809 (chrec_fold_multiply_poly_poly): Likewise.
19811 2017-04-27 Tamar Christina <tamar.christina@arm.com>
19813 PR middle-end/79665
19814 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
19815 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19817 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19820 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19821 (aarch64_function_arg_alignment): Return unsigned int again, but still
19822 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19823 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19824 Don't emit -Wpsabi note.
19825 (aarch64_function_arg_boundary): Likewise.
19826 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19829 2017-04-26 Nathan Sidwell <nathan@acm.org>
19831 * tree.h (crc32_unsigned_n): Declare.
19832 (crc32_unsigned, crc32_unsigned): Make inline.
19833 * tree.c (crc32_unsigned_bits): Replace with ...
19834 (crc32_unsigned_n): ... this.
19835 (crc32_unsigned, crc32_byte): Remove.
19836 (crc32_string): Remove unnecessary braces.
19838 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19840 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19841 * ipa-inline-analysis.c (MAX_TIME): Remove.
19842 (account_size_time): Use sreal for time.
19843 (dump_inline_summary): Update.
19844 (estimate_function_body_sizes): Update.
19845 (estimate_edge_size_and_time): Update.
19846 (estimate_calls_size_and_time): Update.
19847 (estimate_node_size_and_time): Update.
19848 (inline_merge_summary): Update.
19849 (inline_update_overall_summary): Update.
19850 (estimate_time_after_inlining): Update.
19851 (inline_read_section): Update.
19852 (inline_write_summary): Update.
19853 * ipa-inline.c (compute_uninlined_call_time): Update.
19854 (compute_inlined_call_time): Update.
19855 (recursive_inlining): Update.
19856 (inline_small_functions): Update.
19857 (dump_overall_stats): Update.
19858 * ipa-inline.h: Include sreal.h.
19859 (size_time_entry): Turn time to sreal.
19860 (inline_summary): Turn self_time nad time to sreal.
19862 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19864 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19866 (sreal::stream_out, sreal::stream_in): New.
19867 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19869 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19871 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19874 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19877 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19879 (dimode_scalar_chain::compute_convert_gain): Ditto.
19880 (dimode_scalar_chain::make_vector_copies): Ditto.
19881 (dimode_scalar_chain::convert_reg): Ditto.
19882 (dimode_scalar_chain::convert_insn): Ditto.
19883 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19884 (VI248_AVX512BW_1): New mode iterator.
19885 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19886 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19889 2017-04-25 Martin Sebor <msebor@redhat.com>
19891 PR tree-optimization/80497
19892 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19893 constants are representable in HOST_WIDE_INT.
19894 (parse_directive): Ditto.
19896 2017-04-25 Martin Sebor <msebor@redhat.com>
19899 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19900 (new_zero_array): Adjust signature.
19901 (dom_info::dom_init): Used unsigned rather that size_t.
19902 (dom_info::dom_info): Same.
19904 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19905 Jakub Jelinek <jakub@redhat.com>
19908 * config/arm/arm.c: Include gimple.h.
19909 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19910 returns negative, increment ncrn only if it returned positive.
19911 (arm_needs_doubleword_align): Return int instead of bool,
19912 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19913 members, but if there is any such non-FIELD_DECL
19914 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19915 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19916 returns negative, increment nregs only if it returned positive.
19917 (arm_setup_incoming_varargs): Likewise.
19918 (arm_function_arg_boundary): Emit -Wpsabi note if
19919 arm_needs_doubleword_align returns negative, return
19920 DOUBLEWORD_ALIGNMENT only if it returned positive.
19922 2017-04-25 Marek Polacek <polacek@redhat.com>
19925 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19926 first argument to type.
19928 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19931 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19932 type checks to test for compatibility instead of equality.
19934 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19935 Jakub Jelinek <jakub@redhat.com>
19938 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19940 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19941 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19942 the alignment computation, but return their maximum in warn_alignment.
19943 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19944 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19946 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19947 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19950 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19952 * config/arc/simdext.md (dmpyh): Fix typo.
19954 2017-04-25 Richard Biener <rguenther@suse.de>
19956 PR tree-optimization/80492
19957 * alias.c (compare_base_decls): Handle registers with asm
19958 specification conservatively.
19959 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19960 compare_base_decls returning dont-know properly.
19962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19964 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19965 (legitimate_offset_address_p): New function.
19966 (arc_legitimate_address_p): Use above function.
19968 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19970 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19972 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19974 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19975 ACCH registers whenever they are available.
19977 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19979 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19980 double regs fix when not used.
19982 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19984 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19986 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19987 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19989 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19991 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19992 when emitting short ADD instructions.
19994 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19996 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19998 (cmpsi_cc_c_insn): Likewise.
19999 (cbranchsi4_scratch): Compute proper instruction length using
20000 compact_hreg_operand.
20001 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20003 2017-04-25 Richard Biener <rguenther@suse.de>
20005 PR middle-end/80509
20006 * passes.c (pass_manager::pass_manager): Initialize
20007 m_name_to_pass_map.
20009 2017-04-25 Richard Biener <rguenther@suse.de>
20011 PR tree-optimization/79201
20012 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20014 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20017 * config/s390/vector.md: Split MEM->GPR vector moves for
20018 non-s_operand addresses.
20020 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20023 * config/s390/predicates.md (reload_const_wide_int_operand): New
20025 * config/s390/s390.md ("movti"): Remove d/P alternative.
20026 ("movti_bigconst"): New pattern definition.
20028 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20031 * s390-protos.h (s390_expand_cs_hqi): Removed.
20032 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20033 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20034 modes as well as CCZ1mode and CCZmode.
20035 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20036 signature of s390_emit_compare_and_swap.
20037 (s390_expand_cs_hqi): Likewise, make static.
20038 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20039 compare-and-swap, in some cases.
20040 (s390_expand_cs): Wrapper function.
20041 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20043 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20044 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20045 patterns for small and large integers. Forbid symref memory operands.
20046 Move expander to s390.c. Require cc register.
20047 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20048 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20049 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20050 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20051 symref memory operands. Remove CC mode and call s390_match_ccmode
20053 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20055 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20057 * config/s390/s390.md (define_peephole2): New peephole to help
20058 combining the load-and-test pattern with volatile memory.
20060 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20062 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20063 with CCZmode for TARGET_Z196.
20065 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20067 PR rtl-optimization/80501
20068 * combine.c (make_compound_operation_int): Set subreg_code to SET
20069 even for AND with mask of the sign bit of mode.
20071 PR rtl-optimization/80500
20072 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20073 sum's initial value.
20075 2017-04-25 Julian Brown <julian@codesourcery.com>
20076 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20078 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20080 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20082 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20084 2017-04-25 Julian Brown <julian@codesourcery.com>
20085 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20087 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20088 (thunderx2t99_sha): New Reservation.
20090 2017-04-25 Julian Brown <julian@codesourcery.com>
20091 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20093 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20094 type for 1-element load.
20096 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20098 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20100 2017-04-24 Martin Jambor <mjambor@suse.cz>
20102 PR tree-optimization/80293
20103 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20104 char arrays not totally scalarizable if it is false.
20105 (analyze_all_variable_accesses): Pass correct value in the new
20106 parameter. Add a statistics counter.
20108 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20110 PR middle-end/79931
20111 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20113 2017-04-24 Richard Biener <rguenther@suse.de>
20115 PR tree-optimization/80494
20116 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20117 out for complex types.
20119 2017-04-24 Richard Biener <rguenther@suse.de>
20121 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20122 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20123 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20124 (DFS): Adjust and never fail.
20125 (sccvn_dom_walker::fail): Remove.
20126 (sccvn_dom_walker::before_dom_children): Adjust.
20127 (run_scc_vn): Likewise and never fail.
20128 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20129 (pass_fre::execute): Likewise.
20131 2017-04-24 Richard Biener <rguenther@suse.de>
20133 PR tree-optimization/79725
20134 * tree-ssa-sink.c (statement_sink_location): Return whether
20135 failure reason was zero uses. Move that check later.
20136 (sink_code_in_bb): Deal with zero uses by removing the stmt
20139 2017-04-24 Richard Biener <rguenther@suse.de>
20142 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20143 pointer-based references.
20145 2017-04-24 Richard Biener <rguenther@suse.de>
20148 * pass_manager.h (pass_manager::operator new): Remove.
20149 (pass_manager::operator delete): Likewise.
20150 * passes.c (pass_manager::operator new): Remove.
20151 (pass_manager::operator delete): Likewise.
20152 (pass_manager::pass_manager): Zero individual pass members.
20154 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20157 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20158 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20159 Check "XEXP (src, 1)" operand here.
20160 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20161 Check "XEXP (src, 1)" operand here.
20162 (dimode_scalar_chain::make_vector_copies): Detect count register
20163 of a shift instruction. Zero extend count register from QImode
20164 to DImode to satisfy vector shift pattern count operand predicate.
20165 Substitute vector shift count operand with a DImode copy.
20166 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20169 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20171 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20172 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20173 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20174 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20175 (UNSPEC_NOREX_MEM): Remove definition.
20177 2017-04-21 Richard Biener <rguenther@suse.de>
20179 PR tree-optimization/79547
20180 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20181 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20182 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20183 without any constraints.
20185 2017-04-21 Richard Biener <rguenther@suse.de>
20187 PR tree-optimization/78847
20188 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20190 2017-04-21 Richard Biener <rguenther@suse.de>
20192 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20193 (build_distinct_type_copy): Likewise.
20194 (build_variant_type_copy): Likewise.
20195 * tree.c (build_qualified_type): Pass down mem-stat info.
20196 (build_distinct_type_copy): Likewise.
20197 (build_variant_type_copy): Likewise.
20199 2017-04-21 Richard Biener <rguenther@suse.de>
20201 PR tree-optimization/80237
20202 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20204 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20205 for a simplified result.
20207 2016-04-21 Richard Biener <rguenther@suse.de>
20209 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20210 sth as strict as a simple_iv but a chrec without symbols and an
20211 operand defined in the loop we are peeling (and not some subloop).
20212 (propagate_constants_for_unrolling): Propagate all constants.
20214 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20217 * config/i386/i386.c (print_reg): Remove assert for disalowed
20218 regno values, call output_operand_lossage instead.
20220 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20223 * config/i386/constraints.md (Yc): New register constraint.
20224 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20225 Use Yc constraint for alternative 2 of operand 0. Remove
20226 preferred_for_speed attribute.
20228 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20230 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20231 lastprivate clauses in SIMT case.
20233 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20235 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20237 2017-04-20 Richard Biener <rguenther@suse.de>
20239 PR tree-optimization/57796
20240 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20241 as N scalar stores.
20242 (vect_model_load_cost): Cost gathers as N scalar loads.
20244 2017-04-20 Richard Biener <rguenther@suse.de>
20246 * ggc-page.c (ggc_allocated_p): Rename to ...
20247 (safe_lookup_page_table_entry): ... this and return the lookup
20249 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20251 2017-04-20 Richard Biener <rguenther@suse.de>
20253 PR tree-optimization/80453
20254 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20255 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20256 from the conditions.
20257 (vn_phi_eq): Pass them down.
20258 (vn_phi_lookup): Record them.
20259 (vn_phi_insert): Likewise.
20261 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20263 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20264 uninitialized variable warning to avoid buffer overrun.
20266 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20269 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20270 is suppressed for '{ 0 }' in C.
20272 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20274 * BASE-VER: Set to 8.0.0.
20276 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20278 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20279 priority .init_array and .fini_array section with SECTION_NOTYPE
20282 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20284 PR middle-end/80423
20285 * tree.h (build_array_type): Add typeless_storage default argument.
20286 * tree.c (type_cache_hasher::equal): Also compare
20287 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20288 (build_array_type): Add typeless_storage argument, set
20289 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20291 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20292 (build_array_type): Likewise. Add typeless_storage argument.
20294 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20295 Jakub Jelinek <jakub@redhat.com>
20297 PR tree-optimization/80426
20298 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20299 operation on symbolic operands, also compute the overflow for the
20300 invariant part when the operation degenerates into a negation.
20302 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20305 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20306 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20309 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20311 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20314 * config/avr/avr.c (tree.h): Include it.
20315 (cgraph.h): Include it.
20316 (avr_encode_section_info): Don't warn for uninitialized progmem
20317 variable if it's just an alias.
20319 2017-04-19 Richard Biener <rguenther@suse.de>
20322 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20323 when needed by AutoPGO.
20325 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20328 * doc/lto.texi: Remove an extra 'that'.
20330 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20332 PR rtl-optimization/80429
20333 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20334 are only used in debug insns.
20336 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20337 Vladimir Makarov <vmakarov@redhat.com>
20339 * config/sparc/predicates.md (input_operand): Add comment. Return
20340 true for any memory operand when LRA is in progress.
20341 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20343 2017-04-18 Jeff Law <law@redhat.com>
20346 * mips.md ({return,simple_return}_internal): Do not overwrite
20349 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20351 PR tree-optimization/80443
20352 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20353 instead of adding 1, subtract -1 and similarly instead of subtracting
20356 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20358 PR rtl-optimization/80357
20359 * haifa-sched.c (tmp_bitmap): New variable.
20360 (model_recompute): Handle duplicate use records.
20361 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20362 (free_global_sched_pressure_data): Free it.
20364 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20367 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20368 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20369 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20370 instead of SYSTEM_HEADER_DIR.
20372 2017-04-18 Jeff Law <law@redhat.com>
20374 PR middle-end/80422
20375 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20376 predecessors after walking up the insn chain.
20378 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20381 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20382 sizetype type into debug info.
20384 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20387 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20388 unneeded test for TARGET_UPPER_REGS_SF.
20389 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20391 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20394 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20395 instead of gsi_after_labels.
20397 2017-04-18 Jeff Law <law@redhat.com>
20399 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20403 2017-04-13 Jeff Law <law@redhat.com>
20404 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20405 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20407 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20410 * config/avr/avr.c (intl.h): Include it.
20411 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20413 2017-04-18 Martin Liska <mliska@suse.cz>
20415 PR gcov-profile/78783
20416 * gcov-tool.c (gcov_output_files): Validate that destination
20417 file is either removed by the tool or by a user.
20419 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20420 Guy Benyei <guybe@mellanox.com>
20422 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20423 block, and do not negate it, the stored id is already negative.
20425 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20427 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20429 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20432 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20433 masks of options that should be turned off if the VSX vector
20434 options are turned off.
20435 (OTHER_P8_VECTOR_MASKS): Likewise.
20436 (OTHER_VSX_VECTOR_MASKS): Likewise.
20437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20438 rs6000_disable_incompatible_switches to validate no type switches
20440 (rs6000_incompatible_switch): New function to disallow turning on
20441 other vector options if -mno-vsx, -mno-power8-vector, or
20442 -mno-power9-vector are specified.
20444 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20446 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20448 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20450 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20451 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20452 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20453 (ARG_POINTER_CFA_OFFSET): Likewise.
20455 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20457 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20458 conditions to take advantage of various optimizations.
20460 2017-04-13 Jeff Law <law@redhat.com>
20462 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20463 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20464 (zero_extendsidi2_dext): Likewise.
20466 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20469 * fold-const.c (fold_ternary_loc): Revert
20470 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20473 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20475 PR rtl-optimization/80343
20476 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20477 new scratch pseudo.
20479 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20482 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20483 to ubsan_encode_value.
20485 2017-04-13 Jeff Law <law@redhat.com>
20487 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20488 appearing in DEBUG_INSNs.
20490 2017-04-13 Martin Liska <mliska@suse.cz>
20492 PR gcov-profile/80413
20493 * gcov-io.c (gcov_write_string): Copy to buffer just when
20494 allocated size is greater than zero.
20496 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20499 * dwarf2out.c (decls_for_scope): Ignore declarations of
20500 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20502 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20505 * ipa-visibility.c (non_local_p): Fix typos.
20506 (localize_node): When localizing symbol in same comdat group,
20507 dissolve the group only when we know external symbols are going
20509 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20511 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20513 PR tree-optimization/79390
20514 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20515 order does not result in usable sequence, retry with reversed operand
20521 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20522 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20523 op0 instead of fold_convert_loc (loc, type, arg0).
20525 2017-04-12 Jeff Law <law@redhat.com>
20527 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20528 has a delay slot in the generated code.
20530 * config/cris/cris.md (cris_preferred_reload_class): Return
20531 GENNONACR_REGS rather than GENERAL_REGS.
20533 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20536 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20537 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20538 signedness of the result type.
20540 2017-04-12 Richard Biener <rguenther@suse.de>
20541 Jeff Law <law@redhat.com>
20543 PR tree-optimization/80359
20544 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20545 trim stores to TARGET_MEM_REFs.
20547 2017-04-12 Richard Biener <rguenther@suse.de>
20549 PR tree-optimization/79390
20550 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20551 threading case even more.
20553 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20556 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20557 for quad_address_p for TImode, instead of just not indexed_address.
20559 2017-04-12 Richard Biener <rguenther@suse.de>
20560 Bernd Edlinger <bernd.edlinger@hotmail.de>
20562 PR middle-end/79671
20563 * alias.c (component_uses_parent_alias_set_from): Handle
20564 TYPE_TYPELESS_STORAGE.
20565 (get_alias_set): Likewise.
20566 * tree-core.h (tree_type_common): Add typeless_storage flag.
20567 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20568 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20569 for types containing members with TYPE_TYPELESS_STORAGE.
20570 (place_field): Likewise.
20571 (layout_type): Likewise for ARRAY_TYPE.
20572 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20573 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20574 TYPE_TYPELESS_STORAGE.
20575 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20577 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20580 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20581 first argument to type.
20583 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20587 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20588 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20589 (rs6000_expand_binop_builtin): Likewise.
20590 (rs6000_expand_ternop_builtin): Likewise; also add missing
20591 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20592 vshasigma built-ins.
20593 * doc/extend.texi: Document that vec_xxpermdi's third argument
20594 must be a constant.
20596 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20598 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20599 Use shift_const cost parameter when calculating gain of STV shifts.
20601 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20603 PR rtl-optimization/70478
20604 * lra-constraints.c (process_alt_operands): Check memory for
20605 disfavoring memory insn operand.
20607 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20609 PR middle-end/80100
20610 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20611 left shift in unsigned HOST_WIDE_INT type.
20613 PR rtl-optimization/80385
20614 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20615 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20618 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20619 if they have any depend clauses.
20621 2017-04-11 Martin Liska <mliska@suse.cz>
20624 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20625 * ipa-split.c (split_function): Create a local comdat symbol
20626 if caller is in a comdat group.
20628 2017-04-11 Martin Liska <mliska@suse.cz>
20631 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20634 2017-04-11 Martin Sebor <msebor@redhat.com>
20636 PR middle-end/80364
20637 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20638 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20640 (directive::set_width, directive::set_precision, format_character):
20642 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20645 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20648 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20649 conflict, set target->arch_name instead of target->cpu_name.
20651 2017-04-11 Richard Biener <rguenther@suse.de>
20653 PR tree-optimization/80374
20654 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20655 build_zero_cst, remove fold_convertible_p check again.
20657 2017-04-11 Martin Liska <mliska@suse.cz>
20660 * ubsan.c (instrument_object_size): Do not instrument register
20663 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20666 * config/i386/i386-builtin-types.def
20667 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20668 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20669 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20670 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20671 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20672 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20673 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20674 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20675 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20676 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20677 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20678 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20679 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20680 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20681 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20682 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20683 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20684 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20685 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20686 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20687 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20688 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20689 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20690 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20691 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20692 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20693 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20694 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20695 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20696 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20697 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20698 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20699 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20700 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20701 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20702 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20703 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20704 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20705 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20706 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20707 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20708 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20709 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20710 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20711 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20712 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20714 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20715 flag to second_arg_count, handle 4 argument function type _COUNT
20716 aliases, handle second_arg_count on second argument rather than last.
20718 2017-04-10 Jeff Law <law@redhat.com>
20720 PR tree-optimization/80374
20721 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20722 record anything if we can not convert integer_zero_node to the
20725 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20728 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20729 Enhance special handling given to the TARGET_P9_MINMAX option in
20730 relation to certain other options.
20732 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20734 PR tree-optimization/80153
20735 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20736 remove POINTER_PLUS_EXPR's base part directly, rather than through
20739 2017-04-10 Richard Biener <rguenther@suse.de>
20740 Bin Cheng <bin.cheng@arm.com>
20742 PR tree-optimization/80153
20743 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20744 the first element of pointer type aff_tree. Build result expr in
20746 (add_elt_to_tree): Convert to type unconditionally. Remove other
20747 fold_convert calls.
20748 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20749 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20751 2017-04-10 Richard Biener <rguenther@suse.de>
20753 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20756 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20758 PR rtl-optimization/70478
20759 * lra-constraints.c (curr_small_class_check): New.
20760 (update_and_check_small_class_inputs): New.
20761 (process_alt_operands): Update curr_small_class_check. Disfavor
20762 alternative insn memory operands. Check available regs for small
20765 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20768 * config/mips/mips.opt (-mvirt): Update description.
20769 * doc/invoke.texi (-mvirt): Likewise.
20771 2017-04-10 Richard Biener <rguenther@suse.de>
20773 PR middle-end/80362
20774 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20775 looking for NEGATE_EXPR in -A / -B to A / B folding.
20777 2017-04-10 Martin Liska <mliska@suse.cz>
20779 PR gcov-profile/80224
20780 * gcov.c (print_usage): Fix usage string.
20781 (get_gcov_intermediate_filename): Remove.
20782 (output_gcov_file): Use both for normal and intermediate format.
20783 (generate_results): Do not initialize special file for
20784 intermediate format.
20786 2017-04-10 Richard Biener <rguenther@suse.de>
20788 PR tree-optimization/80304
20789 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20792 2017-04-10 Nathan Sidwell <nathan@acm.org>
20795 * config/rs6000/rs6000.c (rs6000_vector_type): New.
20796 (rs6000_init_builtins): Use it.
20798 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20800 * config/arm/arm.md (<mrc>): Add mode to SET source.
20801 (<mrrc>): Likewise.
20803 2017-04-10 Richard Biener <rguenther@suse.de>
20805 PR middle-end/80344
20806 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
20808 2017-04-10 Jakub Jelinek <jakub@redhat.com>
20811 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
20812 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
20813 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
20814 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
20815 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
20816 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20817 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20818 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20819 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20820 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20821 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20822 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20823 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20824 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20825 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20826 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20827 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20828 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20829 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20830 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20831 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20832 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20833 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20835 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20837 PR rtl-optimization/70478
20838 * lra-constraints.c: Reverse the last patch.
20840 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20842 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20843 Add comment for WCHAR_T.
20845 2017-04-08 Martin Liska <mliska@suse.cz>
20848 2017-04-07 Martin Liska <mliska@suse.cz>
20851 * ipa-split.c (split_function): Add function part to a same comdat
20854 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20857 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20859 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20861 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20862 (vec_pairing): ...to this.
20863 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20864 (rs6000_sched_init): Adjust for name change.
20865 (struct rs6000_sched_context): Likewise.
20866 (rs6000_init_sched_context): Likewise.
20867 (rs6000_set_sched_context): Likewise.
20869 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20875 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20877 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20878 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20879 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20881 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20883 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20885 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20887 PR rtl-optimization/70703
20888 * ira-color.c (update_conflict_hard_regno_costs): Use
20889 int64_t instead of HOST_WIDE_INT.
20891 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20893 PR rtl-optimization/70478
20894 * lra-constraints.c (process_alt_operands): Disfavor alternative
20895 insn memory operands.
20897 2017-04-07 Jeff Law <law@redhat.com>
20899 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20900 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20902 2017-04-07 Martin Liska <mliska@suse.cz>
20905 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20906 Show error message instead of an ICE.
20908 2017-04-07 Martin Liska <mliska@suse.cz>
20911 * ipa-split.c (split_function): Add function part to a same comdat
20914 2017-04-07 Richard Biener <rguenther@suse.de>
20916 PR middle-end/80341
20917 * tree.c (get_unwidened): Also handle ! for_type case for
20919 * convert.c (do_narrow): Split out from ...
20920 (convert_to_integer_1): ... here. Do not pass final truncation
20921 type to get_unwidened for TRUNC_DIV_EXPR.
20923 2017-04-07 Richard Biener <rguenther@suse.de>
20925 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20927 (aff_combination_const): Adjust.
20928 (aff_combination_scale): Likewise.
20929 (aff_combination_add_elt): Likewise.
20930 (aff_combination_add_cst): Likewise.
20931 (aff_combination_convert): Likewise.
20932 (add_elt_to_tree): Likewise. Remove unused argument.
20933 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20935 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20937 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20939 * config/arm/arm.c (arm_default_short_enums): Use
20940 ARM_DEFAULT_SHORT_ENUMS.
20941 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20943 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20946 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20947 members with redundant out-of-class redeclaration.
20949 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20952 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20953 * config/i386/i386.md (*zero_extendsidi2):
20954 Add (?*x,*x) and (?*v,*v) alternatives.
20956 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20959 * config/i386/i386.c (ix86_expand_builtin)
20960 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20961 mode from insn data. Convert operands to insn operand mode.
20962 Copy operands that don't satisfy insn predicate to a register.
20964 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20966 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20969 2017-04-06 Richard Biener <rguenther@suse.de>
20971 PR tree-optimization/80334
20972 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20973 preserve alignment of accesses.
20975 2017-04-06 Richard Biener <rguenther@suse.de>
20977 PR tree-optimization/80262
20978 * tree-sra.c (build_ref_for_offset): Preserve address-space
20980 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20981 Drop useless address-space information on MEM_REF offsets.
20983 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20985 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20987 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20989 PR rtl-optimization/70703
20990 * ira-color.c (update_conflict_hard_regno_costs): Use
20991 HOST_WIDE_INT instead of long.
20993 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20996 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20997 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20999 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21000 for x86_64 target. Handle -m3dnowa option.
21002 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21004 PR rtl-optimization/70703
21005 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21006 (update_conflict_hard_regno_costs): Use long instead of unsigned
21007 arithmetic for cost calculation.
21009 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21010 Bernd Edlinger <bernd.edlinger@hotmail.de>
21013 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21016 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21019 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21020 ptr_mode with Pmode throughout.
21021 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21022 into probe_stack_range and use DImode.
21024 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21027 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21028 call_eh_return is true.
21030 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21032 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21033 Initialize last_match_fntype_index.
21035 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21038 * tree-nvr.c: Include internal-fn.h.
21039 (pass_return_slot::execute): Ignore internal calls without
21042 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21043 Richard Biener <rguenther@suse.de>
21046 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21047 captures used multiple times, except for the last use.
21048 * generic-match-head.c: Include gimplify.h.
21050 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21052 PR tree-optimization/79390
21053 * target.h (struct noce_if_info): Declare.
21054 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21055 * target.def (noce_conversion_profitable_p): New target hook.
21056 * ifcvt.h (struct noce_if_info): New type, moved from ...
21057 * ifcvt.c (struct noce_if_info): ... here.
21058 (noce_conversion_profitable_p): Renamed to ...
