1 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3 PR rtl-optimization/77729
4 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5 to (X&(C1&~C2))|C2 transformations.
7 2017-10-03 Martin Jambor <mjambor@suse.cz>
9 PR tree-optimization/82363
10 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11 mismatch, mark lacc written regardless of racc.
13 2017-10-03 Jakub Jelinek <jakub@redhat.com>
15 PR tree-optimization/82381
16 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
17 stmt_to_insert nor wheather SSA_NAMEs are default defs.
18 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
19 fallthrough into reassoc_stmt_dominates_stmt_p.
22 * combine.c (combine_instructions): Don't combine in unreachable
25 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
28 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
29 function to not use the have_cpu variable. Do not set cpu_index,
30 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
32 (rs6000_valid_attribute_p): Remove duplicate initializations of
33 old_optimize and func_optimize.
34 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
35 (rs6000_activate_target_options): Make global.
36 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
39 2017-10-02 Jakub Jelinek <jakub@redhat.com>
41 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
42 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
43 Set *poffset to 0 and *psize and *pmax_size to -1 if
44 *poffset + *psize overflows in HOST_WIDE_INT.
46 PR tree-optimization/82387
47 PR tree-optimization/82388
48 PR tree-optimization/82389
49 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
50 instead of live_bytes non-NULL.
52 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
55 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
58 2017-10-02 Richard Biener <rguenther@suse.de>
60 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
61 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
62 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
64 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
66 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
67 requested precision matches the type's.
68 * calls.c (alloc_max_size): Calculate the new candidate size as
69 a widest_int and use wi::to_widest when comparing it with the
70 current candidate size.
71 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
72 zero rather than integer_zero_node.
73 * match.pd: Check for a no-op conversion before using wi::add
74 rather than after. Use tree_to_uhwi when summing small shift
75 counts into an unsigned int.
77 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
78 Alan Hayward <alan.hayward@arm.com>
79 David Sherwood <david.sherwood@arm.com>
82 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
83 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
84 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
87 2017-10-02 Richard Biener <rguenther@suse.de>
89 PR tree-optimization/82355
90 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
91 a mapping for the enclosing loop but avoid generating one for
93 (copy_bb_and_scalar_dependences): Remove premature codegen
94 error on PHIs in blocks duplicated into multiple places.
95 * graphite-scop-detection.c
96 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
97 in the region use it as loop and nest to analyze the DR in.
98 (try_generate_gimple_bb): Likewise.
99 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
100 (add_loop_constraints): For blocks in a loop not in the region
101 create a dimension with a single iteration.
102 * sese.h (gbb_loop_at_index): Remove assert.
104 2017-10-01 Kevin Buettner <kevinb@redhat.com>
106 * omp-expand.c (adjust_context_scope): New function.
107 (expand_parallel_call): Call adjust_context_scope.
109 2017-10-01 Jeff Law <law@redhat.com>
111 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
112 dom_opt_dom_walker class with direct access to private members.
113 Add comments. Call test_for_singularity.
114 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
115 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
116 m_dummy_cond anymore.
117 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
119 (pass_dominator:execute): Build the dummy_cond here and pass it
120 to the dom_opt_dom_walker ctor.
121 (test_for_singularity): New function.
123 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
124 Maya Rashish <coypu@sdf.org>
126 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
127 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
128 (alpha*-*-netbsd*) Use nbsd_tm_file.
129 (arm*-*-netbsdelf*) Likewise.
130 (i[34567]86-*-netbsdelf*) Likewise.
131 (x86_64-*-netbsd*) Likewise.
132 (mips*-*-netbsd*) Likewise.
133 (powerpc-*-netbsd*) Likewise.
134 (sh*-*-netbsd*) Likewise.
135 (sparc-*-netbsdelf*) Likewise.
136 (sparc64-*-netbsd*) Likewise.
137 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
139 (vax-*-netbsdelf*) Likewise.
140 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
141 (UINT_FAST8_TYPE) Likewise.
142 (INT_FAST16_TYPE) Check CHAR_FAST16.
143 (UINT_FAST16_TYPE) Likewise.
145 2017-09-30 Jakub Jelinek <jakub@redhat.com>
148 * config/i386/i386.md
149 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
150 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
151 *divmodsi4_zext_2): New define_insn_and_split.
152 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
153 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
154 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
155 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
156 New define_insn_and_split.
157 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
158 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
159 operands[1] having DImode when mode is SImode.
161 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
162 always SImode for DIV and MOD in REG_EQUAL notes.
164 2017-09-29 Yury Gribov <tetra2005@gmail.com>
167 * match.pd: Fix handling of NaNs in pattern.
169 2017-09-29 Jeff Law <law@redhat.com>
171 * sbitmap.c (bitmap_bit_in_range_p): New function.
172 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
173 * tree-ssa-dse.c (live_bytes_read): New function.
174 (dse_classify_store): Ignore reads of dead bytes.
176 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
177 typos and whitespace errors.
178 * config/i386/predicates.md (address_no_seg_operand): Likewise.
179 * config/s390/s390.c (s390_emit_prologue): Likewise.
181 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
184 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
185 with a symbol for LRA.
187 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
189 PR rtl-optimization/82338
190 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
192 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
194 * genmodes.c (calc_wider_mode): Suppress qsort macro.
195 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
196 (qsort_chk): Declare.
197 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
198 (qsort_chk): New function.
200 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
202 PR tree-optimization/82337
203 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
204 phi definition if the PHI result appears in an abnormal PHI.
205 (find_basis_for_base_expr): Don't record a basis if the LHS of the
206 basis appears in an abnormal PHI.
208 2017-09-29 Richard Biener <rguenther@suse.de>
210 * graphite-isl-ast-to-gimple.c
211 (translate_isl_ast_to_gimple::set_codegen_error): New function.
212 (binary_op_to_tree): Use it.
213 (get_rename_from_scev): Likewise.
214 (copy_loop_phi_nodes): Likewise.
215 (copy_bb_and_scalar_dependences): Likewise.
216 (translate_pending_phi_nodes): Likewise.
218 2017-09-29 Jakub Jelinek <jakub@redhat.com>
221 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
222 for movabsq $(i32 << shift), r64.
224 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
226 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
227 index when encoding %esp as %rsp to avoid 0x67 prefix.
229 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
231 * config/i386/i386.md (*movsf_internal, *movdf_internal):
232 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
234 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
236 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
237 Extensions with more than 16 double VFP registers.
238 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
239 to_clear_mask and all code related to it. Replace the remaining
240 entry by a sbitmap and adapt code accordingly.
242 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
244 * brig-builtins.def: Change pure attributes to const.
246 2017-09-28 Joseph Myers <joseph@codesourcery.com>
248 * config.gcc (default_gnu_indirect_function): Default to yes for
249 sparc*-*-linux* with glibc.
251 2017-09-28 Joseph Myers <joseph@codesourcery.com>
253 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
254 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
255 when creating .init_array and .fini_array sections with priority
258 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
261 * config/aarch64/aarch64.c
262 (aarch64_builtin_support_vector_misalignment): Always return false
263 when misalignment is unknown.
265 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
267 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
268 this function to return false if the definition used by the swap
269 instruction is artificial, or if the memory address from which the
270 constant value is loaded is not represented by a base address held
271 in a register or if the base address register is a frame or stack
272 pointer. Additionally, return false if the base address of the
273 loaded constant is a SYMBOL_REF but is not considered to be a
275 (replace_swapped_load_constant): New function.
276 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
277 loaded constant vector with a load of a swapped constant vector.
279 2017-09-27 Carl Love <cel@us.ibm.com>
281 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
282 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
283 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
286 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
288 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
289 first, always call autopref_rank_data otherwise.
291 2017-09-27 Richard Biener <rguenther@suse.de>
293 * graphite-scop-detection.c (find_scop_parameters): Move
294 loop bound handling ...
295 (gather_bbs::before_dom_children): ... here, avoiding the need
296 to build scop_info->loop_nest.
297 (record_loop_in_sese): Remove.
298 * sese.h (sese_info_t::loop_nest): Remove.
299 * sese.c (new_sese_info): Do not allocate loop_nest.
300 (free_sese_info): Do not free loop_nest.
302 2017-09-27 Jakub Jelinek <jakub@redhat.com>
305 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
306 lhs from calls if the lhs has addressable type.
308 2017-09-27 Richard Biener <rguenther@suse.de>
310 * graphite.h (scop::max_alias_set): New member.
311 * graphite-scop-detection.c: Remove references to non-existing
313 (build_alias_sets): Record the maximum alias set used for drs.
314 (build_scops): Support zero as unlimited for
315 --param graphite-max-arrays-per-scop.
316 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
318 (build_poly_sr_1): ... here. Compute alias set based on the
319 maximum alias set used for drs rather than
320 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
322 2017-09-27 Richard Biener <rguenther@suse.de>
324 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
325 --param loop-block-tile-size=0 to disable tiling.
327 2017-09-27 Richard Biener <rguenther@suse.de>
329 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
330 (graphite-max-nb-scop-params): Document special value zero.
331 * domwalk.h (dom_walker::STOP): New symbolical constant.
332 (dom_walker::dom_walker): Add optional parameter for bb to
334 (dom_walker::~dom_walker): Declare.
335 (dom_walker::before_dom_children): Document STOP return value.
336 (dom_walker::m_user_bb_to_rpo): New member.
337 (dom_walker::m_bb_to_rpo): Likewise.
338 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
339 mapping here if not provided by the user.
340 (dom_walker::~dom_walker): Free bb to RPO mapping if not
341 provided by the user.
342 (dom_walker::STOP): Define.
343 (dom_walker::walk): Do not compute bb to RPO mapping here.
344 Support STOP return value from before_dom_children to stop
346 * graphite-optimize-isl.c (optimize_isl): If the schedule
347 is the same still generate code if -fgraphite-identity
348 or -floop-parallelize-all are given.
349 * graphite-scop-detection.c: Include cfganal.h.
350 (gather_bbs::gather_bbs): Get and pass through bb to RPO
352 (gather_bbs::before_dom_children): Return STOP for BBs
354 (build_scops): Compute bb to RPO mapping and pass it to
355 the domwalk. Treat --param graphite-max-nb-scop-params=0
356 as not limiting the number of params.
357 * graphite.c (graphite_initialize): Remove limit on the
358 number of basic-blocks in a function.
359 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
360 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
363 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
365 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
366 Adjust code to eliminate needing to do the shift right 32-bits
367 operation after XSCVDPSPN.
369 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
371 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
372 ((X / Y) != 0 -> X >= Y): Likewise.
374 2017-09-26 Carl Love <cel@us.ibm.com>
376 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
377 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
378 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
379 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
380 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
381 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
382 definitions and overloading.
383 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
384 statement for P9V_BUILTIN_XST_LEN_R.
385 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
386 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
387 define_expand and define_insn for the instructions and builtins.
388 * doc/extend.texi: Update the built-in documentation file for the new
390 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
391 define_insn for the instructions
393 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
396 * gcc/config/netbsd-protos.h: New file.
397 * gcc/config/netbsd.c: New file.
398 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
399 * gcc/config/t-netbsd: New file.
400 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
401 (tmake_file) Add t-netbsd.
402 (extra_objs) Add netbsd.o.
404 2017-09-26 Janus Weil <janus@gcc.gnu.org>
408 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
410 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
412 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
413 sign extension from a vector register to a GPR by doing a 32-bit
414 direct move and then an EXTSW.
415 (extendsi<mode>2 splitter): Likewise.
416 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
417 right or vector extract after doing XSCVDPSPN. Use
418 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
420 (movdi_from_sf_zero_ext): Likewise.
421 (reload_gpr_from_vsxsf): Likewise.
422 (p8_mfvsrd_4_disf): Delete, no longer used.
423 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
424 then needing to move the SFmode to a GPR to use the XSCVDPSP
425 instruction instead of FRSP and XSCVDPSPN.
426 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
427 it is adjacent to the other XSCVSPDP insns.
428 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
429 SFmode to be in traditional Altivec registers.
430 (vsx_xscvdpspn): Eliminate useless alternative constraint.
431 (vsx_xscvspdpn): Likewise.
432 (vsx_xscvspdpn_scalar): Likewise.
434 2017-09-26 Martin Jambor <mjambor@suse.cz>
436 * tree-sra.c (compare_access_positions): Put integral types first,
437 stabilize sorting of integral types, remove conditions putting
438 non-full-precision integers last.
439 (sort_and_splice_var_accesses): Disable scalarization if a
440 non-integert would be represented by a non-full-precision integer.
442 2017-09-26 Joseph Myers <joseph@codesourcery.com>
444 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
445 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
446 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
447 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
448 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
449 conditionals inside the function instead of around it. Call
450 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
451 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
453 2017-09-26 Richard Biener <rguenther@suse.de>
455 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
457 (scop_detection::build_scop_breadth): ... this. Removed.
458 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
459 (scop_detection::harmful_stmt_in_bb): Likewise.
460 (scop_detection::graphite_can_represent_stmt): Likewise.
461 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
462 (scop_detection::can_represent_loop): Remove recursion, fold in ...
463 (scop_detection::can_represent_loop_1): ... this. Removed.
464 (scop_detection::harmful_loop_in_region): Simplify after inlining
465 the above and remove more quadraticness.
466 (build_scops): Adjust.
467 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
470 2017-09-26 Jakub Jelinek <jakub@redhat.com>
473 * config/i386/i386.c (ix86_print_operand_address_as): Only test
474 REGNO (base) == SP_REG if base is a REG.
477 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
478 if it is different SSA_NAME.
479 (optimize_range_tests_cmp_bitwise): New function.
480 (optimize_range_tests): Call it.
482 2017-09-26 Richard Biener <rguenther@suse.de>
484 PR tree-optimization/82321
485 * graphite.c (canonicalize_loop_closed_ssa): Properly check
486 for the def being inside the loop.
488 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
490 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
492 * config/s390/s390-builtins.def: Fix constraint on op4.
494 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
496 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
497 independent expanders.
498 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
499 ("vec_ordered", "vec_unordered"): New expanders.
501 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
503 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
506 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
508 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
509 vec_unpacks_lo_v16qi.
510 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
512 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
514 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
515 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
516 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
518 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
520 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
522 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
524 ("*vec_slb<mode>"): New insn pattern.
525 ("vec_shr_<mode>"): New expander.
526 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
527 and force the shift count operand to V16QImode.
528 ("vec_srb<mode>"): Set shift count mode to V16QI.
530 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
532 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
533 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
534 ("vec_widen_smult_hi_<mode>"): New expander definitions.
536 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
539 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
541 2017-09-26 Richard Biener <rguenther@suse.de>
543 PR tree-optimization/82320
544 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
547 2017-09-25 Jeff Law <law@redhat.com>
549 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
550 prototype for new argument.
551 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
552 mostly extracted from rs6000_emit_allocate_stack.
553 (rs6000_emit_probe_stack_range_stack_clash): New function.
554 (rs6000_emit_allocate_stack): Call
555 rs6000_emit_probe_stack_range_stack_clash as needed.
556 (rs6000_emit_probe_stack_range): Add additional argument
557 to call to gen_probe_stack_range{si,di}.
558 (output_probe_stack_range): New.
559 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
560 (output_probe_stack_range_stack_clash): New.
561 (rs6000_emit_prologue): Emit notes into dump file as requested.
562 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
563 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
564 Add additional operand and pass it to output_probe_stack_range.
566 2017-09-25 Bin Cheng <bin.cheng@arm.com>
568 PR tree-optimization/82163
569 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
570 (checking_verify_loop_closed_ssa): New parameter.
571 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
572 (check_loop_closed_ssa_stmt): Delete.
573 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
574 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
575 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
578 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
580 * brig-builtins.def: Treat HSAIL barrier builtins as
581 setjmp/longjump style functions.
583 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
585 * target.def (constant_alignment): New hook.
586 * defaults.h (CONSTANT_ALIGNMENT): Delete.
587 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
588 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
589 * doc/tm.texi: Regenerate.
590 * targhooks.h (default_constant_alignment): Declare.
591 (constant_alignment_word_strings): Likewise.
592 * targhooks.c (default_constant_alignment): New function.
593 (constant_alignment_word_strings): Likewise.
594 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
595 instead of CONSTANT_ALIGNMENT.
596 * varasm.c (align_variable, get_variable_align, build_constant_desc)
597 (force_const_mem): Likewise.
598 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
599 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
600 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
601 (TARGET_CONSTANT_ALIGNMENT): Redefine.
602 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
604 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
605 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
606 constant_alignment_word_strings.
607 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
608 (CONSTANT_ALIGNMENT): Likewise.
609 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
610 (arm_constant_alignment): New function.
611 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
612 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
613 constant_alignment_word_strings.
614 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
615 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
616 constant_alignment_word_strings.
617 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
618 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
619 (cris_constant_alignment): New function.
620 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
621 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
622 (epiphany_constant_alignment): New function.
623 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
624 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
625 constant_alignment_word_strings.
626 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
627 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
628 constant_alignment_word_strings.
629 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
630 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
631 constant_alignment_word_strings.
632 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
633 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
634 * config/i386/i386.c (ix86_constant_alignment): Make static.
635 Use the same interface as the target hook.
636 (TARGET_CONSTANT_ALIGNMENT): Redefine.
637 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
638 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
639 constant_alignment_word_strings.
640 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
641 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
642 (TARGET_CONSTANT_ALIGNMENT): Redefine.
643 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
644 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
645 constant_alignment_word_strings.
646 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
647 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
648 constant_alignment_word_strings.
649 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
650 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
651 constant_alignment_word_strings.
652 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
653 * config/microblaze/microblaze.c (microblaze_constant_alignment):
655 (TARGET_CONSTANT_ALIGNMENT): Redefine.
656 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
657 * config/mips/mips.c (mips_constant_alignment): New function.
658 (TARGET_CONSTANT_ALIGNMENT): Redefine.
659 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
660 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
661 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
662 (mmix_constant_alignment): Make static. Use the same interface
664 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
665 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
666 constant_alignment_word_strings.
667 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
668 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
669 constant_alignment_word_strings.
670 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
671 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
672 constant_alignment_word_strings.
673 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
674 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
675 (rs6000_constant_alignment): New function.
676 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
677 * config/riscv/riscv.c (riscv_constant_alignment): New function.
678 (TARGET_CONSTANT_ALIGNMENT): Redefine.
679 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
680 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
681 (rs6000_constant_alignment): New function.
682 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
683 * config/s390/s390.c (s390_constant_alignment): New function.
684 (TARGET_CONSTANT_ALIGNMENT): Redefine.
685 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
686 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
687 constant_alignment_word_strings.
688 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
689 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
690 (sparc_constant_alignment): New function.
691 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
692 * config/spu/spu.c (spu_constant_alignment): New function.
693 (TARGET_CONSTANT_ALIGNMENT): Redefine.
694 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
695 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
696 constant_alignment_word_strings.
697 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
698 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
699 constant_alignment_word_strings.
700 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
701 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
702 constant_alignment_word_strings.
703 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
704 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
705 (visium_constant_alignment): New function.
706 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
707 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
708 (xtensa_constant_alignment): New function.
709 * system.h (CONSTANT_ALIGNMENT): Poison.
711 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
714 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
715 (rs6000_builtin_valid_without_lhs): New helper function.
716 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
717 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
719 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
721 * target.h (vec_perm_indices): Use unsigned short rather than
723 (auto_vec_perm_indices): Likewise.
724 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
725 Use unsigned int rather than unsigned char.
726 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
728 2017-09-25 Richard Biener <rguenther@suse.de>
730 * cfgloop.h (sort_sibling_loops): Declare.
731 * cfgloop.c (sort_sibling_loops_cmp): New helper.
732 (sort_sibling_loops): New function sorting the sibling loop list
734 * graphite.c (graphite_transform_loops): Sort sibling loops.
736 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
738 * target.def (vec_perm_const_ok): Change sel parameter to
740 * optabs-query.c (can_vec_perm_p): Update accordingly.
741 * doc/tm.texi: Regenerate.
742 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
743 auto_vec_perm_indices and remove separate nelt field.
744 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
745 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
746 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
747 (aarch64_expand_vec_perm_const): Update accordingly.
748 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
750 * config/arm/arm.c (expand_vec_perm_d): Change perm to
751 auto_vec_perm_indices and remove separate nelt field.
752 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
753 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
754 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
756 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
758 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
759 sel to vec_perm_indices.
760 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
761 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
762 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
764 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
767 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
770 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
771 on the FUNCTION_DECL function context if it has a DIE that is a
774 2017-09-25 Richard Biener <rguenther@suse.de>
776 PR tree-optimization/82285
777 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
780 2017-09-25 Tom de Vries <tom@codesourcery.com>
784 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
787 2017-09-25 Richard Biener <rguenther@suse.de>
789 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
790 ISL errors other than isl_error_quota happen. Dump if the
791 schedule is the same.
792 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
793 errors instead of aborting inside ISL.
795 2017-09-25 Iain Sandoe <iain@codesourcery.com>
798 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
799 of libgcc_eh for m64.
800 * config/i386/darwin64.h: Likewise.
802 2017-09-25 Richard Biener <rguenther@suse.de>
805 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
806 attribute for incomplete types nor twice for complete ones.
808 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
811 * config/i386/i386.c (ix86_print_operand_address_as): Encode
812 %esp as %rsp to avoid 0x67 prefix if there is no index or base
815 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
818 * config/i386/i386.opt (mprefer-avx256): Use
819 ix86_target_flags variable.
820 * config/i386/i386.c (ix86_target_string): Move
821 -mprefer-avx256 to flag2_opts.
823 2017-09-22 Jakub Jelinek <jakub@redhat.com>
826 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
827 and x != -1 | y != -1 into (x & y) != -1.
829 2017-09-22 Steve Ellcey <sellcey@cavium.com>
831 * config.gcc: Add new case statement to set
832 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
833 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
834 s390x-*-linux* case statements. Added aarch64 to the list of
835 supported architectures.
837 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
839 PR tree-optimization/82289
840 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
841 STMT_VINFO_RELEVANT_P.
843 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
848 for VR_RANGE only; don't allow VR_ANTI_RANGE.
849 (extract_range_from_binary_expr_1): Don't call
850 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
852 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * target.def (preferred_vector_alignment): New hook.
857 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
859 * doc/tm.texi: Regenerate.
860 * targhooks.h (default_preferred_vector_alignment): Declare.
861 * targhooks.c (default_preferred_vector_alignment): New function.
862 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
864 (DR_TARGET_ALIGNMENT): New macro.
865 (aligned_access_p): Update commentary.
866 (vect_known_alignment_in_bytes): New function.
867 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
869 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
870 Calculate the misalignment based on the target alignment rather than
872 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
873 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
874 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
875 the target alignment, rather than masking the element misalignment
876 with the number of elements in a vector. Also use the target
877 alignment when calculating the maximum number of peels.
878 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
879 instead of TYPE_ALIGN_UNIT.
880 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
881 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
882 (vect_create_addr_base_for_vector_ref): Update call accordingly.
883 (vect_create_data_ref_ptr): Likewise.
884 (vect_setup_realignment): Realign by ANDing with
885 -DR_TARGET_MISALIGNMENT.
886 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
887 the number of peels based on DR_TARGET_ALIGNMENT.
888 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
889 with the guaranteed alignment boundary when deciding whether
891 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
892 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
893 (ensure_base_align): Remove stmt_info parameter. Get the
894 target base alignment from DR_TARGET_ALIGNMENT.
895 (vectorizable_store): Update call accordingly. Interpret
896 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
898 (vectorizable_load): Likewise.
900 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
901 Alan Hayward <alan.hayward@arm.com>
902 David Sherwood <david.sherwood@arm.com>
904 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
905 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
906 (vect_enhance_data_refs_alignment): Likewise.
908 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
910 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
911 error. Only quit immediately if parsing is complete.
912 (BEGIN): Initialize fatal_err and parse_done.
913 (begin fpu, end fpu): Check number of arguments.
914 (begin arch, end arch): Likewise.
915 (begin cpu, end cpu): Likewise.
916 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
917 (optalias): Likewise.
919 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
921 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
922 * config/arm/arm-isa.h: Delete. Move definitions to ...
923 * arm-cpus.in: ... here. Use new feature and fgroup values.
924 * config/arm/arm.c (arm_option_override): Use lower case for feature
926 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
927 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
928 * config/arm/parsecpu.awk (END): Add new command 'isa'.
930 (print_isa_bits_for): New function.
931 (gen_isa): New function.
932 (gen_comm_data): Use print_isa_bits_for.
933 (define feature): New keyword.
934 (define fgroup): New keyword.
935 * config/arm/t-arm (TM_H): Remove.
936 (GTM_H): Add arm-isa.h.
937 (arm-isa.h): Add rule to generate file.
938 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
939 case for feature bit names.
941 2017-09-22 Richard Biener <rguenther@suse.de>
943 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
945 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
946 print of no dependency loops ...
947 * graphite.c (graphite_transform_loops): ... here.
948 (canonicalize_loop_closed_ssa_form): Work from inner to outer
950 (same_close_phi_node, remove_duplicate_close_phi,
951 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
952 (canonicalize_loop_closed_ssa): ... here and simplify.
953 * graphite-optimize-isl.c: Include tree-vectorizer.h.
954 (optimize_isl): Use dump_printf_loc to tell when we stopped
955 optimizing because of an ISL timeout.
957 2017-09-22 Richard Biener <rguenther@suse.de>
959 PR tree-optimization/82291
960 * tree-if-conv.c (predicate_mem_writes): Make sure to
961 remove writes in blocks predicated with false.
963 2017-09-22 Richard Biener <rguenther@suse.de>
965 * sese.c: Include cfganal.h.
966 (if_region_set_false_region): Remove.
967 (create_if_region_on_edge): Likewise.
968 (move_sese_in_condition): Re-implement without destroying
971 2017-09-22 Richard Biener <rguenther@suse.de>
973 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
974 Verify both BBs contain loop PHI nodes before dispatching to
976 (graphite_regenerate_ast_isl): Do not recompute dominators,
977 do not verify three times. Restructure for clarity.
978 * graphite-scop-detection.c (same_close_phi_node,
979 remove_duplicate_close_phi, make_close_phi_nodes_unique,
980 defined_in_loop_p, canonicalize_loop_closed_ssa,
981 canonicalize_loop_closed_ssa_form): Simplify, remove excess
982 checking and SSA rewrite, move to ...
983 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
984 (graphite_initialize): Do not pass in ctx, do not reset the
985 SCEV cache, compute only dominators.
986 (graphite_transform_loops): Allocate ISL ctx after
987 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
988 Maintain post-dominators only around build_scops.
989 * sese.c (if_region_set_false_region): Make static. Free
990 and recompute dominators.
991 (move_sese_in_condition): Assert we don't get called with
992 post-dominators computed.
993 * sese.h (if_region_set_false_region): Remove.
995 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
997 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
998 mode attribute for TARGET_AVX512VL.
1000 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1002 * config/i386/i386.opt (mprefer-avx256): New option.
1003 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1005 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1006 for TARGET_PREFER_AVX256.
1007 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1009 2017-09-21 Jeff Law <law@redhat.com>
1011 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1012 Fix dump output if the only stack space is for pushed registers.
1014 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1016 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1019 2017-09-21 Martin Sebor <msebor@redhat.com>
1022 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1023 code (in C++) or code that triggers warnings.
1025 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1027 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1029 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1031 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1032 * sched-rgn.c: Ditto.
1033 * sel-sched-ir.c: Ditto.
1035 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1037 * toplev.h (set_random_seed): Adjust return type.
1038 * toplev.c (init_local_tick): Move eager initialization of random_seed
1039 to get_random_seed. Adjust comment.
1040 (init_random_seed): Inline to get_random_seed, delete.
1041 (get_random_seed): Initialize random_seed lazily.
1042 (set_random_seed): Do not return previous value.
1043 (print_switch_value): Do not call get_random_seed.
1045 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1047 * cgraph.c (delete_function_version): New, broken out from...
1048 (cgraph_node::delete_function_version): ...here. Rename to
1049 cgraph_node::delete_function_version_by_decl. Update all uses.
1050 (cgraph_node::remove): Call delete_function_version.
1052 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1055 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1056 VAR_DECLs to which addressable non-volatile parameters are mapped
1057 and for id->retvar after the return value assignment. Clear
1058 id->retval and id->retbnd after inlining.