21059 (default_noce_conversion_profitable_p): ... this. No longer
21061 (noce_try_store_flag_constants, noce_try_addcc,
21062 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21063 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21064 instead of noce_conversion_profitable_p.
21065 * config/i386/i386.c: Include ifcvt.h.
21066 (ix86_option_override_internal): Don't override
21067 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21068 (ix86_noce_conversion_profitable_p): New function.
21069 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21070 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21071 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21072 * doc/tm.texi: Regenerated.
21074 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21076 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21079 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21082 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21083 instructions for small multiply cores.
21085 2017-04-04 Jeff Law <law@redhat.com>
21087 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21089 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21090 that are not set by the loop over the elements.
21092 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21095 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21096 int mode, convert_modes it to mode as unsigned, otherwise use
21097 lowpart_subreg to mode rather than SImode.
21098 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21099 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21100 Use DImode instead of SImode for the shift count operand.
21101 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21104 2017-04-04 Richard Biener <rguenther@suse.de>
21106 PR middle-end/80281
21107 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21108 arithmetic done for the negate or the plus. Simplify.
21109 (A - (-B) -> A + B): Likewise.
21110 * fold-const.c (split_tree): Make sure to not negate pointers.
21112 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21114 PR rtl-optimization/60818
21115 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21116 a compare of comparisons with the thing compared if this results
21117 in a different machine mode.
21119 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21121 * alias.c (base_alias_check): Fix typo in comment.
21122 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21123 * cgraphunit.c (symbol_table::compile): Likewise.
21124 * collect2.c (maybe_run_lto_and_relink): Likewise.
21125 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21126 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21127 * config/avr/avr.c (avr_map_op_t): Likewise.
21128 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21129 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21130 * config/epiphany/epiphany.md (movcc): Likewise.
21131 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21132 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21134 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21135 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21136 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21137 * config/sh/sh.c (sh_rtx_costs): Likewise.
21138 * fold-const.c (fold_truth_andor): Likewise.
21139 * genautomata.c (collapse_flag): Likewise.
21140 * gengtype.h (struct type::u::s): Likewise.
21141 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21142 * input.c (FORMAT_AMOUNT): Likewise.
21143 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21144 (known_aggs_to_agg_replacement_list): Likewise.
21145 * ipa-inline-analysis.c: Likewise.
21146 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21147 * ipa-polymorphic-call.c
21148 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21149 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21150 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21152 * modulo-sched.c (apply_reg_moves): Likewise.
21153 * omp-expand.c (build_omp_regions_1): Likewise.
21154 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21155 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21156 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21157 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21158 * value-prof.c: Likewise.
21159 * var-tracking.c (val_reset): Likewise.
21161 2017-04-03 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/80275
21164 * fold-const.c (split_address_to_core_and_offset): Handle
21167 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21169 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21170 descriptors is at least equal to that of functions.
21172 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21174 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21176 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21179 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21180 (mov<IMOD4:mode>): New expander.
21181 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21183 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21185 PR rtl-optimization/79405
21186 * fwprop.c (propagations_left): New variable.
21187 (forward_propagate_into): Decrement it.
21188 (fwprop_init): Initialize it.
21189 (fw_prop): If the variable has reached zero, stop propagating.
21190 (fwprop_addr): Ditto.
21192 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21195 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21196 a FUNCTION_DECL, pass it as decl instead of origin to
21199 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21201 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21204 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21207 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21208 TARGET_VSX_SMALL_INTEGER.
21210 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21212 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21213 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21215 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21217 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21218 extraction from odd-numbered MSA register.
21220 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21222 PR middle-end/80173
21223 * expmed.c (store_bit_field_1): Don't attempt to create
21224 a word subreg out of hard registers wider than word if they
21225 have HARD_REGNO_NREGS of 1 for their mode.
21227 PR middle-end/80163
21228 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21229 conversions to integer types wider than word and pointer.
21232 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21233 (rtx_equal_for_cselib_p): Pass 0 to it.
21234 * cselib.c (cselib_hasher::equal): Likewise.
21235 (rtx_equal_for_cselib_1): Add depth argument. If depth
21236 is 128, don't look up VALUE locs and punt. Increment
21237 depth in recursive calls when walking VALUE locs.
21239 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21241 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21242 (make_gcov_file_name): Use the canonical path name for generating
21244 (read_line): Fix handling of files with ascii null bytes.
21246 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21248 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21249 to initialise a vector register instead
21250 of using a const_int.
21252 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21254 PR translation/80189
21255 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21256 diagnostic messages.
21258 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21261 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21262 (dfp_diex_<mode>): Update mode of operand 1.
21263 * doc/extend.texi (dxex, dxexq): Document change to return type.
21264 (diex, diexq): Document change to argument type.
21266 2017-03-30 Martin Jambor <mjambor@suse.cz>
21269 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21270 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21271 it reflects the signature changes performed at the callee side.
21272 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21273 to cgraph_build_function_type_skip_args.
21274 (build_function_decl_skip_args): Adjust call to the above function.
21276 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21279 * config/i386/sse.md
21280 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21281 register as dest whenever it is a MEM not rtx_equal_p to the
21282 corresponding dup operand, and when forcing into reg move the
21283 reg into the memory afterwards.
21284 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21285 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21286 for the force_reg mode.
21287 (avx512vl_vextractf128<mode>): Use register as dest either
21288 always when a MEM, or when it is a MEM not rtx_equal_p to the
21289 corresponding dup operand, or even not when it is a CONST_VECTOR
21290 depending on the mode and lo vs. hi.
21291 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21293 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21294 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21295 Likewise. Require that operands[2] is even.
21296 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21297 Remove extraneous parens. Require that operands[2] is a multiple
21299 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21300 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21301 disallow memory then.
21303 2017-03-30 Richard Biener <rguenther@suse.de>
21305 PR tree-optimization/77498
21306 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21307 to non-constants over backedges.
21309 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21311 PR rtl-optimization/80233
21312 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21313 as last_combined_insn. Do not test for BARRIER_P separately.
21315 2017-03-29 Andreas Schwab <schwab@suse.de>
21318 * calls.c (prepare_call_address): Convert funexp to Pmode before
21319 copying to temp reg.
21321 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21323 PR tree-optimization/80158
21324 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21325 Handle possible future case of more than one alternate
21327 (replace_rhs_if_not_dup): Likewise.
21328 (replace_one_candidate): Likewise.
21330 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21332 PR rtl-optimization/80193
21333 * ira.c (ira): Do not check allocation for LRA.
21335 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21337 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21338 (nvptx_output_simt_exit): Declare.
21339 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21340 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21341 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21342 (nvptx_declare_function_name): ...here. Emit declaration of local
21343 memory space buffer for omp_simt_enter insn.
21344 (nvptx_output_unisimt_switch): New.
21345 (nvptx_output_softstack_switch): New.
21346 (nvptx_output_simt_enter): New.
21347 (nvptx_output_simt_exit): New.
21348 * config/nvptx/nvptx.h (struct machine_function): New fields
21349 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21350 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21351 (UNSPECV_SIMT_EXIT): Ditto.
21352 (omp_simt_enter_insn): New insn.
21353 (omp_simt_enter): New expansion.
21354 (omp_simt_exit): New insn.
21355 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21357 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21358 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21359 (expand_GOMP_SIMT_EXIT): New.
21360 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21361 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21362 (GOMP_SIMT_EXIT): Ditto.
21363 * target-insns.def (omp_simt_enter): New insn.
21364 (omp_simt_exit): Ditto.
21365 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21367 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21368 (lower_rec_input_clauses): Likewise.
21369 (lower_lastprivate_clauses): Handle SIMT privatization.
21371 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21372 (ompdevlow_adjust_simt_enter): New.
21373 (find_simtpriv_var_op): New.
21374 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21375 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21377 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21378 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21379 (copy_decl_for_dup_finish): Ditto.
21381 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21383 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21386 * config/i386/i386.c (ix86_option_override_internal): Always
21387 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21389 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21391 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21393 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21395 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21396 mark new edge's irreducible flag accordign to it.
21397 (vect_do_peeling): Check loop preheader edge's irreducible flag
21398 and pass it to function slpeel_add_loop_guard.
21400 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21402 PR tree-optimization/80218
21403 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21404 Update block frequencies and counts.
21406 2017-03-28 Richard Biener <rguenther@suse.de>
21408 PR tree-optimization/78644
21409 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21410 of a simplification result we may not use it at all.
21412 2017-03-28 Richard Biener <rguenther@suse.de>
21415 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21416 without arguments, generate default definition of a SSA name.
21418 2017-03-28 Richard Biener <rguenther@suse.de>
21420 PR middle-end/80222
21421 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21422 TYPE_REF_CAN_ALIAS_ALL references.
21423 * fold-const.c (fold_indirect_ref_1): Likewise.
21425 2017-03-28 Martin Liska <mliska@suse.cz>
21428 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21429 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21431 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21432 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21434 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21435 (EXTRA_SPECS): Define.
21436 (SUBTARGET_EXTRA_SPECS): Likewise.
21437 (SUBTARGET_CPP_SPEC): Likewise.
21438 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21439 SUBTARGET_EXTRA_SPECS.
21440 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21442 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21444 * config/arc/simdext.md (vst64_insn): Update pattern.
21445 (vld32wh_insn): Likewise.
21446 (vld32wl_insn): Likewise.
21447 (vld64_insn): Likewise.
21448 (vld32_insn): Likewise.
21450 2017-03-28 Marek Polacek <polacek@redhat.com>
21453 * fold-const.c (fold_comparison): Use protected_set_expr_location
21454 instead of SET_EXPR_LOCATION.
21456 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21458 * tree.c (add_expr): Avoid name lookup warning.
21460 2017-03-27 Jeff Law <law@redhat.com>
21462 PR tree-optimization/80216
21463 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21464 function name. Limit recursion depth.
21465 (record_temporary_equivalences): Corresponding changes.
21467 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21469 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21472 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21475 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21477 * cfgcleanup.c (reg_note_cfa_p): New array.
21478 (insns_have_identical_cfa_notes): New function.
21479 (old_insns_match_p): Don't cross-jump in between /f
21480 and non-/f instructions. If both i1 and i2 are frame related,
21481 verify all CFA notes, their order and content.
21483 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21486 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21487 HImode and SImode with zero extend to DImode to one insn.
21488 (bswap<mode>2_extenddi): Likewise.
21489 (bswapsi2_extenddi): Likewise.
21490 (bswaphi2_extendsi): Likewise.
21491 (bswaphi2): Combine bswap HImode and SImode into one insn.
21492 Separate memory insns from swapping register.
21493 (bswapsi2): Likewise.
21494 (bswap<mode>2): Likewise.
21495 (bswaphi2_internal): Delete, no longer used.
21496 (bswapsi2_internal): Likewise.
21497 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21498 store, and gpr<-gpr swap insns.
21499 (bswap<mode>2_store): Likewise.
21500 (bswaphi2_reg): Register only splitter, combine with the splitter.
21501 (bswaphi2 splitter): Likewise.
21502 (bswapsi2_reg): Likewise.
21503 (bswapsi2 splitter): Likewise.
21504 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21505 the insns into load, store, and register/register insns.
21506 (bswapdi2_ldbrx): Likewise.
21507 (bswapdi2_load): Likewise.
21508 (bswapdi2_store): Likewise.
21509 (bswapdi2_reg): Likewise.
21511 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21513 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21514 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21516 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21519 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21521 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21522 special handling for target option conflicts between dform
21523 options (-mpower9-dform, -mpower9-dform-vector,
21524 -mpower9-dform-scalar) and -mno-direct-move.
21526 2017-03-27 Richard Biener <rguenther@suse.de>
21528 PR tree-optimization/80181
21529 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21531 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21533 * config/arc/predicates.md (move_double_src_operand): Replace the
21534 call to move_double_src_operand with a call to address_operand.
21536 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21538 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21539 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21540 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21542 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21544 * config/arc/predicates.md (long_immediate_loadstore_operand):
21545 Consider scaled addresses cases.
21547 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21549 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21550 restored when in interrupt.
21551 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21552 doesn't have delay slot.
21554 2017-03-27 Richard Biener <rguenther@suse.de>
21557 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21558 inlined thunk clones.
21560 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21563 * asan.c (instrument_derefs): Copy over last operand from
21564 original COMPONENT_REF to the new COMPONENT_REF with
21565 DECL_BIT_FIELD_REPRESENTATIVE.
21566 * ubsan.c (instrument_object_size): Likewise.
21568 2017-03-27 Richard Biener <rguenther@suse.de>
21570 PR tree-optimization/80170
21571 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21572 sure DR/SCEV didnt fold in constants we do not see when looking
21573 at the reference base alignment.
21575 2017-03-27 Richard Biener <rguenther@suse.de>
21577 PR middle-end/80171
21578 * gimple-fold.c (fold_ctor_reference): Properly guard against
21579 NULL return value from canonicalize_constructor_val.
21581 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21584 * config/i386/i386.c (ix86_expand_builtin)
21585 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21586 flags reg setting and flags reg using instructions.
21587 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21588 clobbering instructions to zero extend op2.
21590 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21592 * doc/install.texi (Configuration) <--with-aix-soname>:
21593 Update link to AIX ld.
21595 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21597 PR rtl-optimization/80160
21598 PR rtl-optimization/80159
21599 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21600 reg_alternate_class into account.
21602 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21605 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21606 to consider in curr_insn_transform.
21608 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21610 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21611 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21612 and emit_mode_inner.
21614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21616 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21617 argument to the overloaded builtin variants. Use the new flag to
21618 deprecate certain builtin variants.
21619 * config/s390/s390-builtin-types.def: Add new builtin types.
21620 * config/s390/s390-builtins.h: Support new flags field for
21621 overloaded builtins.
21622 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21623 (s390_macro_to_expand): Enable vector float data type.
21624 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21625 builtins by incrementing the __VEC__ version number.
21626 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21628 (s390_resolve_overloaded_builtin): Emit error messages depending
21629 on the builtin flags.
21630 * config/s390/s390.c (s390_expand_builtin): Support additional
21631 flags argument. Change error message to match the messages
21632 emitted in s390-c.c.
21633 * config/s390/s390.md: New UNSPEC_* constants.
21634 (op_type): Add new instruction types.
21635 * config/s390/vecintrin.h: Add new builtins and test data class
21637 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21638 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21639 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21640 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21641 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21642 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21644 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21645 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21646 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21647 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21649 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21650 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21651 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21652 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21654 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21655 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21656 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21657 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21658 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21659 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21660 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21662 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21663 ("vec_scatter_element<V_HW_4:mode>_DI")
21664 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21665 ("vec_fpint<mode>", "vflls")
21666 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21667 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21668 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21669 ("*vec_cmphe<mode>_cc"): ... these.
21671 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21672 mode constant instead of magic value.
21674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21676 * config/s390/s390.c (s390_expand_vec_compare): Support other
21677 vector floating point modes than just V2DF.
21678 (s390_expand_vcond): Likewise.
21679 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21680 (s390_cannot_change_mode_class): Prevent mode changes between TF
21681 and V1TF in vector registers.
21682 * config/s390/s390.md (DF, SF): New mode attributes.
21683 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21684 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21685 SFmode support for VRs.
21686 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21688 (VFT, VF_HW): New mode iterators.
21689 (vw, sdx): New mode attributes.
21690 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21691 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21692 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21693 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21694 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21695 also the new vector floating point modes. Renaming to ...
21697 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21698 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21699 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21700 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21701 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21702 ("vec_unordered<mode>"): ... these.
21704 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21705 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21706 ("*vec_extendv2df"): New insn definitions.
21708 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21710 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21711 ("mulditi3_2", "*muldi3_sign"): New patterns.
21712 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21713 rename the pattern definition.
21715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21717 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21719 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21721 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21723 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21724 instruction if possible.
21725 * config/s390/vector.md (vec_halfnumelts): New mode
21727 ("*vec_vllezlf<mode>"): New pattern.
21729 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21731 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21732 ("popcountv4si2", "popcountv2di2"): Rename to ...
21733 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21734 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21736 ("popcount<mode>2_vxe"): New pattern.
21738 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21740 * common/config/s390/s390-common.c (processor_flags_table): Add
21742 * config.gcc: Add arch12.
21743 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21744 Default to arch12 for unknown CPU model numbers.
21745 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21746 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21747 PROCESSOR_max sanity check.
21748 * config/s390/s390-opts.h (enum processor_type): Add
21750 * config/s390/s390.c (processor_table): Add arch12.
21751 (s390_expand_builtin): Add check for B_VXE flag.
21752 (s390_issue_rate): Add PROCESSOR_ARCH12.
21753 (s390_get_sched_attrmask): Likewise.
21754 (s390_get_unit_mask): Likewise.
21755 (s390_sched_score): Enable z13 scheduling for arch12.
21756 (s390_sched_reorder): Likewise.
21757 (s390_sched_variable_issue): Likewise.
21758 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21760 (s390_tune_attr): Use z13 scheduling also for arch12.
21761 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21762 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21763 (TARGET_VXE_P): New macros.
21764 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21765 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21766 * config/s390/s390.opt: Add arch12 as processor_type.
21768 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21770 * config/s390/s390.md
21771 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21772 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21773 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21775 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21776 Rename expanders to ...
21778 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21779 ("fixuns_truncdddi2_emu"): ... these.
21781 ("fixuns_trunc<mode>si2_emu"): New expander.
21783 ("*fixuns_truncdfdi2_z13"): Rename to ...
21784 ("*fixuns_truncdfdi2_vx"): ... this.
21786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21788 * config/s390/2964.md: Remove the single element vector compare
21789 instructions which are no longer used.
21790 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21792 (s390_canonicalize_comparison): Remove handling of DFmode
21794 (s390_expand_vec_compare_scalar): Remove function.
21795 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
21796 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
21798 ("*cmp<mode>_ccs"): Add wfcdb instruction.
21800 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21802 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
21804 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
21805 will anyway by matched by mov<mode>_64dfp.
21807 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21809 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
21810 vlef/vstef. Add missing operand to vleif.
21812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21814 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
21815 pair for all vector types with 64 bit elements.
21816 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21817 * config/s390/vector.md (V_HW_64): ... here.
21818 (V_128_NOSINGLE): New mode iterator.
21819 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21820 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21821 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21822 ("*vec_load_pairv2di"): Change to ...
21823 ("*vec_load_pair<mode>"): ... this one.
21825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21827 * config/s390/constraints.md: Add comments.
21828 (jKK): Reject element sizes > 8 bytes.
21829 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21831 * config/s390/s390.md: Add the s_operand checks formerly in
21832 s390_split_ok_p to various splitters where they are still
21834 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21835 for 128 bit vectors. Plus two splitters.
21837 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21839 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21845 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21846 error if the boundary argument is not constant.
21848 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21850 PR rtl-optimization/80112
21851 * loop-doloop.c (doloop_condition_get): Don't check condition
21852 if cmp isn't SET with IF_THEN_ELSE src.
21854 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21856 PR tree-optimization/80158
21857 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21858 replacing a candidate statement, also replace it for the
21859 candidate's alternate interpretation.
21860 (replace_rhs_if_not_dup): Likewise.
21861 (replace_one_candidate): Likewise.
21863 2017-03-24 Richard Biener <rguenther@suse.de>
21865 PR tree-optimization/80167
21866 * graphite-isl-ast-to-gimple.c
21867 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21869 (translate_isl_ast_to_gimple::get_rename): Likewise.
21871 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21874 handling of certain combinations of target options, including the
21875 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21876 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21878 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21881 * config/arm/arm.md (*load_multiple): Add reload_completed to
21882 matching condition.
21884 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21885 Richard Biener <rguenth@suse.de>
21887 PR tree-optimization/79908
21888 PR tree-optimization/80136
21889 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21890 been cast away, gimplify_and_add suffices.
21892 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21894 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21896 2017-03-23 Richard Biener <rguenther@suse.de>
21898 PR tree-optimization/80032
21899 * gimplify.c (gimple_push_cleanup): Forced unconditional
21900 cleanups still have to go to the conditional_cleanups
21903 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21905 PR tree-optimization/80072
21906 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21908 (next_operand_entry_id): Change type to unsigned int.
21909 (sort_by_operand_rank): Make sure to return the right return value
21910 even if unsigned fields are bigger than INT_MAX.
21911 (struct oecount): Change cnt and id type to unsigned int.
21912 (oecount_hasher::equal): Formatting fix.
21913 (oecount_cmp): Make sure to return the right return value
21914 even if unsigned fields are bigger than INT_MAX.
21915 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21918 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21919 TREE_READONLY on result if writing it more than once.
21922 * doc/invoke.texi (-fsanitize=thread): Document that with
21923 -fnon-call-exceptions atomics are not able to throw
21927 * tsan.c: Include tree-eh.h.
21928 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21929 maybe_clean_or_replace_eh_stmt where needed.
21930 (instrument_memory_accesses): Add cfg_changed argument.
21931 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21932 if it returned true.
21933 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21935 PR rtl-optimization/63191
21936 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21937 wrapper function, moved the whole old content into ...
21938 (ix86_delegitimize_address_1): ... this. New inline function.
21939 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21940 true as last argument instead of ix86_delegitimize_address.
21942 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21944 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21945 cortexa57_branch_cost.
21947 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21949 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21951 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21954 * doc/md.texi (Constraints): Document wA constraint.
21955 * config/rs6000/constraints.md (wA): New.
21956 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21957 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21958 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21959 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21961 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21964 * gimplify.c (is_oacc_declared): New function.
21965 (oacc_default_clause): Use it to set default flags for acc declared
21966 variables inside parallel regions.
21967 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21968 declared variables.
21969 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21970 declare attribute to any decl as necessary.
21972 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21975 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21976 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21977 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21978 (arm_arch_lpae): This.
21979 * config/arm/arm.c (arm_arch7ve): Rename into ...
21980 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21981 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21984 2017-03-22 Martin Liska <mliska@suse.cz>
21987 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21988 error message instead of an ICE.
21990 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21992 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21994 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21996 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21998 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22001 2017-03-21 Martin Sebor <msebor@redhat.com>
22003 * doc/extend.texi: Use "cannot" instead of "can't."
22004 * doc/hostconfig.texi: Same.
22005 * doc/install.texi: Same.
22006 * doc/invoke.texi: Same.
22007 * doc/loop.texi: Same.
22008 * doc/md.texi: Same.
22009 * doc/objc.texi: Same.
22010 * doc/rtl.texi: Same.
22011 * doc/tm.texi: Same.
22012 * doc/tm.texi.in: Same.
22013 * doc/trouble.texi: Same.
22015 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22018 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22019 (collect_checksum_attributes): Set it.
22020 (die_checksum_ordered): Use it.
22022 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22024 PR tree-optimization/79908
22025 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22026 change: For a VA_ARG whose LHS has been cast away, use
22027 force_gimple_operand to construct the side effects.
22029 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22031 PR translation/80001
22032 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22033 more amenable to translation.
22034 (oacc_loop_auto_partitions): Likewise.
22036 2017-03-21 Marek Polacek <polacek@redhat.com>
22037 Martin Sebor <msebor@redhat.com>
22039 PR tree-optimization/80109
22040 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22041 on INTEGRAL_TYPE_P.
22043 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22044 Segher Boessenkool <segher@kernel.crashing.org>
22047 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22048 check reg_used_between_p between insn and one of succ or succ2
22049 depending on if succ is artificial insn not inserted into insn
22052 2017-03-21 Martin Liska <mliska@suse.cz>
22054 PR gcov-profile/80081
22055 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22056 * doc/gcc.texi: Include gcov-dump stuff.
22057 * doc/gcov-dump.texi: New file.
22059 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22061 PR rtl-optimization/79150
22062 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22063 conditional jump, if the jump is the last insn of the loop.
22065 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22066 Richard Biener <rguenth@suse.de>
22068 PR tree-optimization/79908
22069 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22070 been cast away, use force_gimple_operand to construct the side
22073 2017-03-21 Martin Liska <mliska@suse.cz>
22075 PR libfortran/79956
22076 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22079 2017-03-21 Brad Spengler <spender@grsecurity.net>
22082 * plugin.c (htab_hash_plugin): New function.
22083 (add_new_plugin): Use it and adjust.
22084 (parse_plugin_arg_opt): Adjust.
22085 (init_one_plugin): Likewise.
22087 2017-03-21 Richard Biener <rguenther@suse.de>
22089 PR tree-optimization/80032
22090 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22091 if set force the cleanup to happen unconditionally.
22092 (gimplify_target_expr): Push inserted clobbers with force_uncond
22093 to avoid them being removed by control-dependent DCE.
22095 2017-03-21 Richard Biener <rguenther@suse.de>
22097 PR tree-optimization/80122
22098 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22099 va_arg_pack_len when the inlined call stmt requires pack
22101 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22103 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22106 * tsan.c (instrument_builtin_call): If the memory model argument
22107 is not a constant, assume it is valid.
22110 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22113 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22115 PR rtl-optimization/79910
22116 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22117 if its dest is used by an insn before I2 (other than the combined
22118 insns themselves, which are properly handled already).
22120 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22123 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22125 * combine.c (record_used_regs): New static function.
22126 (try_combine): Handle situations where there is an additional
22127 instruction between I2 and I3 which needs to have a LOG_LINK
22131 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22133 * combine.c (try_combine): Delete redundant i1 test. Call
22134 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22136 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22139 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22140 alternatives 13/14.
22142 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22144 PR tree-optimization/80054
22145 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22146 the optimization if a PHI or any of its arguments is not dominated
22147 by the candidate's basis. Use gphi* rather than gimple* as
22149 (replace_profitable_candidates): Clean up a gimple* variable that
22150 should be a gphi* variable.
22152 2017-03-20 Martin Sebor <msebor@redhat.com>
22155 * doc/extend.texi (attribute constructor): Document present limitation.
22157 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22160 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22161 __POWER9_VECTOR__ #ifdef control, change template definition to
22162 use Power9-specific built-in function.
22163 (vec_any_eq): Likewise.
22164 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22165 to control outcomes from this test.
22166 (vector_ae_<mode>p): For VEC_F modes, likewise.
22168 2017-03-20 Ian Lance Taylor <iant@google.com>
22170 * config/i386/i386.c (ix86_function_regparm): Save an extra
22171 register for -fsplit-stack with DECL_STATIC_CHAIN.
22173 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22176 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22177 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22179 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22181 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22183 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22186 2017-03-20 Marek Polacek <polacek@redhat.com>
22189 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22191 2017-03-20 Richard Biener <rguenther@suse.de>
22193 PR tree-optimization/80113
22194 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22195 allocate extra SSA name for PHI def.
22196 (add_close_phis_to_outer_loops): Likewise.
22197 (add_close_phis_to_merge_points): Likewise.
22198 (copy_loop_close_phi_args): Likewise.
22199 (copy_cond_phi_nodes): Likewise.
22201 2017-03-20 Martin Liska <mliska@suse.cz>
22203 PR middle-end/79753
22204 * tree-chkp.c (chkp_build_returned_bound): Do not build
22205 returned bounds for a LHS that's not a BOUNDED_P type.
22207 2017-03-20 Martin Liska <mliska@suse.cz>
22211 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22212 COMPLEX_CST and VECTOR_CST.
22214 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22217 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22218 target operand. A new splitter adds the clobber statement in case
22219 the target operand is dead anyway.
22221 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22223 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22224 to age-old versions of binutils and glibc.
22226 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22228 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22230 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22232 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22234 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22236 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22237 requirement for binutils 2.13.
22239 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22241 * combine.c (try_combine): Delete redundant i1 test. Call
22242 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22244 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22246 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22247 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22249 <riscv64-*-elf>: Re-arrange section
22250 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22251 <riscv32-*-linux>: Likewise.
22252 <riscv64-*-elf>: Likewise
22253 <riscv64-*-linux>: Likewise.
22255 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22258 * aarch64.opt(verbose-cost-dump): Fix typo.
22260 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22263 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22264 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22266 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22268 * reload.c (find_reloads): When reloading a nonoffsettable address,
22269 use RELOAD_OTHER for it and its address reloads.
22271 PR rtl-optimization/79910
22272 * combine.c (record_used_regs): New static function.
22273 (try_combine): Handle situations where there is an additional
22274 instruction between I2 and I3 which needs to have a LOG_LINK
22277 2017-03-17 Jeff Law <law@redhat.com>
22279 PR tree-optimization/71437
22280 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22281 conditional in the hash table first.
22282 (vrp_dom_walker::before_dom_children): Extract condition from
22283 ASSERT_EXPR. Record condition, its inverion and any implied
22284 conditions as well.
22286 2017-03-17 Marek Polacek <polacek@redhat.com>
22287 Markus Trippelsdorf <markus@trippelsdorf.de>
22289 PR tree-optimization/80079
22290 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22293 2017-03-17 Richard Biener <rguenther@suse.de>
22295 PR middle-end/80075
22296 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22297 Properly verify the LHS before the RHS possibly claims to be
22299 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22302 2017-03-17 Martin Jambor <mjambor@suse.cz>
22304 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22305 (List of -O2 options): Likewise.
22306 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22309 2017-03-17 Tom de Vries <tom@codesourcery.com>
22311 * gcov-dump.c (print_usage): Print bug_report_url.
22313 2017-03-17 Richard Biener <rguenther@suse.de>
22315 PR middle-end/80050
22316 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22317 (parser::peek): Likewise.
22319 2017-03-17 Richard Biener <rguenther@suse.de>
22321 PR tree-optimization/80048
22322 * sese.c (free_sese_info): Properly release rename_map and
22323 copied_bb_map elements.
22325 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22327 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22328 Add linked-list forward and backlinks. Insert on
22329 construction, remove on destruction.
22330 (class pass_store_merging): Add m_stores_head field.
22331 (pass_store_merging::terminate_and_process_all_chains):
22332 Iterate over m_stores_head list.
22333 (pass_store_merging::terminate_all_aliasing_chains):
22335 (pass_store_merging::execute): Check for debug stmts first.
22336 Push new chains onto the m_stores_head stack.
22338 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22341 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22342 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22343 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22345 2017-03-16 Jeff Law <law@redhat.com>
22347 PR tree-optimization/71437
22348 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22349 member function. Implementation moved into after_dom_children
22350 member function and into the threader's thread_outgoing_edges
22352 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22353 some code into new thread_outgoing_edges.
22354 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22355 definition. Simplify marker handling (do it here). Assume we always
22356 have the available expression and the const/copies tables.
22357 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22359 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22360 * tree-vrp.c (equiv_stack): No longer file scoped.
22361 (vrp_dom_walker): New class.
22362 (vrp_dom_walker::before_dom_children): New member function.
22363 (vrp_dom_walker::after_dom_children): Likewise.
22364 (identify_jump_threads): Setup domwalker. Use it rather than
22365 walking edges in a random order by hand. Simplify setup/finalization.
22366 (finalize_jump_threads): Remove.
22367 (vrp_finalize): Do not call identify_jump_threads here.
22368 (execute_vrp): Do it here instead and call thread_through_all_blocks
22371 PR tree-optimization/71437
22372 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22374 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22376 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22377 (dom_opt_dom_walker::thread_across_edge): Remove
22378 handle_dominating_asserts argument. All callers changed.
22379 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22380 changes. Remove calls to lhs_of_dominating_assert. Other
22381 uses of handle_dominating_asserts turn into unconditional code
22382 (simplify_control_stmt_condition_1): Likewise.
22383 (simplify_control_stmt_condition): Likewise.
22384 (thread_through_normal_block, thread_across_edge): Likewise.
22385 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22386 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22387 object if it is not an SSA_NAME.
22388 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22389 before calling into the VRP specific simplifiers.
22390 (identify_jump_threads): Remove handle_dominating_asserts
22393 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22396 * tree-diagnostic.c (default_tree_printer): No longer static.
22397 * tree-diagnostic.h (default_tree_printer): New prototype.
22399 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22401 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22402 Change ins into fmov.
22404 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22406 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22407 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22408 Use h_con constraint for operand 1.
22409 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22410 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22412 2017-03-15 Jeff Law <law@redhat.com>
22414 PR tree-optimization/71437
22415 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22416 (record_temporary_equivalences): Use it.
22418 PR tree-optimization/71437
22419 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22420 tree-ssa-scopedtables.
22421 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22422 (record_conditions, record_cond, vuse_eq): Likewise.
22423 (record_edge_info): Adjust to API tweak of record_conditions.
22424 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22425 (record_temporary_equivalences, optimize_stmt): Likewise.
22426 (eliminate_redundant_computations): Likewise.
22427 (record_equivalences_from_stmt): Likewise.
22428 * tree-ssa-scopedtables.c: Include options.h and params.h.
22429 (vuse_eq): New function, moved from tree-ssa-dom.c
22430 (build_and_record_new_cond): Likewise.
22431 (record_conditions): Likewise. Accept vector of conditions rather
22432 than edge_equivalence structure for first argument.
22433 for the first argument.
22434 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22435 from tree-ssa-dom.c.
22436 (avail_exprs_stack::record_cond): Likewise.
22437 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22438 from tree-ssa-dom.c.
22439 (avail_exprs_stack): Add new member functions lookup_avail_expr
22441 (record_conditions): Declare.
22443 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22446 * lra-constraints.c (process_alt_operands): Increase reject for
22447 reloading an input/output operand.
22449 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22452 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22453 insns to convert from signed/unsigned char/short to IEEE 128-bit
22455 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22457 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22460 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22461 subreg of inner mode for values already in registers.
22463 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22465 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22466 iteration reg is used after the loop.
22468 2017-03-14 Martin Sebor <msebor@redhat.com>
22470 PR tree-optimization/79800
22471 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22472 precision in negative-positive range.
22473 (format_floating): Call non-const overload with adjusted precision.
22475 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22478 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22481 2017-03-14 Martin Sebor <msebor@redhat.com>
22483 PR middle-end/80020
22484 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22485 * builtins.def (aligned_alloc): Use it.
22488 * Makefile.in (GTFILES): Add calls.c.
22489 * calls.c: Include "gt-calls.h".
22491 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22493 PR rtl-optimization/79728
22494 * regs.h (struct target_regs): New field
22495 x_contains_allocatable_regs_of_mode.
22496 (contains_allocatable_regs_of_mode): New macro.
22497 * reginfo.c (init_reg_sets_1): Initialize it, and change
22498 contains_reg_of_mode so it includes global regs as well.
22499 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22500 rather than contains_regs_of_mode.
22502 2017-03-14 Martin Liska <mliska@suse.cz>
22504 * doc/invoke.texi: Document options that can't be combined with
22505 -fcheck-pointer-bounds.
22507 2017-03-14 Martin Liska <mliska@suse.cz>
22509 PR middle-end/79831
22510 * doc/invoke.texi (-Wchkp): Document the option.
22512 2017-03-14 Martin Liska <mliska@suse.cz>
22514 * Makefile.in: Install gcov-dump.
22516 2017-03-14 Martin Liska <mliska@suse.cz>
22518 * multiple_target.c (expand_target_clones): Bail out for
22519 an invalid attribute.
22521 2017-03-14 Richard Biener <rguenther@suse.de>
22523 * alias.c (struct alias_set_entry): Pack properly.
22524 * cfgloop.h (struct loop): Likewise.
22525 * cse.c (struct set): Likewise.
22526 * ipa-utils.c (struct searchc_env): Likewise.
22527 * loop-invariant.c (struct invariant): Likewise.
22528 * lra-remat.c (struct cand): Likewise.
22529 * recog.c (struct change_t): Likewise.
22530 * rtl.h (struct address_info): Likewise.
22531 * symbol-summary.h (function_summary): Likewise.
22532 * tree-loop-distribution.c (struct partition): Likewise.
22533 * tree-object-size.c (struct object_size_info): Likewise.
22534 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22535 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22536 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22537 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22538 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22539 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22540 (struct _stmt_vec_info): Likewise.
22542 2017-03-14 Martin Liska <mliska@suse.cz>
22545 * multiple_target.c (create_dispatcher_calls): Check that
22546 a target can create a function dispatcher.
22548 2017-03-14 Martin Liska <mliska@suse.cz>
22551 * multiple_target.c (expand_target_clones): Drop local.local
22552 flag for default implementation.
22554 2017-03-14 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/80030
22557 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22559 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22561 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22562 gcc_fallthrough() instead of __attribute__((fallthrough));
22564 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22566 * doc/gcc.texi: Remove "up" link to (DIR).
22567 * doc/gccint.texi: Ditto.
22569 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22571 * doc/install.texi (Specific) <avr>: Remove reference to
22574 2017-03-13 Jeff Law <law@redhat.com>
22576 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22577 attribute rather than comments.
22579 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22580 match_scratch operand is highest.
22582 2017-03-13 Martin Liska <mliska@suse.cz>
22584 PR middle-end/78339
22585 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22586 is a CHKP clone, use original declaration.
22588 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22590 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22591 (arc_conditional_register_usage): Use a different allocation order
22592 when optimizing for size.
22593 * common/config/arc/arc-common.c (arc_option_optimization_table):
22594 Section anchors default on when optimizing for size.
22596 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22598 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22600 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22602 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22603 * config/arc/arc.md (cpu_facility): Add cd variant.
22604 (*movqi_insn): Add code density variant.
22605 (*movhi_insn): Likewise.
22606 (*movqi_insn): Likewise.
22607 (*addsi3_mixed): Likewise.
22608 (subsi3_insn): Likewise.
22610 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22612 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22614 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22616 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22617 expressions with MINUS and UNARY ops.
22619 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22622 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22624 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22625 between vec_select and vector argument.
22626 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22627 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22628 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22629 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22630 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22631 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22633 2017-03-13 Richard Biener <rguenther@suse.de>
22636 * params.def (vect-max-peeling-for-alignment): Fix typo.
22638 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22640 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22641 issue that only occurred with binutils below 2.18.
22643 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22645 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22646 refer to binutils 2.11/2.12 minimum.
22648 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22650 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22651 ftp.kernel.org and simplify binutils requirement.
22653 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22655 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22657 (Optimize Options): Ditto. Also remove redundancy.
22659 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22661 PR translation/79848
22662 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22664 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22665 to G_ to avoid double translation.
22667 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22669 PR translation/79923
22670 * auto-profile.c (get_combined_location): Convert leading
22671 character of diagnostics to lower case and remove trailing period.
22672 (read_profile): Likewise for various diagnostics.
22673 * config/arm/arm.c (arm_option_override): Remove trailing period
22674 from various diagnostics.
22675 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22676 (msp430_expand_delay_cycles): Likewise.
22678 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22681 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22682 full command-line argument, rather than just "str".
22683 (aarch64_validate_march): Likewise.
22684 (aarch64_validate_mtune): Likewise.
22686 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22688 PR rtl-optimization/78911
22689 * lra-assigns.c (must_not_spill_p): New function.
22690 (spill_for): Use it.
22692 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22694 PR tree-optimization/79981
22695 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22696 ATOMIC_COMPARE_EXCHANGE ifn result.
22697 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22698 IFN_ATOMIC_COMPARE_EXCHANGE.
22700 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22703 * opts.c (parse_sanitizer_options): Add missing question mark to
22704 "did you mean" message.
22706 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22708 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22710 (VMULEUH_UNS): Likewise.
22711 (VMULOUB_UNS): Likewise.
22712 (VMULOUH_UNS): Likewise.
22713 * config/rs6000/rs6000.c (builtin_function_type): Remove
22714 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22716 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22719 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22720 x with result of extra_parsing_for_operand_code_0.
22721 (function_reader::extra_parsing_for_operand_code_0): Convert
22722 return type from void to rtx, returning x. When reading
22723 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22724 larger size containing struct block_symbol.
22726 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22728 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22729 -mfloat128-hardware without -m64.
22731 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22734 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22735 entries to the case statement that marks unsigned arguments to
22736 overloaded functions.
22738 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22740 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22741 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22743 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22747 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22749 2017-03-10 Martin Liska <mliska@suse.cz>
22753 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22754 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22757 2017-03-10 Olivier Hainque <hainque@adacore.com>
22759 * tree-switch-conversion (array_value_type): Start by resetting
22760 candidate type to it's main variant.
22762 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22764 PR rtl-optimization/79909
22765 * combine.c (try_combine): Use simplify_replace_rtx on individual
22766 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22767 of the whole CALL_INSN_FUNCTION_USAGE.
22769 PR tree-optimization/79972
22770 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22771 get_range_info on SSA_NAMEs. Formatting fixes.
22773 2017-03-10 Richard Biener <rguenther@suse.de>
22774 Jakub Jelinek <jakub@redhat.com>
22776 PR tree-optimization/77975
22777 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22778 edge to be constant.
22779 (get_val_for): For constant x return it. Formatting fix.
22780 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22781 would use the same bases as the current one.
22783 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22785 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22786 instead of vec_select for V1TImode.
22787 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22789 (VSX_LE_128): Add V1TI to this mode iterator.
22790 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22791 (*vsx_le_perm_store_<mode>): Likewise.
22792 (pre-reload splitter for VSX stores): Likewise.
22793 (post-reload splitter for VSX stores): Likewise.
22794 (*vsx_xxpermdi2_le_<mode>): Likewise.
22795 (*vsx_lxvd2x2_le_<mode>): Likewise.
22796 (*vsx_stxvd2x2_le_<mode>): Likewise.
22798 2017-03-09 Michael Eager <eager@eagercon.com>
22800 Correct failures with --enable-checking=yes,rtl.
22802 * config/microblaze/microblaze.c (microblaze_expand_shift):
22803 Replace GET_CODE test with CONST_INT_P and INTVAL test with
22804 test for const0_rtx.
22805 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
22806 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
22808 2017-03-09 Richard Biener <rguenther@suse.de>
22810 PR tree-optimization/79977
22811 * graphite-scop-detection.c (scop_detection::merge_sese):
22812 Handle the case of extra exits to blocks dominating the entry.
22814 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
22816 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22819 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22821 PR rtl-optimization/79949
22822 * lra-constraints.c (process_alt_operands): Check memory when
22823 trying to predict a cycle. Print about the overall increase.
22825 2017-03-09 Richard Biener <rguenther@suse.de>
22827 PR middle-end/79971
22828 * gimple-expr.c (useless_type_conversion_p): Preserve
22829 TYPE_SATURATING for fixed-point types.
22831 2017-03-09 Richard Biener <rguenther@suse.de>
22834 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22835 alignment of BLKmode params.
22837 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22840 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22841 (VALL_NO_V2Q): Likewise.
22843 * config/aarch64/aarch64-simd.md
22844 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22846 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22847 VALL_NO_V2Q mode iterator.
22848 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22850 2017-03-09 Martin Liska <mliska@suse.cz>
22852 PR tree-optimization/79631
22853 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22854 tree_int_cst_sign_bit just for INTEGER constants.
22856 2017-03-09 Martin Liska <mliska@suse.cz>
22860 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22863 2017-03-09 Marek Polacek <polacek@redhat.com>
22866 * tree.c (inchash::add_expr): Handle TREE_VEC.
22868 2017-03-09 Martin Liska <mliska@suse.cz>
22871 (chkp_narrow_size_and_offset): New function.
22872 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22873 (void chkp_parse_bit_field_ref): New function.
22874 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22875 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22877 2017-03-09 Martin Liska <mliska@suse.cz>
22880 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22881 (chkp_find_bounds_1): Remove gcc_unreachable.
22883 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22886 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22887 BUILT_IN_SYNC*, determine the access type from the size suffix and
22888 always build a MEM_REF with that type. Handle forgotten
22889 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22892 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22893 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22894 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22895 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22896 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22897 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22898 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22899 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22900 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22901 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22902 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22903 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22904 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22905 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22906 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22907 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22908 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22909 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22910 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22911 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22912 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22913 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22914 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22915 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22916 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22917 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22918 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22919 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22920 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22921 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22922 definitions outside of __OPTIMIZE__ guarded section.
22925 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22926 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22927 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22928 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22931 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22933 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22934 ("vfenez<mode>"): Add missing constraints.
22936 2017-03-08 Martin Sebor <msebor@redhat.com>
22939 * config/nds32/nds32.c (nds32_option_override):
22940 Fix misspelled diagnostic.
22942 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22945 * gimplify.c (gimplify_omp_for): Replace index var in outer
22946 taskloop statement with an artificial variable and add
22947 OMP_CLAUSE_PRIVATE clause for it.
22949 2017-03-08 Richard Biener <rguenther@suse.de>
22951 PR tree-optimization/79955
22952 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22953 for accesses that are completely outside of the variable.
22955 2017-03-08 Andrew Haley <aph@redhat.com>
22957 PR tree-optimization/79943
22958 * tree-ssa-loop-split.c (compute_new_first_bound): When
22959 calculating the new upper bound, (END-BEG) should be added, not
22962 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22964 * config/avr/avr.md (setmemhi): Make sure match_dup
22965 operand number comes before match_scratch.
22967 2017-03-08 Richard Biener <rguenther@suse.de>
22969 PR tree-optimization/79920
22970 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22971 with ncopies == 1 to ...
22972 (vect_transform_slp_perm_load): ... here. Properly compute
22973 all element loads by iterating VF times over the group. Do
22974 not handle ncopies (computed in a broken way) in
22975 vect_create_mask_and_perm.
22977 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22980 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22981 is a uniform vector, use uniform_vector_p return value instead of
22982 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22984 2017-03-07 Marek Polacek <polacek@redhat.com>
22986 PR middle-end/79809
22987 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22988 (alloca_call_type): Likewise.
22990 2017-03-07 Martin Liska <mliska@suse.cz>
22992 * gcov.c (process_args): Put comment to correct location.
22994 2017-03-07 Martin Liska <mliska@suse.cz>
22996 PR middle-end/68270
22997 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22998 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22999 (chkp_narrow_bounds_for_field): Likewise.
23000 (chkp_parse_array_and_component_ref): Pass one more argument to
23003 2017-03-07 Richard Biener <rguenther@suse.de>
23005 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23008 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23010 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23011 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23013 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23016 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23017 to end of description.
23018 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23020 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23022 PR rtl-optimization/79901
23023 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23025 (*avx512f_<code><mode>3<mask_name>): ... this.
23026 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23027 iterator instead of VI8_AVX2_AVX512BW.
23029 PR rtl-optimization/79901
23030 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23031 min/max expander, expand it using expand_vec_cond_expr.
23034 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23037 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23040 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23041 to void * for PCH reasons.
23042 * dwarf2out.c (output_loc_operands, output_die): Cast
23043 v.val_vec.array to unsigned char *.
23045 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23048 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23051 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23053 PR rtl-optimization/79571
23054 * lra-constraints.c (process_alt_operands): Calculate static
23055 reject and subtract it from overall when only addresses will be
23058 2017-03-06 Julia Koval <julia.koval@intel.com>
23061 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23062 incoming stack boundary to 128 for 64-bit targets.
23064 2017-03-06 Richard Biener <rguenther@suse.de>
23066 PR tree-optimization/79894
23067 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23068 to NULL after folding it.
23070 2017-03-06 Richard Biener <rguenther@suse.de>
23072 PR tree-optimization/79824
23073 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23074 check disabling peeling for gaps.
23076 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23078 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23079 attributes): Document gettimeofday.
23081 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23083 * config/s390/s390.c (s390_option_override_internal): Set
23084 PARAM_MIN_VECT_LOOP_BOUND
23086 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23088 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23089 * config/s390/s390.md: Likewise.
23091 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23094 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23095 (<avx2_avx512>_perm<mode>): Rename to ...
23096 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23098 (<avx512>_perm<mode>_mask): Rename to ...
23099 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23101 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23102 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23103 instead of VI8F_256_512.
23104 (avx512f_perm<mode>): New define_expand.
23105 (avx512f_perm<mode>_mask): Likewise.
23106 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23107 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23109 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23111 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23112 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23114 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23116 2017-03-06 Martin Liska <mliska@suse.cz>
23119 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23120 when having a SSA NAME w/o VAR_DECL assigned to it.
23122 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23124 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23125 msa_dpsub_<su>_d): Fix MODE for vec_select.
23127 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23129 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23131 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23133 2017-03-06 Richard Biener <rguenther@suse.de>
23135 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23136 * plugin.c (register_plugin_info): Likewise.
23137 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23139 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23141 * config/i386/sse.md (sse_storehps, sse_storelps,
23142 avx_<castmode><avxsizesuffix>_<castmode>,
23143 avx512f_<castmode><avxsizesuffix>_<castmode>,
23144 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23145 in condition that at least one operand is not a MEM.
23147 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23149 PR middle-end/79805
23150 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23151 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23153 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23154 gimple_call_nothrow_p flag based on whether original builtin can throw.
23155 If it can, emit following stmts on the fallthrough edge.
23156 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23157 don't create new bb if inserting just debug stmts on the edge, try to
23158 insert them on the fallthru bb or just reset debug stmts.
23160 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23163 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23164 restore recog_data (including the operand rtxes inside it) around
23165 the call to get_insn_template.
23167 2017-03-03 Martin Sebor <msebor@redhat.com>
23169 PR tree-optimization/79699
23170 * context.c (context::~context): Free MPFR caches to avoid
23171 a memory leak on program exit.
23173 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23175 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23176 Use wide_int::ulow () instead of .elt (0).
23178 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23180 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23181 (*pushxf): Limit oF constraint to 32bit targets and add oC
23182 constraint for 64bit targets.
23183 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23184 (*pushdf): Change rmF constraint to rmC.
23186 2017-03-03 Martin Liska <mliska@suse.cz>
23188 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23189 Remove unused variable.
23191 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23194 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23195 is a memory operand, increase num_memory.
23196 (ix86_expand_args_builtin): Likewise.
23198 2017-03-03 Jan Hubicka <jh@suse.cz>
23201 * ipa-devirt.c (maybe_record_node): Properly handle
23202 __cxa_pure_virtual visibility.
23204 2017-03-03 Martin Liska <mliska@suse.cz>
23206 PR tree-optimization/79803
23207 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23209 (pass_loop_prefetch::execute): Disabled optimization if an
23210 assumption about L1 cache size is not met.
23212 2017-03-03 Martin Liska <mliska@suse.cz>
23214 PR rtl-optimization/79574
23215 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23216 (hash_scan_set): Likewise.
23217 (dump_hash_table): Likewise.
23218 (hoist_code): Likewise.
23220 2017-03-03 Richard Biener <rguenther@suse.de>
23222 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23223 in place of uhigh ().
23224 (fixed_convert_from_real): Likewise.
23226 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23229 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23231 2017-03-03 Richard Biener <rguenther@suse.de>
23233 PR middle-end/79818
23234 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23235 TYPE_OVERFLOW_UNDEFINED check.
23237 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23239 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23241 (vector_ae_<mode>_p): Likewise.
23242 (vector_nez_<mode>_p): Likewise.
23243 (vector_ne_v2di_p): Likewise.
23244 (vector_ae_v2di_p): Likewise.
23245 (vector_ne_<mode>_p): Likewise.
23246 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23248 (vsx_tsqrt<mode>2_fe): Likewise.
23250 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23253 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23255 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23257 PR rtl-optimization/79780
23258 * cprop.c (one_cprop_pass): When second and further conditional trap
23259 in a single basic block is turned into an unconditional trap, turn it
23260 into a deleted note to avoid RTL verification failures.
23262 2017-03-02 Richard Biener <rguenther@suse.de>
23264 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23266 2017-03-02 Richard Biener <rguenther@suse.de>
23268 PR tree-optimization/79345
23270 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23271 param and abort the walk, returning -1 if it is hit.
23272 (walk_aliased_vdefs): Take a limit param and pass it on.
23273 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23274 defaulting to 0 and return a signed int.
23275 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23276 (check_defs): New helper.
23277 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23278 about uninitialized memory.
23279 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23280 bogus uninitialized warning.
23281 (fixed_convert_from_real): Likewise.
23283 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23285 PR tree-optimization/66768
23286 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23287 iv_use if base object can't be determined.
23289 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23291 PR tree-optimization/79345
23292 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23293 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23294 (get_pattern_stats): Initialize it.
23295 * genemit.c (gen_expand): Verify match_scratch numbers come after
23296 match_operand/match_dup numbers.
23297 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23298 match_scratch numbers.
23299 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23301 * config/s390/s390.md (trunctdsd2): Likewise.
23303 2017-03-02 Richard Biener <rguenther@suse.de>
23305 * wide-int.h (wide_int_storage::operator=): Implement in terms
23308 2017-03-02 Richard Biener <rguenther@suse.de>
23310 PR tree-optimization/79777
23311 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23312 the to insert expression to sth existing.
23314 2017-03-01 Martin Sebor <msebor@redhat.com>
23316 PR middle-end/79692
23317 * gimple-ssa-sprintf.c
23318 (directive::known_width_and_precision): New function.
23319 (format_integer): Use it.
23320 (get_mpfr_format_length): Consider the full range of precision
23321 when computing %g output with the # flag. Set the likely byte
23322 count to 3 rather than 1 when precision is indeterminate.
23323 (format_floating): Correct the lower bound of precision.
23325 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23327 * doc/invoke.texi: Document default code model for 64-bit Linux.
23329 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23332 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23333 udiv rather than div since input pattern is unsigned.
23335 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23337 * config/i386/i386.c (print_reg): Warn for values of
23338 unsupported size in integer register.
23340 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23343 * config/rs6000/predicates.md (current_file_function_operand): Do
23344 not allow self calls to be local if the function is replaceable.
23346 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23349 * config/rs6000/altivec.h (vec_ctz and others): Change the
23350 preprocessor macro that controls conditional compilation from
23351 _ARCH_PWR9 to __POWER9_VECTOR__.
23352 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23353 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23354 control (instead of _ARCH_PWR9 control) so that template
23355 definition uses power9-specific function.
23356 (vec_any_eq): Likewise.
23357 (vec_all_ne): Change macro definition to use a power9-specific
23358 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23359 _ARCH_PWR9 control).
23360 (vec_any_eq) Likewise.
23361 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23362 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23363 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23364 support for xvcmpnedp instruction.