1060 2017-09-21 Richard Biener <rguenther@suse.de>
1062 PR tree-optimization/82276
1063 PR tree-optimization/82244
1064 * tree-vrp.c (build_assert_expr_for): Set
1065 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1067 (remove_range_assertions): Revert earlier change.
1069 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1072 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1074 2017-09-21 Richard Biener <rguenther@suse.de>
1076 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1077 Restore valid IL after code generation errors.
1078 * graphite.c (graphite_transform_loops): Diagnose code
1079 generation issues as MSG_MISSED_OPTIMIZATION and continue
1080 with processing SCOPs.
1082 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1087 rather than choosing between plus_constant and gen_rtx_<CODE>.
1088 * expr.c (emit_push_insn): Likewise.
1089 (expand_expr_real_2): Likewise.
1091 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1092 Alan Hayward <alan.hayward@arm.com>
1093 David Sherwood <david.sherwood@arm.com>
1095 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1097 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1098 Alan Hayward <alan.hayward@arm.com>
1099 David Sherwood <david.sherwood@arm.com>
1101 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1102 calling tree_to_uhwi.
1104 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1109 INTEGER_CST rather than a negative test for ADDR_EXPR.
1111 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1112 Alan Hayward <alan.hayward@arm.com>
1113 David Sherwood <david.sherwood@arm.com>
1115 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1116 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1118 2017-09-21 Richard Biener <rguenther@suse.de>
1120 PR tree-optimization/71351
1121 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1122 graphite_create_new_loop_guard): Remove, fold remaining parts
1124 (translate_isl_ast_node_for): ... here and simplify.
1126 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1129 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1130 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1131 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1132 alternative always use QI mode, for -Os imov (=R,R) alternative
1133 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1136 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1137 Jeff Law <law@redhat.com>
1139 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1140 (allocate_stack_space): New function, partially extracted from
1142 (s390_emit_prologue): Track offset to most recent stack probe.
1143 Code to allocate space moved into allocate_stack_space.
1144 Dump actions when no stack is allocated.
1145 (s390_prologue_plus_offset): New function.
1146 (s390_emit_stack_probe): Likewise.
1148 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1150 * common.opt (Wa, Wl, Wp, g, gz=): Add
1152 (gno-column-info): Remove.
1153 (gcolumn-info): Drop RejectNegative.
1155 (gno-record-gcc-switches): Remove.
1156 (grecord-gcc-switches): Drop RejectNegative.
1157 (gno-split-dwarf): Remove.
1158 (gsplit-dwarf): Drop RejectNegative.
1159 (gno-strict-dwarf): Remove.
1160 (gstrict-dwarf): Drop RejectNegative.
1161 * config/darwin.opt (gfull, gused): Add RejectNegative.
1162 * dwarf2out.c (gen_producer_string): Drop
1163 gno-record-gcc-switches handler.
1164 * optc-gen.awk: Add g to prefixes with negative forms.
1165 * opts-common.c (remapping_prefix_p): New.
1166 (find_opt): Check it.
1167 (generate_canonical_option): Test g prefix.
1168 (option_map): Add -gno- mapping.
1169 (add_misspelling_candidates): Check remapping_prefix_p.
1171 2017-09-20 Jeff Law <law@redhat.com>
1173 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1174 thinko in stack clash protection support.
1176 * explow.c (compute_stack_clash_protection_loop_data): Use
1177 CONST_INT_P instead of explicit test. Verify object is a
1178 CONST_INT_P before looking at INTVAL.
1179 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1180 instead of explicit test.
1182 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1185 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1186 address instead of to r1 and r11.
1188 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1190 * config.gcc: Support "knm".
1191 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1192 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1194 * config/i386/i386.c (m_KNM): Define.
1195 (processor_target_table): Add "knm".
1197 (ix86_option_override_internal): Add "knm".
1198 (ix86_issue_rate): Add PROCESSOR_KNM.
1199 (ix86_adjust_cost): Ditto.
1200 (ia32_multipass_dfa_lookahead): Ditto.
1201 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1202 (fold_builtin_cpu): Add M_INTEL_KNM.
1203 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1204 (processor_type): Add PROCESSOR_KNM.
1205 * config/i386/x86-tune.def: Add m_KNM.
1206 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1208 2017-09-20 Richard Biener <rguenther@suse.de>
1210 PR tree-optimization/80213
1211 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1212 are allowed in empty BBs as well.
1213 (canonicalize_loop_closed_ssa): Also look for other complex
1215 (scop_detection::get_sese): Include the loop-closed PHI block
1217 (scop_detection::merge_sese): Remove code adding extra blocks.
1218 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1219 (build_scops): Assert the final returned scop is invalid.
1221 2017-09-20 Richard Biener <rguenther@suse.de>
1223 PR tree-optimization/82264
1224 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1226 (vn_phi_lookup): Likewise.
1227 (vn_phi_insert): Likewise.
1229 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1231 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1232 that fits into uhwi or shwi, add DW_AT_const_value regardless
1233 of early_dwarf without going through RTL, using add_AT_unsigned
1236 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1237 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1238 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1239 (DEBUG_MACRO_SECTION): Likewise.
1240 (DEBUG_LTO_MACRO_SECTION): Likewise.
1241 (DEBUG_STR_DWO_SECTION): Likewise.
1242 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1243 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1244 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1245 (DEBUG_LTO_LINE_STR_SECTION): Define.
1246 (init_sections_and_labels): Initialize debug_line_str_section
1247 variable. Initialize debug_loc_section for -gdwarf-5 to
1248 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1250 2017-09-20 Richard Biener <rguenther@suse.de>
1252 * graphite-sese-to-poly.c (extract_affine): Properly handle
1253 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1255 2017-09-20 Richard Biener <rguenther@suse.de>
1257 PR tree-optimization/81373
1258 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1259 Force SESE live-out defs to be handled even if they are
1262 2017-09-19 Jeff Law <law@redhat.com>
1264 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1265 nothing for stack adjustments with REG_STACK_CHECK.
1266 * sched-deps.c (parse_add_or_inc): Reject insns with
1267 REG_STACK_CHECK from dependency breaking.
1268 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1269 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1270 * reg-notes.def (STACK_CHECK): New note.
1272 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1273 (ix86_expand_prologue): Dump stack clash info as needed.
1274 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1276 * function.c (dump_stack_clash_frame_info): New function.
1277 * function.h (dump_stack_clash_frame_info): Prototype.
1278 (enum stack_clash_probes): New enum.
1280 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1281 flag_stack_clash_protection.
1282 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1283 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1284 (arm_frame_pointer_required): Likewise.
1285 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1286 (ia64_expand_prologue): Likewise.
1287 * config/mips/mips.c (mips_expand_prologue): Likewise.
1288 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1289 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1290 (sparc_flat_expand_prologue): Likewise.
1291 * config/spu/spu.c (spu_expand_prologue): Likewise.
1293 * explow.c: Include "params.h".
1294 (anti_adjust_stack_and_probe_stack_clash): New function.
1295 (get_stack_check_protect): Likewise.
1296 (compute_stack_clash_protection_loop_data): Likewise.
1297 (emit_stack_clash_protection_loop_start): Likewise.
1298 (emit_stack_clash_protection_loop_end): Likewise.
1299 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1300 Use anti_adjust_stack_and_probe_stack_clash.
1301 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1302 (emit_stack_clash_protection_loop_start): Likewise.
1303 (emit_stack_clash_protection_loop_end): Likewise.
1304 * rtl.h (get_stack_check_protect): Prototype.
1305 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1306 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1307 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1309 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1311 * doc/tm.texi: Rebuilt.
1312 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1313 get_stack_check_protect.
1314 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1315 * config/arm/arm.c (arm_expand_prologue): Likewise.
1316 (arm_frame_pointer_required): Likewise.
1317 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1318 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1319 * config/mips/mips.c (mips_expand_prologue): Likewise.
1320 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1321 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1322 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1323 (sparc_flat_expand_prologue): Likewise.
1325 * common.opt (-fstack-clash-protection): New option.
1326 * flag-types.h (enum stack_check_type): Note difference between
1327 -fstack-check= and -fstack-clash-protection.
1328 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1329 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1330 * toplev.c (process_options): Issue warnings/errors for cases
1331 not handled with -fstack-clash-protection.
1332 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1333 (-fstack-check): Note additional problem with -fstack-check=generic.
1334 Note that -fstack-check is primarily for Ada and refer users
1335 to -fstack-clash-protection for stack-clash-protection.
1336 Document new params for stack clash protection.
1338 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1340 * config/i386/i386.c (ix86_split_long_move): Do not handle
1341 address used for LEA in a special way.
1343 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1345 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1347 2017-09-19 Martin Sebor <msebor@redhat.com>
1350 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1351 of incompatible types.
1353 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1355 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1356 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1357 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1358 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1360 2017-09-19 Richard Biener <rguenther@suse.de>
1362 PR tree-optimization/82244
1363 * tree-vrp.c (remove_range_assertions): Do not propagate
1364 a constant to abnormals but replace the assert with a copy.
1366 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1368 PR rtl-optimization/57878
1369 PR rtl-optimization/68988
1370 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1371 avoidance test involving non_reload_pseudos. Move frequency test
1372 below the general fragmentation avoidance test.
1374 2017-09-19 Richard Biener <rguenther@suse.de>
1376 PR tree-optimization/69728
1377 * graphite-sese-to-poly.c (schedule_error): New global.
1378 (add_loop_schedule): Handle empty domain by failing the
1380 (build_original_schedule): Handle schedule_error.
1382 2017-09-19 Richard Biener <rguenther@suse.de>
1384 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1385 Do not iterate to sibling loops but only to siblings of inner
1388 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1391 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1392 that the registers are different.
1394 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1396 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1397 to processor_model and "amdfam17h" to arch_names_table.
1398 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1400 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1403 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1405 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1406 Alan Hayward <alan.hayward@arm.com>
1407 David Sherwood <david.sherwood@arm.com>
1409 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1411 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1413 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1414 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1415 vect_schedule_slp_instance.
1416 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1417 Update call to vect_slp_analyze_node_operations. Simplify return
1419 (vect_slp_analyze_bb_1): Update call accordingly.
1420 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1421 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1422 (vect_schedule_slp): Update call accordingly.
1424 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1425 Alan Hayward <alan.hayward@arm.com>
1426 David Sherwood <david.sherwood@arm.com>
1428 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1429 with types that aren't in fact scalar.
1431 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1433 * tree-vect-slp.c (vect_record_max_nunits): New function,
1435 (vect_build_slp_tree_1): ...here.
1436 (vect_build_slp_tree_2): Call it for phis too.
1438 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1440 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1441 to vect_get_vec_def_for_operand when getting the mask operand.
1443 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1444 Alan Hayward <alan.hayward@arm.com>
1445 David Sherwood <david.sherwood@arm.com>
1447 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1450 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1451 Alan Hayward <alan.hayward@arm.com>
1452 David Sherwood <david.sherwood@arm.com>
1454 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1455 calculation for vector booleans.
1457 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1458 Alan Hayward <alan.hayward@arm.com>
1459 David Sherwood <david.sherwood@arm.com>
1461 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1463 (vect_transform_stmt): ...here.
1464 (vect_analyze_stmt): Use it instead of calling
1465 vectorizable_live_operation directly.
1467 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1469 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1470 non-SIMT targets in acc vector loops.
1472 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1474 * configure.ac: Add arc and check if assembler supports gdwarf2.
1475 * configure: Regenerate.
1477 2017-09-18 Richard Biener <rguenther@suse.de>
1479 PR tree-optimization/82220
1480 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1481 epilogue niters from the min_profitable_iters compute.
1483 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1486 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1487 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1488 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1490 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1493 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1494 switching to a new text section.
1496 2017-09-18 Richard Biener <rguenther@suse.de>
1498 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1500 (build_alias_set): Reject aliases with no access function.
1502 2017-09-18 Richard Biener <rguenther@suse.de>
1504 PR tree-optimization/79622
1505 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1507 (build_cross_bb_scalars_use): Likewise.
1509 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1511 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1513 2017-09-18 Alan Modra <amodra@gmail.com>
1516 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1517 stack_pointer_rtx for count 0. Update comments. Break up
1518 large rtl expression.
1520 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1522 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1523 Increase to 20 bytes.
1524 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1525 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1526 or avx version of the stub.
1528 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1531 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1532 compute the minimum stack alignment. Also update preferred stack
1533 boundary for leaf functions.
1535 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1537 PR tree-optimization/82228
1538 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1541 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1543 * common/config/nds32/nds32-common.c
1544 (nds32_option_optimization_table): Refine formatting.
1545 (nds32_option_optimization_table): Use -fsched-pressure and
1546 -fomit-frame-pointer for specific optimization level.
1548 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1550 * config/nds32/nds32.c: Refine formatting and comments.
1551 * config/nds32/nds32.h: Likewise.
1552 * config/nds32/nds32.md: Likewise.
1553 * config/nds32/nds32-cost.c: Likewise.
1554 * config/nds32/nds32-isr.c: Likewise.
1555 * config/nds32/nds32-md-auxiliary.c: Likewise.
1556 * config/nds32/nds32-multiple.md: Likewise.
1557 * config/nds32/nds32-predicates.c: Likewise.
1559 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1560 Jakub Jelinek <jakub@redhat.com>
1562 Add support for -std=c++2a.
1563 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1565 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1567 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1570 * doc/extend.texi (Common Function Attributes): Add
1571 references to ARM, AArch64, and S/390 specific attributes.
1572 (Function Specific Option Pragmas): Add AArch64 and S/390
1573 to list of back ends that support the target pragma.
1575 2017-09-15 Nathan Sidwell <nathan@acm.org>
1577 * doc/standards.texi: Fix C++17 description. Update URLs for
1580 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1582 * common.opt (Wcast-align=strict): New warning option.
1583 * doc/invoke.texi: Document -Wcast-align=strict.
1585 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1587 * cgraph.h (cgraph_thunk_info): Add comments.
1588 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1589 assert for VIRTUAL_* arguments stricter.
1591 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1593 PR tree-optimization/71026
1594 * match.pd: Move RDIV patterns from fold-const.c
1595 * fold-const.c (distribute_real_division): Removed.
1596 (fold_binary_loc): Remove calls to distribute_real_divison.
1598 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1600 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1601 c++1z and gnu++1z as deprecated. Change other references to
1602 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1603 Change -Wc++1z-compat to -Wc++17-compat.
1604 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1605 * dwarf2out.c (highest_c_language): Handle C++17.
1606 (gen_compile_unit_die): Likewise.
1608 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1610 PR rtl-optimization/82192
1611 * combine.c (make_extraction): Don't look through non-paradoxical
1612 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1615 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1616 Alan Hayward <alan.hayward@arm.com>
1617 David Sherwood <david.sherwood@arm.com>
1619 * target.def (function_arg_offset): New hook.
1620 * targhooks.h (default_function_arg_offset): Declare.
1621 * targhooks.c (default_function_arg_offset): New function.
1622 * function.c (locate_and_pad_parm): Use
1623 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1624 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1625 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1626 * doc/tm.texi: Regenerate.
1627 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1628 * config/spu/spu.c (spu_function_arg_offset): New function.
1629 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1630 * system.h (FUNCTION_ARG_OFFSET): Poison.
1632 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayard <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * target.def (truly_noop_truncation): New hook.
1637 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1638 than TRULY_NOOP_TRUNCATION.
1639 * hooks.h (hook_bool_uint_uint_true): Declare.
1640 * hooks.c (hook_bool_uint_uint_true): New function.
1641 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1642 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1643 * doc/tm.texi: Regenerate.
1644 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1645 rather than TRULY_NOOP_TRUNCATION in comments.
1646 (simplify_comparison): Likewise.
1647 (record_truncated_value): Likewise.
1648 * expmed.c (extract_bit_field_1): Likewise.
1649 (extract_split_bit_field): Likewise.
1650 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1651 instead of TRULY_NOOP_TRUNCATION.
1652 * function.c (assign_parm_setup_block): Likewise.
1653 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1654 * rtlhooks.c: Include target.h.
1655 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1656 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1657 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1658 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1659 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1660 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1661 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1662 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1663 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1664 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1665 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1666 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1667 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1668 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1669 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1670 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1671 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1672 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1673 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1674 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1675 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1676 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1677 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1678 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1679 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1680 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1681 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1682 rather than TRULY_NOOP_TRUNCATION in comments.
1683 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1684 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1685 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1686 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1687 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1688 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1689 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1690 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1691 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1692 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1693 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1694 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1695 rather than TRULY_NOOP_TRUNCATION in comments.
1696 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1697 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1698 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1699 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1700 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1701 TRULY_NOOP_TRUNCATION condition.
1702 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1703 (TRULY_NOOP_TRUNCATION): Delete.
1704 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1705 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1706 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1707 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1708 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1709 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1710 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1711 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1712 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1713 rather than TRULY_NOOP_TRUNCATION in comments.
1714 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1715 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1716 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1717 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1718 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1719 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1721 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1724 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1725 (*cmp_ior): Likewise.
1726 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1727 (*ior_scc_scc_cmp): Likewise.
1728 (*and_scc_scc): Likewise.
1729 (*and_scc_scc_cmp): Likewise.
1731 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1732 Alan Hayard <alan.hayward@arm.com>
1733 David Sherwood <david.sherwood@arm.com>
1735 * target.def (can_change_mode_class): New hook.
1736 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1737 (hard_regno_nregs): Likewise.
1738 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1739 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1740 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1741 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1742 (LOAD_EXTEND_OP): Update accordingly.
1743 * doc/tm.texi: Regenerate.
1744 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1745 CANNOT_CHANGE_MODE_CLASS.
1746 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1747 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1748 * combine.c (simplify_set): Update accordingly.
1749 * emit-rtl.c (validate_subreg): Likewise.
1750 * recog.c (general_operand): Likewise.
1751 * regcprop.c (mode_change_ok): Likewise.
1752 * reload1.c (choose_reload_regs): Likewise.
1753 (inherit_piecemeal_p): Likewise.
1754 * rtlanal.c (simplify_subreg_regno): Likewise.
1755 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1756 instead of CANNOT_CHANGE_MODE_CLASS.
1757 (reload_cse_simplify_operands): Likewise.
1758 * reload.c (push_reload): Use targetm.can_change_mode_class
1759 instead of CANNOT_CHANGE_MODE_CLASS.
1760 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1761 REG_CANNOT_CHANGE_MODE_P.
1762 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1763 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1764 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1765 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1766 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1767 (arm_can_change_mode_class): New function.
1768 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1769 than CANNOT_CHANGE_MODE_CLASS in comments.
1770 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1771 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1772 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1773 (ix86_can_change_mode_class): ...this new function, inverting the
1774 sense of the return value.
1775 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1776 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1777 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1778 (ia64_can_change_mode_class): New function.
1779 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1780 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1781 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1782 (m32c_can_change_mode_class): ...this new function, inverting the
1783 sense of the return value.
1784 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1785 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1786 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1787 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1788 (mips_can_change_mode_class): ...this new function, inverting the
1789 sense of the return value.
1790 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1791 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1792 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1793 (msp430_can_change_mode_class): New function.
1794 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1795 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1796 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1797 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1798 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1799 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1800 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1801 (pa_cannot_change_mode_class): Replace with...
1802 (pa_can_change_mode_class): ...this new function, inverting the
1803 sense of the return value.
1804 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1805 than CANNOT_CHANGE_MODE_CLASS in comments.
1806 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1807 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1808 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1809 (pdp11_cannot_change_mode_class): Replace with...
1810 (pdp11_can_change_mode_class): ...this new function, inverting the
1811 sense of the return value.
1812 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1813 * config/powerpcspe/powerpcspe-protos.h
1814 (rs6000_cannot_change_mode_class_ptr): Delete.
1815 * config/powerpcspe/powerpcspe.c
1816 (rs6000_cannot_change_mode_class_ptr): Delete.
1817 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1818 (rs6000_option_override_internal): Assign to
1819 targetm.can_change_mode_class instead of
1820 rs6000_cannot_change_mode_class_ptr.
1821 (rs6000_cannot_change_mode_class): Replace with...
1822 (rs6000_can_change_mode_class): ...this new function, inverting the
1823 sense of the return value.
1824 (rs6000_debug_cannot_change_mode_class): Replace with...
1825 (rs6000_debug_can_change_mode_class): ...this new function.
1826 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1827 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1828 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1829 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1830 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1832 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1833 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1834 (rs6000_option_override_internal): Assign to
1835 targetm.can_change_mode_class instead of
1836 rs6000_cannot_change_mode_class_ptr.
1837 (rs6000_cannot_change_mode_class): Replace with...
1838 (rs6000_can_change_mode_class): ...this new function, inverting the
1839 sense of the return value.
1840 (rs6000_debug_cannot_change_mode_class): Replace with...
1841 (rs6000_debug_can_change_mode_class): ...this new function.
1842 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1843 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1844 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1845 (s390_can_change_mode_class): ...this new function, inverting the
1846 sense of the return value.
1847 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1848 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1849 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1850 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1851 (sh_cannot_change_mode_class): Replace with...
1852 (sh_can_change_mode_class): ...this new function, inverting the
1853 sense of the return value.
1854 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1855 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1856 (sparc_can_change_mode_class): New function.
1857 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1858 * config/spu/spu.c (spu_can_change_mode_class): New function.
1859 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1860 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1861 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1862 (visium_can_change_mode_class): New function.
1863 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1865 2017-09-15 Richard Biener <rguenther@suse.de>
1867 PR tree-optimization/82217
1868 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1869 but not undefined case.
1871 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1874 * postreload.c (reload_cse_simplify_operands): Skip
1875 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1877 2017-09-15 Richard Biener <rguenther@suse.de>
1879 PR tree-optimization/68823
1880 * graphite-scop-detection.c (build_alias_set): If we have a
1881 possible dependence check whether we can handle them by just
1882 looking at the DRs DR_ACCESS_FNs.
1883 (build_scops): If build_alias_set fails, fail the SCOP.
1885 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1887 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1888 to support float128 built-in functions that require the ISA 3.0
1890 (BU_FLOAT128_3_HW): Likewise.
1891 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1893 (FMAF128): Likewise.
1895 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1896 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1897 floating point instructions.
1898 (rs6000_invalid_builtin): Likewise.
1899 (rs6000_builtin_mask_names): Likewise.
1900 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1901 (RS6000_BTM_FLOAT128_HW): Likewise.
1902 (RS6000_BTM_COMMON): Likewise.
1903 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1905 * doc/extend.texi (RS/6000 built-in functions): Document the
1906 IEEE 128-bit floating point square root and fused multiply-add
1909 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1911 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1912 reg (r2) isn't in the set of registers defined in the prologue.
1914 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1915 Alan Hayward <alan.hayward@arm.com>
1916 David Sherwood <david.sherwood@arm.com>
1918 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1919 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1920 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1921 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1922 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1924 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1925 max_vectorization_factor.
1926 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1928 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1929 Alan Hayward <alan.hayward@arm.com>
1930 David Sherwood <david.sherwood@arm.com>
1932 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1933 (vect_worthwhile_without_simd_p): Declare.
1934 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1935 (vectorizable_reduction): Use it.
1936 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1937 (vectorizable_operation): Likewise.
1939 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1940 Alan Hayward <alan.hayward@arm.com>
1941 David Sherwood <david.sherwood@arm.com>
1943 * tree-vectorizer.h (vect_get_num_copies): New function.
1944 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1945 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1946 (vectorizable_induction): Likewise.
1947 (vectorizable_live_operation): Likewise.
1948 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1949 (vectorizable_bswap): Likewise.
1950 (vectorizable_call): Likewise.
1951 (vectorizable_conversion): Likewise.
1952 (vectorizable_assignment): Likewise.
1953 (vectorizable_shift): Likewise.
1954 (vectorizable_operation): Likewise.
1955 (vectorizable_store): Likewise.
1956 (vectorizable_load): Likewise.
1957 (vectorizable_condition): Likewise.
1958 (vectorizable_comparison): Likewise.
1959 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1961 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1962 Alan Hayward <alan.hayward@arm.com>
1963 David Sherwood <david.sherwood@arm.com>
1965 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1966 of vect_init_vector.
1968 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1969 Alan Hayward <alan.hayward@arm.com>
1970 David Sherwood <david.sherwood@arm.com>
1972 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1973 an inline wrapper that provides a location.
1974 (gimple_build_vector): Likewise.
1975 * gimple-fold.c (gimple_build_vector_from_val): New function.
1976 (gimple_build_vector): Likewise.
1977 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1978 functions to build the initial value. Always return a gimple value.
1979 (get_initial_defs_for_reduction): Likewise. Only compute
1981 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1982 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1983 (vectorizable_induction): Use gimple_build_vector rather than
1986 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1987 Alan Hayward <alan.hayward@arm.com>
1988 David Sherwood <david.sherwood@arm.com>
1990 * target.h (vec_perm_indices): New typedef.
1991 (auto_vec_perm_indices): Likewise.
1992 * optabs-query.h: Include target.h
1993 (can_vec_perm_p): Take a vec_perm_indices *.
1994 * optabs-query.c (can_vec_perm_p): Likewise.
1995 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1996 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1997 * tree-vect-generic.c (lower_vec_perm): Likewise.
1998 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1999 (vect_grouped_load_supported): Likewise.
2000 (vect_shift_permute_load_chain): Likewise.
2001 (vect_permute_store_chain): Use auto_vec_perm_indices.
2002 (vect_permute_load_chain): Likewise.
2003 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2004 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2005 Update uses of can_vec_perm_p.
2006 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2007 mode with a number of elements. Take a vec_perm_indices *.
2008 (vect_create_epilog_for_reduction): Update accordingly.
2009 Use auto_vec_perm_indices.
2010 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2011 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2012 (vect_transform_slp_perm_load): Likewise.
2013 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2014 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2015 (vect_gen_perm_mask_checked): Likewise.
2016 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2017 (vect_gen_perm_mask_checked): Likewise.
2018 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2019 (vectorizable_store): Likewise.
2020 (vectorizable_load): Likewise.
2021 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2022 (vectorizable_bswap): Likewise.
2024 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2025 Alan Hayward <alan.hayward@arm.com>
2026 David Sherwood <david.sherwood@arm.com>
2028 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2029 * tree.c (build_vector): Likewise.
2030 (build_vector_from_ctor): Update accordingly.
2031 (build_vector_from_val): Likewise.
2032 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2033 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2034 * tree-vect-generic.c (add_rshift): Likewise.
2035 (expand_vector_divmod): Likewise.
2036 (optimize_vector_constructor): Likewise.
2037 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2038 (vect_transform_slp_perm_load): Likewise.
2039 (vect_schedule_slp_instance): Likewise.
2040 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2041 (vectorizable_call): Likewise.
2042 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2043 * expmed.c (make_tree): Likewise.
2044 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2045 a vector passed to build_vector.
2046 (fold_convert_const): Likewise.
2047 (exact_inverse): Likewise.
2048 (fold_ternary_loc): Likewise.
2049 (fold_relational_const): Likewise.
2050 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2051 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2052 (const_unop): Likewise. Store the reduction accumulator in a
2053 variable rather than an array.
2054 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2055 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2056 the new vector, rather than constructing it after the input arrays.
2057 (native_interpret_vector): Use auto_vec<tree> when building
2058 a vector passed to build_vector. Add elements in order.
2059 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2060 auto_vec<tree> when building a vector passed to build_vector.
2061 (vect_create_epilog_for_reduction): Likewise.
2062 (vectorizable_induction): Likewise.
2063 (get_initial_def_for_reduction): Likewise. Fix indentation of
2065 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2067 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2068 passed to build_vector.
2070 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2071 Alan Hayward <alan.hayward@arm.com>
2072 David Sherwood <david.sherwood@arm.com>
2074 * tree-core.h (tree_base::u): Add an "nelts" field.
2075 (tree_vector): Use VECTOR_CST_NELTS as the length.
2076 * tree.c (tree_size): Likewise.
2077 (make_vector): Initialize VECTOR_CST_NELTS.
2078 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2079 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2080 TYPE_VECTOR_SUBPARTS.
2081 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2082 as the number of units, setting it from VECTOR_CST_NELTS.
2083 (const_vector_from_tree): Likewise.
2084 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2085 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2086 (fold_negate_expr_1): Likewise.
2087 (fold_convert_const): Likewise.
2088 (const_binop): Likewise. Differentiate the number of output and
2090 (const_unop): Likewise.
2091 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2092 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2093 in cases that did the opposite.