23365 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23366 macro expansion so that Power9 implementation of vec_all_ne does
23367 not use the AltiVec predicate framework.
23368 (VCMPNEH_P): Likewise.
23369 (VCMPNEW_P): Likewise.
23370 (VCMPNED_P): Likewise.
23371 (VCMPNEFP_P): Likewise.
23372 (VCMPNEDP_P): Likewise.
23373 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23374 implementation of vec_any_eq to not use AltiVec predicate
23376 (VCMPAEH_P): Likewise.
23377 (VCMPAEW_P): Likewise.
23378 (VCMPAED_P): Likewise.
23379 (VCMPAEFP_P): Likewise.
23380 (VCMPAEDP_P): Likewise.
23381 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23382 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23383 not use the AltiVec predicate framework.
23384 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23385 of vec_any_eq to not use AltiVec predicate framework.
23386 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23387 support for predefined __POWER9_VECTOR__ macro to indicate that
23388 Power9 instruction selection is enabled.
23389 (altivec_overloaded_builtins): Remove extraneous
23390 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23391 function argument types RS6000_BTI_bool_V16QI and
23392 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23393 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23394 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23395 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23396 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23397 Power9 for implementations of vec_cmpne. Change the signature for
23398 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23399 (representing vec_all_ne) to remove the previously described first
23400 argument of type RS6000_BTI_INTSI, as this was an artifact of
23401 reliance on the AltiVec predicate framework, which is no longer
23402 used in the implementation of these functions. Add
23403 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23404 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23405 since, unlike the AltiVec predicate framework implementation, we
23406 do not share function descriptors between vec_alle and vec_anyeq.
23407 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23408 set of modes that receive special treatment even when
23409 TARGET_P9_VECTOR is true. The special treatment emits code that
23410 does not depend on Power9 instructions.
23411 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23412 define_expand to not rely on AltiVec predicate framework.
23413 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23415 (vector_ne_v2di_p): Change this define_expand to not rely on
23416 AltiVec predicate framework.
23417 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23419 (vector_ne_<mode>_p): Change this define_expand to not rely on
23420 AltiVec predicate framework.
23421 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23423 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23424 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23425 define_insn pattern.
23426 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23427 define_insn pattern because the xvcmpne<VSs>. instruction is not
23429 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23430 instruction is not supported.
23432 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23434 * config/nvptx/nvptx.c: Include intl.h.
23436 2017-03-01 Martin Jambor <mjambor@suse.cz>
23439 * ipa-prop.h (ipa_bits): Removed field known.
23440 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23441 to pointers. Adjusted their comments to warn about their sharing.
23442 (ipcp_transformation_summary): Change bits to a vector of pointers.
23443 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23444 (ipa_get_ipa_bits_for_value): Declare.
23445 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23446 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23447 (ipa_bits_hash_table): Likewise.
23448 (ipa_vr_ggc_hash_traits): Likewise.
23449 (ipa_vr_hash_table): Likewise.
23450 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23451 being pointers and vr_known being removed.
23452 (ipa_set_jf_unknown): Likewise.
23453 (ipa_get_ipa_bits_for_value): New function.
23454 (ipa_set_jfunc_bits): Likewise.
23455 (ipa_get_value_range): New overloaded functions.
23456 (ipa_set_jfunc_vr): Likewise.
23457 (ipa_compute_jump_functions_for_edge): Use the above functions to
23458 construct bits and vr parts of jump functions.
23459 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23460 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23462 (ipcp_grow_transformations_if_necessary): Also allocate
23463 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23465 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23466 them. Fix too long lines.
23467 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23468 vr_known being removed.
23469 (ipa_read_jump_function): Use new setter functions to construct bits
23470 and vr parts of jump functions or set them to NULL.
23471 (write_ipcp_transformation_info): Adjust for bits being pointers.
23472 (read_ipcp_transformation_info): Likewise.
23473 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23475 Include gt-ipa-prop.h.
23476 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23478 (ipcp_store_bits_results): Likewise.
23479 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23480 Do not write to existing jump functions but use a temporary instead.
23482 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23485 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23486 attempt to use its first operand as BIT_FIELD_REF base.
23488 2017-03-01 Richard Biener <rguenther@suse.de>
23490 PR middle-end/79721
23491 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23492 interpolating formula in wrapping arithmetic.
23493 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23495 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23497 PR tree-optimization/79734
23498 * tree-vect-generic.c (expand_vector_condition): Optimize
23499 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23500 Handle VEC_COND_EXPR where comparison has different inner width from
23501 type's inner width.
23503 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23505 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23506 markup, and similar issues. Remove @opindex entries for things
23507 that aren't options. Add missing -mmpy-option entries.
23509 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23511 PR tree-optimization/79737
23512 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23513 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23514 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23515 instead of byte_size. Formatting fix.
23516 (shift_bytes_in_array_right): Formatting fix.
23518 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23521 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23522 condition on optimize for the leaf function test.
23524 2017-02-28 Martin Liska <mliska@suse.cz>
23527 * read-rtl-function.c (function_reader::handle_unknown_directive):
23528 Bail out when one uses -flto.
23530 2017-02-28 Martin Liska <mliska@suse.cz>
23532 * common.opt: Replace space with tabular for options of <number>
23534 * config/i386/i386.opt: Show <number> value for
23535 -mlarge-data-threshold.
23536 * opts.c (print_filtered_help): Do not display number in hexadecimal
23539 2017-02-28 Martin Liska <mliska@suse.cz>
23541 * common.opt: Fix --help=option -Q for options which are of
23544 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23546 * config/i386/i386.c (print_reg): Error out for values
23547 of 8-bit size in invalid integer register.
23549 2017-02-28 Martin Sebor <msebor@redhat.com>
23551 PR tree-optimization/79691
23552 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23554 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23557 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23558 gcc_unreachable with output_operand_lossage.
23560 2017-02-28 Richard Biener <rguenther@suse.de>
23562 PR tree-optimization/79740
23563 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23565 (visit_nary_op): Insert the nary into the hashtable if we
23566 pattern-matched sth.
23567 * tree-ssa-pre.c (eliminate_insert): Robustify.
23569 2017-02-28 Richard Biener <rguenther@suse.de>
23571 PR middle-end/79731
23572 * fold-const.c (decode_field_reference): Reject out-of-bound
23575 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23577 * config/i386/i386.c: Include intl.h.
23578 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23579 instead of just cond ? "..." : "...".
23580 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23581 * coverage.c (read_counts_file): Likewise.
23582 * omp-offload.c: Include intl.h.
23583 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23584 of just cond ? "..." : "...".
23585 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23586 of just cond ? "..." : "...".
23588 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23591 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23593 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23594 'tune for' CPU name.
23595 * config/arm/arm-cpu-data.h: Regenerated.
23597 2017-02-28 Richard Biener <rguenther@suse.de>
23599 PR tree-optimization/79732
23600 * tree-inline.c (expand_call_inline): Do not shadow var.
23602 2017-02-28 Richard Biener <rguenther@suse.de>
23604 PR tree-optimization/79723
23605 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23606 address-space properly.
23608 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23610 * doc/optinfo.texi (Optimization groups): Fix option used for
23612 * doc/invoke.texi (-fopt-info): Document "omp".
23613 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23614 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23615 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23616 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23617 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23619 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23621 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23622 OPTGROUP_OMP with "omp".
23624 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23627 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23628 for arithmetic shift of unsigned V2DI.
23630 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23632 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23633 arc/linux.h headers.
23634 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23635 (LINK_SPEC): Likewise.
23636 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23637 (EXTRA_SPECS): Likewise.
23638 (STARTFILE_SPEC): Likewise.
23639 (ENDFILE_SPEC): Likewise.
23640 (LIB_SPEC): Likewise.
23641 (TARGET_SDATA_DEFAULT): Likewise.
23642 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23643 (MULTILIB_DEFAULTS): Likewise.
23644 (DWARF2_UNWIND_INFO): Likewise.
23645 * config/arc/big.h: New file.
23646 * config/arc/elf.h: Likewise.
23647 * config/arc/linux.h: Likewise.
23648 * config/arc/t-uClibc: Remove.
23650 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23652 PR tree-optimization/77536
23653 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23654 (tree_transform_and_unroll_loop): Use above function to compute the
23655 estimated niter of unrolled loop and use it when scaling profile.
23656 Also use count info rather than frequency if it's non-zero.
23657 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23658 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23659 (vect_transform_loop): Call above function.
23661 2017-02-27 Richard Biener <rguenther@suse.de>
23663 PR tree-optimization/45397
23664 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23665 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23666 (visit_nary_op): Add pattern matching for CSEing sign-changed
23667 or truncated operations with wider ones.
23669 2017-02-27 Richard Biener <rguenther@suse.de>
23671 PR tree-optimization/79690
23672 * tree-vect-stmts.c (vectorizable_store): Use vector type
23673 built from the DR with address-space.
23675 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23677 * doc/invoke.texi (Optimize Options): Refine the description
23678 of asan-use-after-return.
23680 2017-02-25 Alan Modra <amodra@gmail.com>
23682 PR rtl-optimization/79584
23683 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23684 base, not ad->base_term, the reg within base. Remove assertion
23685 that ad->base == ad->base_term. Replace gen_int_mode using
23686 bogus mode with const0_rtx.
23688 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23690 PR middle-end/79396
23691 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23692 FMA_EXPR like tcc_binary or tcc_unary.
23694 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23697 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23699 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23700 (output_loc_operands): Handle DW_OP_call_ref and
23701 DW_OP_GNU_variable_value.
23702 (struct variable_value_struct): New type.
23703 (struct variable_value_hasher): Likewise.
23704 (variable_value_hash): New variable.
23705 (string_types): Remove.
23706 (copy_loc_descr): New function.
23707 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23708 (prepend_loc_descr_to_each): New function.
23709 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23710 instead of add_loc_descr_to_each if the first argument is single
23711 location list and the second has multiple.
23712 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23713 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23714 when looking for variable value which doesn't have other location info.
23715 (loc_list_from_tree): Formatting fix.
23716 (gen_array_type_die): Simplify DW_AT_string_length handling.
23717 (adjust_string_types): Remove.
23718 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23719 string_types. Call resolve_variable_values.
23720 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23721 (resolve_addr_in_expr): Likewise. Add A argument.
23722 (copy_deref_exprloc): Remove deref argument. Adjust for the
23723 original expression being DW_OP_GNU_variable_value with optionally
23724 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23725 optionally after it.
23726 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23727 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23728 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23729 (variable_value_hasher::hash, variable_value_hasher::equal): New
23731 (resolve_variable_value_in_expr, resolve_variable_value,
23732 resolve_variable_values, note_variable_value_in_expr,
23733 note_variable_value): New functions.
23734 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23737 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23740 * opts.h (handle_generated_option): Add GENERATED_P argument.
23741 * opts-common.c (handle_option): Adjust function comment.
23742 (handle_generated_option): Add GENERATED_P argument, pass it to
23744 (control_warning_option): Pass false to handle_generated_option
23746 * opts.c (maybe_default_option): Pass true to handle_generated_option
23748 * optc-gen.awk: Likewise.
23750 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23752 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23753 a REG, look at the REG it is a SUBREG of.
23754 (splitter for cmpeqsi_t): Ditto.
23756 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23758 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23759 the special USEs with the pattern of the insn, not the insn itself.
23761 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23764 * doc/invoke.texi: Document -mload-store-pairs.
23766 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23767 Sandra Loosemore <sandra@codesourcery.com>
23769 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23770 argument isn't a CONST_INT.
23771 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23772 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23773 (nios2_validate_compare): Bypass alternate compare logic if *op2
23774 is not a CONST_INT.
23775 (ldstwm_operation_p): Return false if first_base is not a REG or
23776 if first_offset is not a CONST_INT.
23778 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23780 * config/cris/cris.md: Use correct operand in a define_peephole2.
23782 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23784 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23786 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23788 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23789 this_insn if it is an INSN or JUMP_INSN.
23790 (force_offsettable): Look at base, not at addr.
23791 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23792 on things that aren't necessarily CONST_INTs.
23794 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
23796 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
23797 -mfpmath=sse is the default also for x86-32 targets with SSE2
23798 instruction set when @option{-ffast-math} is enabled
23800 2017-02-24 Jeff Law <law@redhat.com>
23802 PR rtl-optimizatoin/79286
23803 * ira.c (update_equiv_regs): Drop may_trap_p exception to
23806 2017-02-24 Richard Biener <rguenther@suse.de>
23808 PR tree-optimization/79389
23809 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
23812 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
23814 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
23815 function comment to reflect reality.
23816 (loop_exits_before_overflow): Fix typo in function description.
23818 2017-02-24 Richard Biener <rguenther@suse.de>
23820 PR tree-optimization/79389
23821 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23822 properly that a threading opportunity exists. Detect conditional
23823 copy/constant propagation opportunities.
23825 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23827 * config/visium/visium.md (type): Add trap.
23828 (b): New mode attribute.
23829 (*btst): Rename into...
23830 (*btst<mode>): ...this and adjust.
23831 (*cbranchsi4_btst_insn): Rename into...
23832 (*cbranch<mode>4_btst_insn): ...this and adjust.
23833 (trap): New define_insn.
23835 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23837 PR tree-optimization/79389
23838 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23839 (noce_reversed_cond_code): New function.
23840 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23841 reversed_comparison_code. Formatting fix.
23842 (noce_try_store_flag): Test rev_cond != NULL in addition to
23843 reversed_comparison_code.
23844 (noce_try_store_flag_constants): Likewise.
23845 (noce_try_store_flag_mask): Likewise.
23846 (noce_try_addcc): Use rev_cond if non-NULL instead of
23847 reversed_comparison_code.
23848 (noce_try_cmove_arith): Likewise. Formatting fixes.
23849 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23850 (noce_find_if_block): Initialize rev_cond.
23851 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23852 instead of false as last argument never attempt to reverse it
23855 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23857 PR tree-optimization/79663
23858 * tree-predcom.c (combine_chains): Process refs in reverse order
23859 only for ZERO length chains, and add explaining comment.
23861 2017-02-23 Jeff Law <law@redhat.com>
23863 PR tree-optimization/79578
23864 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23865 in call to operand_equal_p.
23867 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23870 * config/i386/cpuid.h: Fix another undefined behavior.
23872 2017-02-23 Richard Biener <rguenther@suse.de>
23874 PR tree-optimization/79683
23875 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23876 vector types for data-refs.
23878 2017-02-23 Martin Liska <mliska@suse.cz>
23880 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23882 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23884 PR middle-end/79665
23885 * internal-fn.c (get_range_pos_neg): Moved to ...
23886 * tree.c (get_range_pos_neg): ... here. No longer static.
23887 * tree.h (get_range_pos_neg): New prototype.
23888 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23889 are known to be in between 0 and signed maximum inclusive, try to
23890 expand both unsigned and signed divmod and use the cheaper one from
23893 2017-02-22 Jeff Law <law@redhat.com>
23895 PR tree-optimization/79578
23896 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23897 to compare base operands.
23899 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23902 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23903 gpc_reg_operand instead of fpr_reg_operand.
23905 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23907 * config/mips/mips.c (mips_return_in_memory): Force FP
23908 vector types to be returned in memory for o32 ABI.
23910 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23912 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23913 instead of DW_TAG_member for static data member declarations and don't
23914 set no_linkage_name for static inline data members.
23915 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23918 2017-02-22 Martin Liska <mliska@suse.cz>
23920 * doc/invoke.texi: Replace inequality signs with square brackets
23923 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23925 PR tree-optimization/68644
23926 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23928 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23931 * lra-constraints.c (simplify_operand_subreg): Handle
23932 WORD_REGISTER_OPERATIONS targets.
23934 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23937 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23938 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23939 elimination by swapping fld*.
23941 2017-02-22 Richard Biener <rguenther@suse.de>
23943 PR tree-optimization/79673
23944 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23945 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23946 irrelevant address-space qualifiers and avoiding a
23947 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23949 2017-02-22 Richard Biener <rguenther@suse.de>
23951 PR tree-optimization/79666
23952 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23953 to not symbolically negate if that may introduce undefined
23956 2017-02-22 Martin Liska <mliska@suse.cz>
23959 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23960 * data-streamer-out.c (streamer_write_gcov_count_stream):
23962 * value-prof.c (stream_out_histogram_value): Make assert more
23963 precise based on type of counter.
23965 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23968 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23969 Use nonimmediate_operand instead of memory_operand for operand 1.
23970 (float-extend standard_x87sse_constant_load splitter): Ditto.
23972 2017-02-21 Jeff Law <law@redhat.com>
23974 PR tree-optimization/79621
23975 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23976 blocks with edges to themselves.
23978 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23981 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23982 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23983 Use gimple_call_builtin_p.
23986 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23987 on temporarily removed DEBUG_INSNs.
23989 PR tree-optimization/79649
23990 * tree-loop-distribution.c (classify_partition): Give up on
23991 non-generic address space loads/stores.
23993 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23995 * doc/loop.texi (Loop manipulation): Remove nonexistent
23996 tree_ssa_loop_version from the documentation.
23997 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23999 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24002 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24003 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24004 * config/rs6000/rs6000.c: Include except.h.
24005 (rs6000_expand_split_stack_prologue): Call
24006 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24008 2017-02-21 Martin Jambor <mjambor@suse.cz>
24011 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24012 have been analyzed.
24014 2017-02-21 Martin Jambor <mjambor@suse.cz>
24016 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24017 for backward compatibility only.
24018 * doc/invoke.texi (Option Summary): Remove all references to
24019 -fipa-cp-alignment.
24021 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24025 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24027 * lra-constraints.c (curr_insn_transform): Handle
24028 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24030 2017-02-21 Martin Liska <mliska@suse.cz>
24032 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24034 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24036 PR translation/79638
24037 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24039 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24042 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24043 (arm_function_ok_for_sibcall): Return false for an indirect call by
24044 descriptor if all the argument registers are used.
24045 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24046 alignment of the function.
24048 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24050 PR tree-optimization/61441
24051 * simplify-rtx.c (simplify_const_unary_operation): For
24052 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24053 the sNaN unmodified.
24055 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24057 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24058 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24059 instead of SYSTEM_HEADER_DIR.
24061 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24062 Martin Liška <mliska@suse.cz>
24064 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24065 Fix typos and grammar, use active voice, and clarify.
24067 2017-02-20 Marek Polacek <polacek@redhat.com>
24069 PR middle-end/79537
24070 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24073 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24075 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24078 * config/i386/i386.c (ix86_expand_builtin): Handle
24079 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24080 ix86_builtins_isa[fcode].isa as a requirement of those
24081 flags and any other flag in the bitmask.
24082 (ix86_init_mmx_sse_builtins): Use 0 instead of
24083 ~OPTION_MASK_ISA_64BIT as mask.
24084 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24085 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24086 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24087 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24089 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24092 * lra-constraints.c (split_reg): Check requested split mode
24093 is supported by the register.
24095 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24097 * lra-constraints.c (simplify_operand_subreg): Remove early
24100 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24103 * lra-constraints.c (curr_insn_transform): Tighten condition
24104 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24106 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24109 * lra-constraints.c (curr_insn_transform): Handle
24110 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24112 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24115 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24117 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24119 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24122 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24125 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24127 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24130 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24132 * final.c (last_columnnum, override_columnnum): New variables.
24133 (final_start_function): Set last_columnnum, pass it to begin_prologue
24134 hook and pass 0 to dwarf2out_begin_prologue.
24135 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24136 to source_line debug hook.
24137 (notice_source_line): Compute last_columnnum and for debug_column_info
24138 return true on column changes.
24139 * debug.h (struct gcc_debug_hooks): Add column argument to
24140 source_line and begin_prologue hooks.
24141 (debug_nothing_int_charstar_int_bool): Remove prototype.
24142 (debug_nothing_int_int_charstar,
24143 debug_nothing_int_int_charstar_int_bool): New prototypes.
24144 (dwarf2out_begin_prologue): Add column argument.
24145 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24146 begin_prologue hooks.
24147 (debug_nothing_int_charstar_int_bool): Remove.
24148 (debug_nothing_int_int_charstar,
24149 debug_nothing_int_int_charstar_int_bool): New functions.
24150 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24151 through to dwarf2out_source_line.
24152 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24153 (dwarf2out_source_line): Add column argument, emit it if requested.
24154 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24156 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24157 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24158 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24159 through to dwarf2out_begin_prologue.
24160 (vmsdbgout_source_line): Add column argument, pass it through to
24161 dwarf2out_source_line.
24162 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24163 dbxout_source_line caller.
24164 (dbxout_source_line): Add column argument.
24166 * common.opt (gno-column-info, gcolumn-info): New options.
24167 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24168 (check_die): Also test for multiple DW_AT_decl_column attributes.
24169 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24170 DW_AT_decl_column if requested.
24171 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24173 (gen_variable_die): Likewise.
24174 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24175 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24178 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24179 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24180 (ix86_handle_option): Handle OPT_m3dnowa.
24181 * doc/invoke.texi (-m3dnowa): Document.
24182 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24183 -m3dnowa instead of -m3dnow -march=athlon.
24186 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24187 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24189 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24192 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24193 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24194 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24195 generator for vsx_xxpermdi_<mode>_be.
24196 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24197 force big-endian semantics.
24198 (vsx_xxpermdi_<mode>_be): New define_expand with same
24199 implementation as previous version of vsx_xxpermdi_<mode>.
24201 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24203 PR tree-optimization/79327
24204 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24205 variable, its initialization and use.
24207 2017-02-17 Julia Koval <julia.koval@intel.com>
24209 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24210 (OPTION_MASK_ISA_PKU_UNSET): New.
24211 (ix86_handle_option): Handle -mrdpid.
24212 * config/i386/cpuid.h (bit_RDPID): New.
24213 * config/i386/driver-i386.c (host_detect_local_cpu):
24214 Detect RDPID feature.
24215 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24216 * config/i386/i386-c.c (ix86_target_macros_internal):
24218 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24219 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24220 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24221 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24222 * config/i386/i386.md (define_insn "rdpid"): New.
24223 * config/i386/i386.opt Add -mrdpid.
24224 * config/i386/immintrin.h (_rdpid_u32): New.
24226 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24228 PR rtl-optimization/79541
24229 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24230 instead of transforming it into USE.
24232 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24234 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24235 If HONOR_SNANS (SFmode) force the input to a register.
24236 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24237 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24238 an frsp or similar insn.
24240 2017-02-17 Martin Liska <mliska@suse.cz>
24242 PR rtl-optimization/79577
24243 * params.def (selsched-max-sched-times): Increase minimum to 1.
24245 2017-02-17 Martin Liska <mliska@suse.cz>
24247 PR rtl-optimization/79574
24248 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24250 2017-02-17 Martin Liska <mliska@suse.cz>
24252 PR tree-optimization/79529
24253 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24254 ssa_defined_default_def_p to handle cases which are implicitly
24256 * tree-ssa.c (ssa_defined_default_def_p): New function.
24257 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24258 which are implicitly defined.
24259 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24261 2017-02-17 Richard Biener <rguenther@suse.de>
24263 PR middle-end/79576
24264 * params.def (max-ssa-name-query-depth): Limit to 10.
24266 2017-02-17 Richard Biener <rguenther@suse.de>
24268 PR tree-optimization/79552
24269 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24272 2017-02-17 Richard Biener <rguenther@suse.de>
24275 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24277 2017-02-17 Marek Polacek <polacek@redhat.com>
24279 PR middle-end/79536
24280 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24281 (fold_negate_expr): New wrapper.
24283 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24285 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24286 Correct terminology and de-emphasize pre-standard behavior.
24288 2017-02-16 Alan Modra <amodra@gmail.com>
24290 PR rtl-optimization/79286
24291 * ira.c (def_dominates_uses): New function.
24292 (update_equiv_regs): Don't create an equivalence for insns that
24293 may trap where the register def does not dominate the use.
24295 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24297 PR rtl-optimization/78127
24298 * lra.c (lra): Call lra_eliminate before finish the loop after
24301 2017-02-16 Richard Biener <rguenther@suse.de>
24303 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24305 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24306 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24307 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24308 (isl_val_int_from_wi): New function.
24309 (extract_affine_gmp): Rename to ...
24310 (extract_affine_wi): ... this, take a widest_int.
24311 (extract_affine_int): Just wrap extract_affine_wi.
24312 (add_param_constraints): Use isl_val_int_from_wi.
24313 (add_loop_constraints): Likewise, and extract_affine_wi.
24315 2017-02-15 Jeff Law <law@redhat.com>
24317 PR middle-end/79521
24318 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24319 ira_init_register_move_cost_if_necessary.
24321 2017-02-15 Martin Sebor <msebor@redhat.com>
24323 PR middle-end/32003
24324 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24325 removed in a prior commit.
24327 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24329 PR tree-optimization/79347
24330 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24331 counters during peeling.
24333 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24335 * Makefile.in (site.exp): Remove "set ISLVER".
24337 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24340 * real.c (real_from_integer): Call real_convert even for decimal.
24342 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24345 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24347 2017-02-14 Andrew Pinski <apinski@cavium.com>
24349 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24350 cores and change the partno/implementer to be correct.
24351 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24352 the 'B" as the implementer.
24353 * config/aarch64/aarch64-tune.md: Regenerate.
24355 2017-02-14 Carl Love <cel@us.ibm.com>
24357 * config/rs6000/rs6000.c: Add case statement entry to make the
24358 xvcvuxdsp built-in argument unsigned.
24359 * config/rs6000/vsx.md: Fix the source and return operand types so they
24360 match the instruction definitions from the ISA document. Fix typo
24361 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24364 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24367 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24368 member early_clobber_alts.
24369 * lra-lives.c (reg_early_clobber_p): New.
24370 (process_bb_lives): Use it.
24371 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24372 (debug_operand_data): Initialize early_clobber_alts.
24373 (setup_operand_alternative): Set up early_clobber_alts.
24374 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24375 alternatives to new_insn_reg.
24376 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24378 (lra_update_insn_regno_info): Pass the new arg.
24380 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24382 PR middle-end/79505
24383 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24384 (new_oacc_loop_raw): Don't clear already cleared fields.
24387 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24388 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24389 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24391 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24394 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24396 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24399 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24400 the extra instruction to the right place to store 128-bit constant
24403 2017-02-14 Martin Sebor <msebor@redhat.com>
24405 PR middle-end/79448
24406 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24407 warning for strings of unknown length.
24409 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24411 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24413 2017-02-14 Jeff Law <law@redhat.com>
24416 * ira-costs.c (scan_one_insn): Initialize register move costs
24417 for pseudos seen in USE/CLOBBER insns.
24419 PR tree-optimization/79095
24420 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24421 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24422 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24423 if the operands are known to be not equal, then the resulting range
24425 (intersect_ranges): If the new range is ~[0,0] and the old range is
24426 wide, then prefer ~[0,0].
24427 * tree-vrp.c (overflow_comparison_p_1): New function.
24428 (overflow_comparison_p): New function.
24429 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24430 if NAME is used in an overflow test.