2095 2017-09-14 Richard Biener <rguenther@suse.de>
2097 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2100 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2102 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2104 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2107 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2108 if and where to split a bb, except for splitting before debug insn
2109 sequences followed by non-label real insn. Delete debug insns
2110 in between basic blocks.
2112 * combine.c (make_compound_operation_int): Formatting fixes.
2114 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2115 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2116 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2117 * config/sol2.h (LINK_EH_SPEC): Likewise.
2118 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2119 * config/s390/linux.h (LINK_SPEC): Likewise.
2120 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2121 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2122 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2123 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2124 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2125 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2126 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2127 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2129 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2131 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2133 (ENDFILE_LINUX_SPEC): Likewise.
2134 (LINK_EH_SPEC): Likewise.
2135 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2136 (LINK_OS_LINUX_SPEC32): Likewise.
2137 (LINK_OS_LINUX_SPEC64): Likewise.
2138 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2139 (LINK_OS_LINUX_SPEC): Likewise.
2141 2017-09-13 Martin Liska <mliska@suse.cz>
2144 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2145 CASE_HIGH is NULL_TREE.
2147 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2148 Alan Hayward <alan.hayward@arm.com>
2149 David Sherwood <david.sherwood@arm.com>
2151 * target.def (secondary_memory_needed): New hook.
2152 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2153 instead of SECONDARY_MEMORY_NEEDED.
2154 (secondary_memory_needed_mode): Likewise.
2155 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2156 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2157 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2158 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2159 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2160 * doc/tm.texi: Regenerate.
2161 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2162 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2163 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2164 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2165 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2166 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2167 mode argument first and change the reg_class arguments to reg_class_t.
2168 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2169 Make static. Update the call to inline_secondary_memory_needed.
2170 (ix86_register_move_cost): Update the call to
2171 inline_secondary_memory_needed.
2172 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2173 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2175 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2176 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2178 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2179 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2180 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2181 and match hook interface. Add comment from mips.h.
2182 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2183 * config/mmix/mmix.md (truncdfsf2): Refer to
2184 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2186 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2187 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2188 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2189 (pa_secondary_memory_needed): New function.
2190 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2191 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2192 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2193 (pdp11_secondary_memory_needed): Make static and match hook interface.
2194 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2195 * config/powerpcspe/powerpcspe-protos.h
2196 (rs6000_secondary_memory_needed_ptr): Delete.
2197 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2199 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2200 (rs6000_option_override_internal): Assign to
2201 targetm.secondary_memory_needed rather than
2202 rs6000_secondary_memory_needed_ptr.
2203 (rs6000_secondary_memory_needed): Match hook interface.
2204 (rs6000_debug_secondary_memory_needed): Likewise.
2205 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2206 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2207 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2208 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2209 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2210 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2212 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2213 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2214 (rs6000_option_override_internal): Assign to
2215 targetm.secondary_memory_needed rather than
2216 rs6000_secondary_memory_needed_ptr.
2217 (rs6000_secondary_memory_needed): Match hook interface.
2218 (rs6000_debug_secondary_memory_needed): Likewise.
2219 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2220 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2221 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2222 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2223 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2224 (sparc_secondary_memory_needed): New function.
2225 * lra-constraints.c (check_and_process_move): Refer to
2226 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2228 (curr_insn_transform): Likewise.
2229 (process_alt_operands): Use targetm.secondary_memory_needed
2230 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2231 (check_secondary_memory_needed_p): Likewise.
2232 (choose_split_class): Likewise.
2233 * reload.c: Unconditionally include code that was previously
2234 conditional on SECONDARY_MEMORY_NEEDED.
2235 (push_secondary_reload): Use targetm.secondary_memory_needed
2236 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2237 (push_reload): Likewise.
2238 * reload1.c: Unconditionally include code that was previously
2239 conditional on SECONDARY_MEMORY_NEEDED.
2240 (choose_reload_regs): Use targetm.secondary_memory_needed
2241 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2242 (gen_reload): Likewise.
2243 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2245 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2246 Alan Hayward <alan.hayward@arm.com>
2247 David Sherwood <david.sherwood@arm.com>
2249 * target.def (secondary_memory_needed_mode): New hook:
2250 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2251 * targhooks.h (default_secondary_memory_needed_mode): New function.
2252 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2253 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2254 * doc/tm.texi: Regenerate.
2255 * lra-constraints.c (check_and_process_move): Use
2256 targetm.secondary_memory_needed_mode instead of
2257 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2258 (curr_insn_transform): Likewise.
2259 * reload.c (get_secondary_mem): Likewise.
2260 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2261 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2263 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2264 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2265 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2266 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2267 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2269 * config/powerpcspe/powerpcspe-protos.h
2270 (rs6000_secondary_memory_needed_mode): Delete.
2271 * config/powerpcspe/powerpcspe.c
2272 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2273 (rs6000_secondary_memory_needed_mode): Make static.
2274 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2275 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2277 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2279 (rs6000_secondary_memory_needed_mode): Make static.
2280 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2281 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2282 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2283 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2284 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2286 (sparc_secondary_memory_needed_mode): New function.
2287 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2289 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2291 * config/aarch64/constraints.md (Umq): New constraint.
2292 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2294 (mov<mode>): Update condition.
2296 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2298 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2299 when bitposition is the same.
2301 2017-09-13 Richard Biener <rguenther@suse.de>
2303 * dwarf2out.c (output_die_symbol): Remove.
2304 (output_die): Do not output a DIEs symbol.
2306 2017-09-13 Richard Biener <rguenther@suse.de>
2309 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2310 default-def to avoid breaking iterator update with the weird
2311 interaction with cgraph_update_edges_for_call_stmt_node.
2313 2017-09-13 Richard Biener <rguenther@suse.de>
2315 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2316 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2317 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2318 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2320 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2322 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2323 Disable pc relative literal load irrespective of
2324 TARGET_FIX_ERR_A53_84341 for default.
2326 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2328 * config/sparc/sparc.c (output_return): Output the source location of
2329 the insn in the delay slot, if any.
2330 (output_sibcall): Likewise.
2332 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2335 * common.opt (-static-pie): New alias.
2336 (shared): Negate static-pie.
2337 (-no-pie): Update help text.
2339 (static-pie): New option.
2340 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2341 -static-pie support.
2342 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2343 (LINK_EH_SPEC): Likewise.
2344 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2345 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2346 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2347 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2348 (init_gcc_specs): Likewise.
2349 (init_spec): Likewise.
2350 (display_help): Update help message for -pie.
2351 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2354 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2356 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2357 (movdi_aarch64): Likewise.
2358 (movti_aarch64): Likewise.
2360 2017-09-12 Simon Wright <simon@pushface.org>
2363 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2364 calculation of the minor version, always output as 0.
2366 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2369 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2370 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2371 on it early, rather than manual conversion late. For
2372 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2373 instead of performing manual conversion.
2375 2017-09-12 Carl Love <cel@us.ibm.com>
2377 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2378 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2380 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2381 VMULOSW): Add definitions.
2382 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2383 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2384 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2385 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2386 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2388 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2390 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2392 (movti_aarch64): Likewise.
2393 (movdf_aarch64): Likewise.
2394 (movtf_aarch64): Likewise.
2395 (load_pairdi): Likewise.
2396 (store_pairdi): Likewise.
2397 (load_pairdf): Likewise.
2398 (store_pairdf): Likewise.
2399 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2400 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2401 (ldr_got_small_<mode>): Likewise.
2402 (ldr_got_small_28k_<mode>): Likewise.
2403 (ldr_got_tiny): Likewise.
2404 * config/aarch64/iterators.md (ldst_sz): New.
2405 (ldpstp_sz): Likewise.
2406 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2408 (thunderx_load): Split load_8 to load_16.
2409 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2411 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2412 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2413 (xgene1_store_pair): Split store_8 to store_16.
2414 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2415 (falkor_st_0_st_sd): Split store_8 to store_16.
2417 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2419 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2420 and store1/2/3/4 to store_4/8/12/16.
2421 * config/aarch64/aarch64.md: Update for rename.
2422 * config/arm/arm.md: Likewise.: Likewise.
2423 * config/arm/arm.c: Likewise.
2424 * config/arm/thumb1.md: Likewise.
2425 * config/arm/thumb2.md: Likewise.
2426 * config/arm/vfp.md: Likewise.
2427 * config/arm/arm-generic.md: Likewise.
2428 * config/arm/arm1020e.md: Likewise.
2429 * config/arm/arm1026ejs.md: Likewise.
2430 * config/arm/arm1136jfs.md: Likewise.
2431 * config/arm/arm926ejs.md: Likewise.
2432 * config/arm/cortex-a15.md: Likewise.
2433 * config/arm/cortex-a17.md: Likewise.
2434 * config/arm/cortex-a5.md: Likewise.
2435 * config/arm/cortex-a53.md: Likewise.
2436 * config/arm/cortex-a57.md: Likewise.
2437 * config/arm/cortex-a7.md: Likewise.
2438 * config/arm/cortex-a8.md: Likewise.
2439 * config/arm/cortex-a9.md: Likewise.
2440 * config/arm/cortex-m4.md: Likewise.
2441 * config/arm/cortex-m7.md: Likewise.
2442 * config/arm/cortex-r4.md: Likewise.
2443 * config/arm/exynos-m1.md: Likewise.
2444 * config/arm/fa526.md: Likewise.
2445 * config/arm/fa606te.md: Likewise.
2446 * config/arm/fa626te.md: Likewise.
2447 * config/arm/fa726te.md: Likewise.
2448 * config/arm/fmp626.md: Likewise.
2449 * config/arm/iwmmxt.md: Likewise.
2450 * config/arm/ldmstm.md: Likewise.
2451 * config/arm/marvell-pj4.md: Likewise.
2452 * config/arm/xgene1.md: Likewise.
2453 * config/aarch64/thunderx.md: Likewise.
2454 * config/aarch64/thunderx2t99.md: Likewise.
2455 * config/aarch64/falkor.md: Likewise.
2457 2017-09-12 Martin Liska <mliska@suse.cz>
2459 * attribs.c (private_lookup_attribute): New function.
2460 * attribs.h (private_lookup_attribute): Declared here.
2461 (lookup_attribute): Called from this place.
2463 2017-09-12 Richard Biener <rguenther@suse.de>
2465 PR tree-optimization/82157
2466 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2467 stmts with side-effects.
2469 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2470 Alan Hayward <alan.hayward@arm.com>
2471 David Sherwood <david.sherwood@arm.com>
2473 * target.def (hard_regno_nregs): New hook.
2474 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2475 * targhooks.h (default_hard_regno_nregs): Declare.
2476 * targhooks.c (default_hard_regno_nregs): New function.
2477 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2478 (TARGET_HARD_REGNO_NREGS): ...this hook.
2479 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2480 (CLASS_MAX_NREGS): Likewise.
2481 * doc/tm.texi: Regenerate.
2482 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2483 instead of HARD_REGNO_NREGS.
2484 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2485 HARD_REGNO_NREGS in the comment.
2486 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2487 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2488 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2489 Return an unsigned int.
2490 (TARGET_HARD_REGNO_NREGS): Redefine.
2491 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2492 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2493 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2494 (arc_hard_regno_nregs): New function.
2495 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2496 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2497 (arm_hard_regno_nregs): New function.
2498 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2499 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2500 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2501 (TARGET_HARD_REGNO_NREGS): Redefine.
2502 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2503 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2504 (HARD_REGNO_NREGS): Delete.
2505 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2506 (cr16_hard_regno_nregs): New function.
2507 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2508 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2509 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2510 (cris_hard_regno_nregs): New function.
2511 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2512 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2513 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2514 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2515 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2516 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2517 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2518 (frv_hard_regno_nregs): Make static. Take and return an
2520 (frv_class_max_nregs): Remove outdated copy of documentation.
2521 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2522 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2523 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2524 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2525 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2526 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2527 (TARGET_HARD_REGNO_NREGS): Redefine.
2528 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2529 (CLASS_MAX_NREGS): Update comment.
2530 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2531 (ia64_hard_regno_nregs): New function.
2532 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2533 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2534 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2535 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2536 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2538 (m32c_hard_regno_nregs): Likewise. Make static.
2539 (TARGET_HARD_REGNO_NREGS): Redefine.
2540 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2541 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2542 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2543 (m68k_hard_regno_nregs): New function.
2544 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2545 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2546 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2547 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2548 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2549 Take and return an unsigned int.
2550 (TARGET_HARD_REGNO_NREGS): Redefine.
2551 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2552 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2553 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2554 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2555 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2556 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2557 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2558 (msp430_hard_regno_nregs): Make static. Take and return an
2560 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2561 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2562 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2563 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2564 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2565 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2566 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2567 (TARGET_HARD_REGNO_NREGS): Redefine.
2568 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2569 (PA_HARD_REGNO_NREGS): ...this.
2570 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2571 (PA_HARD_REGNO_NREGS): ...this.
2572 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2573 (pa_hard_regno_nregs): New function.
2574 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2575 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2576 (pdp11_hard_regno_nregs): New function.
2577 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2578 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2579 (rs6000_hard_regno_nregs_hook): New function.
2580 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2581 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2582 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2583 Take and return an unsigned int. Move earlier in file.
2584 (TARGET_HARD_REGNO_NREGS): Redefine.
2585 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2586 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2587 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2588 (rl78_hard_regno_nregs): Make static. Take and return an
2590 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2591 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2592 (rs6000_hard_regno_nregs_hook): New function.
2593 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2594 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2595 (TARGET_HARD_REGNO_NREGS): Redefine.
2596 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2597 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2598 instead of HARD_REGNO_NREGS.
2599 (s390_hard_regno_nregs): New function.
2600 (s390_hard_regno_mode_ok): Add comment from s390.h.
2601 (TARGET_HARD_REGNO_NREGS): Redefine.
2602 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2603 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2604 (sh_hard_regno_nregs): New function.
2605 (sh_pass_in_reg_p): Use it.
2606 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2607 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2608 (sparc_hard_regno_nregs): New function.
2609 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2610 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2611 (spu_function_arg_advance): Use it, supplying a valid register number.
2612 (TARGET_HARD_REGNO_NREGS): Redefine.
2613 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2614 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2615 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2616 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2617 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2618 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2619 (CLASS_MAX_NREGS): Remove copy of old documentation.
2620 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2621 (visium_hard_regno_nregs): New function.
2622 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2623 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2624 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2625 xtensa_hard_regno_nregs): New function.
2626 * system.h (HARD_REGNO_NREGS): Poison.
2628 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2630 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2631 hard_regno_nregs instead of HARD_REGNO_NREGS.
2632 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2633 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2634 (c6x_expand_epilogue): Likewise.
2635 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2636 (frv_read_iacc_argument): Likewise.
2637 * config/sh/sh.c: Include regs.h.
2638 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2639 (regs_used): Likewise.
2640 (output_stack_adjust): Likewise.
2641 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2642 * expmed.c: Include regs.h.
2643 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2644 * ree.c: Include regs.h.
2645 (combine_reaching_defs): Use hard_regno_nregs instead of
2647 (add_removable_extension): Likewise.
2649 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2651 * regs.h (hard_regno_nregs): Turn into a function.
2652 (end_hard_regno): Update accordingly.
2653 * caller-save.c (setup_save_areas): Likewise.
2654 (save_call_clobbered_regs): Likewise.
2655 (replace_reg_with_saved_mem): Likewise.
2656 (insert_restore): Likewise.
2657 (insert_save): Likewise.
2658 * combine.c (can_change_dest_mode): Likewise.
2659 (move_deaths): Likewise.
2660 (distribute_notes): Likewise.
2661 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2662 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2663 (rs6000_split_multireg_move): Likewise.
2664 (rs6000_register_move_cost): Likewise.
2665 (rs6000_memory_move_cost): Likewise.
2666 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2667 (rs6000_split_multireg_move): Likewise.
2668 (rs6000_register_move_cost): Likewise.
2669 (rs6000_memory_move_cost): Likewise.
2670 * cselib.c (cselib_reset_table): Likewise.
2671 (cselib_lookup_1): Likewise.
2672 * emit-rtl.c (set_mode_and_regno): Likewise.
2673 * function.c (aggregate_value_p): Likewise.
2674 * ira-color.c (setup_profitable_hard_regs): Likewise.
2675 (check_hard_reg_p): Likewise.
2676 (calculate_saved_nregs): Likewise.
2677 (assign_hard_reg): Likewise.
2678 (improve_allocation): Likewise.
2679 (calculate_spill_cost): Likewise.
2680 * ira-emit.c (modify_move_list): Likewise.
2681 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2682 (ira_hard_reg_in_set_p): Likewise.
2683 * ira.c (setup_reg_mode_hard_regset): Likewise.
2684 (clarify_prohibited_class_mode_regs): Likewise.
2685 (check_allocation): Likewise.
2686 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2687 (lra_setup_reg_renumber): Likewise.
2688 (setup_try_hard_regno_pseudos): Likewise.
2689 (spill_for): Likewise.
2690 (assign_hard_regno): Likewise.
2691 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2692 * lra-constraints.c (in_class_p): Likewise.
2693 (lra_constraint_offset): Likewise.
2694 (simplify_operand_subreg): Likewise.
2695 (lra_constraints): Likewise.
2696 (split_reg): Likewise.
2697 (split_if_necessary): Likewise.
2698 (invariant_p): Likewise.
2699 (inherit_in_ebb): Likewise.
2700 * lra-lives.c (process_bb_lives): Likewise.
2701 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2702 (get_hard_regs): Likewise.
2703 (do_remat): Likewise.
2704 * lra-spills.c (assign_spill_hard_regs): Likewise.
2705 * mode-switching.c (create_pre_exit): Likewise.
2706 * postreload.c (reload_combine_recognize_pattern): Likewise.
2707 * recog.c (peep2_find_free_register): Likewise.
2708 * regcprop.c (kill_value_regno): Likewise.
2709 (set_value_regno): Likewise.
2710 (copy_value): Likewise.
2711 (maybe_mode_change): Likewise.
2712 (find_oldest_value_reg): Likewise.
2713 (copyprop_hardreg_forward_1): Likewise.
2714 * regrename.c (check_new_reg_p): Likewise.
2715 (regrename_do_replace): Likewise.
2716 * reload.c (push_reload): Likewise.
2717 (combine_reloads): Likewise.
2718 (find_dummy_reload): Likewise.
2719 (operands_match_p): Likewise.
2720 (find_reloads): Likewise.
2721 (find_equiv_reg): Likewise.
2722 (reload_adjust_reg_for_mode): Likewise.
2723 * reload1.c (count_pseudo): Likewise.
2724 (count_spilled_pseudo): Likewise.
2725 (find_reg): Likewise.
2726 (clear_reload_reg_in_use): Likewise.
2727 (free_for_value_p): Likewise.
2728 (allocate_reload_reg): Likewise.
2729 (choose_reload_regs): Likewise.
2730 (reload_adjust_reg_for_temp): Likewise.
2731 (emit_reload_insns): Likewise.
2732 (delete_output_reload): Likewise.
2733 * rtlanal.c (subreg_get_info): Likewise.
2734 * sched-deps.c (sched_analyze_reg): Likewise.
2735 * sel-sched.c (init_regs_for_mode): Likewise.
2736 (mark_unavailable_hard_regs): Likewise.
2737 (choose_best_reg_1): Likewise.
2738 (verify_target_availability): Likewise.
2739 * valtrack.c (dead_debug_insert_temp): Likewise.
2740 * var-tracking.c (track_loc_p): Likewise.
2741 (emit_note_insn_var_location): Likewise.
2742 * varasm.c (make_decl_rtl): Likewise.
2743 * reginfo.c (choose_hard_reg_mode): Likewise.
2744 (init_reg_modes_target): Refer directly to
2745 this_target_regs->x_hard_regno_nregs.
2747 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2749 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2750 instead of hard_regno_nregs.
2752 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2754 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2755 end_hard_regno instead of hard_regno_nregs.
2756 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2757 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2758 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2759 * ira-color.c (improve_allocation): Likewise.
2760 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2761 * lra-lives.c (mark_regno_live): Likewise.
2762 (mark_regno_dead): Likewise.
2763 * lra-remat.c (operand_to_remat): Likewise.
2764 * lra.c (collect_non_operand_hard_regs): Likewise.
2765 * postreload.c (reload_combine_note_store): Likewise.
2766 (move2add_valid_value_p): Likewise.
2767 * reload.c (regno_clobbered_p): Likewise.
2769 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2771 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2773 * config/v850/v850.c (v850_reorg): Likewise.
2774 * reload.c (refers_to_regno_for_reload_p): Likewise.
2775 (find_equiv_reg): Likewise.
2776 * reload1.c (reload_reg_reaches_end_p): Likewise.
2778 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2780 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2782 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2783 * config/arm/arm.c (output_move_neon): Likewise.
2784 (arm_attr_length_move_neon): Likewise.
2785 (neon_split_vcombine): Likewise.
2786 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2787 (c6x_mark_reg_written): Likewise.
2788 (c6x_dwarf_register_span): Likewise.
2789 * config/i386/i386.c (ix86_save_reg): Likewise.
2790 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2791 (rws_access_reg): Likewise.
2792 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2793 * mode-switching.c (create_pre_exit): Likewise.
2794 * ree.c (combine_reaching_defs): Likewise.
2795 (add_removable_extension): Likewise.
2796 * regcprop.c (find_oldest_value_reg): Likewise.
2797 (copyprop_hardreg_forward_1): Likewise.
2798 * reload.c (reload_inner_reg_of_subreg): Likewise.
2799 (push_reload): Likewise.
2800 (combine_reloads): Likewise.
2801 (find_dummy_reload): Likewise.
2802 (reload_adjust_reg_for_mode): Likewise.
2803 * reload1.c (find_reload_regs): Likewise.
2804 (forget_old_reloads_1): Likewise.
2805 (reload_reg_free_for_value_p): Likewise.
2806 (reload_adjust_reg_for_temp): Likewise.
2807 (emit_reload_insns): Likewise.
2808 (delete_output_reload): Likewise.
2809 * sel-sched.c (choose_best_reg_1): Likewise.
2810 (choose_best_pseudo_reg): Likewise.
2812 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2813 Alan Hayward <alan.hayward@arm.com>
2814 David Sherwood <david.sherwood@arm.com>
2816 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2817 * target.def (slow_unaligned_access): New hook.
2818 * targhooks.h (default_slow_unaligned_access): Declare.
2819 * targhooks.c (default_slow_unaligned_access): New function.
2820 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2821 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2822 * doc/tm.texi: Regenerate.
2823 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2824 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2825 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2827 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2828 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2830 (rs6000_slow_unaligned_access): New function.
2831 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2832 (expand_block_compare): Likewise.
2833 (expand_strn_compare): Likewise.
2834 (rs6000_rtx_costs): Likewise.
2835 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2836 (riscv_slow_unaligned_access): Likewise.
2837 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2838 (riscv_slow_unaligned_access_p): ...this and make static.
2839 (riscv_option_override): Update accordingly.
2840 (riscv_slow_unaligned_access): New function.
2841 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2842 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2843 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2844 (rs6000_slow_unaligned_access): New function.
2845 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2846 (rs6000_rtx_costs): Likewise.
2847 * config/rs6000/rs6000-string.c (expand_block_compare)
2848 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2849 of SLOW_UNALIGNED_ACCESS.
2850 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2851 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2852 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2853 of SLOW_UNALIGNED_ACCESS.
2854 * expmed.c (simple_mem_bitfield_p): Likewise.
2855 * expr.c (alignment_for_piecewise_move): Likewise.
2856 (emit_group_load_1): Likewise.
2857 (emit_group_store): Likewise.
2858 (copy_blkmode_from_reg): Likewise.
2859 (emit_push_insn): Likewise.
2860 (expand_assignment): Likewise.
2861 (store_field): Likewise.
2862 (expand_expr_real_1): Likewise.
2863 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2864 * lra-constraints.c (simplify_operand_subreg): Likewise.
2865 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2866 * gimple-ssa-store-merging.c: Likewise in block comment at start
2868 * tree-ssa-strlen.c: Include target.h.
2869 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2870 of SLOW_UNALIGNED_ACCESS.
2871 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2873 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2875 PR rtl-optimization/82185
2876 * expmed.c (emit_store_flag_int): Only test tem if it has been
2879 2017-09-12 Richard Biener <rguenther@suse.de>
2882 * match.pd ((FTYPE) N CMP CST): Fix typo.
2884 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2886 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2888 (mips_near_type_p): Add 'short_call' attribute as a synonym
2890 * doc/extend.texi (short_call): Document new function attribute.
2892 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2895 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2896 assertion check that in the condition.
2897 (get_atomic_generic_size): Likewise. Before testing if parameter
2898 has pointer type, if it has array type, call for C++
2899 default_conversion to perform array-to-pointer conversion.
2901 2017-09-12 Richard Biener <rguenther@suse.de>
2903 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2904 for operations we cannot scalarize.
2906 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2908 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2909 vectors heap vectors. Clean up comments.
2910 Make visited_bbs a reference.
2911 (profitable_jump_thread_path): Make GC
2912 vectors heap vectors. Clean up comments.
2914 (convert_and_register_jump_thread_path): Make GC vectors heap
2916 (check_subpath_and_update_thread_path): Same. Clean up comments.
2917 Make visited_bbs a reference.
2918 (handle_phi): Abstract common code to to
2919 register_jump_thread_path_if_profitable.
2920 Rename VAR_BB to DEF_BB.
2922 Make GC vectors heap vectors.
2923 Make visited_bbs a reference.
2924 (handle_assignment): Same.
2925 (register_jump_thread_path_if_profitable): New.
2926 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2928 Make GC vectors heap vectors. Clean up comments.
2929 Make visited_bbs a reference.
2930 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2931 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2934 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2937 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2938 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2940 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2942 Revert r251800 and r251799.
2944 2017-09-11 Martin Jambor <mjambor@suse.cz>
2947 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2948 arguments in advance.
2949 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2950 use it to find predecessor edges.
2951 (naive_outof_ssa): Collect vector of predecessors.
2953 2017-09-08 Jason Merrill <jason@redhat.com>
2955 PR c++/70029 - ICE with ref-qualifier and -flto
2956 * langhooks.h (struct lang_hooks_for_types): Add
2957 copy_lang_qualifiers.
2958 * attribs.c (build_type_attribute_qual_variant): Use it.
2959 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2961 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2962 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2964 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2967 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2968 (*mulsi3_sp64): New instruction.
2969 (mulsi3): New expander.
2971 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2973 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2975 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2977 * sancov.c: Include memmodel.h.
2979 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2982 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2985 2017-09-07 Carl Love <cel@us.ibm.com>
2987 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2988 the sldi instruction.
2990 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2992 * sancov.c: Include tm_p.h.
2994 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2997 * output.h (switch_to_other_text_partition): New declaration.
2998 * varasm.c (switch_to_other_text_partition): New function.
2999 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3000 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3001 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3002 to the other text partition before emitting LCL label and switch back
3003 after emitting the word after it.
3005 2017-09-07 Richard Biener <rguenther@suse.de>
3007 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3008 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3009 critical edges here, after loop init.
3010 (pass_data_pre): Remove PROP_no_crit_edges flags.
3011 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3012 for valueization of call args to avoid leaking VN_TOP.
3013 (visit_use): Assert we do not visit default defs.
3014 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3015 Use error_mark_node to more easily detect leaking VN_TOP.
3016 All default-defs are varying, not VN_TOP. Mark them visited.
3017 (run_scc_vn): Make code match comment.
3019 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3021 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3022 OPTION_MASK_FLOAT128_KEYWORD.
3023 (POWERPC_MASKS): Likewise.
3024 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3025 support for the -mfloat128-type option, and make -mfloat128
3026 default on PowerPC Linux systems. Define or undefine
3027 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3028 Define __float128 to be __ieee128 if IEEE 128-bit support is
3029 enabled, or undefine it.
3030 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3031 Delete defining __FLOAT128_TYPE__.
3032 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3033 -mfloat128-type option and make -mfloat128 default on PowerPC
3035 (TARGET_FLOAT128_TYPE): Likewise.
3036 (-mfloat128-type): Likewise.
3037 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3038 Delete the -mfloat128-type option and make -mfloat128 default on
3039 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3040 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3041 128-bit floating point is enabled. Change tests from using
3042 -mfloat128-type to -mfloat128.
3043 (rs6000_mangle_type): Use the correct mangling for the __float128
3044 type even if normal long double is restricted to 64-bits.
3045 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3047 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3048 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3049 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3051 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3052 documentation for -mfloat128.
3054 2017-09-06 Olivier Hainque <hainque@adacore.com>
3056 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3058 2017-09-06 Wish Wu <wishwu007@gmail.com>
3059 Jakub Jelinek <jakub@redhat.com>
3061 * asan.c (initialize_sanitizer_builtins): Add
3062 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3063 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3064 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3065 BT_FN_VOID_UINT64_PTR variables.
3066 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3067 (BT_FN_VOID_UINT16_UINT16): Likewise.
3068 (BT_FN_VOID_UINT32_UINT32): Likewise.
3069 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3070 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3071 (BT_FN_VOID_UINT64_PTR): Likewise.
3072 * common.opt (flag_sanitize_coverage): New variable.
3073 (fsanitize-coverage=trace-pc): Remove.
3074 (fsanitize-coverage=): Add.
3075 * flag-types.h (enum sanitize_coverage_code): New enum.
3076 * fold-const.c (fold_range_test): Disable non-short-circuit
3077 optimization if flag_sanitize_coverage.
3078 (fold_truth_andor): Likewise.
3079 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3080 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3081 (coverage_sanitizer_opts): New array.
3082 (get_closest_sanitizer_option): Add OPTS argument, handle also
3083 OPT_fsanitize_coverage_.
3084 (parse_sanitizer_options): Adjusted to also handle
3085 OPT_fsanitize_coverage_.
3086 (common_handle_option): Add OPT_fsanitize_coverage_.
3087 * sancov.c (instrument_comparison, instrument_switch): New function.
3088 (sancov_pass): Add trace-cmp support.
3089 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3090 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3091 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3092 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3093 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3094 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3095 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3096 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3097 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3098 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3100 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3102 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3103 error. Only quit immediately if parsing is complete.
3104 (BEGIN): Initialize fatal_err and parse_done.
3105 (begin fpu, end fpu): Check number of arguments.
3106 (begin arch, end arch): Likewise.
3107 (begin cpu, end cpu): Likewise.
3108 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3109 (optalias): Likewise.
3111 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3113 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3114 * config/arm/arm-isa.h: Delete. Move definitions to ...
3115 * arm-cpus.in: ... here. Use new feature and fgroup values.
3116 * config/arm/arm.c (arm_option_override): Use lower case for feature
3118 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3119 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3120 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3122 (print_isa_bits_for): New function.
3123 (gen_isa): New function.
3124 (gen_comm_data): Use print_isa_bits_for.
3125 (define feature): New keyword.
3126 (define fgroup): New keyword.
3127 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3128 (arm-isa.h): Add rule to generate file.
3129 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3130 case for feature bit names.
3132 2017-09-06 Richard Biener <rguenther@suse.de>
3134 * tree-ssa-pre.c (NECESSARY): Remove.
3135 (create_expression_by_pieces): Do not touch pass-local flags.
3136 (insert_into_preds_of_block): Likewise.
3137 (do_pre_regular_insertion): Likewise.
3138 (eliminate_insert): Likewise.
3139 (eliminate_dom_walker::before_dom_children): Likewise.
3140 (fini_eliminate): Do not look at inserted_exprs.
3141 (mark_operand_necessary): Remove.
3142 (remove_dead_inserted_code): Replace with simple work-list
3143 algorithm based on inserted_exprs and SSA uses.
3144 (pass_pre::execute): Re-order fini_eliminate and
3145 remove_dead_inserted_code.
3147 2017-09-06 Olivier Hainque <hainque@adacore.com>
3149 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3150 for VxWorks 7. Adjust surrounding comments.
3152 2017-09-06 Richard Biener <rguenther@suse.de>
3154 * gimple-ssa-strength-reduction.c
3155 (find_candidates_dom_walker::before_dom_children): Also allow
3158 2017-09-06 Richard Biener <rguenther@suse.de>
3160 PR tree-optimization/82108
3161 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3162 for gap in the non-permutation SLP case.
3164 2017-09-06 Martin Jambor <mjambor@suse.cz>
3166 PR tree-optimization/82078
3167 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3168 add_access_to_work_queue...
3169 (build_accesses_from_assign): ...here.
3170 (propagate_all_subaccesses): Make sure racc is the group
3171 representative, if there is one.
3173 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3176 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3179 2017-09-06 Richard Biener <rguenther@suse.de>
3181 * gimple-ssa-strength-reduction.c
3182 (find_candidates_dom_walker::before_doom_children): Use a
3183 type and not a mode check.
3185 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3188 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3190 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3192 2017-09-05 Jeff Law <law@redhat.com>
3194 PR tree-optimization/64910
3195 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3196 cases where we have 3 or more operands.
3198 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3201 * omp-low.c (lower_omp_for): Recompute tree invariant if
3202 gimple_omp_for_initial/final is ADDR_EXPR.
3205 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3206 into gimple val before gimplification fo the COND_EXPR.
3208 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3210 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3211 REGION_COPY argument.
3212 (thread_through_all_blocks): Remove unused argument to
3213 duplicate_thread_path.
3215 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3216 Alan Hayward <alan.hayward@arm.com>
3217 David Sherwood <david.sherwood@arm.com>
3219 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3220 Take a scalar_mode rather than a machine_mode.
3221 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3222 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3223 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3224 (aarch64_gen_adjusted_ldpstp): Likewise.
3225 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3227 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3232 Take a scalar_int_mode instead of a machine_mode.
3233 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3234 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3235 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3236 (aarch64_simd_attr_length_rglist): Delete.
3237 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3238 a scalar_int_mode instead of a machine_mode.
3239 (aarch64_add_offset): Likewise.
3240 (aarch64_internal_mov_immediate): Likewise
3241 (aarch64_add_constant_internal): Likewise.
3242 (aarch64_add_constant): Likewise.
3243 (aarch64_movw_imm): Likewise.
3244 (aarch64_rtx_arith_op_extract_p): Likewise.
3245 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3246 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3247 Remove assert that the mode isn't a vector.
3248 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3249 (aarch64_expand_mov_immediate): Update calls after above changes.
3250 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3251 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3252 (aarch64_move_imm): Likewise.
3253 (aarch64_can_const_movi_rtx_p): Likewise.
3254 (aarch64_strip_extend): Likewise.
3255 (aarch64_extr_rtx_p): Likewise.
3256 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3257 a CONST_INT when the mode parameter is VOIDmode.
3258 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3260 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3262 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3263 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3264 (bitwise_type_for_mode): Update accordingly.
3266 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3268 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3269 * stor-layout.c (mode_for_size_tree): Likewise.
3270 (mode_for_array): Update accordingly.
3271 (layout_decl): Likewise.
3272 (compute_record_mode): Likewise. Only set the mode once.
3274 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3276 * target.def (get_mask_mode): Change return type to opt_mode.
3278 * doc/tm.texi: Regenerate.
3279 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3280 * targhooks.c (default_get_mask_mode): Likewise.
3281 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3282 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3283 targetm.get_mask_mode.
3284 * tree.c (build_truth_vector_type): Likewise.
3286 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3288 * machmode.h (mode_for_vector): Return an opt_mode.
3289 * stor-layout.c (mode_for_vector): Likewise.
3290 (mode_for_int_vector): Update accordingly.
3291 (layout_type): Likewise.
3292 * config/i386/i386.c (emit_memmov): Likewise.
3293 (ix86_expand_set_or_movmem): Likewise.
3294 (ix86_expand_vector_init): Likewise.
3295 (ix86_get_mask_mode): Likewise.
3296 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3298 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3299 * expmed.c (extract_bit_field_1): Likewise.
3300 * expr.c (expand_expr_real_2): Likewise.
3301 * optabs-query.c (can_vec_perm_p): Likewise.
3302 (can_vec_mask_load_store_p): Likewise.
3303 * optabs.c (expand_vec_perm): Likewise.
3304 * targhooks.c (default_get_mask_mode): Likewise.
3305 * tree-vect-stmts.c (vectorizable_store): Likewise.
3306 (vectorizable_load): Likewise.
3307 (get_vectype_for_scalar_type_and_size): Likewise.
3309 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3311 * machmode.h (mode_for_int_vector): New function.
3312 * stor-layout.c (mode_for_int_vector): Likewise.
3313 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3314 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3315 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3316 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3317 (s390_expand_vcond): Likewise.
3319 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3321 * machmode.h (opt_machine_mode): New type.
3322 (opt_mode<T>): Allow construction from anything that can be
3324 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3325 (mode_for_size): Return an opt_machine_mode.
3326 * stor-layout.c (mode_for_size): Likewise.
3327 (mode_for_size_tree): Update call accordingly.
3328 (bitwise_mode_for_mode): Likewise.
3329 (make_fract_type): Likewise.
3330 (make_accum_type): Likewise.
3331 * caller-save.c (replace_reg_with_saved_mem): Update call
3333 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3334 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3335 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3336 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3337 * expmed.c (extract_bit_field_1): Likewise.
3338 * reload.c (get_secondary_mem): Likewise.
3339 * varasm.c (assemble_integer): Likewise.
3340 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3343 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3345 * machmode.h (decimal_float_mode_for_size): New function.
3346 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3347 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3348 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3349 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3350 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3351 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3353 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3355 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3356 (get_builtin_sync_mode): Likewise.
3357 (expand_ifn_atomic_compare_exchange): Likewise.
3358 (expand_builtin_atomic_clear): Likewise.
3359 (expand_builtin_atomic_test_and_set): Likewise.
3360 (fold_builtin_atomic_always_lock_free): Likewise.
3361 * calls.c (compute_argument_addresses): Likewise.
3362 (emit_library_call_value_1): Likewise.
3363 (store_one_arg): Likewise.
3364 * combine.c (combine_instructions): Likewise.
3365 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3366 * config/arm/arm.c (arm_function_value): Likewise.
3367 (aapcs_allocate_return_reg): Likewise.
3368 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3369 * config/i386/i386.c (construct_container): Likewise.
3370 (ix86_gimplify_va_arg): Likewise.
3371 (ix86_expand_sse_cmp): Likewise.
3372 (emit_memmov): Likewise.
3373 (emit_memset): Likewise.
3374 (expand_small_movmem_or_setmem): Likewise.
3375 (ix86_expand_pextr): Likewise.
3376 (ix86_expand_pinsr): Likewise.
3377 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3378 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3380 * config/mips/mips.c (mips_function_value_1) Likewise.
3381 (mips_block_move_straight): Likewise.
3382 (mips_expand_ins_as_unaligned_store): Likewise.
3383 * config/powerpcspe/powerpcspe.c
3384 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3385 (rs6000_darwin64_record_arg_flush): Likewise.
3386 * config/rs6000/rs6000.c
3387 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3388 (rs6000_darwin64_record_arg_flush): Likewise.
3389 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3390 (sparc_function_value_1): Likewise.
3391 * config/spu/spu.c (adjust_operand): Likewise.
3392 (spu_emit_branch_or_set): Likewise.
3393 (arith_immediate_p): Likewise.
3394 * emit-rtl.c (gen_lowpart_common): Likewise.
3395 * expr.c (expand_expr_real_1): Likewise.
3396 * function.c (assign_parm_setup_block): Likewise.
3397 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3398 * reload1.c (alter_reg): Likewise.
3399 * stor-layout.c (mode_for_vector): Likewise.
3400 (layout_type): Likewise.
3402 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3404 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3405 (spu_convert_move): Likewise.
3406 * lower-subreg.c (resolve_simple_move): Likewise.
3408 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3411 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3412 define_insn to a define_expand.
3413 (altivec_vsum2sws_direct): New define_insn.
3414 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3416 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3418 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3421 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3426 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3427 realign stack if stack alignment needed is less than incoming
3430 2017-09-05 Marek Polacek <polacek@redhat.com>
3433 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3436 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3438 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3440 2017-09-05 Richard Biener <rguenther@suse.de>
3442 PR tree-optimization/82084
3443 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3445 2017-09-05 Richard Biener <rguenther@suse.de>
3447 PR tree-optimization/82102
3448 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3450 2017-09-05 Martin Liska <mliska@suse.cz>
3452 PR tree-optimization/82032
3453 * tree-cfg.c (generate_range_test): New function.
3454 * tree-cfg.h (generate_range_test): Declared here.
3455 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3456 (cleanup_control_expr_graph): Use it.
3457 * tree-switch-conversion.c (try_switch_expansion): Remove
3459 (emit_case_nodes): Use generate_range_test.
3461 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3464 * config/i386/i386.md (*<btsc><mode>_mask): Add
3465 TARGET_USE_BT to insn constraint.
3466 (*btr<mode>_mask): Ditto.
3468 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3470 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3471 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3473 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3476 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3477 TARGET_NEON and TARGET_IWMMXT.
3478 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3479 TARGET_NEON and TARGET_IWMMXT.
3480 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3482 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3484 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3485 (ix86_rewrite_tls_address): Ditto.
3486 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3487 (ix86_rewrite_tls_address_1): Ditto.
3488 (ix86_rewrite_tls_address): Ditto.
3489 * config/i386/predicates.md (tls_address_pattern): New predicate.
3490 * config/i386/i386.md (TLS address splitter): New splitter.
3492 2017-09-04 Richard Biener <rguenther@suse.de>
3494 PR tree-optimization/82084
3495 * fold-const.h (can_native_encode_string_p): Declare.
3496 * fold-const.c (can_native_encode_string_p): Factor out from ...
3497 (native_encode_string): ... here.
3498 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3499 vectorizing stores from constants we later cannot handle.
3501 2017-09-04 Marek Polacek <polacek@redhat.com>
3504 * doc/invoke.texi: Update -Wtautological-compare documentation.
3506 2017-09-04 Jeff Law <law@redhat.com>
3508 PR tree-optimization/64910
3509 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3510 swap the first and last operand if the last is a constant.
3512 2017-09-04 Marek Polacek <polacek@redhat.com>
3515 * convert.c (do_narrow): When sanitizing signed integer overflows,
3516 bail out for signed types.
3517 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3519 2017-09-04 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/82060
3522 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3523 Move devirtualization after stmt folding and before EH/AB/noreturn
3524 cleanup to get the stmt refs canonicalized. Use a bool instead
3525 of gimple_modified_p since that doesn't work for NOPs. Schedule
3526 NOPs generated by folding for removal.
3528 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3529 Alan Hayward <alan.hayward@arm.com>
3530 David Sherwood <david.sherwood@arm.com>
3532 * coretypes.h (pad_direction): New enum.
3533 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3534 (FUNCTION_ARG_PADDING): Likewise.
3535 * target.def (function_arg_padding): New hook.
3536 * targhooks.h (default_function_arg_padding): Declare.
3537 * targhooks.c (default_function_arg_padding): New function.
3538 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3539 (TARGET_FUNCTION_ARG_PADDING): ...this.
3540 * doc/tm.texi: Regenerate.
3541 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3542 instead of direction.
3543 (compute_argument_addresses): Likewise.
3544 (load_register_parameters): Likewise.
3545 (emit_library_call_value_1): Likewise.
3546 (store_one_arg): Use targetm.calls.function_arg_padding instead
3547 of FUNCTION_ARG_PADDING.
3548 (must_pass_in_stack_var_size_or_pad): Likewise.
3549 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3550 (emit_group_store): Likewise.
3551 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3552 instead of FUNCTION_ARG_PADDING.
3553 (emit_push_insn): Likewise, and propagate enum change throughout
3555 * function.h (direction): Delete.
3556 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3558 * function.c (assign_parm_find_stack_rtl): Likewise.
3559 (assign_parm_setup_block_p): Likewise.
3560 (assign_parm_setup_block): Likewise.
3561 (gimplify_parameters): Likewise.
3562 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3563 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3565 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3566 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3567 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3568 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3569 (aarch64_function_arg_padding): ...this new function.
3570 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3571 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3572 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3573 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3574 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3575 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3576 (arm_pad_arg_upward): Replace with...
3577 (arm_function_arg_padding): ...this new function.
3578 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3580 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3581 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3582 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3583 (ia64_hpux_function_arg_padding): Replace with...
3584 (ia64_function_arg_padding): ...this new function. Use pad_direction
3585 instead of direction. Check for TARGET_HPUX.
3586 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3587 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3588 (iq2000_function_arg_padding): New function.
3589 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3590 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3591 (mips_function_arg_padding): ...this new function.
3592 (mips_pad_reg_upward): Update accordingly.
3593 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3594 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3595 targetm.calls.function_arg_padding.
3596 (FUNCTION_ARG_PADDING): Delete.
3597 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3598 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3599 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3600 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3601 (nios2_block_reg_padding): Return pad_direction instead of direction.
3602 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3603 instead of direction.
3604 (nios2_function_arg_padding): Likewise. Make static.
3605 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3606 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3607 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3608 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3609 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3610 (pa_function_arg_padding): Make static. Return pad_direction instead
3612 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3613 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3614 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3615 instead of direction. Use targetm.calls.function_arg_padding.
3616 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3617 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3618 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3619 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3620 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3622 (function_arg_padding): Rename to...
3623 (rs6000_function_arg_padding): ...this. Make static. Return
3624 pad_direction instead of direction.
3625 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3626 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3627 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3628 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3629 instead of direction. Use targetm.calls.function_arg_padding.
3630 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3631 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3632 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3633 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3634 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3635 (function_arg_padding): Rename to...
3636 (rs6000_function_arg_padding): ...this. Make static. Return
3637 pad_direction instead of direction.
3638 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3639 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3640 * config/s390/s390.c (s390_function_arg_padding): New function.
3641 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3642 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3643 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3644 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3645 (function_arg_padding): Rename to...
3646 (sparc_function_arg_padding): ...this. Make static. Return
3647 pad_direction instead of direction.
3648 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3649 * config/spu/spu.c (spu_function_arg_padding): New function.
3650 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3651 * system.h (FUNCTION_ARG_PADDING): Poison.
3653 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3654 Alan Hayward <alan.hayward@arm.com>
3655 David Sherwood <david.sherwood@arm.com>
3657 * target.def (modes_tieable_p): New hook.
3658 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3659 (TARGET_MODES_TIEABLE_P): ...this.
3660 * doc/tm.texi.in: Regenerate.
3661 * hooks.h (hook_bool_mode_mode_true): Declare.
3662 * hooks.c (hook_bool_mode_mode_true): New function.
3663 * combine.c (subst): Use targetm.modes_tieable_p instead of
3665 * dse.c (find_shift_sequence): Likewise.
3666 * expmed.c (extract_low_bits): Likewise.
3667 * lower-subreg.c: Include target.h.
3668 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3670 * rtlanal.c (rtx_cost): Likewise.
3671 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3672 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3673 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3674 (TARGET_MODES_TIEABLE_P): Redefine.
3675 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3676 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3677 (TARGET_MODES_TIEABLE_P): Redefine.
3678 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3679 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3680 (arc_modes_tieable_p): New function.
3681 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3682 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3683 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3684 (arm_modes_tieable_p): Make static.
3685 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3686 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3687 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3688 (TARGET_MODES_TIEABLE_P): Redefine.
3689 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3690 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3691 (TARGET_MODES_TIEABLE_P): Redefine.
3692 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3693 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3694 (cr16_modes_tieable_p): New function.
3695 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3696 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3697 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3698 (TRULY_NOOP_TRUNCATION): Update comment.
3699 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3700 (TRULY_NOOP_TRUNCATION): Update comment.
3701 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3702 (frv_modes_tieable_p): New function.
3703 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3704 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3705 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3706 (TARGET_MODES_TIEABLE_P): Redefine.
3707 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3708 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3709 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3710 (TARGET_MODES_TIEABLE_P): Redefine.
3711 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3712 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3713 (ia64_modes_tieable_p): New function.
3714 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3715 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3716 (iq2000_modes_tieable_p): New function.
3717 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3718 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3719 (lm32_modes_tieable_p): New function.
3720 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3721 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3722 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3723 (TARGET_MODES_TIEABLE_P): Redefine.
3724 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3725 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3726 (m32r_modes_tieable_p): New function.
3727 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3728 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3729 (m68k_modes_tieable_p): New function.
3730 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3731 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3732 (mcore_modes_tieable_p): New function.
3733 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3734 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3736 (TARGET_MODES_TIEABLE_P): Redefine.
3737 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3738 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3739 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3740 (TARGET_MODES_TIEABLE_P): Redefine.
3741 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3742 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3743 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3744 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3745 (mn10300_modes_tieable_p): ...this and make static.
3746 (TARGET_MODES_TIEABLE_P): Redefine.
3747 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3748 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3749 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3750 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3751 (msp430_modes_tieable_p): Make static.
3752 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3753 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3754 (TARGET_MODES_TIEABLE_P): Redefine.
3755 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3756 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3757 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3758 (TARGET_MODES_TIEABLE_P): Redefine.
3759 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3760 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3761 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3762 (TARGET_MODES_TIEABLE_P): Redefine.
3763 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3764 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3765 (pdp11_modes_tieable_p): New function.
3766 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3767 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3768 (rs6000_modes_tieable_p): New function.
3769 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3770 * config/powerpcspe/powerpcspe.md: Update comment.
3771 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3772 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3773 (TARGET_MODES_TIEABLE_P): Redefine.
3774 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3775 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3776 (rl78_modes_tieable_p): New function.
3777 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3778 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3779 (rs6000_modes_tieable_p): New function.
3780 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3781 * config/rs6000/rs6000.md: Update comment.
3782 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3783 * config/rx/rx.c (rx_modes_tieable_p): New function.
3784 (TARGET_MODES_TIEABLE_P): Redefine.
3785 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3786 * config/s390/s390.c (s390_modes_tieable_p): New function.
3787 (TARGET_MODES_TIEABLE_P): Redefine.
3788 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3789 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3790 (sh_modes_tieable_p): New function.
3791 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3792 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3793 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3794 (sparc_modes_tieable_p): Make static.
3795 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3796 * config/spu/spu.c (spu_modes_tieable_p): New function.
3797 (TARGET_MODES_TIEABLE_P): Redefine.
3798 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3799 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3800 (TARGET_MODES_TIEABLE_P): Redefine.
3801 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3802 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3803 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3804 * config/v850/v850.c (v850_modes_tieable_p): New function.
3805 (TARGET_MODES_TIEABLE_P): Redefine.
3806 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3807 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3808 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3809 (visium_modes_tieable_p): New function.
3810 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3811 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3812 (xtensa_modes_tieable_p): New function.
3813 * system.h (MODES_TIEABLE_P): Poison.
3815 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3816 Alan Hayward <alan.hayward@arm.com>
3817 David Sherwood <david.sherwood@arm.com>
3819 * target.def (hard_regno_mode_ok): New hook.
3820 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3821 (TARGET_HARD_REGNO_MODE_OK): ...this.
3822 * doc/tm.texi.in: Regenerate.
3823 * hooks.h (hook_bool_uint_mode_true): Declare.
3824 * hooks.c (hook_bool_uint_mode_true): New function.
3825 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3827 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3828 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3829 instead of HARD_REGNO_MODE_OK.
3830 * caller-save.c: Include target.h.
3831 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3833 * combine.c (can_combine_p): Likewise.
3834 (combinable_i3pat): Likewise.
3835 (can_change_dest_mode): Likewise.
3836 * expr.c (init_expr_target): Likewise.
3837 (convert_move): Likewise.
3838 (convert_modes): Likewise.
3839 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3840 (setup_prohibited_mode_move_regs): Likewise.
3841 * ira.h (target_ira): Likewise.
3842 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3843 * lra-constraints.c (process_alt_operands): Likewise.
3844 (split_reg): Likewise.
3845 * recog.c (peep2_find_free_register): Likewise.
3846 * ree.c (combine_reaching_defs): Likewise.
3847 * regcprop.c (maybe_mode_change): Likewise.
3848 * reginfo.c (init_reg_sets_1): Likewise.
3849 (choose_hard_reg_mode): Likewise.
3850 (simplifiable_subregs): Likewise.
3851 * regrename.c (check_new_reg_p): Likewise.
3852 * reload.c (find_valid_class): Likewise.
3853 (find_valid_class_1): Likewise.
3854 (reload_inner_reg_of_subreg): Likewise.
3855 (push_reload): Likewise.
3856 (combine_reloads): Likewise.
3857 (find_dummy_reload): Likewise.
3858 (find_reloads): Likewise.
3859 * reload1.c (find_reg): Likewise.
3860 (set_reload_reg): Likewise.
3861 (allocate_reload_reg): Likewise.
3862 (choose_reload_regs): Likewise.
3863 (reload_adjust_reg_for_temp): Likewise.
3864 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3865 (simplify_subreg_regno): Likewise.
3866 * sel-sched.c (init_regs_for_mode): Likewise.
3867 * varasm.c (make_decl_rtl): Likewise.
3868 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3869 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3871 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3872 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3873 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3874 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3875 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3876 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3877 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3878 (arc_mode_class): Delete.
3879 (HARD_REGNO_MODE_OK): Delete.
3880 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3881 (arc_hard_regno_mode_ok): Rename old array to...
3882 (arc_hard_regno_mode_ok_modes): ...this.
3883 (arc_conditional_register_usage): Update accordingly.
3884 (arc_mode_class): Make static.
3885 (arc_hard_regno_mode_ok): New function.
3886 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3887 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3888 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3889 (arm_hard_regno_mode_ok): Make static.
3890 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3892 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3893 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3894 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3896 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3897 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3898 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3899 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3900 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3901 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3902 * config/bfin/predicates.md (valid_reg_operand): Use
3903 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3904 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3905 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3907 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3908 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3909 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3910 (cr16_hard_regno_mode_ok): Make static and return a bool.
3911 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3912 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3913 (cris_hard_regno_mode_ok): New function.
3914 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3915 (epiphany_mode_class): Delete.
3916 (HARD_REGNO_MODE_OK): Delete.
3917 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3918 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3919 (hard_regno_mode_ok): Rename to...
3920 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3921 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3922 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3924 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3925 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3926 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3927 (frv_hard_regno_mode_ok): Make static and return a bool.
3928 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3930 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3931 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3932 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3933 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3935 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3936 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3937 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3938 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3940 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3941 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3942 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3943 (ia64_hard_regno_mode_ok): New function.
3944 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3945 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3946 (iq2000_hard_regno_mode_ok): New function.
3947 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3948 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3949 (lm32_hard_regno_mode_ok): New function.
3950 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3951 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3952 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3953 instead of HARD_REGNO_MODE_OK.
3954 (m32c_hard_regno_ok): Rename to...
3955 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3956 (m32c_cannot_change_mode_class): Update accordingly.
3957 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3958 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3959 (m32r_mode_class): Delete.
3960 (HARD_REGNO_MODE_OK): Delete.
3961 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3962 (m32r_hard_regno_mode_ok): Rename to...
3963 (m32r_hard_regno_modes): ...this.
3964 (m32r_mode_class): Make static.
3965 (m32r_hard_regno_mode_ok): New function.
3966 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3967 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3968 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3969 (m68k_hard_regno_mode_ok): Make static.
3970 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3971 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3972 (mcore_hard_regno_mode_ok): New function.
3973 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3974 (HARD_REGNO_MODE_OK): Delete.
3975 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3977 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3978 (microblaze_hard_regno_mode_ok): New function.
3979 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3980 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3981 (mips_hard_regno_mode_ok): Delete.
3982 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3983 (mips_hard_regno_mode_ok_p): ...this and make static.
3984 (mips_hard_regno_mode_ok_p): Rename to...
3985 (mips_hard_regno_mode_ok_uncached): ...this.
3986 (mips_hard_regno_mode_ok): New function.
3987 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3988 of HARD_REGNO_MODE_OK.
3989 (mips_option_override): Update after above name changes.
3990 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3991 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3992 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3993 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3994 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3996 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3997 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3998 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3999 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4000 (msp430_hard_regno_mode_ok): Make static and return a bool.
4001 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4002 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4003 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4005 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4006 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4007 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4008 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4009 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4010 (PA_HARD_REGNO_MODE_OK): ...this
4011 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4012 (PA_HARD_REGNO_MODE_OK): ...this.
4013 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4014 (pa_hard_regno_mode_ok): New function.
4015 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4016 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4017 (pdp11_hard_regno_mode_ok): New function.
4018 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4019 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4021 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4023 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4024 (rs6000_hard_regno_mode_ok): Rename to...
4025 (rs6000_hard_regno_mode_ok_uncached): ...this.
4026 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4027 (rs6000_hard_regno_mode_ok): New function.
4028 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4029 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4030 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4031 (riscv_hard_regno_mode_ok): ...this and make static.
4032 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4033 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4034 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4035 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4036 (rl78_hard_regno_mode_ok): Make static and return bool.