24431 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24432 overflow check that can be expressed as an equality test, then adjust
24433 ops to be that equality test.
24435 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24437 * config/s390/s390-builtin-types.def: Remove flags argument.
24438 * config/s390/s390.c (s390_init_builtins): Likewise.
24440 2017-02-14 Martin Liska <mliska@suse.cz>
24442 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24443 vector. Fix trailing white spaces.
24445 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24447 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24450 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24452 PR rtl-optimization/68664
24453 * config/arm/arm.c (arm_sched_can_speculate_insn):
24454 New function. Declare prototype.
24455 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24457 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24459 PR rtl-optimization/68664
24460 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24462 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24464 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24466 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24467 max skip bytes for function, loop and jump.
24469 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24471 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24472 ABS_EXPR for gimple dump.
24474 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24477 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24479 PR tree-optimization/79408
24480 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24481 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24482 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24483 also if rhs1 is INTEGER_CST.
24485 2017-02-14 Richard Biener <rguenther@suse.de>
24487 PR middle-end/79432
24488 * tree-into-ssa.c (insert_phi_nodes): When the function can
24489 have abnormal edges rewrite SSA names with broken use-def
24490 dominance out of SSA and register them for PHI insertion.
24492 2017-02-13 Martin Sebor <msebor@redhat.com>
24494 PR middle-end/79496
24495 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24496 clearing info.nowrite flag when snprintf size argument is a range.
24498 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24500 * cprop.c (cprop_jump): Add missing space in string literal.
24501 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24502 (get_constraint_for_component_ref): Likewise.
24503 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24504 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24505 * lra-constraints.c (process_alt_operands): Likewise.
24506 * ipa-inline.c (inline_small_functions): Likewise.
24507 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24508 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24509 * trans-mem.c (diagnose_tm_1_op): Likewise.
24510 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24511 (grid_parallel_clauses_gridifiable): Likewise.
24513 * config/nvptx/mkoffload.c (process): Add space in between
24516 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24517 "MOD4_SSE_REGS" and "ALL_REGS".
24519 * spellcheck.c (test_data): Add , in between "foo" and "food".
24521 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24524 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24525 boundary crossing check and subsequent code generation agree.
24527 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24529 * config/aarch64/aarch64.c (has_memory_op): Delete.
24530 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24533 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24535 PR rtl-optimization/79388
24536 PR rtl-optimization/79450
24537 * combine.c (distribute_notes): When removing TEM_INSN for which
24538 corresponding dest has last value recorded, invalidate that last
24541 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24543 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24544 of explicit '@'. Add missing assembly comment marker on branch costs
24547 2017-02-13 Nathan Sidwell <nathan@acm.org>
24549 * gengtype-lex.l (<in_struct>): Add '/'.
24551 2017-02-13 Martin Liska <mliska@suse.cz>
24554 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24556 2017-02-13 Richard Biener <rguenther@suse.de>
24558 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24560 * configure: Re-generate.
24561 * config.in: Likewise.
24562 * graphite-dependences.c: Simplify as if
24563 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24564 * graphite-isl-ast-to-gimple.c: Likewise.
24565 * graphite-optimize-isl.c: Likewise.
24566 * graphite-poly.c: Likewise.
24567 * graphite-sese-to-poly.c: Likewise.
24568 * graphite.h: Likewise.
24569 * toplev.c: Include isl/version.h and use isl_version () for
24570 printing the ISL version.
24571 * doc/install.texi: Update ISL requirement.
24573 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24575 * doc/standards.texi (Standards): Update reference to
24578 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24580 * doc/extend.texi (Named Address Spaces): sourceware.org now
24582 * doc/install.texi (Binaries): Ditto.
24585 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24587 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24588 standard terminology "stringize"/"stringizing" throughout.
24589 * doc/cppinternals.texi: Likewise.
24591 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24593 * doc/extend.texi: Fix some spelling mistakes and typos.
24594 * doc/invoke.texi: Likewise.
24596 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24599 * params.def (inline-min-speedup) Change from 10 to 8.
24601 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24603 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24606 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24609 * ipa-inline-analysis.c (get_minimal_bb): New function.
24610 (record_modified): Use it.
24611 (remap_edge_change_prob): Handle also ancestor functions.
24613 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24615 * doc/contrib.texi (Contributors): Remove broken link into
24616 the Mauve CVS repository.
24618 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24620 PR middle-end/79454
24621 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24622 result computation whenever lhs doesn't have vector mode, not
24623 just when it has BLKmode.
24625 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24627 * doc/makefile.texi (profiledbootstrap): Refer to the
24628 installation instructions only in textual form.
24630 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24633 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24635 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24637 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24638 (Specific): Update mingw-w64 reference.
24640 (Specific): Remove broken link to Renesas RX processor.
24642 2017-02-10 Richard Biener <rguenther@suse.de>
24644 * toplev.c (process_options): Do not mention obsolete graphite
24645 options when printing sorry message about missing graphite support.
24646 Mention -floop-nest-optimize.
24648 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24650 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24651 (vtst_p16): Likewise.
24652 (vtstq_p8): Likewise.
24653 (vtstq_p16): Likewise.
24655 (vtstq_p64): Likewise.
24656 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24657 (vset_lane_p64): New.
24658 (vsetq_lane_p64): New.
24660 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24662 PR tree-optimization/79411
24663 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24664 stmt operands are SSA_NAMEs used in abnormal phis.
24665 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24668 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24671 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24674 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24676 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24678 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24680 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24684 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24685 not arbitrary TREE_CONSTANT.
24688 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24689 "omp declare target link" attribute unless is_global_var.
24690 * omp-offload.c (find_link_var_op): Likewise.
24692 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24693 Chung-Lin Tang <cltang@codesourcery.com>
24695 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24697 (gimplify_adjust_omp_clauses): Don't delete TILE.
24698 (gimplify_omp_for): Deal with TILE.
24699 * internal-fn.c (expand_GOACC_TILE): New function.
24700 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24702 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24703 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24705 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24706 avoid DIV for outermost collapse var.
24707 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24708 Remove out of date comments, fix whitespace.
24709 * omp-general.c (omp_extract_for_data): Deal with tiling.
24710 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24711 adjust OLF_DIM_BASE value.
24712 (struct omp_for_data): Add tiling field.
24713 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24714 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24715 for auto loops. Remove default auto determining, moved to
24716 oacc_loop_fixed_partitions.
24717 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24718 stmts, add e_mask field.
24719 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24720 (oacc_thread_numbers): Use oacc_dim_call.
24721 (oacc_xform_tile): New.
24722 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24723 (finish_oacc_loop): Adjust for ifns vector.
24724 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24725 add case for GOACC_TILE fns.
24726 (oacc_loop_xform_loop): Delete.
24727 (oacc_loop_process): Iterate over call list directly, and add
24728 handling for GOACC_TILE fns.
24729 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24731 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24732 vector partitioning to outer loops. Assign 2 partitions to loops
24733 when available. Add TILE handling.
24734 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24735 (execite_oacc_device_lower): Process GOACC_TILE fns,
24736 ignore unknown specs.
24737 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24738 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24739 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24741 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24743 * configure.ac (ACX_BUGURL): Update.
24744 * configure: Regenerate.
24746 2017-02-09 Richard Biener <rguenther@suse.de>
24748 PR tree-optimization/69823
24749 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24750 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24752 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24754 * config/arc/arc-c.def: Add __NPS400__ definition.
24755 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24756 (TARGET_NPS400): Define.
24758 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24760 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24762 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24763 pointer, arch_info.
24764 (arc_cpu_types): Fill the arch_info field with a pointer into the
24765 arc_arch_types table.
24766 (arc_selected_cpu): Declare.
24767 * config/arc/arc.c (arc_selected_cpu): Make global.
24768 (arc_selected_arch): Delete.
24769 (arc_base_cpu): Delete.
24770 (arc_override_options): Remove references to deleted variables,
24771 update access to arch information.
24772 (ARC_OPT): Update access to arch information.
24773 (ARC_OPTX): Likewise.
24774 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24775 (TARGET_ARC600): Update access to arch information.
24776 (TARGET_ARC601): Likewise.
24777 (TARGET_ARC700): Likewise.
24778 (TARGET_EM): Likewise.
24779 (TARGET_HS): Likewise.
24780 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24783 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24786 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24787 condition/operands for integer GE/LE/GEU/LEU operations.
24789 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24791 PR translation/79397
24792 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
24795 2017-02-08 Martin Jambor <mjambor@suse.cz>
24798 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
24799 whether allocation happened.
24800 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
24801 nothing was allocated.
24803 2017-02-08 Jakub Jelinek <jakub@redhat.com>
24805 PR tree-optimization/79408
24806 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
24807 constant, but SSA_NAME with a known integer range, use the minimum
24808 of that range instead of op1 to determine if modulo can be replaced
24809 with its first operand.
24811 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24813 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
24815 2017-02-08 Richard Biener <rguenther@suse.de>
24817 PR tree-optimization/71824
24818 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24819 Check all loops contained in the merged region.
24821 2017-02-07 Andrew Pinski <apinski@cavium.com>
24823 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24825 2017-02-07 Andrew Pinski <apinski@cavium.com>
24827 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24828 (thunderxt88): Likewise.
24829 (thunderxt81): Disable LSE and change v8.1 to v8.
24830 (thunderxt83): Likewise.
24832 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24833 Richard Biener <rguenther@suse.de>
24835 PR middle-end/79399
24836 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24837 type from int to size_t.
24838 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24840 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24842 PR rtl-optimization/79386
24843 * cprop.c (bypass_conditional_jumps): Initialize
24844 bypass_last_basic_block already before splitting bbs after
24845 unconditional traps...
24846 (bypass_conditional_jumps): ... rather than here.
24849 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24850 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24851 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24852 fix -masm=intel patterns.
24854 2017-02-07 Richard Biener <rguenther@suse.de>
24856 PR tree-optimization/79256
24857 PR middle-end/79278
24858 * builtins.c (get_object_alignment_2): Use min_align_of_type
24859 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24860 and ADJUST_FIELD_ALIGN.
24862 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24864 * doc/tm.texi: Regenerate.
24865 * stor-layout.c (layout_decl): Adjust.
24866 (update_alignment_for_field): Likewise.
24867 (place_field): Likewise.
24868 (min_align_of_type): Likewise.
24869 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24870 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24871 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24872 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24873 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24874 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24875 * config/i386/i386.c (x86_field_alignment): Likewise.
24876 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24877 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24878 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24879 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24880 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24881 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24885 2017-01-30 Richard Biener <rguenther@suse.de>
24887 PR tree-optimization/79256
24888 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24889 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24892 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24894 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24895 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24896 builtins to SImode and emit a zero-extend, if necessary.
24898 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24900 * docs/invoke.texi (RISC-V Options): Alphabetize.
24902 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24904 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24907 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24909 * config/riscv/riscv.c: New file.
24910 * common/config/riscv/riscv-common.c: Likewise.
24911 * config.gcc: Likewise.
24912 * config/riscv/constraints.md: Likewise.
24913 * config/riscv/elf.h: Likewise.
24914 * config/riscv/generic.md: Likewise.
24915 * config/riscv/linux.h: Likewise.
24916 * config/riscv/multilib-generator: Likewise.
24917 * config/riscv/peephole.md: Likewise.
24918 * config/riscv/pic.md: Likewise.
24919 * config/riscv/predicates.md: Likewise.
24920 * config/riscv/riscv-builtins.c: Likewise.
24921 * config/riscv/riscv-c.c: Likewise.
24922 * config/riscv/riscv-ftypes.def: Likewise.
24923 * config/riscv/riscv-modes.def: Likewise.
24924 * config/riscv/riscv-opts.h: Likewise.
24925 * config/riscv/riscv-protos.h: Likewise.
24926 * config/riscv/riscv.h: Likewise.
24927 * config/riscv/riscv.md: Likewise.
24928 * config/riscv/riscv.opt: Likewise.
24929 * config/riscv/sync.md: Likewise.
24930 * config/riscv/t-elf-multilib: Likewise.
24931 * config/riscv/t-linux: Likewise.
24932 * config/riscv/t-linux-multilib: Likewise.
24933 * config/riscv/t-riscv: Likewise.
24934 * configure.ac: Likewise.
24935 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24936 Waterman as RISC-V maintainers.
24937 * doc/install.texi: Add RISC-V entries.
24938 * doc/invoke.texi: Add RISC-V options section.
24939 * doc/md.texi: Add RISC-V constraints section.
24940 * configure: Regenerated.
24942 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24945 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24946 false values to be constant vectors with all 0 or all 1 bits set.
24947 (vcondu<mode><mode>): Likewise.
24948 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24950 (fpmask_comparison_operator): Update comment.
24951 (vecint_comparison_operator): New predicate.
24952 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24953 vector conditionals when the true and false values are constant
24954 vectors with all 0 bits or all 1 bits set.
24956 2017-02-06 Martin Sebor <msebor@redhat.com>
24958 PR tree-optimization/79376
24959 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24961 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24963 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24964 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24965 to simplify split condition.
24967 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24969 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24970 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24973 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24975 PR rtl-optimization/68664
24976 * target.def (can_speculate_insn): New hook.
24977 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24978 * doc/tm.texi: Regenerate.
24979 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24980 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24981 (rs6000_sched_can_speculate_insn): New function.
24983 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24985 PR tree-optimization/79284
24986 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24987 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24988 vectorizable_mask_load_store, vectorizable_operation,
24989 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24990 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24991 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24992 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24993 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24994 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24995 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24996 is_gimple_assign (stmt). Replace another such test with
24997 is_gimple_assign (stmt).
24999 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25002 * config/avr/avr.c (rtl-iter.h): Include it.
25003 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25004 (avr_legitimate_combined_insn): ...and implementation.
25006 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25008 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25009 * config/s390/s390.c (s390_const_operand_ok)
25010 (s390_canonicalize_comparison, s390_extract_part)
25011 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25012 (s390_contiguous_bitmask_p, s390_rtx_costs)
25013 (legitimize_pic_address): Likewise.
25014 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25015 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25016 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25017 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25018 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25020 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25022 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25023 REGNO($0) == REGNO($1).
25025 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25027 * config/s390/linux.h(SIZE_TYPE): Add comment.
25029 2017-02-06 Julian Brown <julian@codesourcery.com>
25030 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25031 Virendra Pathak <virendra.pathak@broadcom.com>
25033 * config/aarch64/aarch64-cores.def: Change the scheduler
25035 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25036 * config/aarch64/thunderx2t99.md: New file.
25038 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25040 * doc/standards.texi (Go Language): Update link to language
25043 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25045 * tree-eh.c (lower_resx): Sanitize profile.
25046 (cleanup_empty_eh_move_lp): Likewise.
25048 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25051 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25053 * cfgloopmanip.h (loop_version): Update prototype.
25054 * modulo-sched.c (sms_schedule): Update call of loop_version.
25055 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25056 * tree-parloops.c (gen_parallel_loop): Likewise.
25057 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25058 * tree-ssa-loop-split.c (split_loop): Likewise.
25059 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25060 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25062 2017-02-05 Martin Liska <mliska@suse.cz>
25065 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25067 (print_operand_address): Initialize a struct to zero.
25069 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25071 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25072 garbage collector only in textual form.
25074 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25076 * doc/extend.texi (x86 specific memory model extensions for
25077 transactional memory): Simplify a phrase.
25079 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25082 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25083 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25084 (atomic_storedi_1): Likewise.
25086 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25088 PR tree-optimization/79338
25089 * tree-parloops.c (gather_scalar_reductions): Don't call
25090 vect_analyze_loop_form for loop->inner before destroying loop's
25093 2017-02-03 Martin Sebor <msebor@redhat.com>
25095 PR tree-optimization/79327
25096 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25097 when precision has resulted in leading zeros.
25098 (format_integer): Adjust the likely counter to assume an unknown
25099 argument that may be zero is non-zero.
25101 2017-02-03 Jason Merrill <jason@redhat.com>
25104 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25105 avoid copying non-taken branch.
25107 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25109 PR tree-optimization/79340
25110 * tree-vect-loop.c (vectorizable_reduction): Release
25111 vec_defs elements after safe_splicing them into other vectors.
25114 PR tree-optimization/79327
25115 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25116 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25118 (format_integer): Use wide_int_to_tree instead of build_int_cst
25119 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25120 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25121 of shortest and longest sequence.
25123 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25125 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25126 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25128 2017-02-03 Walter Lee <walt@tilera.com>
25131 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25132 after initial stackframe link reg save.
25133 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25135 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25138 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25139 wu for stxssp alternative.
25141 2017-02-03 Martin Sebor <msebor@redhat.com>
25143 PR tree-optimization/79352
25144 * gimple-fold.c (get_range_strlen): Add argument.
25145 (get_range_strlen): Change return type to bool.
25146 (get_maxval_strlen): Pass in a dummy argument.
25147 * gimple-fold.h (get_range_strlen): Change return type to bool.
25148 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25149 * tree.h (array_at_struct_end_p): Add argument.
25150 * tree.c (array_at_struct_end_p): Handle it.
25152 2017-02-03 Martin Liska <mliska@suse.cz>
25155 * multiple_target.c (create_dispatcher_calls): Redirect edge
25156 from a caller of a dispatcher.
25157 (expand_target_clones): Make the clones local.
25158 (ipa_target_clone): Do both target clones and resolvers.
25159 (ipa_dispatcher_calls): Remove the pass.
25160 (pass_dispatcher_calls::gate): Likewise.
25161 (make_pass_dispatcher_calls): Likewise.
25162 * passes.def (pass_target_clone): Put as very first IPA early
25165 2017-02-03 Martin Liska <mliska@suse.cz>
25167 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25168 in case of a function with ifunc attribute.
25170 2017-02-03 Martin Liska <mliska@suse.cz>
25172 * cgraph.c (cgraph_node::dump): Dump function version info.
25173 * symtab.c (symtab_node::dump_base): Add missing new line.
25175 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25177 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25178 (ifcombine_ifandif): Use it.
25180 2017-02-03 Martin Liska <mliska@suse.cz>
25182 * doc/invoke.texi: Document default value for
25183 use-after-scope-direct-emission-threshold.
25185 2017-02-03 Martin Liska <mliska@suse.cz>
25187 PR tree-optimization/79339
25188 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25189 (format_floating): Likewise.
25191 2017-02-03 Martin Liska <mliska@suse.cz>
25194 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25196 (ipa_node_params_t::remove): Likewise.
25197 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25198 initialization from removed ipa_node_params_t::insert.
25199 (ipa_node_params::~ipa_node_params): Move from removed
25200 ipa_node_params_t::release.
25201 * symbol-summary.h (symbol_summary::m_released): New member.
25202 Do not release a summary twice. Do not allow to call finalizer
25203 for types of a summary that live in GGC memory.
25205 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25207 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25210 2017-02-02 Martin Sebor <msebor@redhat.com>
25212 PR middle-end/79275
25213 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25214 (format_string): Tighten up the range of output for non-constant
25215 strings and correct the expected range for wide non-constant strings.
25217 2017-02-02 Martin Sebor <msebor@redhat.com>
25219 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25221 PR middle-end/32003
25222 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25224 (-fdump-tree-@var): Add to index and document how to come up
25225 with pass-specific option and dump file names.
25226 (-fdump-passes): Clarify where to look for output.
25228 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25230 PR middle-end/77445
25231 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25232 statistics of the analyzed path; allow threading for speed when
25233 any of BBs along the path are optimized for speed.
25235 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25237 PR middle-end/78468
25238 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25239 settings of the virtual registers.
25242 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25244 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25245 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25248 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25250 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25251 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25253 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25255 * config/s390/s390.md: Add missing comments with the expanded
25257 * config/s390/vector.md: Likewise.
25258 * config/s390/vx-builtins.md: Likewise.
25260 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25263 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25264 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25265 conditions on a single line.
25267 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25269 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25270 __S390_VX__ to __VX__.
25272 2017-02-01 Andrew Pinski <apinski@cavium.com>
25274 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25275 stmt_info to record_stmt_cost.
25276 (vect_get_known_peeling_cost): Pass stmt_info if known to
25278 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25279 cpu_vector_cost field into
25280 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25281 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25282 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25283 splitting of scalar_stmt_cost and vec_stmt_cost.
25284 (thunderx_vector_cost): Likewise.
25285 (cortexa57_vector_cost): LIkewise.
25286 (exynosm1_vector_cost): Likewise.
25287 (xgene1_vector_cost): Likewise.
25288 (thunderx2t99_vector_cost): Improve after the splitting of the two
25290 (aarch64_builtin_vectorization_cost): Update for the splitting of
25291 scalar_stmt_cost and vec_stmt_cost.
25293 2017-02-01 Torvald Riegel <triegel@redhat.com>
25294 Richard Henderson <rth@redhat.com>
25296 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25297 conditional on existance of a fast atomic load.
25298 * optabs-query.c (can_atomic_load_p): New function.
25299 * optabs-query.h (can_atomic_load_p): Declare it.
25300 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25301 no fast atomic load is available for the particular size of access.
25302 (expand_atomic_compare_and_swap): Likewise.
25303 (expand_atomic_load): Likewise.
25304 (expand_atomic_store): Likewise.
25305 (expand_atomic_fetch_op): Likewise.
25306 * testsuite/lib/target-supports.exp
25307 (check_effective_target_sync_int_128): Remove x86 because it provides
25308 no fast atomic load.
25309 (check_effective_target_sync_int_128_runtime): Likewise.
25311 2017-02-01 Richard Biener <rguenther@suse.de>
25313 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25314 (graphite_transform_loops): Provide opt-info for optimized nests.
25315 * tree-parloop.c (parallelize_loops): Provide opt-info for
25316 parallelized loops.
25318 2017-02-01 Richard Biener <rguenther@suse.de>
25320 PR middle-end/79315
25321 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25322 was not set before.
25324 2017-02-01 Richard Biener <rguenther@suse.de>
25326 PR tree-optimization/71824
25327 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25328 Verify the loops are valid in the merged SESE region.
25329 (scop_detection::can_represent_loop_1): Check analyzing the
25330 evolution of the number of iterations in the region succeeds.
25332 2017-01-31 Ian Lance Taylor <iant@golang.org>
25334 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25335 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25337 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25339 PR preprocessor/79210
25340 * input.c (get_substring_ranges_for_loc): Replace line_width
25341 assertion with error-handling.
25343 2017-01-31 Richard Biener <rguenther@suse.de>
25345 PR tree-optimization/77318
25346 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25347 (create_pw_aff_from_tree): Take loop parameter.
25348 (add_condition_to_pbb): Pass loop of the condition to
25349 create_pw_aff_from_tree.
25351 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25353 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25354 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25356 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25360 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25362 (convert_int_to_float128): Likewise.
25363 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25364 (convert_int_to_float128): Likewise.
25365 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25366 (UNSPEC_IEEE128_CONVERT): Likewise.
25367 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25368 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25369 Use local variables for IBM extended format.
25370 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25371 (fix_trunc<mode>si2_fprs): Likewise.
25372 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25373 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25374 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25375 to know that we can now have integers of all sizes in vector
25377 (fix<uns>_<mode>di2_hw): Likewise.
25378 (float<uns>_<mode>si2_hw): Likewise.
25379 (fix_<mode>si2_hw): Likewise.
25380 (fixuns_<mode>si2_hw): Likewise.
25381 (float<uns>_<mode>di2_hw): Likewise.
25382 (float_<mode>di2_hw): Likewise.
25383 (float_<mode>si2_hw): Likewise.
25384 (floatuns_<mode>di2_hw): Likewise.
25385 (floatuns_<mode>si2_hw): Likewise.
25386 (xscvqp<su>wz_<mode>): Delete, no longer used.
25387 (xscvqp<su>dz_<mode>): Likewise.
25388 (xscv<su>dqp_<mode>): Likewise.
25389 (ieee128_mfvsrd_64bit): Likewise.
25390 (ieee128_mfvsrd_32bit): Likewise.
25391 (ieee128_mfvsrwz): Likewise.
25392 (ieee128_mtvsrw): Likewise.
25393 (ieee128_mtvsrd_64bit): Likewise.
25394 (ieee128_mtvsrd_32bit): Likewise.
25396 2017-01-31 Martin Liska <mliska@suse.cz>
25399 * ipa-prop.c (ipa_free_all_node_params): Call release method
25400 instead of ~sumbol_summary to not to trigger double times
25403 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25405 PR tree-optimization/71691
25406 * bitmap.h (class auto_bitmap): New.
25407 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25408 is_maybe_undefined instead of ssa_undefined_value_p.
25410 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25412 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25413 __S390_ARCH_LEVEL__ to __ARCH__.
25415 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25417 PR tree-optimization/79267
25418 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25419 if should_remove_lhs_p is true.
25421 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25424 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25425 (add_alignment_attribute): New.
25426 (base_type_die): Add alignment attribute.
25427 (subrange_type_die): Likewise.
25428 (modified_type_die): Likewise.
25429 (gen_array_type_die): Likewise.
25430 (gen_descr_array_type_die: Likewise.
25431 (gen_enumeration_type_die): Likewise.
25432 (gen_subprogram_die): Likewise.
25433 (gen_variable_die): Likewise.
25434 (gen_field_die): Likewise.
25435 (gen_ptr_to_mbr_type_die): Likewise.
25436 (gen_struct_or_union_type_die): Likewise.
25437 (gen_subroutine_type_die): Likewise.
25438 (gen_typedef_die): Likewise.
25439 (base_type_cmp): Compare alignment attribute.
25441 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25444 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25445 (setb_unsigned) New pattern for setb with CCUNS.
25446 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25447 subfc./subfe sequence to avoid overflow problems. Generate a
25448 shorter sequence with cmpld/setb for power9.
25449 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25450 for generating subfc. instruction.
25451 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25452 now uses this instruction.
25454 2017-01-30 Ian Lance Taylor <iant@google.com>
25457 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25458 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25460 2017-01-30 Martin Sebor <msebor@redhat.com>
25462 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25463 Move constant to the right of a relational operator.
25464 (get_mpfr_format_length, format_character, format_string): Ditto.
25465 (should_warn_p, maybe_warn): Same.
25467 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25469 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25472 * asan.c (get_translation_unit_decl): Remove function.
25473 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25475 2017-01-30 Martin Liska <mliska@suse.cz>
25477 PR gcov-profile/79259
25478 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25479 -fprofile-generate.
25481 2017-01-30 Martin Liska <mliska@suse.cz>
25484 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25485 Initialize variables with NULL value.
25487 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25490 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25492 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25494 2017-01-30 Richard Biener <rguenther@suse.de>
25496 PR tree-optimization/79276
25497 * tree-vrp.c (process_assert_insertions): Properly adjust common
25498 when removing a duplicate.
25500 * gcc.dg/torture/pr79276.c: New testcase.
25502 2017-01-30 Richard Biener <rguenther@suse.de>
25504 PR tree-optimization/79256
25505 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25506 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25508 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25510 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25513 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25514 ("*r<noxa>sbg_<mode>_sll_bitmask")
25515 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25516 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25517 Use contiguous_bitmask_nowrap_operand.