4037 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4038 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4040 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4041 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4042 (rs6000_hard_regno_mode_ok): Rename to...
4043 (rs6000_hard_regno_mode_ok_uncached): ...this.
4044 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4045 (rs6000_hard_regno_mode_ok): New function.
4046 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4047 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4048 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4049 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4050 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4051 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4053 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4054 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4055 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4056 (sh_hard_regno_mode_ok): Make static.
4057 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4058 instead of HARD_REGNO_MODE_OK.
4059 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4060 (sparc_mode_class): Delete.
4061 (HARD_REGNO_MODE_OK): Delete.
4062 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4063 (hard_regno_mode_classes): Make static.
4064 (sparc_mode_class): Likewise.
4065 (sparc_hard_regno_mode_ok): New function.
4066 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4067 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4068 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4070 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4071 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4072 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4073 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4074 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4075 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4076 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4077 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4078 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4079 (visium_hard_regno_mode_ok): New function.
4080 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4081 instead of HARD_REGNO_MODE_OK.
4082 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4083 (HARD_REGNO_MODE_OK): Delete.
4084 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4085 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4086 (xtensa_option_override): Update accordingly.
4087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4088 (xtensa_hard_regno_mode_ok): New function.
4089 * system.h (HARD_REGNO_MODE_OK): Poison.
4091 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4092 Alan Hayward <alan.hayward@arm.com>
4093 David Sherwood <david.sherwood@arm.com>
4095 * target.def (hard_regno_call_part_clobbered): New hook.
4096 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4098 * doc/tm.texi: Regenerate.
4099 * hooks.h (hook_bool_uint_mode_false): Declare.
4100 * hooks.c (hook_bool_uint_mode_false): New function.
4101 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4102 * cselib.c (cselib_process_insn): Use
4103 targetm.hard_regno_call_part_clobbered instead of
4104 HARD_REGNO_CALL_PART_CLOBBERED.
4105 * ira-conflicts.c (ira_build_conflicts): Likewise.
4106 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4107 * lra-constraints.c (need_for_call_save_p): Likewise.
4108 * lra-lives.c: Include target.h.
4109 (check_pseudos_live_through_calls): Use
4110 targetm.hard_regno_call_part_clobbered instead of
4111 HARD_REGNO_CALL_PART_CLOBBERED.
4112 * regcprop.c: Include target.h.
4113 (copyprop_hardreg_forward_1): Use
4114 targetm.hard_regno_call_part_clobbered instead of
4115 HARD_REGNO_CALL_PART_CLOBBERED.
4116 * reginfo.c (choose_hard_reg_mode): Likewise.
4117 * regrename.c (check_new_reg_p): Likewise.
4118 * reload.c (find_equiv_reg): Likewise.
4119 * reload1.c (emit_reload_insns): Likewise.
4120 * sched-deps.c (deps_analyze_insn): Likewise.
4121 * sel-sched.c (init_regs_for_mode): Likewise.
4122 (mark_unavailable_hard_regs): Likewise.
4123 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4124 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4125 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4127 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4128 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4129 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4131 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4133 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4134 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4135 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4137 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4138 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4139 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4141 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4142 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4144 * config/powerpcspe/powerpcspe.c
4145 (rs6000_hard_regno_call_part_clobbered): New function.
4146 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4147 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4148 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4150 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4151 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4152 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4154 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4155 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4156 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4158 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4159 Alan Hayward <alan.hayward@arm.com>
4160 David Sherwood <david.sherwood@arm.com>
4162 * rtl.h (subreg_memory_offset): Declare.
4163 * emit-rtl.c (subreg_memory_offset): New function.
4164 * expmed.c (store_bit_field_1): Use it.
4165 * expr.c (undefined_operand_subword_p): Likewise.
4166 * simplify-rtx.c (simplify_subreg): Likewise.
4168 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4170 PR rtl-optimization/57448
4173 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4174 using atomic_load pattern.
4175 (expand_atomic_store): Likewise.
4177 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4180 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4181 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4182 handling. Use replace_call_with_value with NULL instead of
4183 gsi_replace, unlink_stmt_vdef and release_defs.
4185 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4188 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4190 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4193 * rtl.h (emit_library_call_value_1): Declare.
4194 (emit_library_call): Replace declaration with a series of overloads.
4195 Remove the parameter count argument.
4196 (emit_library_call_value): Likewise.
4197 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4198 with an "rtx_mode_t *".
4199 (emit_library_call_value): Delete.
4200 (emit_library_call): Likewise.
4201 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4202 (asan_emit_allocas_unpoison): Likewise.
4203 * builtins.c (expand_builtin_powi): Likewise.
4204 (expand_asan_emit_allocas_unpoison): Likewise.
4205 * cfgexpand.c (expand_main_function): Likewise.
4206 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4207 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4208 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4209 * config/arm/arm.c (arm_trampoline_init): Likewise.
4210 (arm_call_tls_get_addr): Likewise.
4211 (arm_expand_divmod_libfunc): Likewise.
4212 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4213 (smulsi3_highpart): Likewise.
4214 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4215 (c6x_expand_compare): Likewise.
4216 (c6x_expand_movmem): Likewise.
4217 * config/frv/frv.c (frv_trampoline_init): Likewise.
4218 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4219 (ix86_expand_divmod_libfunc): Likewise.
4220 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4221 (ia64_expand_compare): Likewise.
4222 (ia64_profile_hook): Likewise.
4223 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4224 (nonlocal_goto): Likewise.
4225 (restore_stack_nonlocal): Likewise.
4226 * config/m32r/m32r.c (block_move_call): Likewise.
4227 (m32r_trampoline_init): Likewise.
4228 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4229 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4230 (m68k_call_m68k_read_tp): Likewise.
4231 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4232 (microblaze_expand_divide): Likewise.
4233 * config/mips/mips.h (mips_args): Likewise.
4234 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4235 (MIPS_ICACHE_SYNC): Likewise.
4236 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4237 (nios2_trampoline_init): Likewise.
4238 * config/pa/pa.c (hppa_tls_call): Likewise.
4239 (pa_trampoline_init): Likewise.
4240 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4241 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4242 (expand_strn_compare): Likewise.
4243 (rs6000_generate_compare): Likewise.
4244 (rs6000_expand_float128_convert): Likewise.
4245 (output_profile_hook): Likewise.
4246 (rs6000_trampoline_init): Likewise.
4247 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4248 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4249 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4250 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4251 (rs6000_generate_compare): Likewise.
4252 (rs6000_expand_float128_convert): Likewise.
4253 (output_profile_hook): Likewise.
4254 (rs6000_trampoline_init): Likewise.
4255 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4256 * config/sh/sh.c (sh_trampoline_init): Likewise.
4257 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4258 (sparc_emit_float_lib_cmp): Likewise.
4259 (sparc32_initialize_trampoline): Likewise.
4260 (sparc64_initialize_trampoline): Likewise.
4261 (sparc_profile_hook): Likewise.
4262 * config/spu/spu.c (ea_load_store): Likewise.
4263 * config/spu/spu.md (floatunssidf2): Likewise.
4264 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4265 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4266 * config/visium/visium.c (expand_block_move_4): Likewise.
4267 (expand_block_move_2): Likewise.
4268 (expand_block_move_1): Likewise.
4269 (expand_block_set_4): Likewise.
4270 (expand_block_set_2): Likewise.
4271 (expand_block_set_1): Likewise.
4272 (visium_trampoline_init): Likewise.
4273 (visium_profile_hook): Likewise.
4274 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4275 (xtensa_setup_frame_addresses): Likewise.
4276 (xtensa_trampoline_init): Likewise.
4277 * except.c (sjlj_emit_function_enter): Likewise.
4278 (sjlj_emit_function_exit): Likewise.
4279 * explow.c (allocate_dynamic_stack_space): Likewise.
4280 (probe_stack_range): Likewise.
4281 * expr.c (convert_mode_scalar): Likewise.
4282 * optabs.c (expand_binop): Likewise.
4283 (expand_twoval_binop_libfunc): Likewise.
4284 (expand_unop): Likewise.
4285 (prepare_cmp_insn): Likewise.
4286 (prepare_float_lib_cmp): Likewise.
4287 (expand_float): Likewise.
4288 (expand_fix): Likewise.
4289 (expand_fixed_convert): Likewise.
4290 (maybe_emit_sync_lock_test_and_set): Likewise.
4291 (expand_atomic_compare_and_swap): Likewise.
4292 (expand_mem_thread_fence): Likewise.
4293 (expand_atomic_fetch_op): Likewise.
4295 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4297 * doc/generic.texi (OpenACC): Adjust URL.
4298 * doc/invoke.texi (C Dialect Options): Ditto.
4300 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4302 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4303 predicate for operand 1. Add (m,<S>) constraint.
4304 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4305 Prevent memory operand 1 with register operand 2.
4307 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4309 PR rtl-optimization/82024
4310 * combine.c (try_combine): If the combination result is a PARALLEL,
4311 and we only need to retain the SET in there that would be placed
4312 at I2, check that we can place that at I3 instead, before doing so.
4314 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4317 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4319 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4320 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4323 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4324 Jeff Law <law@redhat.com>
4326 * varasm.c (bss_initializer_p): Do not put constants into .bss
4327 (categorize_decl_for_section): Handle bss_initializer_p returning
4328 false when DECL_INITIAL is NULL.
4330 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4333 * config/s390/s390.c (s390_can_inline_p): New function.
4335 2017-09-01 Jeff Law <law@redhat.com>
4337 PR tree-optimization/82052
4338 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4339 Always initialize the returned slot after a hash table miss
4340 when INSERT is true.
4342 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4344 * config/s390/s390.md (mem_signal_fence): Remove.
4345 * doc/md.texi (mem_signal_fence): Remove.
4346 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4348 * target-insns.def (mem_signal_fence): Remove.
4350 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4353 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4356 * asan.c (create_odr_indicator): Strip name encoding from assembler
4357 name before appending it after __odr_asan_.
4359 2017-09-01 Martin Liska <mliska@suse.cz>
4361 PR tree-optimization/82059
4362 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4363 frequency only when an edge is redirected.
4365 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4367 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4368 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4369 (arc_conditional_register_usage): Remove ARC600 lp_count
4371 (arc_file_start): Emit Tag_ARC_CPU_variation.
4372 (arc_can_use_doloop_p): New conditions to use ZOLs.
4373 (hwloop_fail): New function.
4374 (hwloop_optimize): Likewise.
4375 (hwloop_pattern_reg): Likewise.
4376 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4377 (arc_reorg_loops): New function, calls reorg_loops.
4378 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4379 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4381 (arc_loop_hazard): Remove function, functionality moved into
4383 (arc_hazard): Remove arc_loop_hazard call.
4384 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4385 into hwloop_optimize.
4386 (arc_label_align): Remove ZOL handling.
4387 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4388 * config/arc/arc.md (doloop_begin): Remove pattern.
4389 (doloop_begin_i): Likewise.
4390 (doloop_end_i): Likewise.
4391 (doloop_fallback): Likewise.
4392 (doloop_fallback_m): Likewise.
4393 (doloop_end): Reimplement expand.
4394 (arc_lp): New pattern for LP instruction.
4395 (loop_end): New pattern.
4396 (loop_fail): Likewise.
4397 (decrement_and_branch_until_zero): Likewise.
4398 * config/arc/arc.opt (mlpc-width): New option.
4399 * doc/invoke.texi (mlpc-width): Document option.
4401 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4403 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4404 (arc_ccfsm_advance): Fix checking for delay slots.
4405 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4407 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4409 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4411 (movhi_insn): Update store instruction constraint which are saving
4412 6-bit short immediates.
4413 (movsi_insn): Consider also short scaled load operations.
4414 (zero_extendhisi2_i): Use Usd constraint instead of T.
4415 (extendhisi2_i): Add q constraint.
4416 (arc_clzsi2): Add type and length attributes.
4417 (arc_ctzsi2): Likewise.
4418 * config/arc/constraints.md (Usc): Update constraint, the
4419 assembler can parse two relocations for a single instruction.
4421 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4423 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4424 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4426 2017-08-31 Olivier Hainque <hainque@adacore.com>
4428 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4429 match as powerpc-wrs-vxworks*.
4431 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4433 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4434 register constraint for by-element operand.
4435 (aarch64_mls_elt_merge<mode>): Likewise.
4437 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4439 * config/arc/arc.c (arc_can_follow_jump): Check for short
4442 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4444 * config.gcc: Use g.opt for arc.
4445 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4446 functionality moved to ...
4447 (legitimate_scaled_address_p): New function, ...here.
4448 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4449 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4450 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4452 (arc_override_options): Handle G option.
4453 (arc_output_pic_addr_const): Correct function definition.
4454 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4455 (arc_decl_anon_ns_mem_p): Delete.
4456 (arc_in_small_data_p): Overhaul this function to take into
4457 consideration the value given via G option.
4458 (arc_rewrite_small_data_1): Renamed and corrected old
4459 arc_rewrite_small_data function.
4460 (arc_rewrite_small_data): New function.
4461 (small_data_pattern): Don't use pic_offset_table_rtx.
4462 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4463 * config/arc/simdext.md (movmisalignv2hi): Use
4464 prepare_move_operands function.
4466 (movmisalign*): Likewise.
4467 * doc/invoke.texi (ARC options): Document -G option.
4469 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4471 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4473 * config/arc/arc.c (arc_print_operand): Output scalled address for
4474 sdata whenever is possible.
4475 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4476 load/stores are available.
4477 (compact_sda_memory_operand): Check for the alignment required by
4478 code density instructions.
4479 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4481 * config/arc/constraints.md (Usd): Update constraint.
4482 (Us0): New constraint.
4483 (Usc): Update constraint.
4485 2017-08-31 Richard Biener <rguenther@suse.de>
4488 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4491 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4493 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4494 Resize type_signature.
4496 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4497 Alan Hayward <alan.hayward@arm.com>
4498 David Sherwood <david.sherwood@arm.com>
4500 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4501 subregs whose inner modes can be stored in GPRs.
4502 (aarch64_classify_index): Likewise.
4504 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4505 Alan Hayward <alan.hayward@arm.com>
4506 David Sherwood <david.sherwood@arm.com>
4508 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4509 (V_INT_EQUIV): ...this.
4510 (v_cmp_result): Rename to...
4511 (v_int_equiv): ...this.
4512 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4513 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4514 (copysign<mode>3): Likewise.
4515 (aarch64_simd_bsl<mode>_internal): Likewise.
4516 (aarch64_simd_bsl<mode>): Likewise.
4517 (vec_cmp<mode><mode>): Likewise.
4518 (vcond<mode><mode>): Likewise.
4519 (vcond<v_cmp_mixed><mode>): Likewise.
4520 (vcondu<mode><v_cmp_mixed>): Likewise.
4521 (aarch64_cm<optab><mode>): Likewise.
4522 (aarch64_cmtst<mode>): Likewise.
4523 (aarch64_fac<optab><mode>): Likewise.
4524 (vec_perm_const<mode>): Likewise.
4525 (vcond_mask_<mode><v_cmp_result>): Rename to...
4526 (vcond_mask_<mode><v_int_equiv>): ...this.
4527 (vec_cmp<mode><v_cmp_result>): Rename to...
4528 (vec_cmp<mode><v_int_equiv>): ...this.
4530 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4531 Alan Hayward <alan.hayward@arm.com>
4532 David Sherwood <david.sherwood@arm.com>
4534 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4536 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4537 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4538 (UNSPEC_LD4_DREG): New unspecs.
4539 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4540 (aarch64_ld2<mode>_dreg_be): Replace with...
4541 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4543 (aarch64_ld3<mode>_dreg_le)
4544 (aarch64_ld3<mode>_dreg_be): Replace with...
4545 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4547 (aarch64_ld4<mode>_dreg_le)
4548 (aarch64_ld4<mode>_dreg_be): Replace with...
4549 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4552 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4554 PR tree-optimization/81987
4555 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4556 insert an initializer in a location not dominated by the stride
4559 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4561 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4562 on the entire header of the finally block in the fallthru case.
4564 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4566 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4568 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4570 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4571 rs6000_emit_move_from_cr and call renamed function.
4572 (rs6000_emit_prologue): Call renamed functions.
4573 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4574 movesi_from_cr, remove volatile CRs.
4576 2017-08-30 Jon Beniston <jon@beniston.com>
4577 Richard Biener <rguenther@suse.de>
4579 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4580 of VECTOR_MODE_P check.
4581 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4582 element vector types.
4584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4586 * df.h (df_read_modify_subreg_p): Remove in favor of...
4587 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4588 const_rtx instead of an rtx.
4589 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4590 * df-problems.c (df_word_lr_mark_ref): Likewise.
4591 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4592 (mark_pseudo_reg_dead): Likewise.
4593 (mark_ref_dead): Likewise.
4594 * reginfo.c (init_subregs_of_mode): Likewise.
4595 * sched-deps.c (sched_analyze_1): Likewise.
4596 * df-scan.c (df_def_record_1): Likewise.
4597 (df_uses_record): Likewise.
4598 (df_read_modify_subreg_p): Remove in favor of...
4599 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4600 const_rtx instead of an rtx.
4602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4603 Alan Hayward <alan.hayward@arm.com>
4604 David Sherwood <david.sherwood@arm.com>
4606 * rtl.h (partial_subreg_p): New function.
4607 * caller-save.c (save_call_clobbered_regs): Use it.
4608 * calls.c (expand_call): Likewise.
4609 * combine.c (combinable_i3pat): Likewise.
4610 (simplify_set): Likewise.
4611 (make_extraction): Likewise.
4612 (make_compound_operation_int): Likewise.
4613 (gen_lowpart_or_truncate): Likewise.
4614 (force_to_mode): Likewise.
4615 (make_field_assignment): Likewise.
4616 (reg_truncated_to_mode): Likewise.
4617 (record_truncated_value): Likewise.
4618 (move_deaths): Likewise.
4619 * cse.c (record_jump_cond): Likewise.
4620 (cse_insn): Likewise.
4621 * cselib.c (cselib_lookup_1): Likewise.
4622 * expmed.c (extract_bit_field_using_extv): Likewise.
4623 * function.c (assign_parm_setup_reg): Likewise.
4624 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4625 * ira-build.c (create_insn_allocnos): Likewise.
4626 * lra-coalesce.c (merge_pseudos): Likewise.
4627 * lra-constraints.c (match_reload): Likewise.
4628 (simplify_operand_subreg): Likewise.
4629 (curr_insn_transform): Likewise.
4630 * lra-lives.c (process_bb_lives): Likewise.
4631 * lra.c (new_insn_reg): Likewise.
4632 (lra_substitute_pseudo): Likewise.
4633 * regcprop.c (mode_change_ok): Likewise.
4634 (maybe_mode_change): Likewise.
4635 (copyprop_hardreg_forward_1): Likewise.
4636 * reload.c (push_reload): Likewise.
4637 (find_reloads): Likewise.
4638 (find_reloads_subreg_address): Likewise.
4639 * reload1.c (alter_reg): Likewise.
4640 (eliminate_regs_1): Likewise.
4641 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4643 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4645 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4646 back to if statements, including unpack.
4648 2017-08-30 Martin Liska <mliska@suse.cz>
4651 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4653 (func_checker::compare_asm_inputs_outputs): ... this function.
4654 (func_checker::compare_gimple_asm): Use the function to compare
4655 also ASM constrains.
4656 * ipa-icf-gimple.h: Rename the function.
4658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4659 Alan Hayward <alan.hayward@arm.com>
4660 David Sherwood <david.sherwood@arm.com>
4662 * coretypes.h (complex_mode): New type.
4663 * gdbhooks.py (build_pretty_printer): Handle it.
4664 * machmode.h (complex_mode): New class.
4665 (complex_mode::includes_p): New function.
4666 (is_complex_int_mode): Likewise.
4667 (is_complex_float_mode): Likewise.
4668 * genmodes.c (get_mode_class): Handle complex mode classes.
4669 * function.c (expand_function_end): Use is_complex_int_mode.
4671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4672 Alan Hayward <alan.hayward@arm.com>
4673 David Sherwood <david.sherwood@arm.com>
4675 * coretypes.h (scalar_mode_pod): New typedef.
4676 * gdbhooks.py (build_pretty_printer): Handle it.
4677 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4678 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4679 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4680 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4684 Alan Hayward <alan.hayward@arm.com>
4685 David Sherwood <david.sherwood@arm.com>
4687 * machmode.h (mode_for_vector): Take a scalar_mode instead
4689 * stor-layout.c (mode_for_vector): Likewise.
4690 * explow.c (promote_mode): Use as_a <scalar_mode>.
4691 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4694 Alan Hayward <alan.hayward@arm.com>
4695 David Sherwood <david.sherwood@arm.com>
4697 * target.def (preferred_simd_mode): Take a scalar_mode
4698 instead of a machine_mode.
4699 * targhooks.h (default_preferred_simd_mode): Likewise.
4700 * targhooks.c (default_preferred_simd_mode): Likewise.
4701 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4702 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4703 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4704 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4705 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4706 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4707 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4708 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4710 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4711 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4712 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4713 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4714 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4715 * doc/tm.texi: Regenerate.
4716 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4720 Alan Hayward <alan.hayward@arm.com>
4721 David Sherwood <david.sherwood@arm.com>
4723 * target.def (scalar_mode_supported_p): Take a scalar_mode
4724 instead of a machine_mode.
4725 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4726 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4727 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4728 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4729 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4730 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4731 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4732 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4733 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4734 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4735 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4736 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4737 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4738 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4740 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4741 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4742 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4743 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4744 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4746 * doc/tm.texi: Regenerate.
4748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4749 Alan Hayward <alan.hayward@arm.com>
4750 David Sherwood <david.sherwood@arm.com>
4752 * coretypes.h (opt_scalar_mode): New typedef.
4753 * gdbhooks.py (build_pretty_printers): Handle it.
4754 * machmode.h (mode_iterator::get_2xwider): Add overload for
4756 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4758 * expr.c (convert_mode_scalar): Likewise.
4759 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4760 * optabs.c (expand_float): Likewise.
4761 (expand_fix): Likewise.
4762 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4765 Alan Hayward <alan.hayward@arm.com>
4766 David Sherwood <david.sherwood@arm.com>
4768 * optabs.c (expand_float): Explicitly check for scalars before
4769 using a branching expansion.
4770 (expand_fix): Likewise.
4772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4773 Alan Hayward <alan.hayward@arm.com>
4774 David Sherwood <david.sherwood@arm.com>
4776 * expr.c (convert_mode): Split scalar handling out into...
4777 (convert_mode_scalar): ...this new function. Treat the modes
4780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4781 Alan Hayward <alan.hayward@arm.com>
4782 David Sherwood <david.sherwood@arm.com>
4784 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4786 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4789 Alan Hayward <alan.hayward@arm.com>
4790 David Sherwood <david.sherwood@arm.com>
4792 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4793 rather than a machine_mode.
4794 (fixed_from_string): Likewise.
4795 (fixed_convert): Likewise.
4796 (fixed_convert_from_int): Likewise.
4797 (fixed_convert_from_real): Likewise.
4798 (real_convert_from_fixed): Likewise.
4799 * fixed-value.c (fixed_from_double_int): Likewise.
4800 (fixed_from_string): Likewise.
4801 (fixed_convert): Likewise.
4802 (fixed_convert_from_int): Likewise.
4803 (fixed_convert_from_real): Likewise.
4804 (real_convert_from_fixed): Likewise.
4805 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4808 Alan Hayward <alan.hayward@arm.com>
4809 David Sherwood <david.sherwood@arm.com>
4811 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4812 of separate mode class checks. Do not allow vector modes here.
4813 (immed_wide_int_const): Use as_a <scalar_mode>.
4814 * explow.c (trunc_int_for_mode): Likewise.
4815 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4816 (wi::shwi): Likewise.
4817 (wi::min_value): Likewise.
4818 (wi::max_value): Likewise.
4819 * dwarf2out.c (loc_descriptor): Likewise.
4820 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4824 Alan Hayward <alan.hayward@arm.com>
4825 David Sherwood <david.sherwood@arm.com>
4827 * tree.h (SCALAR_TYPE_MODE): New macro.
4828 * expr.c (expand_expr_addr_expr_1): Use it.
4829 (expand_expr_real_2): Likewise.
4830 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4831 (fold_convert_const_fixed_from_int): Likewise.
4832 (fold_convert_const_fixed_from_real): Likewise.
4833 (native_encode_fixed): Likewise
4834 (native_encode_complex): Likewise
4835 (native_encode_vector): Likewise.
4836 (native_interpret_fixed): Likewise.
4837 (native_interpret_real): Likewise.
4838 (native_interpret_complex): Likewise.
4839 (native_interpret_vector): Likewise.
4840 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4841 (simd_clone_adjust_argument_types): Likewise.
4842 (simd_clone_init_simd_arrays): Likewise.
4843 (simd_clone_adjust): Likewise.
4844 * stor-layout.c (layout_type): Likewise.
4845 * tree.c (build_minus_one_cst): Likewise.
4846 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4847 * tree-inline.c (estimate_move_cost): Likewise.
4848 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4849 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4850 (vectorizable_reduction): Likewise.
4851 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4852 (vect_recog_mixed_size_cond_pattern): Likewise.
4853 (check_bool_pattern): Likewise.
4854 (adjust_bool_pattern): Likewise.
4855 (search_type_for_mask_1): Likewise.
4856 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4857 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4858 (vectorizable_load): Likewise.
4859 (vectorizable_store): Likewise.
4860 * ubsan.c (ubsan_encode_value): Likewise.
4861 * varasm.c (output_constant): Likewise.
4863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4864 Alan Hayward <alan.hayward@arm.com>
4865 David Sherwood <david.sherwood@arm.com>
4867 * coretypes.h (scalar_mode): New class.
4868 * machmode.h (scalar_mode): Likewise.
4869 (scalar_mode::includes_p): New function.
4870 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4871 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4872 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4873 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4874 * expmed.c (store_bit_field_1): Likewise.
4875 (extract_bit_field_1): Likewise.
4876 * expr.c (write_complex_part): Likewise.
4877 (read_complex_part): Likewise.
4878 (emit_move_complex_push): Likewise.
4879 (expand_expr_real_2): Likewise.
4880 * function.c (assign_parm_setup_reg): Likewise.
4881 (assign_parms_unsplit_complex): Likewise.
4882 * optabs.c (expand_binop): Likewise.
4883 * rtlanal.c (subreg_get_info): Likewise.
4884 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4885 * varasm.c (output_constant_pool_2): Likewise.
4887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4888 Alan Hayward <alan.hayward@arm.com>
4889 David Sherwood <david.sherwood@arm.com>
4891 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4893 (expmed_mult_highpart_optab): Likewise.
4894 (expmed_mult_highpart): Likewise.
4896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4897 Alan Hayward <alan.hayward@arm.com>
4898 David Sherwood <david.sherwood@arm.com>
4900 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4901 instead of a machine_mode.
4902 (builtin_memset_read_str): Likewise.
4903 * builtins.c (c_readstr): Likewise.
4904 (builtin_memcpy_read_str): Likewise.
4905 (builtin_strncpy_read_str): Likewise.
4906 (builtin_memset_read_str): Likewise.
4907 (builtin_memset_gen_str): Likewise.
4908 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4909 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4910 instead of a machine_mode.
4911 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4913 (make_extraction): Likewise.
4914 (try_widen_shift_mode): Take and return scalar_int_modes instead
4916 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4917 a scalar_int_mode instead of a machine_mode.
4918 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4919 (avr_addr_space_pointer_mode): Likewise.
4920 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4921 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4922 (msp430_unwind_word_mode): Likewise.
4923 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4924 (spu_addr_space_pointer_mode): Likewise.
4925 (spu_addr_space_address_mode): Likewise.
4926 (spu_libgcc_cmp_return_mode): Likewise.
4927 (spu_libgcc_shift_count_mode): Likewise.
4928 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4929 (rl78_addr_space_pointer_mode): Likewise.
4930 (fl78_unwind_word_mode): Likewise.
4931 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4933 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4934 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4935 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4936 (mips_valid_pointer_mode): Likewise.
4937 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4938 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4939 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4941 (ft32_addr_space_address_mode): Likewise.
4942 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4943 scalar_int_mode instead of a machine_mode.
4944 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4946 (m32c_addr_space_address_mode): Likewise.
4947 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4948 (rs6000_eh_return_filter_mode): Likewise.
4949 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4950 (rs6000_eh_return_filter_mode): Likewise.
4951 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4952 (s390_libgcc_shift_count_mode): Likewise.
4953 (s390_unwind_word_mode): Likewise.
4954 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4956 * target.def (mode_rep_extended): Likewise.
4957 (valid_pointer_mode): Likewise.
4958 (addr_space.valid_pointer_mode): Likewise.
4959 (eh_return_filter_mode): Return a scalar_int_mode rather than
4961 (libgcc_cmp_return_mode): Likewise.
4962 (libgcc_shift_count_mode): Likewise.
4963 (unwind_word_mode): Likewise.
4964 (addr_space.pointer_mode): Likewise.
4965 (addr_space.address_mode): Likewise.
4966 * doc/tm.texi: Regenerate.
4967 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4969 (do_jump): Use scalar_int_mode for local variables.
4970 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4971 rather than a machine_mode.
4972 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4973 (scompare_loc_descriptor_wide): Likewise.
4974 (scompare_loc_descriptor_narrow): Likewise.
4975 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4977 * except.c (sjlj_emit_dispatch_table): Likewise.
4978 (expand_builtin_eh_copy_values): Likewise.
4979 * explow.c (convert_memory_address_addr_space_1): Likewise.
4980 Take a scalar_int_mode rather than a machine_mode.
4981 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4982 than a machine_mode.
4983 (memory_address_addr_space): Use scalar_int_mode for local variables.
4984 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4985 rather than a machine_mode.
4986 * expmed.c (mask_rtx): Likewise.
4987 (init_expmed_one_conv): Likewise.
4988 (expand_mult_highpart_adjust): Likewise.
4989 (extract_high_half): Likewise.
4990 (expmed_mult_highpart_optab): Likewise.
4991 (expmed_mult_highpart): Likewise.
4992 (expand_smod_pow2): Likewise.
4993 (expand_sdiv_pow2): Likewise.
4994 (emit_store_flag_int): Likewise.
4995 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4997 (extract_low_bits): Likewise.
4998 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5000 * expr.c (pieces_addr::adjust): Likewise.
5001 (can_store_by_pieces): Likewise.
5002 (store_by_pieces): Likewise.
5003 (clear_by_pieces_1): Likewise.
5004 (expand_expr_addr_expr_1): Likewise.
5005 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5006 (expand_expr_real_1): Likewise.
5007 (try_casesi): Likewise.
5008 * final.c (shorten_branches): Likewise.
5009 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5010 type of "mode" to machine_mode.
5011 * internal-fn.c (expand_arith_overflow_result_store): Take a
5012 scalar_int_mode rather than a machine_mode.
5013 (expand_mul_overflow): Use scalar_int_mode for local variables.
5014 * loop-doloop.c (doloop_modify): Likewise.
5015 (doloop_optimize): Likewise.
5016 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5017 than a machine_mode.
5018 (expand_doubleword_shift_condmove): Likewise.
5019 (expand_doubleword_shift): Likewise.
5020 (expand_doubleword_clz): Likewise.
5021 (expand_doubleword_popcount): Likewise.
5022 (expand_doubleword_parity): Likewise.
5023 (expand_absneg_bit): Use scalar_int_mode for local variables.
5024 (prepare_float_lib_cmp): Likewise.
5025 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5026 rather than a machine_mode.
5027 (convert_memory_address_addr_space): Likewise.
5028 (get_mode_bounds): Likewise.
5029 (get_address_mode): Return a scalar_int_mode rather than a
5031 * rtlanal.c (get_address_mode): Likewise.
5032 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5033 than a machine_mode.
5034 * targhooks.c (default_mode_rep_extended): Likewise.
5035 (default_valid_pointer_mode): Likewise.
5036 (default_addr_space_valid_pointer_mode): Likewise.
5037 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5038 than a machine_mode.
5039 (default_libgcc_cmp_return_mode): Likewise.
5040 (default_libgcc_shift_count_mode): Likewise.
5041 (default_unwind_word_mode): Likewise.
5042 (default_addr_space_pointer_mode): Likewise.
5043 (default_addr_space_address_mode): Likewise.
5044 * targhooks.h (default_eh_return_filter_mode): Likewise.
5045 (default_libgcc_cmp_return_mode): Likewise.
5046 (default_libgcc_shift_count_mode): Likewise.
5047 (default_unwind_word_mode): Likewise.
5048 (default_addr_space_pointer_mode): Likewise.
5049 (default_addr_space_address_mode): Likewise.
5050 (default_mode_rep_extended): Take a scalar_int_mode rather than
5052 (default_valid_pointer_mode): Likewise.
5053 (default_addr_space_valid_pointer_mode): Likewise.
5054 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5056 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5057 rather than a machine_mode.
5058 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5059 for local variables.
5060 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5061 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5062 than a machine_mode.
5064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5065 Alan Hayward <alan.hayward@arm.com>
5066 David Sherwood <david.sherwood@arm.com>
5068 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5069 the mode argument to scalar_int_mode.
5070 (do_jump_by_parts_zero_rtx): Likewise.
5071 (do_jump_by_parts_equality_rtx): Likewise.
5072 (do_jump_by_parts_greater): Take a mode argument.
5073 (do_jump_by_parts_equality): Likewise.
5074 (do_jump_1): Update calls accordingly.
5076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5077 Alan Hayward <alan.hayward@arm.com>
5078 David Sherwood <david.sherwood@arm.com>
5080 * is-a.h (safe_dyn_cast): New function.
5081 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5082 (jump_table_for_label): Likewise.
5083 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5084 instead of an rtx_insn *.
5085 (shorten_branches): Use dyn_cast instead of LABEL_P and
5086 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5087 rtx_jump_table_data::get_data_mode.
5088 (final_scan_insn): Likewise.
5090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5091 Alan Hayward <alan.hayward@arm.com>
5092 David Sherwood <david.sherwood@arm.com>
5094 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5095 trying to combine a full-register integer set with a subreg
5098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5099 Alan Hayward <alan.hayward@arm.com>
5100 David Sherwood <david.sherwood@arm.com>
5102 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5103 that is always either address_mode or pointer_mode.
5105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5106 Alan Hayward <alan.hayward@arm.com>
5107 David Sherwood <david.sherwood@arm.com>
5109 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5110 when the two are known to be equal.
5112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5113 Alan Hayward <alan.hayward@arm.com>
5114 David Sherwood <david.sherwood@arm.com>
5116 * simplify-rtx.c (simplify_const_unary_operation): Use
5117 is_a <scalar_int_mode> instead of checking for a nonzero
5118 precision. Forcibly convert op_mode to a scalar_int_mode
5119 in that case. More clearly differentiate the operand and
5120 result modes and use the former when deciding what the value
5121 of a count-bits operation should be. Use is_int_mode instead
5122 of checking for a MODE_INT. Remove redundant check for whether
5123 this mode has a zero precision.
5125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5126 Alan Hayward <alan.hayward@arm.com>
5127 David Sherwood <david.sherwood@arm.com>
5129 * optabs.c (widen_leading): Change the type of the mode argument
5130 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5131 (widen_bswap): Likewise.
5132 (expand_parity): Likewise.
5133 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5134 (expand_ffs): Likewise.
5135 (epand_unop): Check for scalar integer modes before calling the
5138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5139 Alan Hayward <alan.hayward@arm.com>
5140 David Sherwood <david.sherwood@arm.com>
5142 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5144 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5151 value_mode arguments. Use scalar_int_mode internally.
5152 (store_bit_field_1): Rename the new integer mode from imode
5153 to op0_mode and use it instead of GET_MODE (op0). Update calls
5154 to store_split_bit_field, store_bit_field_using_insv and
5155 store_fixed_bit_field.
5156 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5157 Use scalar_int_mode internally. Use a bit count rather than a mode
5158 when calculating the largest bit size for get_best_mode.
5159 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5160 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5161 Remove assertion that OP0 has a scalar integer mode.
5162 (store_split_bit_field): Add op0_mode and value_mode arguments.
5163 Update calls to extract_fixed_bit_field.
5164 (extract_bit_field_using_extv): Add an op0_mode argument.
5165 Use scalar_int_mode internally.
5166 (extract_bit_field_1): Rename the new integer mode from imode to
5167 op0_mode and use it instead of GET_MODE (op0). Update calls to
5168 extract_split_bit_field, extract_bit_field_using_extv and
5169 extract_fixed_bit_field.
5170 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5171 to extract_split_bit_field and extract_fixed_bit_field_1.
5172 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5173 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5175 (extract_split_bit_field): Add an op0_mode argument. Update call
5176 to extract_fixed_bit_field.
5178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5179 Alan Hayward <alan.hayward@arm.com>
5180 David Sherwood <david.sherwood@arm.com>
5182 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5183 * explow.c (hard_function_value): Likewise.
5184 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5185 convert_to_mode call outside the loop.
5186 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5187 for the mode iterator. Require the mode specified by max_pieces
5189 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5191 (copy_blkmode_to_reg): Likewise.
5192 (set_storage_via_setmem): Likewise.
5193 * optabs.c (prepare_cmp_insn): Likewise.
5194 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5195 * stor-layout.c (finish_bitfield_representative): Likewise.
5197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5198 Alan Hayward <alan.hayward@arm.com>
5199 David Sherwood <david.sherwood@arm.com>
5201 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5202 * expr.c (convert_move): Use them.
5203 (convert_modes): Likewise.
5204 (store_expr_with_bounds): Likewise.
5206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5207 Alan Hayward <alan.hayward@arm.com>
5208 David Sherwood <david.sherwood@arm.com>
5210 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5211 parameter for the mode of "x". Remove the "known_x", "known_mode"
5212 and "known_ret" arguments. Change the type of the mode argument
5214 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5215 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5216 (reg_num_sign_bit_copies_for_combine): Likewise.
5217 * rtlanal.c (nonzero_bits1): Likewise.
5218 (num_sign_bit_copies1): Likewise.
5219 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5220 (reg_num_sign_bit_copies_general): Likewise.
5221 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5222 (reg_nonzero_bits_general): Likewise.
5224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5225 Alan Hayward <alan.hayward@arm.com>
5226 David Sherwood <david.sherwood@arm.com>
5228 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5229 than in subroutines. Return 1 for non-integer modes.
5230 (cached_num_sign_bit_copies): Change the type of the mode parameter
5232 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5233 classes. Handle CONST_INT_P first and then check whether X also
5234 has a scalar integer mode. Check the same thing for inner registers
5235 of a SUBREG and for values that are being extended or truncated.
5237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5238 Alan Hayward <alan.hayward@arm.com>
5239 David Sherwood <david.sherwood@arm.com>
5241 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5242 in subroutines. Return the mode mask for non-integer modes.
5243 (cached_nonzero_bits): Change the type of the mode parameter
5245 (nonzero_bits1): Likewise. Remove early exit for other mode
5246 classes. Handle CONST_INT_P first and then check whether X
5247 also has a scalar integer mode.
5249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5250 Alan Hayward <alan.hayward@arm.com>
5251 David Sherwood <david.sherwood@arm.com>
5253 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5254 Return a scalar_int_mode and assert that the size is greater than
5256 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5257 (op_by_pieces_d::op_by_pieces_d): Likewise.
5258 (op_by_pieces_d::run): Likewise.
5259 (can_store_by_pieces): Likewise.
5261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5262 Alan Hayward <alan.hayward@arm.com>
5263 David Sherwood <david.sherwood@arm.com>
5265 * combine.c (extract_left_shift): Add a mode argument and update
5267 (make_compound_operation_int): Change the type of the mode parameter
5268 to scalar_int_mode and update the call to extract_left_shift.
5270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5271 Alan Hayward <alan.hayward@arm.com>
5272 David Sherwood <david.sherwood@arm.com>
5274 * combine.c (simplify_and_const_int): Change the type of the mode
5275 parameter to scalar_int_mode.
5276 (simplify_and_const_int_1): Likewise. Update recursive call.
5278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5279 Alan Hayward <alan.hayward@arm.com>
5280 David Sherwood <david.sherwood@arm.com>
5282 * combine.c (simplify_compare_const): Check that the mode is a
5283 scalar_int_mode (rather than VOIDmode) before testing its
5285 (simplify_comparison): Move COMPARISON_P handling out of the
5286 loop and restrict the latter part of the loop to scalar_int_modes.
5287 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5288 and when considering SUBREG_REGs. Use is_int_mode instead of
5289 checking GET_MODE_CLASS against MODE_INT.
5291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5296 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5297 shift_unit_mode and for modes involved in scalar shifts.
5299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5300 Alan Hayward <alan.hayward@arm.com>
5301 David Sherwood <david.sherwood@arm.com>
5303 * combine.c (force_int_to_mode): New function, split out from...
5304 (force_to_mode): ...here. Keep xmode up-to-date and use it
5305 instead of GET_MODE (x).
5307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5308 Alan Hayward <alan.hayward@arm.com>
5309 David Sherwood <david.sherwood@arm.com>
5311 * optabs-query.h (extraction_insn::struct_mode): Change type to
5312 opt_scalar_int_mode and update comment.
5313 (extraction_insn::field_mode): Change type to scalar_int_mode.
5314 (extraction_insn::pos_mode): Likewise.
5315 * combine.c (make_extraction): Update accordingly.
5316 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5317 (get_optab_extraction_insn): Likewise.
5318 * recog.c (simplify_while_replacing): Likewise.
5319 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5320 parameter to opt_scalar_int_mode.
5322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5323 Alan Hayward <alan.hayward@arm.com>
5324 David Sherwood <david.sherwood@arm.com>
5326 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5327 to a scalar_int_mode instead of a machine_mode.
5328 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5329 (get_best_mode): Return a boolean and use a pointer argument to store
5330 the selected mode. Replace the limit mode parameter with a bit limit.
5331 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5332 for the values returned by bit_field_mode_iterator::next_mode.
5333 (store_bit_field): Update call to get_best_mode.
5334 (store_fixed_bit_field): Likewise.
5335 (extract_fixed_bit_field): Likewise.
5336 * expr.c (optimize_bitfield_assignment_op): Likewise.
5337 * fold-const.c (optimize_bit_field_compare): Likewise.
5338 (fold_truth_andor_1): Likewise.
5339 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5340 Update for new type of m_mode.
5341 (get_best_mode): As above.
5343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5344 Alan Hayward <alan.hayward@arm.com>
5345 David Sherwood <david.sherwood@arm.com>
5347 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5348 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5349 (store_bit_field): Check is_a <scalar_int_mode> before calling
5350 strict_volatile_bitfield_p.
5351 (extract_bit_field): Likewise.
5353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5354 Alan Hayward <alan.hayward@arm.com>
5355 David Sherwood <david.sherwood@arm.com>
5357 * target.def (cstore_mode): Return a scalar_int_mode.
5358 * doc/tm.texi: Regenerate.
5359 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5360 * targhooks.h (default_cstore_mode): Likewise.
5361 * targhooks.c (default_cstore_mode): Likewise, using a forced
5363 * expmed.c (emit_cstore): Expect the target of the cstore to be
5366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5367 Alan Hayward <alan.hayward@arm.com>
5368 David Sherwood <david.sherwood@arm.com>
5370 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5372 (niter_desc): Likewise mode.
5373 (iv_analyze): Add a mode parameter.
5375 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5376 and change its type to scalar_int_mode.
5377 * loop-iv.c: Update commentary at head of file.
5378 (iv_constant): Pass the mode paraeter before the rtx it describes
5379 and change its type to scalar_int_mode. Remove VOIDmode handling.
5380 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5381 (iv_extend): Likewise.
5382 (shorten_into_mode): Likewise.
5383 (iv_add): Use scalar_int_mode.
5384 (iv_mult): Likewise.
5385 (iv_shift): Likewise.
5386 (canonicalize_iv_subregs): Likewise.
5387 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5388 it describes and change its mode to scalar_int_mode. Also change
5389 the type of the returned inner_mode to scalar_int_mode.
5390 (get_biv_step): Likewise, turning outer_mode from a pointer
5391 into a direct parameter. Update call to get_biv_step_1.
5392 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5393 iv_constant and get_biv_step.
5394 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5395 and change its type to scalar_int_mode. Don't initialise iv->mode
5396 to VOIDmode and remove later checks for its still being VOIDmode.
5397 Update calls to iv_analyze_op and iv_analyze_expr. Check
5398 is_a <scalar_int_mode> when changing the mode under consideration.
5399 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5400 Update call to iv_analyze_expr.
5401 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5402 inner register is not also a scalar_int_mode. Update call to
5404 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5405 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5406 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5407 separate mode class checks. Update calls to iv_analyze. Remove
5408 fix-up of VOIDmodes after iv_analyze_biv.
5409 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5410 don't have a scalar_int_mode. Update call to biv_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 * cfgexpand.c (convert_debug_memory_address): Use
5417 as_a <scalar_int_mode>.
5418 * combine.c (expand_compound_operation): Likewise.
5419 (make_extraction): Likewise.
5420 (change_zero_ext): Likewise.
5421 (simplify_comparison): Likewise.
5422 * cse.c (cse_insn): Likewise.
5423 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5424 (mem_loc_descriptor): Likewise.
5425 (loc_descriptor): Likewise.
5426 * expmed.c (init_expmed_one_mode): Likewise.
5427 (synth_mult): Likewise.
5428 (emit_store_flag_1): Likewise.
5429 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5430 of a comparison with size.
5431 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5432 (reduce_to_bit_field_precision): Likewise.
5433 * function.c (expand_function_end): Likewise.
5434 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5435 * loop-doloop.c (doloop_modify): Likewise.
5436 * optabs.c (expand_binop): Likewise.
5437 (expand_unop): Likewise.
5438 (expand_copysign_absneg): Likewise.
5439 (prepare_cmp_insn): Likewise.
5440 (maybe_legitimize_operand): Likewise.
5441 * recog.c (const_scalar_int_operand): Likewise.
5442 * rtlanal.c (get_address_mode): Likewise.
5443 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5444 (simplify_cond_clz_ctz): Likewise.
5445 * tree-nested.c (get_nl_goto_field): Likewise.
5446 * tree.c (build_vector_type_for_mode): Likewise.
5447 * var-tracking.c (use_narrower_mode): Likewise.
5449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5450 Alan Hayward <alan.hayward@arm.com>
5451 David Sherwood <david.sherwood@arm.com>
5453 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5454 * builtins.c (expand_builtin_signbit): Use it.
5455 * cfgexpand.c (expand_debug_expr): Likewise.
5456 * dojump.c (do_jump): Likewise.
5457 (do_compare_and_jump): Likewise.
5458 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5459 * expmed.c (make_tree): Likewise.
5460 * expr.c (expand_expr_real_2): Likewise.
5461 (expand_expr_real_1): Likewise.
5462 (try_casesi): Likewise.
5463 * fold-const-call.c (fold_const_call_ss): Likewise.
5464 * fold-const.c (unextend): Likewise.
5465 (extract_muldiv_1): Likewise.
5466 (fold_single_bit_test): Likewise.
5467 (native_encode_int): Likewise.
5468 (native_encode_string): Likewise.
5469 (native_interpret_int): Likewise.
5470 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5471 * internal-fn.c (expand_addsub_overflow): Likewise.
5472 (expand_neg_overflow): Likewise.
5473 (expand_mul_overflow): Likewise.
5474 (expand_arith_overflow): Likewise.
5475 * match.pd: Likewise.
5476 * stor-layout.c (layout_type): Likewise.
5477 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5478 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5479 * tree-ssanames.c (get_range_info): Likewise.
5480 * tree-switch-conversion.c (array_value_type) Likewise.
5481 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5482 (vect_recog_divmod_pattern): Likewise.
5483 (vect_recog_mixed_size_cond_pattern): Likewise.
5484 * tree-vrp.c (extract_range_basic): Likewise.
5485 (simplify_float_conversion_using_ranges): Likewise.
5486 * tree.c (int_fits_type_p): Likewise.
5487 * ubsan.c (instrument_bool_enum_load): Likewise.
5488 * varasm.c (mergeable_string_section): Likewise.
5489 (narrowing_initializer_constant_valid_p): Likewise.
5490 (output_constant): Likewise.
5492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5493 Alan Hayward <alan.hayward@arm.com>
5494 David Sherwood <david.sherwood@arm.com>
5496 * machmode.h (NARROWEST_INT_MODE): New macro.
5497 * expr.c (alignment_for_piecewise_move): Use it instead of
5498 GET_CLASS_NARROWEST_MODE (MODE_INT).
5499 (push_block): Likewise.
5500 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5502 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5505 Alan Hayward <alan.hayward@arm.com>
5506 David Sherwood <david.sherwood@arm.com>
5508 * postreload.c (move2add_valid_value_p): Change the type of the
5509 mode parameter to scalar_int_mode.
5510 (move2add_use_add2_insn): Add a mode parameter and use it instead
5512 (move2add_use_add3_insn): Likewise.
5513 (reload_cse_move2add): Update accordingly.
5515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5516 Alan Hayward <alan.hayward@arm.com>
5517 David Sherwood <david.sherwood@arm.com>
5519 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5521 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5522 * optabs.c (expand_unop): Likewise.
5524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5525 Alan Hayward <alan.hayward@arm.com>
5526 David Sherwood <david.sherwood@arm.com>
5528 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5529 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5530 (popcount_loc_descriptor): Likewise.
5531 (bswap_loc_descriptor): Likewise.
5532 (rotate_loc_descriptor): Likewise.
5533 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5534 calling the functions above.
5536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5537 Alan Hayward <alan.hayward@arm.com>
5538 David Sherwood <david.sherwood@arm.com>
5540 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5542 (try_combine): Likewise.
5543 (simplify_if_then_else): Likewise.
5544 * cse.c (cse_insn): Likewise.
5545 * dwarf2out.c (mem_loc_descriptor): Likewise.
5546 * emit-rtl.c (gen_lowpart_common): Likewise.
5547 * simplify-rtx.c (simplify_truncation): Likewise.
5548 (simplify_binary_operation_1): Likewise.
5549 (simplify_const_relational_operation): Likewise.
5550 (simplify_ternary_operation): Likewise.
5551 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5554 Alan Hayward <alan.hayward@arm.com>
5555 David Sherwood <david.sherwood@arm.com>
5557 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5558 * reload.c (push_reload): Likewise.
5559 (find_reloads): Likewise.
5561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5562 Alan Hayward <alan.hayward@arm.com>
5563 David Sherwood <david.sherwood@arm.com>
5565 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5566 (make_compound_operation_int): Likewise.
5567 (change_zero_ext): Likewise.
5568 * expr.c (convert_move): Likewise.
5569 (convert_modes): Likewise.
5570 * fwprop.c (forward_propagate_subreg): Likewise.
5571 * loop-iv.c (get_biv_step_1): Likewise.
5572 * optabs.c (widen_operand): Likewise.
5573 * postreload.c (move2add_valid_value_p): Likewise.
5574 * recog.c (simplify_while_replacing): Likewise.
5575 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5576 (simplify_binary_operation_1): Likewise. Remove redundant
5577 mode equality check.
5579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5580 Alan Hayward <alan.hayward@arm.com>
5581 David Sherwood <david.sherwood@arm.com>
5583 * combine.c (combine_simplify_rtx): Add checks for
5584 is_a <scalar_int_mode>.
5585 (simplify_if_then_else): Likewise.
5586 (make_field_assignment): Likewise.
5587 (simplify_comparison): Likewise.
5588 * ifcvt.c (noce_try_bitop): Likewise.
5589 * loop-invariant.c (canonicalize_address_mult): Likewise.
5590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5593 Alan Hayward <alan.hayward@arm.com>
5594 David Sherwood <david.sherwood@arm.com>
5596 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5597 is_a <scalar_int_mode> instead of != BLKmode.
5599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5600 Alan Hayward <alan.hayward@arm.com>
5601 David Sherwood <david.sherwood@arm.com>
5603 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5604 instead of != VOIDmode.
5605 * combine.c (if_then_else_cond): Likewise.
5606 (change_zero_ext): Likewise.
5607 * dwarf2out.c (mem_loc_descriptor): Likewise.
5608 (loc_descriptor): Likewise.
5609 * rtlanal.c (canonicalize_condition): Likewise.
5610 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
5617 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5620 Alan Hayward <alan.hayward@arm.com>
5621 David Sherwood <david.sherwood@arm.com>
5623 * wide-int.h (int_traits<unsigned char>) New class.
5624 (int_traits<unsigned short>) Likewise.
5625 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5626 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5628 * combine.c (set_nonzero_bits_and_sign_copies): Use
5629 is_a <scalar_int_mode>.
5630 (find_split_point): Likewise.
5631 (combine_simplify_rtx): Likewise.
5632 (simplify_logical): Likewise.
5633 (expand_compound_operation): Likewise.
5634 (expand_field_assignment): Likewise.
5635 (make_compound_operation): Likewise.
5636 (extended_count): Likewise.
5637 (change_zero_ext): Likewise.
5638 (simplify_comparison): Likewise.
5639 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5640 (ucompare_loc_descriptor): Likewise.
5641 (minmax_loc_descriptor): Likewise.
5642 (mem_loc_descriptor): Likewise.
5643 (loc_descriptor): Likewise.
5644 * expmed.c (init_expmed_one_mode): Likewise.
5645 * lra-constraints.c (lra_constraint_offset): Likewise.
5646 * optabs.c (prepare_libcall_arg): Likewise.
5647 * postreload.c (move2add_note_store): Likewise.
5648 * reload.c (operands_match_p): Likewise.
5649 * rtl.h (load_extend_op): Likewise.
5650 * rtlhooks.c (gen_lowpart_general): Likewise.
5651 * simplify-rtx.c (simplify_truncation): Likewise.
5652 (simplify_unary_operation_1): Likewise.
5653 (simplify_binary_operation_1): Likewise.
5654 (simplify_const_binary_operation): Likewise.
5655 (simplify_const_relational_operation): Likewise.
5656 (simplify_subreg): Likewise.
5657 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5658 * var-tracking.c (adjust_mems): Likewise.
5659 (prepare_call_arguments): Likewise.
5661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5662 Alan Hayward <alan.hayward@arm.com>
5663 David Sherwood <david.sherwood@arm.com>
5665 * machmode.h (is_int_mode): New fuction.
5666 * combine.c (find_split_point): Use it.
5667 (combine_simplify_rtx): Likewise.
5668 (simplify_if_then_else): Likewise.
5669 (simplify_set): Likewise.
5670 (simplify_shift_const_1): Likewise.
5671 (simplify_comparison): Likewise.
5672 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5673 * cse.c (notreg_cost): Likewise.
5674 (cse_insn): Likewise.
5675 * cselib.c (cselib_lookup_1): Likewise.
5676 * dojump.c (do_jump_1): Likewise.
5677 (do_compare_rtx_and_jump): Likewise.
5678 * dse.c (get_call_args): Likewise.
5679 * dwarf2out.c (rtl_for_decl_init): Likewise.
5680 (native_encode_initializer): Likewise.
5681 * expmed.c (emit_store_flag_1): Likewise.
5682 (emit_store_flag): Likewise.
5683 * expr.c (convert_modes): Likewise.
5684 (store_field): Likewise.
5685 (expand_expr_real_1): Likewise.
5686 * fold-const.c (fold_read_from_constant_string): Likewise.
5687 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5688 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5689 * optabs.c (expand_binop): Likewise.
5690 (expand_unop): Likewise.
5691 (expand_abs_nojump): Likewise.
5692 (expand_one_cmpl_abs_nojump): Likewise.
5693 * simplify-rtx.c (mode_signbit_p): Likewise.
5694 (val_signbit_p): Likewise.
5695 (val_signbit_known_set_p): Likewise.
5696 (val_signbit_known_clear_p): Likewise.
5697 (simplify_relational_operation_1): Likewise.
5698 * tree.c (vector_type_mode): Likewise.
5700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5701 Alan Hayward <alan.hayward@arm.com>
5702 David Sherwood <david.sherwood@arm.com>
5704 * machmode.h (smallest_mode_for_size): Fix formatting.
5705 (smallest_int_mode_for_size): New function.
5706 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5707 instead of smallest_mode_for_size.
5708 * combine.c (make_extraction): Likewise.
5709 * config/arc/arc.c (arc_expand_movmem): Likewise.
5710 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5711 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5712 * config/s390/s390.c (s390_expand_insv): Likewise.
5713 * config/sparc/sparc.c (assign_int_registers): Likewise.
5714 * config/spu/spu.c (spu_function_value): Likewise.
5715 (spu_function_arg): Likewise.
5716 * coverage.c (get_gcov_type): Likewise.