25519 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25522 * config/rs6000/altivec.h (vec_xl): Revise #define.
25523 (vec_xst): Likewise.
25525 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25527 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25529 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25531 PR rtl-optimization/79194
25532 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25533 traps before call to bypass_conditional_jumps.
25535 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25537 PR tree-optimization/71374
25538 * lra-constraints.c (check_conflict_input_operands): New.
25539 (match_reload): Use it.
25541 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25544 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25545 account to calculate conflict_set.
25547 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25549 PR rtl-optimization/78559
25550 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25551 other_insn in combine.
25553 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25555 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25556 uint16_type_node for BT_UINT16.
25558 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25560 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25561 "RTL Tests" to menu.
25562 (GIMPLE Tests): New node.
25563 (RTL Tests): New node.
25565 2017-01-27 Richard Biener <rguenther@suse.de>
25567 PR tree-optimization/79245
25568 * tree-loop-distribution.c (distribute_loop): Apply cost
25569 modeling also to detected patterns.
25571 2017-01-27 Richard Biener <rguenther@suse.de>
25573 PR tree-optimization/71433
25574 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25575 (compare_assert_loc): New function.
25576 (process_assert_insertions): Sort and optimize assert locations
25577 to remove duplicates and push down identical assertions on
25578 edges to their destination block.
25580 2017-01-27 Richard Biener <rguenther@suse.de>
25582 PR tree-optimization/79244
25583 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25584 out SSA names even if abnormal.
25586 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25588 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25589 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25590 instead of MPFR_RNDN.
25592 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25595 * arm.c (arm_option_override): Don't call build_target_option_node
25596 until after doing all option overrides.
25597 (arm_valid_target_attribute_tree): Likewise.
25599 2017-01-27 Martin Liska <mliska@suse.cz>
25601 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25602 for {cd}tors and C++ {cd}tors.
25604 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25606 * config/s390/s390.md ("*setmem_long_and")
25607 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25609 2017-01-26 Martin Sebor <msebor@redhat.com>
25611 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25614 2017-01-26 Martin Sebor <msebor@redhat.com>
25616 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25617 HAVE_DFmode before using XFmode or DFmode.
25618 (parse_directive): Avoid using the z length modifier to avoid
25619 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25621 PR middle-end/78703
25622 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25623 to accept adjustment as an array.
25624 (get_int_range): New function.
25625 (struct directive): Make width and prec arrays.
25626 (directive::set_width, directive::set_precision): Call get_int_range.
25627 (format_integer, format_floating): Handle width and precision ranges.
25628 (format_string, parse_directive): Same.
25630 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25633 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25634 comdat_type_p set, just clone them, but keep the children in the
25638 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25639 which have direct callers with -fvar-tracking-assignments enabled
25641 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25642 inside of type units.
25644 2017-01-26 Martin Sebor <msebor@redhat.com>
25646 PR middle-end/78703
25647 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25649 (struct format_result): Replace number_chars, number_chars_min,
25650 and number_chars_max with a single member of struct result_range.
25652 (format_result::operator+=): Adjust.
25653 (struct fmtresult): Remove bounded. Handle likely and unlikely
25655 (fmtresult::adjust_for_width_or_precision): New function.
25656 (fmtresult:type_max_digits): New function.
25657 (bytes_remaining): Handle likely and unlikely counters.
25658 (min_bytes_remaining): Remove.
25659 (format_percent): Simplify.
25660 (format_integer, format_floating): Set likely and unlikely counters.
25661 (get_string_length, format_character, format_string): Same.
25662 (format_plain, should_warn_p): New function.
25663 (maybe_warn): Call should_warn_p. Update diagnostic messages
25664 and handle those for all directives, including plain strings.
25665 (format_directive): Handle likely and unlikely counters.
25666 Remove unnecessary quoting from diagnostics. Add an informational
25668 (add_bytes): Remove.
25669 (pass_sprintf_length::compute_format_length): Simplify.
25670 (try_substitute_return_value): Handle likely and unlikely counters.
25672 2017-01-26 Carl Love <cel@us.ibm.com>
25674 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25675 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25677 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25680 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25681 endianess for subregs into account.
25682 * lra-constraints.c (lra_constraints): Do risky transformations
25683 always on the first iteration.
25684 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25685 last_call_used_reg_set.
25686 (process_bb_lives): Define and use last_call_used_reg_set.
25687 * lra.c (lra): Always continue after lra_constraints on the first
25690 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25692 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25694 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25696 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25698 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25699 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25700 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25701 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25702 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25703 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25704 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25705 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25706 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25708 2017-01-26 Marek Polacek <polacek@redhat.com>
25711 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25712 for the third operand.
25714 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25716 PR middle-end/79236
25717 * omp-low.c (struct omp_context): Add simt_stmt field.
25718 (scan_omp_for): Return omp_context *.
25719 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25720 context to the _simt_ SIMD stmt.
25721 (lower_omp_for): For combined SIMD with sibling _simt_
25722 SIMD, make sure to use the same decls in _looptemp_
25723 clauses as in the sibling.
25725 2017-01-26 David Sherwood <david.sherwood@arm.com>
25727 PR middle-end/79212
25728 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25731 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25734 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25735 emit fld b; fld a; if possible.
25737 * brig-builtins.def: Update copyright years.
25738 * config/arm/arm_acle_builtins.def: Update copyright years.
25740 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25743 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25744 constraint instead of o for the stxsd instruction.
25746 2017-01-25 Carl Love <cel@us.ibm.com>
25748 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25749 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25751 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25753 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25755 2017-01-25 Richard Biener <rguenther@suse.de>
25757 PR tree-optimization/69264
25758 * target.def (vector_alignment_reachable): Improve documentation.
25759 * doc/tm.texi: Regenerate.
25760 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25762 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25763 earlier changes with respect to TYPE_USER_ALIGN.
25764 (vector_alignment_reachable_p): Likewise. Improve dumping.
25766 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25769 * config/arm/arm.md (xordi3): Force constant operand into a register
25772 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25774 * doc/invoke.texi (-fstore-merging): Correct default optimization
25775 levels at which it is enabled.
25776 (-O): Move -fstore-merging from list to...
25779 2017-01-25 Richard Biener <rguenther@suse.de>
25782 * omp-expand.c: Include debug.h.
25783 (expand_omp_taskreg): Make sure to generate early debug before
25784 outlining anything from a function.
25785 (expand_omp_target): Likewise.
25786 (grid_expand_target_grid_body): Likewise.
25788 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25791 * asan.c (get_translation_unit_decl): New function.
25792 (asan_add_global): Extract modules file name from globals
25793 TRANSLATION_UNIT_DECL name.
25795 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
25798 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
25799 for long calls with APCS frame and VFP.
25801 2017-01-24 David Malcolm <dmalcolm@redhat.com>
25803 * cfg.c (original_copy_tables_initialized_p): New function.
25804 * cfg.h (original_copy_tables_initialized_p): New decl.
25805 * cfgrtl.c (relink_block_chain): Guard the call to
25806 free_original_copy_tables with a call to
25807 original_copy_tables_initialized_p.
25808 * cgraph.h (symtab_node::native_rtl_p): New decl.
25809 * cgraphunit.c (symtab_node::native_rtl_p): New function.
25810 (symtab_node::needed_p): Don't assert for early assembly output
25811 for __RTL functions.
25812 (cgraph_node::finalize_function): Set "force_output" for __RTL
25814 (cgraph_node::analyze): Bail out early for __RTL functions.
25815 (analyze_functions): Update assertion to support __RTL functions.
25816 (cgraph_node::expand): Bail out early for __RTL functions.
25817 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25819 * function.h (struct function): Update comment for field
25821 * gimple-expr.c: Include "tree-pass.h".
25822 (gimple_has_body_p): Return false for __RTL functions.
25823 * Makefile.in (OBJS): Add run-rtl-passes.o.
25824 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25826 (gcc::pass_manager::get_clean_slate): New accessor.
25827 * passes.c: Include "insn-addr.h".
25828 (should_skip_pass_p): Add logging. Update logic for running
25829 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25830 property-provider override so it is only done for gimple passes.
25832 (skip_pass): New function.
25833 (execute_one_pass): Call skip_pass when skipping passes.
25834 * read-md.c (md_reader::read_char): Support filtering
25835 the input to a subset of line numbers.
25836 (md_reader::md_reader): Initialize fields
25837 m_first_line and m_last_line.
25838 (md_reader::read_file_fragment): New function.
25839 * read-md.h (md_reader::read_file_fragment): New decl.
25840 (md_reader::m_first_line): New field.
25841 (md_reader::m_last_line): New field.
25842 * read-rtl-function.c (function_reader::create_function): Only
25843 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25844 curr_properties. Set DECL_INITIAL to a dummy block.
25845 (read_rtl_function_body_from_file_range): New function.
25846 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25848 * run-rtl-passes.c: New file.
25849 * run-rtl-passes.h: New file.
25851 2017-01-24 Jeff Law <law@redhat.com>
25853 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25856 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25858 PR tree-optimization/79159
25859 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25860 (record_nonwrapping_iv): Improve boundary using above function if no
25861 value range information.
25863 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25864 Martin Jambor <mjambor@suse.cz>
25866 * brig-builtins.def: New file.
25867 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25868 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25869 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25870 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25871 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25872 * builtin-types.def (BT_INT8): New.
25873 (BT_INT16): Likewise.
25874 (BT_UINT8): Likewise.
25875 (BT_UINT16): Likewise.
25876 (BT_FN_ULONG): Likewise.
25877 (BT_FN_UINT_INT): Likewise.
25878 (BT_FN_UINT_ULONG): Likewise.
25879 (BT_FN_UINT_LONG): Likewise.
25880 (BT_FN_UINT_PTR): Likewise.
25881 (BT_FN_ULONG_PTR): Likewise.
25882 (BT_FN_INT8_FLOAT): Likewise.
25883 (BT_FN_INT16_FLOAT): Likewise.
25884 (BT_FN_UINT32_FLOAT): Likewise.
25885 (BT_FN_UINT16_FLOAT): Likewise.
25886 (BT_FN_UINT8_FLOAT): Likewise.
25887 (BT_FN_UINT64_FLOAT): Likewise.
25888 (BT_FN_UINT16_UINT32): Likewise.
25889 (BT_FN_UINT32_UINT16): Likewise.
25890 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25891 (BT_FN_INT_PTR_INT): Likewise.
25892 (BT_FN_UINT_PTR_UINT): Likewise.
25893 (BT_FN_LONG_PTR_LONG): Likewise.
25894 (BT_FN_ULONG_PTR_ULONG): Likewise.
25895 (BT_FN_VOID_UINT64_UINT64): Likewise.
25896 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25897 (BT_FN_INT8_INT8_INT8): Likewise.
25898 (BT_FN_INT16_INT16_INT16): Likewise.
25899 (BT_FN_INT_INT_INT): Likewise.
25900 (BT_FN_UINT_FLOAT_UINT): Likewise.
25901 (BT_FN_FLOAT_UINT_UINT): Likewise.
25902 (BT_FN_ULONG_UINT_UINT): Likewise.
25903 (BT_FN_ULONG_UINT_PTR): Likewise.
25904 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25905 (BT_FN_UINT_UINT_UINT): Likewise.
25906 (BT_FN_VOID_UINT_PTR): Likewise.
25907 (BT_FN_UINT_UINT_PTR: Likewise.
25908 (BT_FN_UINT32_UINT64_PTR): Likewise.
25909 (BT_FN_INT_INT_UINT_UINT): Likewise.
25910 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25911 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25912 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25913 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25914 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25915 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25916 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25917 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25918 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25919 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25920 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25921 * doc/frontends.texi: List BRIG FE.
25922 * doc/install.texi (Testing): Add BRIG tesring requirements.
25923 * doc/invoke.texi (Overall Options): Mention BRIG.
25924 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25926 2017-01-24 Richard Biener <rguenther@suse.de>
25928 PR translation/79208
25929 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25931 2017-01-24 Martin Jambor <mjambor@suse.cz>
25934 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25935 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25936 and known_contexts.
25938 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25940 PR middle-end/79123
25941 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25942 casts from signed to unsigned really don't have a range.
25944 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25946 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25947 GMP_RNDx for compatiblity.
25949 2017-01-24 Martin Liska <mliska@suse.cz>
25952 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25953 that would prevent us to call alloca with -1 as argument.
25955 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25957 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25958 Avoid -Wformat-security warning.
25960 2017-01-23 Andrew Pinski <apinski@cavium.com>
25962 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25965 2017-01-23 Martin Sebor <msebor@redhat.com>
25967 PR middle-end/78703
25968 * gimple-ssa-sprintf.c (warn_level): New global.
25969 (format_integer): Use it here and throughout the rest of the file.
25970 Use the same switch to compute sign as base.
25971 (maybe_warn): New function.
25972 (format_directive): Factor out warnings into maybe_warn.
25973 Add debugging output. Use warn_level.
25974 (add_bytes): Use warn_level.
25975 (pass_sprintf_length::compute_format_length): Add debugging output.
25976 (try_substitute_return_value): Same.
25977 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25979 PR middle-end/78703
25980 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25981 (struct fmtresult, format_integer, format_floating): Adjust.
25982 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25983 (get_string_length, format_string,format_directive): Same.
25984 (pass_sprintf_length::compute_format_length): Same.
25985 (try_substitute_return_value): Simplify slightly.
25987 PR middle-end/78703
25988 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25989 (fmtresult::operator+=): Outlined.
25990 (struct fmtresult): Add ctors.
25991 (struct conversion_spec): Rename...
25992 (struct directive): ...to this. Add and remove data members.
25993 (directive::set_width, directive::set_precision): New functions.
25994 (format_percent): Use fmtresult ctor.
25995 (get_width_and_precision): Remove.
25996 (format_integer): Make naming changes. Avoid computing width and
25998 (format_floating): Same. Adjust indentation.
25999 (format_character, format_none): New functions.
26000 (format_string): Moved character handling to format_character.
26001 (format_directive): Remove arguments, change return type.
26002 (parse_directive): New function.
26003 (pass_sprintf_length::compute_format_length): Move directive
26004 parsing to parse_directive.
26006 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26008 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26009 (assign_assembler_name_if_needed): ... this.
26010 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26011 (assign_assembler_name_if_needed): ... this.
26012 (free_lang_data_in_cgraph): Adjust callers.
26013 * cgraphunit.c (cgraph_node::analyze): Likewise.
26014 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26017 2017-01-23 Richard Biener <rguenther@suse.de>
26019 PR tree-optimization/79088
26020 PR tree-optimization/79188
26021 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26022 resetting loop bounds after last path deletion. Reset loop
26023 bounds of the target loop, make code match the comments.
26024 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26025 Make sure loops need no fixups.
26027 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26029 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26030 exponent support with double type for first argument.
26031 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26032 type returned by __builtin_vec_extract_sig,
26033 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26034 functions from "vector int" to "vector unsigned int" or from
26035 "vector long long int" to "vector unsigned long long int".
26036 Changed type returned by __builtin_vec_extract_exp,
26037 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26038 functions from "vector int" to "vector unsigned int" or from
26039 "vector long long int" to "vector unsigned long long int".
26040 Changed return type of __builtin_vec_test_data_class,
26041 __builtin_vec_test_data_class_sp, and
26042 __builtin_vec_test_data_class_dp from "vector int" to
26043 "vector bool int" or from "vector long long int" to "vector bool
26044 long long int" and changed second argument type from "unsigned
26045 int" to "int". Added new overloaded function forms "vector float
26046 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26047 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26048 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26049 double, vector unsigned long long int)" and "vector double
26050 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26051 long int)". Changed return type of
26052 __builtin_scalar_test_data_class and
26053 __builtin_scalar_test_data_class_sp and
26054 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26055 int" and changed second argument from "unsigned int" to "int".
26056 Changed type returned by __builtin_scalar_test_neg,
26057 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26058 from "int" to "bool int". Added new overloaded function form
26059 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26060 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26061 exponent double-precision with floating point first argument.
26062 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26063 documentation of scalar_test_data_class, scalar_test_neg,
26064 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26065 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26066 vec_test_data_class built-in functions to reflect refinements in
26067 their type signatures.
26069 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26071 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26073 (aarch64_elf_asm_destructor): Likewise.
26075 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26077 PR rtl-optimization/78634
26078 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26079 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26080 * ifcvt.c (noce_try_cmove): Add missing cost check.
26082 PR rtl-optimization/71724
26083 * combine.c (if_then_else_cond): Look for situations where it is
26084 beneficial to undo the work of one of the recursive calls.
26086 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26088 PR tree-optimization/70754
26089 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26090 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26091 combined stmt before it if not NULL.
26092 (combine_chains): Process refs reversely and compute dominance point
26095 2017-01-23 Martin Liska <mliska@suse.cz>
26097 PR tree-optimization/79196
26098 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26099 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26101 (strlen_optimize_stmt): Call the renamed function.
26103 2017-01-23 Michael Matz <matz@suse.de>
26105 PR tree-optimization/78384
26106 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26108 2017-01-23 Richard Biener <rguenther@suse.de>
26110 PR tree-optimization/79186
26111 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26112 both incoming edges before moving an assert.
26114 2017-01-23 Martin Jambor <mjambor@suse.cz>
26116 * ipa-prop.c (load_from_param_1): Removed.
26117 (load_from_unmodified_param): Bits from load_from_param_1 put back
26119 (load_from_param): Removed.
26120 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26121 with stmt. Reverted back to use of load_from_unmodified_param.
26123 2017-01-23 Martin Jambor <mjambor@suse.cz>
26126 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26127 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26128 field a pointer to garbage collected vector, mark lattices and
26129 ipcp_orig_node with GTY((skip)).
26130 (ipa_get_param_count): Adjust to descriptors being a pointer.
26131 (ipa_get_param): Likewise.
26132 (ipa_get_type): Likewise.
26133 (ipa_get_param_move_cost): Likewise.
26134 (ipa_set_param_used): Likewise.
26135 (ipa_get_controlled_uses): Likewise.
26136 (ipa_set_controlled_uses): Likewise.
26137 (ipa_is_param_used): Likewise.
26138 (ipa_node_params_t): Move into garbage collector. New methods insert
26140 (ipa_node_params_sum): Annotate wth GTY(()).
26141 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26143 (ipa_load_from_parm_agg): Adjust declaration.
26144 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26145 * ipa-profile.c (ipa_profile): Likewise.
26146 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26147 (ipa_populate_param_decls): Make descriptors parameter garbage
26149 (ipa_dump_param): Adjust to descriptors being a pointer.
26150 (ipa_alloc_node_params): Likewise.
26151 (ipa_initialize_node_params): Likewise.
26152 (load_from_param_1): Make descriptors parameter garbage collected.
26153 (load_from_unmodified_param): Likewise.
26154 (load_from_param): Likewise.
26155 (ipa_load_from_parm_agg): Likewise.
26156 (ipa_node_params::~ipa_node_params): Removed.
26157 (ipa_free_all_node_params): Remove call to delete operator.
26158 (ipa_node_params_t::insert): New.
26159 (ipa_node_params_t::remove): Likewise.
26160 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26161 copy known_csts and known_contexts vectors.
26162 (ipa_read_node_info): Adjust to descriptors being a pointer.
26163 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26165 (ipcp_transform_function): Make descriptors variable garbage
26168 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26170 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26171 * config/i386/avx512dqintrin.h: Ditto.
26172 * config/i386/avx512fintrin.h: Ditto.
26173 * config/i386/i386.c: Handle new builtins.
26174 * config/i386/i386-builtin.def: Add new builtins.
26175 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26176 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26178 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26179 Martin Liska <mliska@suse.cz>
26181 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26182 * asan.c (asan_expand_poison_ifn): Support stores and use
26183 appropriate ASAN report function.
26184 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26185 * internal-fn.def (ASAN_POISON_USE): Declare.
26186 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26187 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26188 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26189 ASAN_POISON calls w/o LHS.
26190 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26191 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26192 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26193 * gimplify.c (asan_poison_variables): Add attribute
26194 use_after_scope_memory to variables that really needs to live
26196 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26197 having the attribute.
26199 2017-01-23 Martin Liska <mliska@suse.cz>
26201 * asan.c (create_asan_shadow_var): New function.
26202 (asan_expand_poison_ifn): Likewise.
26203 * asan.h (asan_expand_poison_ifn): New declaration.
26204 * internal-fn.c (expand_ASAN_POISON): Likewise.
26205 * internal-fn.def (ASAN_POISON): New builtin.
26206 * sanopt.c (pass_sanopt::execute): Expand
26207 asan_expand_poison_ifn.
26208 * tree-inline.c (copy_decl_for_dup_finish): Make function
26210 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26211 * tree-ssa.c (is_asan_mark_p): New function.
26212 (execute_update_addresses_taken): Rewrite local variables
26213 (identified just by use-after-scope as addressable) into SSA.
26215 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26217 * doc/install.texi (Specific): opensource.apple.com uses https
26218 now. Remove trailing slash.
26220 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26222 * README.Portability: Remove note on an Irix compatibility issue.
26224 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26226 * gcov.c (INCLUDE_ALGORITHM): Define.
26227 (INCLUDE_VECTOR): Define.
26228 No longer include <vector> and <algorithm> directly.
26230 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26232 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26234 * doc/invoke.texi (Code Gen Options): Ditto.
26236 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26239 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26241 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26243 rtl-optimization/79125
26244 * cprop.c (local_cprop_pass): Handle cases where we make an
26245 unconditional trap.
26247 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26251 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26252 read from, for big endian.
26254 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26256 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26257 register pauth builtins for LP64 only.
26259 2017-01-20 Marek Polacek <polacek@redhat.com>
26262 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26265 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26267 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26269 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26270 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26271 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26273 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26276 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26277 in big-endian mode when they are not a single duplicated value.
26279 2017-01-20 Richard Biener <rguenther@suse.de>
26281 * BASE-VER: Bump to 7.0.1.
26283 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26285 * omp-low.c (omplow_simd_context): New struct. Use it...
26286 (lower_rec_simd_input_clauses): ...here and...
26287 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26288 references to idx, lane, max_vf, is_simt.
26290 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26292 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26295 2017-01-20 Martin Jambor <mjambor@suse.cz>
26297 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26298 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26300 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26301 (GTFILES): Rename hsa.c to hsa-common.c.
26302 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26303 * hsa-dump.c: Likewise.
26304 * hsa-gen.c: Likewise.
26305 * hsa-regalloc.c: Likewise.
26306 * ipa-hsa.c: Likewise.
26307 * omp-expand.c: Likewise.
26308 * omp-low.c: Likewise.
26309 * toplev.c: Likewise.
26311 2017-01-20 Marek Polacek <polacek@redhat.com>
26314 * doc/invoke.texi: Document -Wduplicated-branches.
26315 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26316 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26317 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26318 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26319 return 0 only when not OEP_LEXICOGRAPHIC.
26320 (fold_build_cleanup_point_expr): Use the expression
26321 location when building CLEANUP_POINT_EXPR.
26322 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26323 * tree.c (add_expr): Handle error_mark_node.
26325 2017-01-20 Martin Liska <mliska@suse.cz>
26328 * tree-profile.c (init_ic_make_global_vars): Do not call
26330 (gimple_init_gcov_profiler): Likewise.
26332 2017-01-20 Martin Liska <mliska@suse.cz>
26335 * cgraph.h (maybe_create_reference): Remove argument and
26337 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26339 * ipa-cp.c (create_specialized_node): Likewise.
26340 * symtab.c (symtab_node::maybe_create_reference): Handle
26341 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26343 2017-01-20 Martin Liska <mliska@suse.cz>
26345 * read-rtl-function.c (function_reader::create_function): Use
26346 build_decl instread of build_decl_stat.
26348 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26350 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26351 * config/i386/avx512dqintrin.h: Ditto.
26352 * config/i386/avx512fintrin.h: Ditto.
26353 * config/i386/i386-builtin-types.def: Add new types.
26354 * config/i386/i386.c: Handle new types.
26355 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26356 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26357 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26358 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26359 (__builtin_ia32_kshiftridi): New.
26360 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26362 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26366 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26367 define to rs6000_init_stack_protect_guard.
26368 (rs6000_init_stack_protect_guard): New function.
26370 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26371 Yunqiang Su <yunqiang.su@imgtec.com>
26373 * config.gcc (supported_defaults): Add madd4.
26374 (with_madd4): Add validation.
26375 (all_defaults): Add madd4.
26376 * config/mips/mips.opt (mmadd4): New option.
26377 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26379 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26381 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26382 (ISA_HAS_FUSED_MADD4): Likewise.
26383 * doc/invoke.texi (-mmadd4): Document the new option.
26384 * doc/install.texi (--with-madd4): Document the new option.
26386 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26388 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26389 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26390 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26391 (aarch64_init_pauth_hint_builtins): New.
26392 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26393 (aarch64_expand_builtin): Expand new builtins.
26395 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26397 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26398 * combine-stack-adj.c (no_unhandled_cfa): Handle
26399 REG_CFA_TOGGLE_RA_MANGLE.
26400 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26401 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26402 info for return address signing.
26403 (aarch64_expand_epilogue): Likewise.
26405 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26407 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26408 * config/aarch64/aarch64-protos.h
26409 (aarch64_return_address_signing_enabled): New declaration.
26410 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26412 (aarch64_expand_prologue): Sign return address before it's pushed onto
26414 (aarch64_expand_epilogue): Authenticate return address fetched from
26416 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26417 (aarch64_attributes): New function attributes for "sign-return-address".
26418 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26419 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26420 ("*do_return"): Generate combined instructions according to key index.
26421 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26422 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26424 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26425 * config/aarch64/aarch64.opt (msign-return-address=): New.
26426 * doc/extend.texi (AArch64 Function Attributes): Documents
26427 "sign-return-address=".
26428 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26430 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26432 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26433 overall option summary.
26435 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26437 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26438 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26439 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26440 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26442 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26444 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26445 -mpower9-minmax by default for -mcpu=power9.
26446 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26447 128-bit floating point.
26449 2017-01-20 Alan Modra <amodra@gmail.com>
26451 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26452 optimizing for size.
26454 2017-01-20 Alan Modra <amodra@gmail.com>
26457 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26458 for strcmp and strncmp from corresponding builtin decl.
26460 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26462 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26463 instead of i386/rtems-64.h.
26464 * config/i386/rtems-64.h: Remove.
26466 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26470 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26472 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26474 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26476 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26477 Change int to HOST_WIDE_INT.
26478 * config/aarch64/aarch64-protos.h
26479 (aarch64_simd_gen_const_vector_dup): Likewise.
26480 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26482 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26484 * langhooks-def.h (lhd_type_for_size): New decl.
26485 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26486 * langhooks.c (lhd_type_for_size): New function, taken from
26489 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26491 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26492 define_bypass for CR latency.
26493 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26494 (power9-alu2): Add define_bypass for CR latency.
26496 (power9-mul): Update insn latency.
26497 (power9-mul-compare): Update insn latency, bypass latency and remove
26500 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26502 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26504 * config/aarch64/aarch64.md
26505 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26506 aarch64_nopcrelative_literal_loads.