5717 (get_gcov_unsigned_t): Likewise.
5718 * dse.c (find_shift_sequence): Likewise.
5719 * expmed.c (store_bit_field_1): Likewise.
5720 * expr.c (convert_move): Likewise.
5721 (store_field): Likewise.
5722 * internal-fn.c (expand_arith_overflow): Likewise.
5723 * optabs-query.c (get_best_extraction_insn): Likewise.
5724 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5725 * stor-layout.c (layout_type): Likewise.
5726 (initialize_sizetypes): Likewise.
5727 * targhooks.c (default_get_mask_mode): Likewise.
5728 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5731 Alan Hayward <alan.hayward@arm.com>
5732 David Sherwood <david.sherwood@arm.com>
5734 * machmode.h (opt_mode::else_blk): New function.
5735 (int_mode_for_mode): Declare.
5736 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5737 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5739 * cfgexpand.c (expand_debug_expr): Likewise.
5740 * combine.c (gen_lowpart_or_truncate): Likewise.
5741 (gen_lowpart_for_combine): Likewise.
5742 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5743 * config/avr/avr.c (avr_to_int_mode): Likewise.
5744 (avr_out_plus_1): Likewise.
5745 (avr_out_plus): Likewise.
5746 (avr_out_round): Likewise.
5747 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5748 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5749 (s390_expand_vcond): Likewise.
5750 * config/spu/spu.c (spu_split_immediate): Likewise.
5751 (spu_expand_mov): Likewise.
5752 * dse.c (get_stored_val): Likewise.
5753 * expmed.c (store_bit_field_1): Likewise.
5754 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5756 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5757 (extract_low_bits): Likewise.
5758 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5759 handling rather than repeating the check.
5760 (emit_group_store): Likewise.
5761 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5762 * optabs.c (expand_absneg_bit): Likewise.
5763 (expand_copysign_absneg): Likewise.
5764 (expand_copysign_bit): Likewise.
5765 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5766 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5767 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5768 * var-tracking.c (prepare_call_arguments): Likewise.
5769 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5770 int_mode_for_mode instead of mode_for_size.
5771 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5774 Alan Hayward <alan.hayward@arm.com>
5775 David Sherwood <david.sherwood@arm.com>
5777 * machmode.h (int_mode_for_size): New function.
5778 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5779 instead of mode_for_size.
5780 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5782 * combine.c (expand_field_assignment): Use int_mode_for_size
5783 instead of mode_for_size.
5784 (make_extraction): Likewise.
5785 (simplify_shift_const_1): Likewise.
5786 (simplify_comparison): Likewise.
5787 * dojump.c (do_jump): Likewise.
5788 * dwarf2out.c (mem_loc_descriptor): Likewise.
5789 * emit-rtl.c (init_derived_machine_modes): Likewise.
5790 * expmed.c (flip_storage_order): Likewise.
5791 (convert_extracted_bit_field): Likewise.
5792 * expr.c (copy_blkmode_from_reg): Likewise.
5793 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5794 * internal-fn.c (expand_mul_overflow): Likewise.
5795 * lower-subreg.c (simple_move): Likewise.
5796 * optabs-libfuncs.c (init_optabs): Likewise.
5797 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5798 * tree.c (vector_type_mode): Likewise.
5799 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5800 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5801 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5802 * tree-vect-stmts.c (vectorizable_load): Likewise.
5803 (vectorizable_store): Likewise.
5805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5806 Alan Hayward <alan.hayward@arm.com>
5807 David Sherwood <david.sherwood@arm.com>
5809 * coretypes.h (pod_mode): New type.
5810 (scalar_int_mode_pod): New typedef.
5811 * machmode.h (pod_mode): New class.
5812 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5813 * genmodes.c (emit_mode_int_n): Update accordingly.
5814 * lower-subreg.h (target_lower_subreg): Change type to
5815 scalar_int_mode_pod.
5816 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5817 scalar_int_mode_pod.
5819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5820 Alan Hayward <alan.hayward@arm.com>
5821 David Sherwood <david.sherwood@arm.com>
5823 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5824 machine_mode to scalar_int_mode.
5825 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5826 (rs6000_option_override_internal): Remove cast to int.
5827 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5828 machine_mode to scalar_int_mode.
5829 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5830 (rs6000_option_override_internal): Remove cast to int.
5831 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5832 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5834 * config/s390/s390.c (s390_expand_builtin): Likewise.
5835 * coretypes.h (scalar_int_mode): New type.
5836 (opt_scalar_int_mode): New typedef.
5837 * machmode.h (scalar_int_mode): New class.
5838 (scalar_int_mode::includes_p): New function.
5839 (byte_mode): Change type to scalar_int_mode.
5840 (word_mode): Likewise.
5841 (ptr_mode): Likewise.
5842 * emit-rtl.c (byte_mode): Likewise.
5843 (word_mode): Likewise.
5844 (ptr_mode): Likewise.
5845 (init_derived_machine_modes): Update accordingly.
5846 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5847 and MODE_PARTIAL_INT.
5848 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5849 opt_scalar_int_mode.
5851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5852 Alan Hayward <alan.hayward@arm.com>
5853 David Sherwood <david.sherwood@arm.com>
5855 * target.def (libgcc_floating_mode_supported_p): Take a
5857 * doc/tm.texi: Regenerate.
5858 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5860 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5861 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5865 Alan Hayward <alan.hayward@arm.com>
5866 David Sherwood <david.sherwood@arm.com>
5868 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5869 * doc/tm.texi: Regenerate.
5870 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5871 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5872 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5873 * targhooks.h (default_floatn_mode): Likewise.
5874 * targhooks.c (default_floatn_mode): Likewise.
5875 * tree.c (build_common_tree_nodes): Update accordingly.
5877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5882 (mode_iterator::iterate_p): Likewise.
5883 (mode_iterator::get_wider): Likewise.
5884 * expr.c (init_expr_target): Use opt_scalar_float_mode.
5886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5887 Alan Hayward <alan.hayward@arm.com>
5888 David Sherwood <david.sherwood@arm.com>
5890 * coretypes.h (opt_scalar_float_mode): New typedef.
5891 * machmode.h (float_mode_for_size): New function.
5892 * emit-rtl.c (double_mode): Delete.
5893 (init_emit_once): Use float_mode_for_size.
5894 * stor-layout.c (layout_type): Likewise.
5895 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5898 Alan Hayward <alan.hayward@arm.com>
5899 David Sherwood <david.sherwood@arm.com>
5901 * output.h (assemble_real): Take a scalar_float_mode.
5902 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5903 * config/arm/arm.md (consttable_4): Likewise.
5904 (consttable_8): Likewise.
5905 (consttable_16): Likewise.
5906 * config/mips/mips.md (consttable_float): Likewise.
5907 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5908 * varasm.c (assemble_real): Take a scalar_float_mode.
5909 (output_constant_pool_2): Update accordingly.
5910 (output_constant): Likewise.
5912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5913 Alan Hayward <alan.hayward@arm.com>
5914 David Sherwood <david.sherwood@arm.com>
5916 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5917 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5918 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5919 (native_encode_real): Likewise.
5920 (native_interpret_real): Likewise.
5921 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5922 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5929 <scalar_float_mode>. Simplify.
5930 (gen_extend_conv_libfunc): Likewise.
5932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5933 Alan Hayward <alan.hayward@arm.com>
5934 David Sherwood <david.sherwood@arm.com>
5936 * coretypes.h (scalar_float_mode): New type.
5937 * machmode.h (mode_traits::from_int): Use machine_mode if
5938 USE_ENUM_MODES is defined.
5939 (is_a): New function.
5941 (dyn_cast): Likewise.
5942 (scalar_float_mode): New class.
5943 (scalar_float_mode::includes_p): New function.
5944 (is_float_mode): Likewise.
5945 * gdbhooks.py (MachineModePrinter): New class.
5946 (build_pretty_printer): Use it for scalar_float_mode.
5947 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5948 (format_helper::format_helper): Turn into a template.
5949 * genmodes.c (get_mode_class): New function.
5950 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5951 or machine_mode if none.
5952 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5953 as_a <scalar_float_mode>.
5954 * dwarf2out.c (mem_loc_descriptor): Likewise.
5955 (insert_float): Likewise.
5956 (add_const_value_attribute): Likewise.
5957 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5958 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5959 (expand_unop): Update accordingly.
5960 (expand_abs_nojump): Likewise.
5961 (expand_copysign_absneg): Take a scalar_float_mode.
5962 (expand_copysign_bit): Likewise.
5963 (expand_copysign): Update accordingly.
5965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5966 Alan Hayward <alan.hayward@arm.com>
5967 David Sherwood <david.sherwood@arm.com>
5969 * coretypes.h (opt_mode): New class.
5970 * machmode.h (opt_mode): Likewise.
5971 (opt_mode::else_void): New function.
5972 (opt_mode::require): Likewise.
5973 (opt_mode::exists): Likewise.
5974 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5975 (GET_MODE_2XWIDER_MODE): Likewise.
5976 (mode_iterator::get_wider): Update accordingly.
5977 (mode_iterator::get_2xwider): Likewise.
5978 (mode_iterator::get_known_wider): Likewise, turning into a template.
5979 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5980 forcing a wider mode to exist.
5981 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5982 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5983 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5984 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5985 * lower-subreg.c (init_lower_subreg): Likewise.
5986 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5987 on the final iteration.
5988 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5989 a wider mode exists before asking for a move pattern.
5990 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5991 forcing a wider mode to exist.
5992 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5993 returning false if no such mode exists.
5994 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5995 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5996 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5997 Avoid checking for a MODE_INT if we already know the mode is not a
5999 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6000 forcing a wider mode to exist.
6001 (expmed_mult_highpart_optab): Likewise.
6002 (expmed_mult_highpart): Likewise.
6003 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6005 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6006 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6007 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6008 * internal-fn.c (expand_mul_overflow): Update use of
6009 GET_MODE_2XWIDER_MODE.
6010 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6011 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6012 GET_MODE_WIDER_MODE.
6013 (convert_plusminus_to_widen): Likewise.
6014 * tree-switch-conversion.c (array_value_type): Likewise.
6015 * var-tracking.c (emit_note_insn_var_location): Likewise.
6016 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6017 Return false inside rather than outside the loop if no wider mode
6019 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6020 and GET_MODE_2XWIDER_MODE
6021 (can_compare_p): Use else_void.
6022 * gdbhooks.py (OptMachineModePrinter): New class.
6023 (build_pretty_printer): Use it for opt_mode.
6025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6026 Alan Hayward <alan.hayward@arm.com>
6027 David Sherwood <david.sherwood@arm.com>
6029 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6030 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6033 Alan Hayward <alan.hayward@arm.com>
6034 David Sherwood <david.sherwood@arm.com>
6036 * machmode.h (mode_traits): New structure.
6037 (get_narrowest_mode): New function.
6038 (mode_iterator::start): Likewise.
6039 (mode_iterator::iterate_p): Likewise.
6040 (mode_iterator::get_wider): Likewise.
6041 (mode_iterator::get_known_wider): Likewise.
6042 (mode_iterator::get_2xwider): Likewise.
6043 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6044 (FOR_EACH_MODE): Likewise.
6045 (FOR_EACH_MODE_FROM): Likewise.
6046 (FOR_EACH_MODE_UNTIL): Likewise.
6047 (FOR_EACH_WIDER_MODE): Likewise.
6048 (FOR_EACH_2XWIDER_MODE): Likewise.
6049 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6050 * combine.c (simplify_comparison): Likewise
6051 * config/i386/i386.c (type_natural_mode): Likewise.
6052 * cse.c (cse_insn): Likewise.
6053 * dse.c (find_shift_sequence): Likewise.
6054 * emit-rtl.c (init_derived_machine_modes): Likewise.
6055 (init_emit_once): Likewise.
6056 * explow.c (hard_function_value): Likewise.
6057 * expmed.c (extract_fixed_bit_field_1): Likewise.
6058 (extract_bit_field_1): Likewise.
6059 (expand_divmod): Likewise.
6060 (emit_store_flag_1): Likewise.
6061 * expr.c (init_expr_target): Likewise.
6062 (convert_move): Likewise.
6063 (alignment_for_piecewise_move): Likewise.
6064 (widest_int_mode_for_size): Likewise.
6065 (emit_block_move_via_movmem): Likewise.
6066 (copy_blkmode_to_reg): Likewise.
6067 (set_storage_via_setmem): Likewise.
6068 (compress_float_constant): Likewise.
6069 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6070 * optabs-query.c (get_best_extraction_insn): Likewise.
6071 * optabs.c (expand_binop): Likewise.
6072 (expand_twoval_unop): Likewise.
6073 (expand_twoval_binop): Likewise.
6074 (widen_leading): Likewise.
6075 (widen_bswap): Likewise.
6076 (expand_parity): Likewise.
6077 (expand_unop): Likewise.
6078 (prepare_cmp_insn): Likewise.
6079 (prepare_float_lib_cmp): Likewise.
6080 (expand_float): Likewise.
6081 (expand_fix): Likewise.
6082 (expand_sfix_optab): Likewise.
6083 * postreload.c (move2add_use_add2_insn): Likewise.
6084 * reg-stack.c (reg_to_stack): Likewise.
6085 * reginfo.c (choose_hard_reg_mode): Likewise.
6086 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6087 * stor-layout.c (mode_for_size): Likewise.
6088 (smallest_mode_for_size): Likewise.
6089 (mode_for_vector): Likewise.
6090 (finish_bitfield_representative): Likewise.
6091 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6092 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6093 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6094 * var-tracking.c (prepare_call_arguments): Likewise.
6096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6097 Alan Hayward <alan.hayward@arm.com>
6098 David Sherwood <david.sherwood@arm.com>
6100 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6101 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6102 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6103 * machmode.h (mode_size): Move earlier in file.
6104 (mode_precision): Likewise.
6105 (mode_inner): Likewise.
6106 (mode_nunits): Likewise.
6107 (mode_unit_size): Likewise.
6108 (unit_unit_precision): Likewise.
6109 (mode_wider): Likewise.
6110 (mode_2xwider): Likewise.
6111 (machine_mode): New class.
6112 (mode_to_bytes): New function.
6113 (mode_to_bits): Likewise.
6114 (mode_to_precision): Likewise.
6115 (mode_to_inner): Likewise.
6116 (mode_to_unit_size): Likewise.
6117 (mode_to_unit_precision): Likewise.
6118 (mode_to_nunits): Likewise.
6119 (GET_MODE_SIZE): Use mode_to_bytes.
6120 (GET_MODE_BITSIZE): Use mode_to_bits.
6121 (GET_MODE_PRECISION): Use mode_to_precision.
6122 (GET_MODE_INNER): Use mode_to_inner.
6123 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6124 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6125 (GET_MODE_NUNITS): Use mode_to_nunits.
6126 * system.h (ALWAYS_INLINE): New macro.
6127 * config/powerpcspe/powerpcspe-c.c
6128 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6129 int for arg1_mode and arg2_mode.
6131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6132 Alan Hayward <alan.hayward@arm.com>
6133 David Sherwood <david.sherwood@arm.com>
6135 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6136 Prefix mode names with E_ in case statements.
6137 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6138 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6139 (aarch64_split_simd_move): Likewise.
6140 (aarch64_gen_storewb_pair): Likewise.
6141 (aarch64_gen_loadwb_pair): Likewise.
6142 (aarch64_gen_store_pair): Likewise.
6143 (aarch64_gen_load_pair): Likewise.
6144 (aarch64_get_condition_code_1): Likewise.
6145 (aarch64_constant_pool_reload_icode): Likewise.
6146 (get_rsqrte_type): Likewise.
6147 (get_rsqrts_type): Likewise.
6148 (get_recpe_type): Likewise.
6149 (get_recps_type): Likewise.
6150 (aarch64_gimplify_va_arg_expr): Likewise.
6151 (aarch64_simd_container_mode): Likewise.
6152 (aarch64_emit_load_exclusive): Likewise.
6153 (aarch64_emit_store_exclusive): Likewise.
6154 (aarch64_expand_compare_and_swap): Likewise.
6155 (aarch64_gen_atomic_cas): Likewise.
6156 (aarch64_emit_bic): Likewise.
6157 (aarch64_emit_atomic_swap): Likewise.
6158 (aarch64_emit_atomic_load_op): Likewise.
6159 (aarch64_evpc_trn): Likewise.
6160 (aarch64_evpc_uzp): Likewise.
6161 (aarch64_evpc_zip): Likewise.
6162 (aarch64_evpc_ext): Likewise.
6163 (aarch64_evpc_rev): Likewise.
6164 (aarch64_evpc_dup): Likewise.
6165 (aarch64_gen_ccmp_first): Likewise.
6166 (aarch64_gen_ccmp_next): Likewise.
6167 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6168 (alpha_emit_xfloating_libcall): Likewise.
6169 (emit_insxl): Likewise.
6170 (alpha_arg_type): Likewise.
6171 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6172 (arc_preferred_simd_mode): Likewise.
6173 (arc_secondary_reload): Likewise.
6174 (get_arc_condition_code): Likewise.
6175 (arc_print_operand): Likewise.
6176 (arc_legitimate_constant_p): Likewise.
6177 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6178 * config/arc/arc.md (casesi_load): Likewise.
6179 (casesi_compact_jump): Likewise.
6180 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6181 (cc_use_register): Likewise.
6182 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6183 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6184 (arm_init_iwmmxt_builtins): Likewise.
6185 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6186 (neon_expand_vector_init): Likewise.
6187 (arm_attr_length_move_neon): Likewise.
6188 (maybe_get_arm_condition_code): Likewise.
6189 (arm_emit_vector_const): Likewise.
6190 (arm_preferred_simd_mode): Likewise.
6191 (arm_output_iwmmxt_tinsr): Likewise.
6192 (thumb1_output_casesi): Likewise.
6193 (thumb2_output_casesi): Likewise.
6194 (arm_emit_load_exclusive): Likewise.
6195 (arm_emit_store_exclusive): Likewise.
6196 (arm_expand_compare_and_swap): Likewise.
6197 (arm_evpc_neon_vuzp): Likewise.
6198 (arm_evpc_neon_vzip): Likewise.
6199 (arm_evpc_neon_vrev): Likewise.
6200 (arm_evpc_neon_vtrn): Likewise.
6201 (arm_evpc_neon_vext): Likewise.
6202 (arm_validize_comparison): Likewise.
6203 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6204 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6205 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6206 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6207 (c6x_preferred_simd_mode): Likewise.
6208 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6209 (epiphany_rtx_costs): Likewise.
6210 * config/epiphany/predicates.md (proper_comparison_operator):
6212 * config/frv/frv.c (condexec_memory_operand): Likewise.
6213 (frv_emit_move): Likewise.
6214 (output_move_single): Likewise.
6215 (output_condmove_single): Likewise.
6216 (frv_hard_regno_mode_ok): Likewise.
6217 (frv_matching_accg_mode): Likewise.
6218 * config/h8300/h8300.c (split_adds_subs): Likewise.
6219 (h8300_rtx_costs): Likewise.
6220 (h8300_print_operand): Likewise.
6221 (compute_mov_length): Likewise.
6222 (output_logical_op): Likewise.
6223 (compute_logical_op_length): Likewise.
6224 (compute_logical_op_cc): Likewise.
6225 (h8300_shift_needs_scratch_p): Likewise.
6226 (output_a_shift): Likewise.
6227 (compute_a_shift_length): Likewise.
6228 (compute_a_shift_cc): Likewise.
6229 (expand_a_rotate): Likewise.
6230 (output_a_rotate): Likewise.
6231 * config/i386/i386.c (classify_argument): Likewise.
6232 (function_arg_advance_32): Likewise.
6233 (function_arg_32): Likewise.
6234 (function_arg_64): Likewise.
6235 (function_value_64): Likewise.
6236 (ix86_gimplify_va_arg): Likewise.
6237 (ix86_legitimate_constant_p): Likewise.
6238 (put_condition_code): Likewise.
6239 (split_double_mode): Likewise.
6240 (ix86_avx256_split_vector_move_misalign): Likewise.
6241 (ix86_expand_vector_logical_operator): Likewise.
6242 (ix86_split_idivmod): Likewise.
6243 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6244 (ix86_build_const_vector): Likewise.
6245 (ix86_build_signbit_mask): Likewise.
6246 (ix86_match_ccmode): Likewise.
6247 (ix86_cc_modes_compatible): Likewise.
6248 (ix86_expand_branch): Likewise.
6249 (ix86_expand_sse_cmp): Likewise.
6250 (ix86_expand_sse_movcc): Likewise.
6251 (ix86_expand_int_sse_cmp): Likewise.
6252 (ix86_expand_vec_perm_vpermi2): Likewise.
6253 (ix86_expand_vec_perm): Likewise.
6254 (ix86_expand_sse_unpack): Likewise.
6255 (ix86_expand_int_addcc): Likewise.
6256 (ix86_split_to_parts): Likewise.
6257 (ix86_vectorize_builtin_gather): Likewise.
6258 (ix86_vectorize_builtin_scatter): Likewise.
6259 (avx_vpermilp_parallel): Likewise.
6260 (inline_memory_move_cost): Likewise.
6261 (ix86_tieable_integer_mode_p): Likewise.
6262 (x86_maybe_negate_const_int): Likewise.
6263 (ix86_expand_vector_init_duplicate): Likewise.
6264 (ix86_expand_vector_init_one_nonzero): Likewise.
6265 (ix86_expand_vector_init_one_var): Likewise.
6266 (ix86_expand_vector_init_concat): Likewise.
6267 (ix86_expand_vector_init_interleave): Likewise.
6268 (ix86_expand_vector_init_general): Likewise.
6269 (ix86_expand_vector_set): Likewise.
6270 (ix86_expand_vector_extract): Likewise.
6271 (emit_reduc_half): Likewise.
6272 (ix86_emit_i387_round): Likewise.
6273 (ix86_mangle_type): Likewise.
6274 (ix86_expand_round_sse4): Likewise.
6275 (expand_vec_perm_blend): Likewise.
6276 (canonicalize_vector_int_perm): Likewise.
6277 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6278 (expand_vec_perm_1): Likewise.
6279 (expand_vec_perm_interleave3): Likewise.
6280 (expand_vec_perm_even_odd_pack): Likewise.
6281 (expand_vec_perm_even_odd_1): Likewise.
6282 (expand_vec_perm_broadcast_1): Likewise.
6283 (ix86_vectorize_vec_perm_const_ok): Likewise.
6284 (ix86_expand_vecop_qihi): Likewise.
6285 (ix86_expand_mul_widen_hilo): Likewise.
6286 (ix86_expand_sse2_abs): Likewise.
6287 (ix86_expand_pextr): Likewise.
6288 (ix86_expand_pinsr): Likewise.
6289 (ix86_preferred_simd_mode): Likewise.
6290 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6291 * config/i386/sse.md (*andnot<mode>3): Likewise.
6292 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6293 (*<code><mode>3): Likewise.
6294 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6295 (ia64_expand_atomic_op): Likewise.
6296 (ia64_arg_type): Likewise.
6297 (ia64_mode_to_int): Likewise.
6298 (ia64_scalar_mode_supported_p): Likewise.
6299 (ia64_vector_mode_supported_p): Likewise.
6300 (expand_vec_perm_broadcast): Likewise.
6301 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6302 (iq2000_function_arg_advance): Likewise.
6303 (iq2000_function_arg): Likewise.
6304 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6305 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6306 (m68k_libcall_value): Likewise.
6307 (m68k_function_value): Likewise.
6308 (sched_attr_op_type): Likewise.
6309 * config/mcore/mcore.c (mcore_output_move): Likewise.
6310 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6312 (microblaze_function_arg): Likewise.
6313 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6314 (mips_print_operand): Likewise.
6315 (mips_mode_ok_for_mov_fmt_p): Likewise.
6316 (mips_vector_mode_supported_p): Likewise.
6317 (mips_preferred_simd_mode): Likewise.
6318 (mips_expand_vpc_loongson_even_odd): Likewise.
6319 (mips_expand_vec_unpack): Likewise.
6320 (mips_expand_vi_broadcast): Likewise.
6321 (mips_expand_vector_init): Likewise.
6322 (mips_expand_vec_reduc): Likewise.
6323 (mips_expand_msa_cmp): Likewise.
6324 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6325 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6326 (cc_flags_for_mode): Likewise.
6327 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6328 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6329 (nds32_output_casesi_pc_relative): Likewise.
6330 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6331 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6332 (nvptx_gen_unpack): Likewise.
6333 (nvptx_gen_pack): Likewise.
6334 (nvptx_gen_shuffle): Likewise.
6335 (nvptx_gen_wcast): Likewise.
6336 (nvptx_preferred_simd_mode): Likewise.
6337 * config/pa/pa.c (pa_secondary_reload): Likewise.
6338 * config/pa/predicates.md (base14_operand): Likewise.
6339 * config/powerpcspe/powerpcspe-c.c
6340 (altivec_resolve_overloaded_builtin): Likewise.
6341 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6343 (rs6000_preferred_simd_mode): Likewise.
6344 (output_vec_const_move): Likewise.
6345 (rs6000_expand_vector_extract): Likewise.
6346 (rs6000_split_vec_extract_var): Likewise.
6347 (reg_offset_addressing_ok_p): Likewise.
6348 (rs6000_legitimate_offset_address_p): Likewise.
6349 (rs6000_legitimize_address): Likewise.
6350 (rs6000_emit_set_const): Likewise.
6351 (rs6000_const_vec): Likewise.
6352 (rs6000_emit_move): Likewise.
6353 (spe_build_register_parallel): Likewise.
6354 (rs6000_darwin64_record_arg_recurse): Likewise.
6355 (swap_selector_for_mode): Likewise.
6356 (spe_init_builtins): Likewise.
6357 (paired_init_builtins): Likewise.
6358 (altivec_init_builtins): Likewise.
6359 (do_load_for_compare): Likewise.
6360 (rs6000_generate_compare): Likewise.
6361 (rs6000_expand_float128_convert): Likewise.
6362 (emit_load_locked): Likewise.
6363 (emit_store_conditional): Likewise.
6364 (rs6000_output_function_epilogue): Likewise.
6365 (rs6000_handle_altivec_attribute): Likewise.
6366 (rs6000_function_value): Likewise.
6367 (emit_fusion_gpr_load): Likewise.
6368 (emit_fusion_p9_load): Likewise.
6369 (emit_fusion_p9_store): Likewise.
6370 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6371 (fusion_gpr_mem_load): Likewise.
6372 (fusion_addis_mem_combo_load): Likewise.
6373 (fusion_addis_mem_combo_store): Likewise.
6374 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6375 (fusion_gpr_mem_load): Likewise.
6376 (fusion_addis_mem_combo_load): Likewise.
6377 (fusion_addis_mem_combo_store): Likewise.
6378 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6380 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6381 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6382 (rs6000_preferred_simd_mode): Likewise.
6383 (output_vec_const_move): Likewise.
6384 (rs6000_expand_vector_extract): Likewise.
6385 (rs6000_split_vec_extract_var): Likewise.
6386 (reg_offset_addressing_ok_p): Likewise.
6387 (rs6000_legitimate_offset_address_p): Likewise.
6388 (rs6000_legitimize_address): Likewise.
6389 (rs6000_emit_set_const): Likewise.
6390 (rs6000_const_vec): Likewise.
6391 (rs6000_emit_move): Likewise.
6392 (rs6000_darwin64_record_arg_recurse): Likewise.
6393 (swap_selector_for_mode): Likewise.
6394 (paired_init_builtins): Likewise.
6395 (altivec_init_builtins): Likewise.
6396 (rs6000_expand_float128_convert): Likewise.
6397 (emit_load_locked): Likewise.
6398 (emit_store_conditional): Likewise.
6399 (rs6000_output_function_epilogue): Likewise.
6400 (rs6000_handle_altivec_attribute): Likewise.
6401 (rs6000_function_value): Likewise.
6402 (emit_fusion_gpr_load): Likewise.
6403 (emit_fusion_p9_load): Likewise.
6404 (emit_fusion_p9_store): Likewise.
6405 * config/rx/rx.c (rx_gen_move_template): Likewise.
6406 (flags_from_mode): Likewise.
6407 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6408 (s390_slb_comparison): Likewise.
6409 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6410 (s390_vector_mode_supported_p): Likewise.
6411 (s390_cc_modes_compatible): Likewise.
6412 (s390_match_ccmode_set): Likewise.
6413 (s390_canonicalize_comparison): Likewise.