26507 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26509 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26511 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26512 TARGET_LOONGSON_3A.
26513 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26515 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26518 * config.gcc (supported_defaults): Add lxc1-sxc1.
26519 (with_lxc1_sxc1): Add validation.
26520 (all_defaults): Add lxc1-sxc1.
26521 * config/mips/mips.opt (mlxc1-sxc1): New option.
26522 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26524 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26525 __mips_no_lxc1_sxc1.
26526 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26527 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26528 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26530 2017-01-19 Richard Biener <rguenther@suse.de>
26532 PR tree-optimization/72488
26533 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26534 sure to restore SSA info.
26535 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26537 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26539 PR rtl-optimization/79121
26540 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26541 of the inner type when shifting an extended value.
26543 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26546 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26547 interposable aliases.
26549 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26552 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26553 Use the evmergelohi instruction.
26554 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26555 (mov_sitf_e500_subreg8_2_be): Likewise.
26556 (mov_sitf_e500_subreg12_2_le): Likewise.
26557 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26558 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26559 (mov_sitf_e500_subreg8_2_le): Likewise.
26560 (mov_sitf_e500_subreg12_2_be): Likewise.
26562 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26564 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26565 attribute from vecsimple to vecperm.
26566 (altivec_vbpermq2): Likewise.
26568 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26571 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26573 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26574 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26575 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26576 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26577 case where N arg is SIZE_MAX.
26578 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26579 (cmpstrsi): Add pattern.
26581 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26583 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26584 __builtin_vec_revb builtins.
26585 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26586 built-in functions to support generation of the ISA 3.0 XXBR<x>
26587 vector byte reverse instructions.
26588 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26589 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26590 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26591 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26592 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26593 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26594 (P9V_BUILTIN_VEC_REVB): Likewise.
26595 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26596 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26597 (p9_xxbrq_v16qi): Likewise.
26598 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26599 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26600 (p9_xxbrh_v8hi): Likewise.
26601 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26602 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26603 vec_revb built-in functions.
26605 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26607 PR rtl-optimization/78952
26608 * config/i386/i386.md (any_extract): New code iterator.
26609 (*insvqi_2): Use any_extract for source operand.
26610 (*insvqi_3): Use any_shiftrt for source operand.
26612 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26614 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26616 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26618 2017-01-18 Matthias Klose <doko@ubuntu.com>
26620 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26622 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26624 * config/rs6000/altivec.h (vec_bperm): Change #define.
26625 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26626 (altivec_vbpermq2): New define_insn.
26627 (altivec_vbpermd): Likewise.
26628 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26629 function interface.
26630 (VBPERMD): Likewise.
26631 (VBPERM): New polymorphic function interface.
26632 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26633 Add entries for P9V_BUILTIN_VEC_VBPERM.
26634 * doc/extend.texi: Add interfaces for vec_bperm.
26636 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26638 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26639 first letter of error messages.
26640 (s390_resolve_overloaded_builtin): Likewise.
26641 * config/s390/s390.c (s390_expand_builtin): Likewise.
26642 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26643 (s390_valid_target_attribute_inner_p): Likewise.
26644 * config/s390/s390.md ("tabort"): Likewise.
26646 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26648 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26649 (ISA_AVOID_DIV_HILO): New macro.
26650 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26651 (ISA_HAS_DDIV): Likewise.
26653 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26655 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26657 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26659 * doc/invoke.texi (fabi-version): Spelling fix.
26661 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26664 * doc/invoke.texi (fabi-version): Mention mangling fix for
26667 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26670 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26672 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26675 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26678 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26680 (stack_protect_test): Ditto.
26681 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26682 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26684 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26685 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26686 -mstack-protector-guard-offset=.
26687 (RS/6000 and PowerPC Options): Ditto.
26689 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26691 * config/i386/i386.h (MASK_CLASS_P): New define.
26692 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26693 there are no registers from different register sets also when
26694 mask registers are used. Update function comment.
26695 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26696 to (*k/*r) and (*k/*km) alternatives.
26698 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26700 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26701 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26702 (EH_RETURN_HANDLER_RTX): New define.
26703 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26704 Force frame pointer in EH return functions.
26705 (aarch64_expand_epilogue): Add barrier for eh_return.
26706 (aarch64_final_eh_return_addr): Remove.
26707 (aarch64_eh_return_handler_rtx): New function.
26708 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26710 (aarch64_eh_return_handler_rtx): New prototype.
26712 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26714 * config/rs6000/altivec.h (vec_rlmi): New #define.
26715 (vec_vrlnm): Likewise.
26716 (vec_rlnm): Likewise.
26717 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26718 (UNSPEC_VRLNM): Likewise.
26719 (VIlong): New mode iterator.
26720 (altivec_vrl<VI_char>mi): New define_insn.
26721 (altivec_vrl<VI_char>nm): Likewise.
26722 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26724 (VRLDNM): Likewise.
26725 (RLNM): New polymorphic function entry.
26726 (VRLWMI): New monomorphic function entry.
26727 (VRLDMI): Likewise.
26728 (RLMI): New polymorphic function entry.
26729 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26730 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26731 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26734 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26737 * dwarf2out.c (field_byte_offset): Restore the
26738 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26739 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26740 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26743 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26746 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26748 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26751 * dwarf2out.c (add_data_member_location_attribute): For constant
26752 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26753 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26754 DW_AT_byte_size attributes.
26756 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26758 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26759 after forcing to constant memory when the code model is medium.
26761 2017-01-17 Julia Koval <julia.koval@intel.com>
26764 * config/i386/avx512fintrin.h
26765 (_mm512_i32gather_ps): Change __addr type to void const*.
26766 (_mm512_mask_i32gather_ps): Ditto.
26767 (_mm512_i32gather_pd): Ditto.
26768 (_mm512_mask_i32gather_pd): Ditto.
26769 (_mm512_i64gather_ps): Ditto.
26770 (_mm512_mask_i64gather_ps): Ditto.
26771 (_mm512_i64gather_pd): Ditto.
26772 (_mm512_mask_i64gather_pd): Ditto.
26773 (_mm512_i32gather_epi32): Ditto.
26774 (_mm512_mask_i32gather_epi32): Ditto.
26775 (_mm512_i32gather_epi64): Ditto.
26776 (_mm512_mask_i32gather_epi64): Ditto.
26777 (_mm512_i64gather_epi32): Ditto.
26778 (_mm512_mask_i64gather_epi32): Ditto.
26779 (_mm512_i64gather_epi64): Ditto.
26780 (_mm512_mask_i64gather_epi64): Ditto.
26781 (_mm512_i32scatter_ps): Change __addr type to void*.
26782 (_mm512_mask_i32scatter_ps): Ditto.
26783 (_mm512_i32scatter_pd): Ditto.
26784 (_mm512_mask_i32scatter_pd): Ditto.
26785 (_mm512_i64scatter_ps): Ditto.
26786 (_mm512_mask_i64scatter_ps): Ditto.
26787 (_mm512_i64scatter_pd): Ditto.
26788 (_mm512_mask_i64scatter_pd): Ditto.
26789 (_mm512_i32scatter_epi32): Ditto.
26790 (_mm512_mask_i32scatter_epi32): Ditto.
26791 (_mm512_i32scatter_epi64): Ditto.
26792 (_mm512_mask_i32scatter_epi64): Ditto.
26793 (_mm512_i64scatter_epi32): Ditto.
26794 (_mm512_mask_i64scatter_epi32): Ditto.
26795 (_mm512_i64scatter_epi64): Ditto.
26796 (_mm512_mask_i64scatter_epi64): Ditto.
26797 * config/i386/avx512pfintrin.h
26798 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
26799 (_mm512_mask_prefetch_i32gather_ps): Ditto.
26800 (_mm512_mask_prefetch_i64gather_pd): Ditto.
26801 (_mm512_mask_prefetch_i64gather_ps): Ditto.
26802 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
26803 (_mm512_prefetch_i32scatter_ps): Ditto.
26804 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
26805 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
26806 (_mm512_prefetch_i64scatter_pd): Ditto.
26807 (_mm512_prefetch_i64scatter_ps): Ditto.
26808 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
26809 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
26810 * config/i386/avx512vlintrin.h
26811 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
26812 (_mm_mmask_i32gather_ps): Ditto.
26813 (_mm256_mmask_i32gather_pd): Ditto.
26814 (_mm_mmask_i32gather_pd): Ditto.
26815 (_mm256_mmask_i64gather_ps): Ditto.
26816 (_mm_mmask_i64gather_ps): Ditto.
26817 (_mm256_mmask_i64gather_pd): Ditto.
26818 (_mm_mmask_i64gather_pd): Ditto.
26819 (_mm256_mmask_i32gather_epi32): Ditto.
26820 (_mm_mmask_i32gather_epi32): Ditto.
26821 (_mm256_mmask_i32gather_epi64): Ditto.
26822 (_mm_mmask_i32gather_epi64): Ditto.
26823 (_mm256_mmask_i64gather_epi32): Ditto.
26824 (_mm_mmask_i64gather_epi32): Ditto.
26825 (_mm256_mmask_i64gather_epi64): Ditto.
26826 (_mm_mmask_i64gather_epi64): Ditto.
26827 (_mm256_i32scatter_ps): Change __addr type to void*.
26828 (_mm256_mask_i32scatter_ps): Ditto.
26829 (_mm_i32scatter_ps): Ditto.
26830 (_mm_mask_i32scatter_ps): Ditto.
26831 (_mm256_i32scatter_pd): Ditto.
26832 (_mm256_mask_i32scatter_pd): Ditto.
26833 (_mm_i32scatter_pd): Ditto.
26834 (_mm_mask_i32scatter_pd): Ditto.
26835 (_mm256_i64scatter_ps): Ditto.
26836 (_mm256_mask_i64scatter_ps): Ditto.
26837 (_mm_i64scatter_ps): Ditto.
26838 (_mm_mask_i64scatter_ps): Ditto.
26839 (_mm256_i64scatter_pd): Ditto.
26840 (_mm256_mask_i64scatter_pd): Ditto.
26841 (_mm_i64scatter_pd): Ditto.
26842 (_mm_mask_i64scatter_pd): Ditto.
26843 (_mm256_i32scatter_epi32): Ditto.
26844 (_mm256_mask_i32scatter_epi32): Ditto.
26845 (_mm_i32scatter_epi32): Ditto.
26846 (_mm_mask_i32scatter_epi32): Ditto.
26847 (_mm256_i32scatter_epi64): Ditto.
26848 (_mm256_mask_i32scatter_epi64): Ditto.
26849 (_mm_i32scatter_epi64): Ditto.
26850 (_mm_mask_i32scatter_epi64): Ditto.
26851 (_mm256_i64scatter_epi32): Ditto.
26852 (_mm256_mask_i64scatter_epi32): Ditto.
26853 (_mm_i64scatter_epi32): Ditto.
26854 (_mm_mask_i64scatter_epi32): Ditto.
26855 (_mm256_i64scatter_epi64): Ditto.
26856 (_mm256_mask_i64scatter_epi64): Ditto.
26857 (_mm_i64scatter_epi64): Ditto.
26858 (_mm_mask_i64scatter_epi64): Ditto.
26859 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26860 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26861 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26862 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26863 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26864 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26865 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26866 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26867 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26868 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26869 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26870 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26871 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26872 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26873 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26874 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26875 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26876 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26877 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26878 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26879 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26880 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26881 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26882 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26883 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26884 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26885 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26886 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26887 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26888 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26889 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26890 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26891 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26892 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26893 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26894 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26895 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26896 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26897 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26898 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26899 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26900 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26901 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26902 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26903 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26904 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26905 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26906 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26907 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26908 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26909 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26910 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26911 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26912 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26913 definitions accordingly.
26915 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26916 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26919 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26922 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26925 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26926 hard regs for inner regno.
26928 2017-01-17 Martin Liska <mliska@suse.cz>
26931 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26932 assumption and add comment.
26934 2017-01-17 Nathan Sidwell <nathan@acm.org>
26936 * ipa-visibility.c (localize_node): New function, broken out of ...
26937 (function_and_variable_visibility): ... here. Call it.
26939 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26941 PR middle-end/77445
26942 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26943 correctly set frequency of oudgoing edge.
26944 (duplicate_thread_path): Fix profile updating.
26946 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26949 * configure.ac: Add GCC_BASE_VER.
26950 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26951 version from BASE-VER file.
26952 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26953 (gcc.o): Depend on $(BASEVER).
26954 * common.opt (dumpfullversion): New option.
26955 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26956 * doc/invoke.texi: Document -dumpfullversion.
26957 * doc/install.texi: Document --with-gcc-major-version-only.
26958 * configure: Regenerated.
26960 2017-01-17 Richard Biener <rguenther@suse.de>
26962 PR tree-optimization/71433
26963 * tree-vrp.c (register_new_assert_for): Merge same asserts
26964 on all incoming edges.
26965 (process_assert_insertions_for): Handle insertions at the
26968 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26970 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26971 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26973 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26976 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26979 2017-01-17 Alan Modra <amodra@gmail.com>
26982 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26983 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26984 symbolic stack limit when pic.
26986 2017-01-16 Martin Sebor <msebor@redhat.com>
26988 PR tree-optimization/78608
26989 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26991 2017-01-16 Jeff Law <law@redhat.com>
26994 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26995 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26996 for several include directories that may be relative to sysroot.
26997 * config/i386/x-mingw32 (gplus_includedir): Define.
26998 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26999 (native_system_includedir): Likewise.
27000 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27001 override if TARGET_SYSTEM_ROOT is defined.
27002 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27004 PR tree-optimization/79090
27005 PR tree-optimization/33562
27006 PR tree-optimization/61912
27007 PR tree-optimization/77485
27008 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27009 and computed trims into the dump file.
27011 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27013 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27015 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27018 * gimplify.c (gimplify_init_constructor): If want_value and
27019 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27023 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27024 sequence. Formatting fixes.
27025 (doloop_optimize): Formatting fixes.
27028 * gcc.c (debug_level_greater_than_spec_func): New function.
27029 (static_spec_functions): Add debug-level-gt spec function.
27030 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27032 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27033 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27034 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27035 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27036 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27037 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27039 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27041 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27042 QImode fixups to general and mask registers only.
27044 2017-01-16 Carl Love <cel@us.ibm.com>
27046 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27047 for built-in functions
27048 vector signed char vec_nabs (vector signed char)
27049 vector signed short vec_nabs (vector signed short)
27050 vector signed int vec_nabs (vector signed int)
27051 vector signed long long vec_nabs (vector signed long long)
27052 vector float vec_nabs (vector float)
27053 vector double vec_nabs (vector double)
27054 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27056 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27057 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27058 * doc/extend.texi: Update the documentation file for the new built-in
27061 2017-01-16 Martin Sebor <msebor@redhat.com>
27063 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27066 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27068 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27069 UNSPEC_VSX__XXSPLTD to require special splat handling.
27071 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27074 * system.h: Poison strndup.
27076 2017-01-16 Alan Modra <amodra@gmail.com>
27079 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27082 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27084 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27086 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27088 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27089 call recog here. Assert that INSN_CODE (insn) is non-negative.
27091 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27094 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27096 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27097 in the currently scheduled RTL fragment.
27099 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27101 PR rtl-optimization/78751
27102 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27105 2017-01-14 Jeff Law <law@redhat.com>
27107 PR tree-optimization/79090
27108 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27109 variable length stores.
27110 (compute_trims): Delete dead assignment to *trim_tail.
27111 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27114 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27116 PR rtl-optimization/78626
27117 PR rtl-optimization/78727
27118 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27119 of a block, and split such blocks after everything else is finished.
27121 2017-01-14 Alan Modra <amodra@gmail.com>
27124 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27125 target legitimate_combined_insn.
27126 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27127 (rs6000_legitimate_combined_insn): New function.
27128 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27130 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27131 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27132 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27134 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27136 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27138 2017-01-13 Jeff Law <law@redhat.com>
27140 PR tree-optimization/33562
27141 PR tree-optimization/61912
27142 PR tree-optimization/77485
27143 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27145 (delete_dead_assignment): Likewise.
27146 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27147 statement to delete_dead_call and delete_dead_assignment.
27149 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27152 * substring-locations.c (format_warning_va): Strengthen case 1 so
27153 that both endpoints of the substring must be within the format
27154 range for just the substring to be printed.
27156 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27158 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27159 * config/i386/i386.c (ix86_target_string): Add missing options
27160 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27161 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27162 flags_other and ix86_target_other to flags2_other. Display unknown
27164 (ix86_valid_target_attribute_inner_p): Add missing options and
27165 reorder options by implied ISAs, as in ix86_target_string.
27167 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27169 * hash-table.h (hash_table::too_empty_p): New function.
27170 (hash_table::expand): Use it.
27171 (hash_table::traverse): Likewise.
27172 (hash_table::empty_slot): Use sizeof (value_type) instead of
27173 sizeof (PTR) to convert bytes to elements. Shrink the table
27174 if the current size is excessive for the current number of
27177 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27179 * ira-costs.c (record_reg_classes): Break from the inner loop
27180 early once alt_fail is known to be true. Update outer loop
27181 handling accordingly.
27183 2017-01-13 Jeff Law <law@redhat.com>
27185 * tree-ssa-dse.c (decrement_count): New function.
27186 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27187 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27188 when we know the partially dead statement is a mem* function.
27190 PR tree-optimization/61912
27191 PR tree-optimization/77485
27192 * tree-ssa-dse.c: Include expr.h.
27193 (maybe_trim_constructor_store): New function.
27194 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27196 PR tree-optimization/33562
27197 PR tree-optimization/61912
27198 PR tree-optimization/77485
27199 * doc/invoke.texi: Document new dse-max-object-size param.
27200 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27201 * tree-ssa-dse.c: Include params.h.
27202 (dse_store_status): New enum.
27203 (initialize_ao_ref_for_dse): New, partially extracted from
27205 (valid_ao_ref_for_dse, normalize_ref): New.
27206 (setup_live_bytes_from_ref, compute_trims): Likewise.
27207 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27208 (maybe_trim_partially_dead_store): Likewise.
27209 (maybe_trim_complex_store): Likewise.
27210 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27211 Track what bytes live from the original store. Return tri-state
27212 for dead, partially dead or live.
27213 (dse_dom_walker): Add constructor, destructor and new private members.
27214 (delete_dead_call, delete_dead_assignment): New extracted from
27216 (dse_optimize_stmt): Make a member of dse_dom_walker.
27217 Use initialize_ao_ref_for_dse.
27219 PR tree-optimization/33562
27220 PR tree-optimization/61912
27221 PR tree-optimization/77485
27222 * sbitmap.h (bitmap_count_bits): Prototype.
27223 (bitmap_clear_range, bitmap_set_range): Likewise.
27224 * sbitmap.c (bitmap_clear_range): New function.
27225 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27227 2017-01-13 Martin Liska <mliska@suse.cz>
27230 * function.c (set_cfun): Add new argument force.
27231 * function.h (set_cfun): Likewise.
27232 * ipa-inline-transform.c (inline_call): Use the function when
27233 strict alising from is dropped for function we inline to.
27235 2017-01-13 Richard Biener <rguenther@suse.de>
27237 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27238 for dumping GIMPLE INTEGER_CSTs.
27240 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27242 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27243 to 201112L since C++17.
27245 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27248 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27249 if -fsanitize=kernel-address is present.
27251 2017-01-13 Richard Biener <rguenther@suse.de>
27253 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27254 as _Literal ( type ) number in case usual suffixes do not
27255 preserve all information.
27257 2017-01-13 Richard Biener <rguenther@suse.de>
27259 PR tree-optimization/77283
27260 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27261 and ssa-iterators.h.
27262 (is_feasible_trace): Implement a cost model based on joiner
27265 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27268 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27269 char or short to __float128/_Float128 directly.
27271 2017-01-12 Martin Sebor <msebor@redhat.com>
27273 to -Wformat-overflow.
27274 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27275 (min_bytes_remaining): Same.
27276 (get_string_length): Same.
27277 (format_string): Same.
27278 (format_directive): Same.
27280 (pass_sprintf_length::handle_gimple_call): Same.
27282 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27284 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27285 info.nowrite calls with no lhs that can't throw. Return bool
27286 whether gsi_remove has been called or not.
27287 (pass_sprintf_length::handle_gimple_call): Return bool whether
27288 try_substitute_return_value called gsi_remove. Formatting fix.
27289 (pass_sprintf_length::execute): Don't use gsi_remove if
27290 handle_gimple_call returned true.
27293 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27294 be removed due to side-effects, don't remove following barrier nor
27295 turn the successor edge into fallthru edge.
27297 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27300 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27301 element-reversing loads and stores as not swappable.
27303 2017-01-12 Nathan Sidwell <nathan@acm.org>
27304 Nicolai Stange <nicstange@gmail.com>
27306 * combine.c (try_combine): Don't ignore result of overlap checking
27307 loop. Combine overlap & asm check into single loop.
27309 2017-01-12 Richard Biener <rguenther@suse.de>
27311 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27312 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27314 2017-01-12 Richard Biener <rguenther@suse.de>
27316 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27317 and TS_TARGET_OPTION directly derive from TS_BASE.
27318 * tree-core.h (tree_optimization_option): Derive from tree_base.
27319 (tree_target_option): Likewise.
27321 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27323 * config/i386/i386.c (memory_address_length): Increase len
27324 only when rip_relative_addr_p returns false.
27326 2017-01-11 Julia Koval <julia.koval@intel.com>
27328 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27329 (OPTION_MASK_ISA_SGX_SET): New.
27330 (ix86_handle_option): Handle OPT_msgx.
27331 * config.gcc: Added sgxintrin.h.
27332 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27333 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27334 * config/i386/i386.c (ix86_target_string): Add -msgx.
27336 (ix86_option_override_internal): Handle new options.
27337 (ix86_valid_target_attribute_inner_p): Add sgx.
27338 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27339 * config/i386/i386.opt: Add msgx.
27340 * config/i386/sgxintrin.h: New file.
27341 * config/i386/x86intrin.h: Add sgxintrin.h.
27343 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27346 * fold-const.c (maybe_nonzero_address): Return 1 for function
27348 (tree_single_nonzero_warnv_p): Don't handle function local objects
27352 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27355 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27358 * opts.c: Include "spellcheck.h"
27359 (struct string_fragment): New struct.
27360 (struct edit_distance_traits<const string_fragment &>): New
27362 (get_closest_sanitizer_option): New function.
27363 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27365 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27367 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27369 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27370 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27371 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27372 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27373 for initial die_offset if dwarf_split_debug_info.
27374 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27375 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27377 (output_skeleton_debug_sections): Formatting fix. Use
27378 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27379 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27381 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27383 * config/arm/cortex-a53.md: Add bypasses for
27384 cortex_a53_r2f_cvt.
27385 (cortex_a53_r2f): Only use for transfers.
27386 (cortex_a53_f2r): Likewise.
27387 (cortex_a53_r2f_cvt): Add reservation for conversions.
27388 (cortex_a53_f2r_cvt): Likewise.
27390 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27392 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27393 to all inlined functions, change static to extern.
27395 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27398 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27400 (arm_assemble_integer): Likewise.
27402 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27404 * config.gcc: Use new awk script to check CPU, FPU and architecture
27405 parameters for --with-... options.
27406 * config/arm/parsecpu.awk: New file
27407 * config/arm/arm-cpus.in: New file.
27408 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27410 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27412 * config/arm/t-arm: Update dependency rules.
27413 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27414 of processing .def files.
27415 * config/arm/genopt.sh: Deleted.
27416 * config/arm/gentune.sh: Deleted.
27417 * config/arm/arm-cores.def: Deleted.
27418 * config/arm/arm-arches.def: Deleted.
27419 * config/arm/arm-fpus.def: Deleted.
27420 * config/arm/arm-tune.md: Regenerated.
27421 * config/arm/arm-tables.opt: Regenerated.
27422 * config/arm/arm-cpu.h: New generated file.
27423 * config/arm/arm-cpu-data.h: New generated file.
27424 * config/arm/arm-cpu-cdata.h: New generated file.
27426 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27429 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27431 (input_varpool_node): Unpack dynamically_initialized bit.
27433 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27435 PR rtl-optimization/79032
27436 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27437 the alignment of the adjusted memory reference against that of MODE,
27438 instead of the alignment of the original memory reference.
27440 2017-01-11 Martin Jambor <mjambor@suse.cz>
27442 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27444 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27445 decorated functions.
27447 2017-01-11 Richard Biener <rguenther@suse.de>
27449 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27450 set range/nonnull info for PHI results. Do not set it on
27451 stmts marked for removal.
27453 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27455 * expr.c (store_field): In the bitfield case, fetch the return value
27456 from the registers before applying a single big-endian adjustment.
27457 Always do a final load for a BLKmode value not larger than a word.
27459 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27462 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27463 that we correctly handle column numbers greater than
27464 LINE_MAP_MAX_COLUMN_NUMBER.
27466 2017-01-10 Martin Sebor <msebor@redhat.com>
27468 PR middle-end/78245
27469 * gimple-ssa-sprintf.c (get_destination_size): Call
27470 {init,fini}object_sizes.
27471 * tree-object-size.c (addr_object_size): Adjust.
27472 (pass_through_call): Adjust.
27473 (pass_object_sizes::execute): Adjust.
27474 * tree-object-size.h (fini_object_sizes): Declare.
27476 2017-01-10 Martin Sebor <msebor@redhat.com>
27478 PR tree-optimization/78775
27479 * builtins.c (get_size_range): Move...
27480 * calls.c: ...to here.
27481 (alloc_max_size): Accept zero argument.
27482 (operand_signed_p): Remove.
27483 (maybe_warn_alloc_args_overflow): Call get_size_range.
27484 * calls.h (get_size_range): Declare.
27486 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27488 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27489 from TI's devices.csv file as of September 2016.
27490 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27492 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27494 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27495 * doc/invoke.texi: Likewise.
27496 * doc/md.texi: Likewise.
27497 * doc/objc.texi: Likewise.
27499 2017-01-10 Joshua Conner <joshconner@google.com>
27501 * config/arm/fuchsia-elf.h: New file.
27502 * config/fuchsia.h: New file.
27503 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27504 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27506 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27508 2016-01-10 Richard Biener <rguenther@suse.de>
27510 PR tree-optimization/79034
27511 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27512 Propagate out degenerate PHIs in the joiner.
27514 2017-01-10 Martin Liska <mliska@suse.cz>
27516 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27517 (sort_congruence_classes_by_decl_uid): Likewise.
27518 (sort_congruence_class_groups_by_decl_uid): Likewise.
27519 (sem_item_optimizer::merge_classes): Sort class, groups in these
27520 classes and members in the groups by DECL_UID of declarations.
27521 This would make merge operations stable.
27523 2017-01-10 Martin Liska <mliska@suse.cz>
27525 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27526 usage of m_classes_vec.
27527 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27528 (sem_item_optimizer::get_group_by_hash): Likewise.
27529 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27530 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27531 (sem_item_optimizer::verify_classes): Likewise.
27532 (sem_item_optimizer::process_cong_reduction): Likewise.
27533 (sem_item_optimizer::dump_cong_classes): Likewise.
27534 (sem_item_optimizer::merge_classes): Likewise.
27535 * ipa-icf.h (congruence_class_hash): Rename from
27536 congruence_class_group_hash. Remove declaration of m_classes_vec.
27538 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27540 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27541 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27542 * config.gcc: Add avx512vpopcntdqintrin.h.
27543 * config/i386/avx512vpopcntdqintrin.h: New.
27544 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27545 * config/i386/i386-builtin-types.def: Add new types.
27546 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27547 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27548 __builtin_ia32_vpopcountq_v8di_mask): New.
27549 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27550 __AVX512VPOPCNTDQ__.
27551 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27552 (PTA_AVX512VPOPCNTDQ): Define.
27553 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27554 TARGET_AVX512VPOPCNTDQ_P): Define.
27555 * config/i386/i386.opt: Add mavx512vpopcntdq.
27556 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27557 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27559 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27561 PR middle-end/77484
27562 * predict.def (PRED_CALL): Set to 67.
27564 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27566 * expr.c (store_field): In the bitfield case, if the value comes from
27567 a function call and is of an aggregate type returned in registers, do
27568 not modify the field mode; extract the value in all cases if the mode
27569 is BLKmode and the size is not larger than a word.
27571 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27574 * config/i386/cpuid.h: Fix undefined behavior.
27576 2017-01-04 Jeff Law <law@redhat.com>
27578 PR tree-optimization/79007
27579 PR tree-optimization/67955
27580 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27581 conservative for pt.null when flag_non_call_exceptions is on.
27583 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27585 PR translation/79019
27586 PR translation/79020
27587 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27588 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27589 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27591 * config/avr/avr.opt (maccumulate-args): Likewise.
27592 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27593 * common.opt (freport-bug): Likewise.
27594 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27595 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27596 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27597 translatable string.
27598 * config/i386/i386.c (function_value_32): Likewise.
27599 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27600 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27602 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27603 * common/config/msp430/msp430-common.c (msp430_handle_option):
27605 * symtab.c (symtab_node::verify_base): Likewise.
27606 * opts.c (set_debug_level): Likewise.
27607 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27608 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27609 missing whitespace to translatable strings.
27610 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27611 * config/sh/superh.h: Likewise.
27612 * config/i386/xopintrin.h: Likewise.
27613 * config/i386/znver1.md: Likewise.
27614 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27615 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27616 * double-int.h (struct double_int): Likewise.
27617 * double-int.c (div_and_round_double): Likewise.
27618 * wide-int.cc: Likewise.
27619 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27620 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27621 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27622 (crossjumps_occurred): ... this.
27623 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27626 PR tree-optimization/78899
27627 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27628 returning bool return struct loop *, NULL for failure and the new
27630 (versionable_outer_loop_p): Don't version outer loop if it has
27631 dont_vectorized bit set.
27632 (tree_if_conversion): When versioning outer loop, ensure
27633 tree_if_conversion is performed also on the inner loop of the
27634 non-vectorizable outer loop copy.
27635 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27636 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27637 prevent vectorization of it.
27638 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27639 the outer loop vectorization of the non-scalar version is attempted
27640 before vectorization of the inner loop in scalar version. If
27641 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27642 vectorization of its inner loop.
27643 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27644 has 2 inner loops, rename also on edges from bb whose single pred
27645 is outer_loop->header. Fix typo in function comment.
27647 2017-01-09 Martin Sebor <msebor@redhat.com>
27650 * asan.c (asan_emit_stack_protection): Increase local buffer size
27651 to avoid snprintf truncation warning.
27653 2017-01-09 Andrew Pinski <apinski@cavium.com>
27655 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27656 to reference thunderx2t99 for the tuning structure
27657 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27659 (thunderx2t99_extra_costs): This.
27660 * config/aarch64/aarch64-tune.md: Regenerate.
27661 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27662 (vulcan_addrcost_table): This.
27663 (vulcan_regmove_cost): Rename to ...
27664 (thunderx2t99_regmove_cost): This.
27665 (vulcan_vector_cost): Rename to ...
27666 (thunderx2t99_vector_cost): this.
27667 (vulcan_branch_cost): Rename to ...
27668 (thunderx2t99_branch_cost): This.
27669 (vulcan_tunings): Rename to ...
27670 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27671 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27673 2017-01-09 Martin Jambor <mjambor@suse.cz>
27677 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27678 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27679 (propagate_vr_accross_jump_function): Use the above function for all
27680 value range computations for pass-through jump functions and type
27681 converasion from explicit value range values.
27682 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27683 parameters from TYPE_ARG_TYPES.
27684 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27685 (ipa_write_node_info): Stream type of the actual argument.
27686 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27688 2017-01-09 Martin Liska <mliska@suse.cz>
27691 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27692 (lookup_compiler): Do not show error message with have_E.
27694 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27696 PR tree-optimization/78938
27697 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27698 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27699 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27700 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27703 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27705 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27708 2017-01-09 Richard Biener <rguenther@suse.de>
27710 PR tree-optimization/78997
27711 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27712 name condition properly.
27714 2017-01-09 Richard Biener <rguenther@suse.de>
27717 * dwarf2out.c (is_cxx): New overload with context.
27718 (is_naming_typedef_decl): Use it.
27720 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27722 * invoke.texi (Option Summary): Correct spacing in option lists
27723 and add line breaks to fix over-long lines.
27725 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27727 PR middle-end/17660
27729 * extend.texi (Common Variable Attributes): Add xref to GCC
27730 Internals manual to explain mode attribute keywords.
27732 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27735 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27736 and Preprocessor Options.
27737 (Options for Linking): Document -pthread here....
27738 (RS/6000 and PowerPC Options): ...not here.
27739 (Solaris 2 Options): ...or here.
27740 * doc/cppopts.texi: Document -pthread.
27742 2017-01-08 Martin Sebor <msebor@redhat.com>
27744 PR middle-end/77708
27745 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27746 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27747 New member functions.
27748 (format_directive): Used them.
27750 (pass_sprintf_length::handle_gimple_call): Same.
27751 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27752 to avoid truncation for any argument.
27753 (extract_affine_mul): Same.
27754 * tree.c (get_file_function_name): Same.
27756 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27758 PR middle-end/77484
27759 * predict.def (PRED_INDIR_CALL): Set to 86.
27761 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27763 PR preprocessor/54124
27764 * doc/cppopts.texi: Reformat -d subtable to list the full name
27765 of the options. Add cross-reference to the docs for the general
27766 compiler -d options.
27767 * doc/invoke.texi (Developer Options): Add cross-reference to the
27768 preprocessor-specific -d option documentation.
27770 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27772 PR preprocessor/13498
27773 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27774 redudant material, and reflect new command-line options.
27775 (System Headers): Likewise.
27777 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27779 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27780 -isystem, and -idirafter. Copy-edit.
27781 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27782 default for -ftrack-macro-expansion. Delete obsolete and
27783 badly-formatted implementation details about -fdebug-cpp output.
27784 * doc/cppwarnopts.texi: Copy-edit.
27786 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27789 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27790 that the transition from a max line width >= 1<<10 to narrower
27791 lines works correctly.
27793 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
27795 * doc/options.texi (PerFunction): New.
27796 * opt-functions.awk (switch_flags): Map both Optimization and
27797 PerFunction to CL_OPTIMIZATION.
27798 * opth-gen.awk: Test for PerFunction flag along with
27800 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
27801 it only when the latter is present. Skip those that don't in
27802 the hash function generator.
27803 * common.opt (fvar-tracking): Mark as PerFunction instead of
27805 (fvar-tracking-assignments): Likewise.
27806 (fvar-tracking-assignments-toggle): Likewise.
27807 (fvar-tracking-uninit): Likewise.
27809 2017-01-07 Jakub Jelinek <jakub@redhat.com>
27811 PR translation/79018
27812 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
27815 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27818 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27819 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27820 TARGET_LONG_JUMP_TABLE_OFFSETS.
27821 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27822 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27823 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27824 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27825 * config/m68k/m68k.md (tablejump expander): Likewise.
27826 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27827 TARGET_LONG_JUMP_TABLE_OFFSETS.
27828 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27829 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27831 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27832 David Holsgrove <david.holsgrove@xilinx.com>
27834 * common/config/microblaze/microblaze-common.c
27835 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27836 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27838 * config/microblaze/microblaze.c (microblaze_must_save_register)
27839 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27841 (microblaze_eh_return): New function.
27842 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27843 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27844 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27845 * config/microblaze/microblaze.md (eh_return): New pattern.
27847 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27849 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27850 GCC_DIAGNOSTIC_STRINGIFY): Define.
27852 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27854 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27856 * config/arm/arm.md (<mcrr>): New.
27858 * config/arm/arm.c (arm_arch5te): New.
27859 (arm_option_override): Set arm_arch5te.
27860 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27862 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27863 (arm_mcrr_qualifiers): ... this. New.
27864 (MRRC_QUALIFIERS): Define to...
27865 (arm_mrrc_qualifiers): ... this. New.
27866 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27868 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27869 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27870 (MRRCI, mrrc, MRRC): New.
27871 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27872 VUNSPEC_MRRC2): New.
27874 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27876 * config/arm/arm.md (<mcr>): New.
27878 * config/arm/arm.c (arm_coproc_builtin_available): Add
27879 support for mcr, mrc, mcr2 and mrc2.
27880 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27881 (arm_mcr_qualifiers): ... this. New.
27882 (MRC_QUALIFIERS): Define to ...
27883 (arm_mrc_qualifiers): ... this. New.
27884 (MCR_QUALIFIERS): Define to ...
27885 (arm_mcr_qualifiers): ... this. New.
27886 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27888 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27889 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27890 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27891 VUNSPEC_MRC2): New.
27893 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27895 * config/arm/arm.md (*ldc): New.
27899 * config/arm/arm.c (arm_coproc_builtin_available): Add
27900 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27901 (arm_coproc_ldc_stc_legitimate_address): New.
27902 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27903 'qualifier_const_pointer'.
27904 (LDC_QUALIFIERS): Define to...
27905 (arm_ldc_qualifiers): ... this. New.
27906 (STC_QUALIFIERS): Define to...
27907 (arm_stc_qualifiers): ... this. New.
27908 * config/arm/arm-protos.h
27909 (arm_coproc_ldc_stc_legitimate_address): New.
27910 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27911 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27912 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27913 stc2, stcl, stc2l): New.
27914 * config/arm/constraints.md (Uz): New.
27915 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27916 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27917 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27918 VUNSPEC_STC2L): New.
27920 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27922 * config/arm/arm.md (<cdp>): New.
27923 * config/arm/arm.c (neon_const_bounds): Rename this ...
27924 (arm_const_bounds): ... this.
27925 (arm_coproc_builtin_available): New.
27926 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27927 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27928 (CDP_QUALIFIERS): Define to...
27929 (arm_cdp_qualifiers): ... this. New.
27931 (arm_expand_builtin_args): Add case for 6 arguments.
27932 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27933 (arm_const_bounds): ... this.
27934 (arm_coproc_builtin_available): New.
27935 * config/arm/arm_acle.h (__arm_cdp): New.
27937 * config/arm/arm_acle_builtins.def (cdp): New.
27939 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27940 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27941 'arm_const_bounds'.
27942 * config/arm/types.md (coproc): New.
27943 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27944 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27945 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27946 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27948 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27950 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27951 (UBINOP_QUALIFIERS): New.
27953 (acle_builtin_data): New. Change comment.
27954 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27955 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27956 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27957 arm_acle_builtins.def.
27958 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27959 (arm_init_acle_builtins): New.
27960 (CRC32_BUILTIN): Remove.
27961 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27962 crc32cb, crc32ch and crc32cw.
27963 (arm_init_crc32_builtins): Remove.
27964 (arm_init_builtins): Use arm_init_acle_builtins rather
27965 than arm_init_crc32_builtins.
27966 (arm_expand_acle_builtin): New.
27967 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27968 * config/arm/arm_acle_builtins.def: New.
27970 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27972 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27973 (arm_builtin_datum): ... this.
27974 (arm_init_neon_builtin): Rename to ...
27975 (arm_init_builtin): ... this. Add a new parameters PREFIX
27976 and USE_SIG_IN_NAME.
27977 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27978 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27979 'arm_builtin_datum'.
27980 (arm_init_vfp_builtins): Likewise.
27981 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27982 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27983 (arm_expand_neon_args): Rename to ...
27984 (arm_expand_builtin_args): ... this. Rename builtin_arg
27985 enum values and differentiate between ARG_BUILTIN_MEMORY
27986 and ARG_BUILTIN_NEON_MEMORY.
27987 (arm_expand_neon_builtin_1): Rename to ...
27988 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27989 values, arm_expand_builtin_args and add bool parameter NEON.
27990 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27991 (arm_expand_vfp_builtin): Likewise.
27992 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27994 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27996 PR middle-end/77484
27997 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27998 * predict.c (tree_estimate_probability_bb): Reverse direction of
27999 polymorphic call predictor.
28001 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28003 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28004 (determine_pass_name_match): ...this new function and...
28005 (should_skip_pass_p): ...this new function.
28007 2017-01-06 Nathan Sidwell <nathan@acm.org>
28009 * ipa-visibility.c (function_and_variable_visibility): Reformat
28010 comments and long lines. Remove extrneous if.
28011 * symtab.c (symtab_node::make_decl_local): Fix code format.
28012 (symtab_node::set_section_for_node): Fix comment typo.
28014 2017-01-06 Martin Liska <mliska@suse.cz>
28017 * lra-constraints.c: Rename invariant to lra_invariant.
28018 * predict.c (set_even_probabilities): Initialize e to NULL.
28020 2017-01-05 Martin Sebor <msebor@redhat.com>
28022 PR tree-optimization/78910
28023 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28024 (format_integer): Correct off-by-one error in the handling
28025 of precision with negative numbers in signed conversions..
28027 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28029 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28031 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28033 PR tree-optimization/71016
28034 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28035 factor_out_conditional_conversion. Formatting fix.
28036 (factor_out_conditional_conversion): Add cond_stmt argument.
28037 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28038 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28041 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28043 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28044 read-rtl-function.o, and selftest-rtl.o.
28045 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28046 (selftest::aarch64_test_loading_full_dump): New function.
28047 (selftest::aarch64_run_selftests): New function.
28048 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28049 selftest::aarch64_run_selftests.
28050 * config/i386/i386.c
28051 (selftest::ix86_test_loading_dump_fragment_1): New function.
28052 (selftest::ix86_test_loading_call_insn): New function.
28053 (selftest::ix86_test_loading_full_dump): New function.
28054 (selftest::ix86_test_loading_unspec): New function.
28055 (selftest::ix86_run_selftests): Call the new functions.
28056 * emit-rtl.c (maybe_set_max_label_num): New function.
28057 * emit-rtl.h (maybe_set_max_label_num): New decl.
28058 * function.c (instantiate_decls): Guard call to
28059 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28060 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28062 * gensupport.c (gen_reader::gen_reader): Pass "false"
28063 for new "compact" param of rtx_reader.
28064 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28065 rather than an empty string for NULL strings.
28066 * read-md.c: Potentially include config.h rather than bconfig.h.
28067 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28068 (have_error): New global, copied from errors.c.
28069 (md_reader::read_name): Rename to...
28070 (md_reader::read_name_1): ...this, adding "out_loc" param,
28071 and converting "missing name or number" to returning false, rather
28073 (md_reader::read_name): Reimplement in terms of read_name_1.
28074 (md_reader::read_name_or_nil): New function.
28075 (md_reader::read_string): Handle "(nil)" by returning NULL.
28076 (md_reader::md_reader): Add new param "compact".
28077 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28078 (md_reader::read_file): New method.
28079 * read-md.h (md_reader::md_reader): Add new param "compact".
28080 (md_reader::read_file): New method.
28081 (md_reader::is_compact): New accessor.
28082 (md_reader::read_name): Convert return type from void to file_location.
28083 (md_reader::read_name_or_nil): New decl.
28084 (md_reader::read_name_1): New decl.
28085 (md_reader::m_compact): New field.
28086 (noop_reader::noop_reader): Pass "false" for new "compact" param
28088 (rtx_reader::rtx_reader): Add new "compact" param.
28089 (rtx_reader::read_rtx_operand): Make virtual and convert return
28090 type from void to rtx.
28091 (rtx_reader::read_until): New decl.
28092 (rtx_reader::handle_any_trailing_information): New virtual function.
28093 (rtx_reader::postprocess): New virtual function.
28094 (rtx_reader::finalize_string): New virtual function.
28095 (rtx_reader::m_in_call_function_usage): New field.
28096 (rtx_reader::m_reuse_rtx_by_id): New field.
28097 * read-rtl-function.c: New file.
28098 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28099 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28100 (selftest::verify_three_block_rtl_cfg): New decl.
28101 * read-rtl-function.h: New file.
28102 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28103 For host, include function.h, memmodel.h, and emit-rtl.h.
28104 (one_time_initialization): New function.
28105 (struct compact_insn_name): New struct.
28106 (compact_insn_names): New array.
28107 (find_code): Handle insn codes in compact dumps.
28108 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28109 (bind_subst_iter_and_attr): Likewise.
28110 (add_condition_to_string): Likewise.
28111 (add_condition_to_rtx): Likewise.
28112 (apply_attribute_uses): Likewise.
28113 (add_current_iterators): Likewise.
28114 (apply_iterators): Likewise.
28115 (initialize_iterators): Guard usage of apply_subst_iterator with
28116 #ifdef GENERATOR_FILE.
28117 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28118 (md_reader::read_mapping): Likewise.
28119 (add_define_attr_for_define_subst): Likewise.
28120 (add_define_subst_attr): Likewise.
28121 (read_subst_mapping): Likewise.
28122 (check_code_iterator): Likewise.
28123 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28125 (one_time_initialization): New function.
28126 (rtx_reader::read_until): New method.
28127 (read_flags): New function.
28128 (parse_reg_note_name): New function.
28129 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28130 Handle reuse_rtx ids.
28131 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28132 Add parsing support for RTL dumps, mirroring the special-cases in
28133 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28134 values, and calling handle_any_trailing_information.
28135 (rtx_reader::read_rtx_operand): Convert return type from void
28136 to rtx, returning return_rtx. Handle case 'e'. Call
28137 finalize_string on XSTR and XTMPL fields.
28138 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28139 "(nil)" values were omitted. Call the postprocess vfunc on the
28141 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28142 class ctor. Initialize m_in_call_function_usage. Call
28143 one_time_initialization.
28144 * rtl-tests.c (selftest::test_uncond_jump): Call
28145 set_new_first_and_last_insn.
28146 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28147 * selftest-rtl.c: New file.
28148 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28149 (selftest::get_insn_by_uid): New decl.
28150 * selftest-run-tests.c (selftest::run_tests): Call
28151 read_rtl_function_c_tests.
28152 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28153 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28156 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28158 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28159 operands in a special way. Assert that pos+len <= mode precision.
28161 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28163 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28164 3 argument Alias with unlimited for the negative form.
28165 (fno-vect-cost-model): Removed.
28167 2017-01-05 Martin Liska <mliska@suse.cz>
28169 * hsa-gen.c (gen_hsa_divmod): New function.
28170 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28172 2017-01-05 Martin Liska <mliska@suse.cz>
28175 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28178 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28180 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28181 small constant length operands.
28183 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28185 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28186 between loop iterations.
28188 2017-01-05 Martin Liska <mliska@suse.cz>
28191 * gimplify.c (gimplify_decl_expr): Compare to
28192 asan_poisoned_variables instread of checking flags.
28193 (gimplify_target_expr): Likewise.
28194 (gimplify_expr): Likewise.
28195 (gimplify_function_tree): Conditionally initialize
28196 asan_poisoned_variables.
28198 2017-01-04 Jeff Law <law@redhat.com>
28200 PR tree-optimizatin/78812
28201 * rtl.h (contains_mem_rtx_p): Prototype.
28202 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28203 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28204 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28205 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28206 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28208 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28210 * input.c (assert_char_at_range): Default-initialize actual_range.
28212 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28214 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28215 to match the caller.
28217 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28219 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28220 insns after final jump in test to emit dummy move.
28222 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28224 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28225 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28227 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28229 * multiple_target.c (create_dispatcher_calls): Init e_next.
28230 * tree-ssa-loop-split.c (split_loop): Init border.
28231 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28234 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28239 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28240 (altivec_register_operand): Do not return true if the operand
28241 contains a SUBREG mixing SImode and SFmode.
28242 (vsx_register_operand): Likewise.
28243 (vsx_reg_sfsubreg_ok): New predicate.
28244 (vfloat_operand): Do not return true if the operand contains a
28245 SUBREG mixing SImode and SFmode.
28246 (vint_operand): Likewise.
28247 (vlogical_operand): Likewise.
28248 (gpc_reg_operand): Likewise.
28249 (int_reg_operand): Likewise.
28250 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28251 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28252 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28254 (rs6000_emit_move_si_sf_subreg): New helper function.
28255 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28256 fixup SUBREGs involving SImode and SFmode.
28257 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28258 numbers for the new peephole2 optimization.
28259 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28260 the GLIBC math library that do AND/IOR/XOR operations on single
28261 precision floating point.
28262 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28263 target macros to say whether we need to avoid SUBREGs mixing
28265 (TARGET_ALLOW_SF_SUBREG): Likewise.
28266 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28267 (UNSPEC_SI_FROM_SF): Likewise.
28268 (iorxor): Change spacing.
28269 (and_ior_xor): New iterator for AND, IOR, and XOR.
28270 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28271 (movdi_from_sf_zero_ext): Likewise.
28272 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28273 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28274 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28275 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28276 (fms<mode>4): Likewise.
28277 (fnma<mode>4): Likewise.
28278 (fnms<mode>4): Likewise.
28279 (nfma<mode>4): Likewise.
28280 (nfms<mode>4): Likewise.
28282 2017-01-04 Marek Polacek <polacek@redhat.com>
28285 * doc/invoke.texi: Document -Wpointer-compare.
28287 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28289 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28292 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28293 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28296 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28299 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28300 documentation of the powerpc_popcntb_ok attribute.
28301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28302 code to issue warning messages if a requested CPU configuration is
28303 not supported by the binary (assembler and loader) toolchain.
28304 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28305 made to define a built-in function that has been disabled.
28306 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28307 made to define a built-in function that has been disabled.
28308 (altivec_init_builtins): Add comment explaining why definition
28309 of the DST built-in functions is not preceded by an assertion
28310 check. Add assertions to prevent ICE if attempts are made to
28311 define an altivec predicate or an abs* built-in function that has
28313 (htm_init_builtins): Add comment explaining why definition of the
28314 htm built-in functions is not preceded by an assertion check.
28316 2017-01-04 Jeff Law <law@redhat.com>
28318 PR tree-optimizatin/67955
28319 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28320 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28321 the points-to solution does not include pt_null. Use DECL_PT_UID
28324 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28326 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28327 Use gen_int_mode instead of gen_lopwart for const_int operands.
28329 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28331 PR tree-optimization/71563
28332 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28333 out of range value - has low bits known to be zero.
28335 2017-01-04 Alan Modra <amodra@gmail.com>
28337 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28338 * configure: Regenerate.
28339 * config.in: Regenerate.
28341 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28344 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28345 a substring of the message, but strcmp with the whole message. Ifdef
28346 ENABLE_NLS, translate the message first using dgettext.
28348 2017-01-03 Jeff Law <law@redhat.com>
28350 PR tree-optimizatin/78856
28351 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28352 (mark_threaded_blocks): Remove code to truncate thread paths that
28353 cross multiple loop headers. Instead invalidate the cached loop
28354 iteration information and handle case of a thread path walking
28355 into an irreducible region.
28357 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28360 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28361 assertions. Add support for doing the signbit if the IEEE 128-bit
28362 floating point value is in a GPR.
28363 * config/rs6000/rs6000.md (Fsignbit): Delete.
28364 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28365 Update the length attribute if the value is in a GPR.
28366 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28367 the sign or zero extension instruction, since the value is always 0/1.
28368 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28371 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28372 extracting SImode to a GPR register so that we can generate a
28373 store, limit the vector to be in a traditional Altivec register
28374 for the vextuwrx instruction.
28376 2017-01-03 Ian Lance Taylor <iant@google.com>
28378 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28380 2017-01-03 Martin Sebor <msebor@redhat.com>
28382 PR tree-optimization/78696
28383 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28384 precision. Use MPFR for %f for greater fidelity. Correct handling
28386 (pass_sprintf_length::compute_format_length): Set width and precision
28387 specified by asrerisk to void_node for vararg functions.
28388 (try_substitute_return_value): Adjust dump output.
28390 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28392 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28394 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28396 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28397 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28398 -mlra/-mno-lra was passed to the compiler.
28400 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28402 PR rtl-optimization/65618
28403 * emit-rtl.c (try_split): Move initialization of "before" and
28404 "after" to just before the call to emit_insn_after_setloc.
28406 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28408 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28410 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28412 * dwarf2out.c (gen_enumeration_type_die): When
28413 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28415 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28417 PR tree-optimization/78965
28418 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28419 Change first argument from const call_info & to call_info &. For %n
28420 set info.nowrite to false.
28422 PR middle-end/78901
28423 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28424 possibly throwing calls.
28426 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28427 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28428 and fns handling, rather than in a separate case SSA_NAME.
28430 2017-01-02 Jeff Law <law@redhat.com>
28432 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28433 fixes for first_period and second_period variables.
28435 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28438 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28439 (*insvqi_1): New insn pattern.
28440 (*insvqi_1_mem_rex64): Ditto.
28441 (*insvqi_2): Ditto.
28442 (*insvqi_3): Rename from *insvqi.
28444 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28446 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28448 * doc/cfg.texi (Edges): Remove reference to Java.
28449 (Maintaining the CFG): Ditto.
28451 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28453 PR middle-end/77674
28454 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28455 transparent aliases.
28457 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28459 PR middle-end/77484
28460 * predict.def (PRED_CALL): Update hitrate.
28461 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28462 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28463 into direct/indirect/polymorphic variants.
28465 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28467 Update copyright years.
28469 * gcc.c (process_command): Update copyright notice dates.
28470 * gcov-dump.c (print_version): Ditto.
28471 * gcov.c (print_version): Ditto.
28472 * gcov-tool.c (print_version): Ditto.
28473 * gengtype.c (create_file): Ditto.
28474 * doc/cpp.texi: Bump @copying's copyright year.
28475 * doc/cppinternals.texi: Ditto.
28476 * doc/gcc.texi: Ditto.
28477 * doc/gccint.texi: Ditto.
28478 * doc/gcov.texi: Ditto.
28479 * doc/install.texi: Ditto.
28480 * doc/invoke.texi: Ditto.
28482 Copyright (C) 2017 Free Software Foundation, Inc.
28484 Copying and distribution of this file, with or without modification,
28485 are permitted in any medium without royalty provided the copyright
28486 notice and this notice are preserved.