6414 (s390_emit_compare_and_swap): Likewise.
6415 (s390_branch_condition_mask): Likewise.
6416 (s390_rtx_costs): Likewise.
6417 (s390_secondary_reload): Likewise.
6418 (__SECONDARY_RELOAD_CASE): Likewise.
6419 (s390_expand_cs): Likewise.
6420 (s390_preferred_simd_mode): Likewise.
6421 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6422 * config/sh/sh.c (sh_print_operand): Likewise.
6423 (dump_table): Likewise.
6424 (sh_secondary_reload): Likewise.
6425 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6426 * config/sh/sh.md (casesi_worker_1): Likewise.
6427 (casesi_worker_2): Likewise.
6428 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6429 (fcc_comparison_operator): Likewise.
6430 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6431 (emit_soft_tfmode_cvt): Likewise.
6432 (sparc_preferred_simd_mode): Likewise.
6433 (output_cbranch): Likewise.
6434 (sparc_print_operand): Likewise.
6435 (sparc_expand_vec_perm_bmask): Likewise.
6436 (vector_init_bshuffle): Likewise.
6437 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6438 (spu_vector_mode_supported_p): Likewise.
6439 (spu_expand_insv): Likewise.
6440 (spu_emit_branch_or_set): Likewise.
6441 (spu_handle_vector_attribute): Likewise.
6442 (spu_builtin_splats): Likewise.
6443 (spu_builtin_extract): Likewise.
6444 (spu_builtin_promote): Likewise.
6445 (spu_expand_sign_extend): Likewise.
6446 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6447 (tilegx_simd_int): Likewise.
6448 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6449 (tilepro_simd_int): Likewise.
6450 * config/v850/v850.c (const_double_split): Likewise.
6451 (v850_print_operand): Likewise.
6452 (ep_memory_offset): Likewise.
6453 * config/vax/vax.c (vax_rtx_costs): Likewise.
6454 (vax_output_int_move): Likewise.
6455 (vax_output_int_add): Likewise.
6456 (vax_output_int_subtract): Likewise.
6457 * config/visium/predicates.md (visium_branch_operator): Likewise.
6458 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6459 (visium_print_operand_address): Likewise.
6460 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6461 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6462 (xtensa_expand_conditional_branch): Likewise.
6463 (xtensa_copy_incoming_a7): Likewise.
6464 (xtensa_output_literal): Likewise.
6465 * dfp.c (decimal_real_maxval): Likewise.
6466 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6469 Alan Hayward <alan.hayward@arm.com>
6470 David Sherwood <david.sherwood@arm.com>
6472 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6473 (mode_nunits_inline): Likewise.
6474 (mode_inner_inline): Likewise.
6475 (mode_unit_size_inline): Likewise.
6476 (mode_unit_precision_inline): Likewise.
6477 (emit_insn_modes_h): Likewise. Also emit a #define of the
6479 (emit_mode_wider): Add an E_ prefix to mode names.
6480 (emit_mode_complex): Likewise.
6481 (emit_mode_inner): Likewise.
6482 (emit_mode_adjustments): Likewise.
6483 (emit_mode_int_n): Likewise.
6484 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6485 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6486 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6487 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6488 (CRC32_BUILTIN, ENTRY): Likewise.
6489 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6490 (aarch64_pop_regs): Likewise.
6491 (aarch64_process_components): Likewise.
6492 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6493 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6494 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6495 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6496 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6497 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6499 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6500 (pcmpistr): Likewise.
6501 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6502 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6503 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6505 * config/rl78/rl78.c (mduc_regs): Likewise.
6506 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6507 (htm_expand_builtin): Likewise.
6508 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6509 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6510 * config/xtensa/xtensa.c (print_operand): Likewise.
6511 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6512 (NUM_MODE_VECTOR_INT): Likewise.
6513 * genoutput.c (null_operand): Likewise.
6514 (output_operand_data): Likewise.
6515 * genrecog.c (print_parameter_value): Likewise.
6516 * lra.c (debug_operand_data): Likewise.
6518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6520 * dwarf2out.c (scompare_loc_descriptor_wide)
6521 (scompare_loc_descriptor_narrow): New functions, split out from...
6522 (scompare_loc_descriptor): ...here.
6523 * expmed.c (emit_store_flag_int): New function, split out from...
6524 (emit_store_flag): ...here.
6526 2017-08-30 Richard Biener <rguenther@suse.de>
6528 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6529 (dwarf2out_early_finish): Move setting of AT_pubnames from
6530 early debug output to early finish.
6532 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6534 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6535 and -mdata-region to the assembler.
6537 2017-08-30 Richard Biener <rguenther@suse.de>
6539 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6541 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6542 (dwarf2out_early_global_decl): Only generate a DIE for the
6543 abstract origin if it doesn't already exist or is a declaration DIE.
6544 (resolve_addr): Do not add the linkage name twice when
6545 generating a stub DIE for the DW_TAG_GNU_call_site target.
6547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6549 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6550 Use machine_mode rather than int for arg1_mode.
6552 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6555 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6556 that the second argument of the built-in functions to unpack
6557 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6558 switch statement instead a lot of if statements.
6559 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6560 Allow 64-bit values to be in Altivec registers as well as
6561 traditional floating point registers.
6562 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6564 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6566 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6567 MAX_REGS_PER_ADDRESS == 1.
6569 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6571 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6572 * config/i386/i386.c (ix86_option_override_internal): Simplify
6573 setting of opts->x_flag_entry.
6575 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6576 Jakub Jelinek <jakub@redhat.com>
6577 Richard Biener <rguenther@suse.de>
6579 PR tree-optimization/81503
6580 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6581 folded constant fits in the target type; reorder tests for clarity.
6583 2017-08-29 Martin Liska <mliska@suse.cz>
6585 * passes.def: Include pass_lower_switch.
6586 * stmt.c (dump_case_nodes): Remove and move to
6587 tree-switch-conversion.
6588 (case_values_threshold): Likewise.
6589 (expand_switch_as_decision_tree_p): Likewise.
6590 (emit_case_decision_tree): Likewise.
6591 (expand_case): Likewise.
6592 (balance_case_nodes): Likewise.
6593 (node_has_low_bound): Likewise.
6594 (node_has_high_bound): Likewise.
6595 (node_is_bounded): Likewise.
6596 (emit_case_nodes): Likewise.
6597 (struct simple_case_node): New struct.
6598 (add_case_node): Remove.
6599 (emit_case_dispatch_table): Use vector instead of case_list.
6600 (reset_out_edges_aux): Remove.
6601 (compute_cases_per_edge): Likewise.
6602 (expand_case): Build list of simple_case_node.
6603 (expand_sjlj_dispatch_table): Use it.
6604 * tree-switch-conversion.c (struct case_node): Moved from
6605 stmt.c and adjusted.
6606 (emit_case_nodes): Likewise.
6607 (node_has_low_bound): Likewise.
6608 (node_has_high_bound): Likewise.
6609 (node_is_bounded): Likewise.
6610 (case_values_threshold): Likewise.
6611 (reset_out_edges_aux): Likewise.
6612 (compute_cases_per_edge): Likewise.
6613 (add_case_node): Likewise.
6614 (dump_case_nodes): Likewise.
6615 (balance_case_nodes): Likewise.
6616 (expand_switch_as_decision_tree_p): Likewise.
6617 (emit_jump): Likewise.
6618 (emit_case_decision_tree): Likewise.
6619 (try_switch_expansion): Likewise.
6620 (do_jump_if_equal): Likewise.
6621 (emit_cmp_and_jump_insns): Likewise.
6622 (fix_phi_operands_for_edge): New function.
6623 (record_phi_operand_mapping): Likewise.
6624 (class pass_lower_switch): New pass.
6625 (pass_lower_switch::execute): New function.
6626 (make_pass_lower_switch): Likewise.
6627 (conditional_probability):
6628 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6629 * tree-pass.h: Add make_pass_lower_switch.
6631 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6634 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6637 2017-08-29 Richard Biener <rguenther@suse.de>
6639 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6640 we do not add a DW_AT_inline attribute twice.
6641 (gen_subprogram_die): Remove code setting DW_AT_inline on
6642 DECL_ABSTRACT_P nodes.
6644 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6646 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6647 calls to internal functions.
6648 (gimplify_modify_expr): Likewise.
6649 * tree-call-cdce.c (use_internal_fn): Likewise.
6650 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6651 (convert_to_divmod): Set the nothrow flag.
6652 * tree-if-conv.c (predicate_mem_writes): Likewise.
6653 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6654 (vectorizable_call): Likewise.
6655 (vectorizable_store): Likewise.
6656 (vectorizable_load): Likewise.
6657 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6658 (vect_recog_mask_conversion_pattern): Likewise.
6660 2017-08-29 Martin Liska <mliska@suse.cz>
6663 * gcc.c (driver_handle_option): Add new argument.
6664 * opts-common.c (handle_option): Pass
6665 target_option_override_hook.
6666 * opts-global.c (lang_handle_option): Add new option.
6667 (set_default_handlers): Add new argument.
6668 (decode_options): Likewise.
6669 * opts.c (target_handle_option): Likewise.
6670 (common_handle_option): Call target_option_override_hook.
6671 * opts.h (struct cl_option_handler_func): Add hook for
6672 target option override.
6673 (struct cl_option_handlers): Likewise.
6674 (set_default_handlers): Add new argument.
6675 (decode_options): Likewise.
6676 (common_handle_option): Likewise.
6677 (target_handle_option): Likewise.
6678 * toplev.c (toplev::main): Pass targetm.target_option.override
6681 2017-08-29 Richard Biener <rguenther@suse.de>
6682 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6684 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6685 life to the active subtree.
6687 2017-08-28 Jeff Law <law@redhat.com>
6689 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6690 derive_equivalences.
6691 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6693 (edge_info::derive_equivalences): New private member function
6695 * tree-ssa-dom.c (class edge_info): Changed from a struct
6696 to a class. Add ctor/dtor, methods and data members.
6697 (edge_info::edge_info): Renamed from allocate_edge_info.
6698 Initialize additional members.
6699 (edge_info::~edge_info): New.
6700 (free_dom_edge_info): Delete the edge info.
6701 (record_edge_info): Use new class & associated member functions.
6702 Tighten forms for testing for edge equivalences.
6703 (record_temporary_equivalences): Iterate over the simple
6704 equivalences rather than assuming there's only one per edge.
6705 (cprop_into_successor_phis): Iterate over the simple
6706 equivalences rather than assuming there's only one per edge.
6707 (optimize_stmt): Use operand_equal_p rather than pointer
6708 equality for mini-DSE code.
6710 2017-08-28 Nathan Sidwell <nathan@acm.org>
6712 * gcc.c (execute): Fold SIGPIPE handling into switch
6713 statement. Adjust internal error message.
6715 2017-08-28 Richard Biener <rguenther@suse.de>
6718 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6719 Do nothing for removed DIEs.
6721 2017-08-28 Richard Biener <rguenther@suse.de>
6723 PR tree-optimization/81977
6724 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6727 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6730 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6731 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6732 using targetm.gen_mem_thread_fence.
6734 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6737 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6738 predicate to register_operand. Reorder operands.
6739 (*btr<mode>): Ditto.
6740 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6741 (*btr<mode>_mask): Ditto.
6743 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6745 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6746 * config/rs6000/xmmintrin.h: New file.
6747 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6749 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6752 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6753 parameter and_insn and return it.
6754 (recombine_lvx_pattern): Insert a copy to ensure availability of
6755 the base register of the copied masking operation at the point of
6756 the instruction replacement.
6757 (recombine_stvx_pattern): Likewise.
6759 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6761 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6762 undocumented switches.
6763 (-mpower9-dform-vector): Likewise.
6764 (-mpower9-dform): Likewise.
6765 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6766 comments to delete references to -mpower9-dform* switches.
6767 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6768 Delete reference to -mpower9-dform* switches, test for
6769 -mpower9-vector instead.
6770 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6771 (OTHER_P9_VECTOR_MASKS): Likewise.
6772 (POWERPC_MASKS): Likewise.
6773 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6774 tests against -mpower9-dform* to -mpower9-vector. Delete code
6775 that checked for -mpower9-dform* consistancy with other options.
6776 Add test for -mpower9-misc to enable other power9 switches.
6777 (rs6000_init_hard_regno_mode_ok): Likewise.
6778 (rs6000_option_override_internal): Likewise.
6779 (rs6000_emit_prologue): Likewise.
6780 (rs6000_emit_epilogue): Likewise.
6781 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6782 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6783 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6785 (emit_fusion_p9_store): Likewise.
6786 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6787 resetting these macros if the assembler does not support ISA 3.0
6789 (TARGET_P9_DFORM_VECTOR): Likewise.
6790 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6791 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6793 2017-08-25 Alan Modra <amodra@gmail.com>
6795 PR rtl-optimization/81747
6796 * cse.c (cse_extended_basic_block): Don't attempt to record
6797 equivalences for degenerate conditional branches that branch
6798 to their fall-through.
6800 2017-08-24 Martin Sebor <msebor@redhat.com>
6803 * gimple-fold.c (size_must_be_zero_p): New function.
6804 (gimple_fold_builtin_memory_op): Call it.
6806 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6808 * config/rs6000/mm_malloc.h: New file.
6810 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6812 PR tree-optimization/81913
6813 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6814 analysis when either IVs in condition can wrap.
6816 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6818 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6819 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6821 2017-08-24 Richard Biener <rguenther@suse.de>
6824 * targhooks.c (default_target_can_inline_p): Properly
6825 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6826 is present and always compare.
6827 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6828 infer -mfpmath=sse from TARGET_SSE_P.
6829 (ix86_can_inline_p): Properly use target_option_default_node when
6830 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6832 2017-08-24 Richard Biener <rguenther@suse.de>
6835 * dwarf2out.c (output_die): Handle flag_generate_offload like
6837 (output_comp_unit): Likewise.
6838 (gen_array_type_die): Likewise.
6839 (dwarf2out_early_finish): Likewise.
6840 (note_variable_value_in_expr): Likewise.
6841 (dwarf2out_finish): Likewise. Adjust assert.
6842 * cgraphunit.c (symbol_table::compile): Move setting of
6843 flag_generate_offload earlier ...
6844 (symbol_table::finalize_compilation_unit): ... here, before
6845 early debug finalization.
6847 2017-08-24 Richard Biener <rguenther@suse.de>
6849 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6850 and ipa-fnsummary.h.
6851 (ix86_can_inline_p): When ix86_fpmath flags do not match
6852 check whether the callee uses FP math at all.
6854 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6857 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6858 instead of TYPE_PRECISION.
6860 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6861 Alan Hayward <alan.hayward@arm.com>
6862 David Sherwood <david.sherwood@arm.com>
6864 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6866 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6868 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6870 (ix86_frame::stack_realign_allocate): New field.
6871 (struct machine_frame_state): Modify comments.
6872 (machine_frame_state::sp_realigned_fp_end): New field.
6873 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6875 (sp_valid_at): Add assertion to assure no attempt to access invalid
6876 offset of a realigned stack.
6877 (fp_valid_at): Likewise.
6878 (choose_baseaddr): Modify comments.
6879 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6880 ix86_expand_prologue.
6881 (ix86_expand_prologue): Modify stack realignment and allocation.
6882 (ix86_expand_epilogue): Modify comments.
6883 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6884 avx2_runtime, avx512f, and avx512f_runtime.
6886 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6888 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6889 (mstackrealign): Do not init to -1.
6890 * config/i386/i386.c (ix86_option_override_internal):
6891 Check opts_set, not opts when setting default value of
6892 opts->x_ix86_force_align_arg_pointer.
6894 2017-08-23 Richard Biener <rguenther@suse.de>
6896 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6897 lang_hooks.decl_printable_name.
6898 * print-rtl-function.c (print_rtx_function): Likewise.
6899 * tree-pretty-print.c (dump_function_header): Likewise.
6901 2017-08-23 Richard Biener <rguenther@suse.de>
6904 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6905 -g0 at compile-time.
6907 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6910 * doc/sourcebuild.texi (Other hardware attributes):
6913 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6916 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6917 Fix single-use check.
6919 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6921 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6923 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6925 * doc/install.texi: Modify to add more details on running selected
6928 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6930 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6931 is combined with -mabi=ms.
6932 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6935 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6937 PR tree-optimization/81488
6938 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6939 and cached_basis fields.
6940 (MAX_SPREAD): New constant.
6941 (alloc_cand_and_find_basis): Initialize new fields.
6942 (clear_visited): New function.
6943 (create_phi_basis_1): Rename from create_phi_basis, set visited
6944 and cached_basis fields.
6945 (create_phi_basis): New wrapper function.
6946 (phi_add_costs_1): Rename from phi_add_costs, add spread
6947 parameter, set visited field, short-circuit when limits reached.
6948 (phi_add_costs): New wrapper function.
6949 (record_phi_increments_1): Rename from record_phi_increments, set
6951 (record_phi_increments): New wrapper function.
6952 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6953 (phi_incr_cost): New wrapper function.
6954 (all_phi_incrs_profitable_1): Rename from
6955 all_phi_incrs_profitable, set visited field.
6956 (all_phi_incrs_profitable): New wrapper function.
6958 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6959 Alan Hayward <alan.hayward@arm.com>
6960 David Sherwood <david.sherwood@arm.com>
6962 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6963 that takes the outer and inner modes.
6964 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6965 comparison as the canonical test for a paradoxical subreg.
6966 * combine.c (simplify_set): Use paradoxical_subreg_p.
6967 (make_extraction): Likewise.
6968 (force_to_mode): Likewise.
6969 (rtx_equal_for_field_assignment_p): Likewise.
6970 (gen_lowpart_for_combine): Likewise.
6971 (simplify_comparison): Likewise.
6972 * cse.c (equiv_constant): Likewise.
6973 * expmed.c (store_bit_field_1): Likewise.
6974 * final.c (alter_subreg): Likewise.
6975 * fwprop.c (propagate_rtx): Likewise.
6976 (forward_propagate_subreg): Likewise.
6977 * ira-conflicts.c (ira_build_conflicts): Likewise.
6978 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6979 * lra-constraints.c (curr_insn_transform): Likewise.
6980 (split_reg): Likewise.
6981 * lra-eliminations.c (move_plus_up): Likewise.
6982 (lra_eliminate_regs_1): Likewise.
6983 * recog.c (general_operand): Likewise.
6984 * ree.c (combine_reaching_defs): Likewise.
6985 * reload.c (push_reload): Likewise.
6986 (find_reloads): Likewise.
6987 * reload1.c (elimination_effects): Likewise.
6988 (compute_reload_subreg_offset): Likewise.
6989 (choose_reload_regs): Likewise.
6990 * rtlanal.c (subreg_lsb_1): Likewise.
6991 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6992 (simplify_subreg): Likewise.
6993 * var-tracking.c (track_loc_p): Likewise.
6994 * emit-rtl.c (byte_lowpart_offset): Likewise.
6995 (paradoxical_subreg_p): Delete out-of-line definition.
6997 2017-08-22 Jeff Law <law@redhat.com>
6999 PR tree-optimization/81741
7000 PR tree-optimization/71947
7001 * tree-ssa-dom.c: Include tree-inline.h.
7002 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7003 equivalences if one is more expensive to compute than the other.
7004 * tree-ssa-scopedtables.h (class const_or_copies): Make
7005 record_const_or_copy_raw method private.
7006 (class avail_exprs_stack): New method simplify_binary_operation.
7007 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7008 avail_exprs_stack::simplify_binary_operation as needed.
7009 (avail_exprs_stack::simplify_binary_operation): New function.
7011 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7013 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7014 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7015 (DOT_SYMBOLS): Likewise.
7016 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7017 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7018 (RS6000_ABI_NAME): Likewise.
7019 (TARGET_CMODEL): Likewise.
7020 (TOC_SECTION_ASM_OP): Likewise.
7021 (SET_CMODEL): New macro.
7022 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7024 2017-08-22 Richard Biener <rguenther@suse.de>
7026 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7027 to allow for free-lang-data replacements similar to verify_type_variant.
7029 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7030 Alan Hayward <alan.hayward@arm.com>
7031 David Sherwood <david.sherwood@arm.com>
7033 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7034 VOIDmode for the LABEL_REF.
7036 2017-08-22 Richard Biener <rguenther@suse.de>
7038 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7040 2017-08-22 Richard Biener <rguenther@suse.de>
7042 * common.opt (feliminate-dwarf2-dups): Ignore.
7043 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7044 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7045 same_die_p_wrap, compute_section_prefix,
7046 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7047 (comdat_symbol_id, comdat_symbol_number): Likewise.
7048 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7050 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7051 (output_die): Mark unreachable path unreachable.
7052 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7053 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7054 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7055 (dwarf2out_early_finish): Likewise.
7057 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7059 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7061 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7064 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7065 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7066 (avr_attribute_table) <io, io_low, address>: Initialize
7067 .decl_required with true.
7069 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7071 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7072 undocumented debugging options.
7073 (-mvsx-scalar-double): Likewise.
7074 (-mallow-df-permute): Likewise.
7075 (-mvectorize-builtins): Likewise.
7076 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7077 (rs6000_builtin_vectorized_function): Likewise.
7078 (rs6000_builtin_md_vectorized_function): Likewise.
7079 (rs6000_opt_vars): Likewise.
7081 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7084 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7085 (*btrq_imm): Rename from *btrq.
7086 (*btcq_imm): Rename from *btcq.
7087 (btsc): New code attribute.
7088 (*<btsc><mode>): New insn pattern.
7089 (*btr<mode>): Ditto.
7090 (*<btsc><mode>_mask): New insn_and_split pattern.
7091 (*btr<mode>_mask): Ditto.
7093 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7094 Alan Hayward <alan.hayward@arm.com>
7095 David Sherwood <david.sherwood@arm.com>
7097 * function.c (pad_below): Simplify padding calculation.
7099 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7100 Alan Hayward <alan.hayward@arm.com>
7101 David Sherwood <david.sherwood@arm.com>
7103 * target.def (function_prologue): Remove frame size argument.
7104 (function_epilogue): Likewise.
7105 * doc/tm.texi: Regenerate.
7106 * final.c (final_start_function): Update call to function_prologue.
7107 (final_end_function): Update call to function_epilogue.
7108 (default_function_pro_epilogue): Remove frame size argument.
7109 * output.h (default_function_pro_epilogue): Likewise.
7110 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7111 (arm_output_function_prologue): Likewise.
7112 * config/frv/frv.c (frv_function_prologue): Likewise.
7113 (frv_function_epilogue): Likewise.
7114 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7115 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7116 (ia64_output_function_epilogue): Likewise.
7117 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7118 (m32r_output_function_epilogue): Likewise.
7119 * config/microblaze/microblaze.c (microblaze_function_prologue)
7120 (microblaze_function_epilogue): Likewise.
7121 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7122 (mips_output_function_epilogue): Likewise.
7123 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7124 (mmix_target_asm_function_epilogue): Likewise.
7125 * config/msp430/msp430.c (msp430_start_function): Likewise.
7126 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7127 (nds32_asm_function_epilogue): Likewise.
7128 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7129 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7130 (pa_output_function_epilogue): Likewise.
7131 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7132 (rs6000_output_function_epilogue): Likewise.
7133 * config/rl78/rl78.c (rl78_start_function): Likewise.
7134 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7135 (rs6000_output_function_epilogue): Likewise.
7136 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7137 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7138 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7139 (sparc_asm_function_epilogue): Likewise.
7141 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7143 * tree.h (type_has_mode_precision_p): New function.
7144 * convert.c (convert_to_integer_1): Use it.
7145 * expr.c (expand_expr_real_2): Likewise.
7146 (expand_expr_real_1): Likewise.
7147 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7148 * match.pd: Likewise.
7149 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7150 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7151 * tree-tailcall.c (process_assignment): Likewise.
7152 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7153 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7154 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7155 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7156 (vectorizable_assignment): Likewise.
7157 (vectorizable_shift): Likewise.
7158 (vectorizable_operation): Likewise.
7159 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7161 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7163 * match.pd: Add pow (C, x) simplification.
7165 2017-08-21 Richard Biener <rguenther@suse.de>
7167 PR tree-optimization/81900
7168 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7169 for blocks with abnormal predecessors.
7170 (compute_antic): Do not set visited flag prematurely.
7172 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7175 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7177 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7179 * stor-layout.h (vector_type_mode): Move to...
7180 * tree.h (vector_type_mode): ...here.
7181 * stor-layout.c (vector_type_mode): Move to...
7182 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7184 2017-08-21 Richard Biener <rguenther@suse.de>
7186 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7187 register_external_die hooks.
7188 (debug_false_tree_charstarstar_uhwistar): Declare.
7189 (debug_nothing_tree_charstar_uhwi): Likewise.
7190 * debug.c (do_nothing_debug_hooks): Adjust.
7191 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7192 (debug_nothing_tree_charstar_uhwi): Likewise.
7193 * dbxout.c (dbx_debug_hooks): Adjust.
7194 (xcoff_debug_hooks): Likewise.
7195 * sdbout.c (sdb_debug_hooks): Likewise.
7196 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7197 * dwarf2out.c (macinfo_label_base): New global.
7198 (dwarf2out_register_external_die): New function for the
7199 register_external_die hook.
7200 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7201 (dwarf2_debug_hooks): Use them.
7202 (dwarf2_lineno_debug_hooks): Adjust.
7203 (struct die_struct): Add with_offset flag.
7204 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7205 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7206 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7207 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7208 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7209 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7210 defining section names for the early LTO debug variants.
7211 (reset_indirect_string): New helper.
7212 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7213 (print_dw_val): Add support for offsetted symbol references.
7214 (get_ultimate_context): Split out from is_cxx.
7215 (is_cxx): Use get_ultimate_context.
7216 (is_fortran): Add decl overload.
7217 (compute_comp_unit_symbol): Split out worker from
7218 compute_section_prefix.
7219 (compute_section_prefix): Call compute_comp_unit_symbol and
7220 set comdat_type_p here.
7221 (output_die): Skip DIE symbol output for the LTO added one.
7222 Handle DIE symbol references with offset.
7223 (output_comp_unit): Guard section name mangling properly.
7224 For LTO debug sections emit a symbol at the section beginning
7225 which we use to refer to its DIEs.
7226 (add_abstract_origin_attribute): For DIEs registered via
7227 dwarf2out_register_external_die directly refer to the early
7228 DIE rather than indirectly through the shadow one we created.
7229 Remove obsolete call to dwarf2out_abstract_function for
7230 non-function/block origins.
7231 (gen_array_type_die): When generating early LTO debug do
7232 not emit DW_AT_string_length.
7233 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7234 late when in LTO. As suggested place a gcc_unreachable for
7235 the DECL_ABSTRACT_P case.
7236 (gen_subprogram_die): Avoid another specification DIE
7237 for early built declarations/definitions for the late LTO case.
7238 (gen_variable_die): Add type references for late duplicated VLA dies
7240 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7241 we have the abstract instance already.
7242 (process_scope_var): Adjust decl DIE contexts in LTO which
7243 first puts them in limbo.
7244 (gen_decl_die): Do not generate type DIEs late apart from
7245 types for VLAs or for decls we do not yet have a DIE. Do not
7246 call dwarf2out_abstract_function late.
7247 (dwarf2out_early_global_decl): Make sure to create DIEs
7248 for abstract instances of a decl first.
7249 (dwarf2out_late_global_decl): Adjust comment.
7250 (output_macinfo_op): With multiple macro sections use
7251 macinfo_label_base to distinguish labels.
7252 (output_macinfo): Likewise. Update macinfo_label_base.
7253 Pass in the line info label.
7254 (note_variable_value_in_expr): When generating LTO resolve
7255 all variable values here by generating DIEs as needed.
7256 (init_sections_and_labels): Add early LTO debug flag parameter
7257 and generate different sections and names if set. Add generation
7258 counter for the labels so we can have multiple of them.
7259 (reset_dies): Helper to allow DIEs to be output multiple times.
7260 (dwarf2out_finish): When outputting DIEs to the fat part of an
7261 LTO object first reset DIEs.
7262 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7263 (modified_type_die): Check for decl_ultimate_origin being self
7265 (gen_type_die_with_usage): Likewise.
7266 (gen_typedef_die): Allow decl_ultimate_origin being self.
7267 (set_decl_abstract_flags): Remove.
7268 (set_block_abstract_flags): Likewise.
7269 (dwarf2out_abstract_function): Treat the early generated DIEs
7270 as the abstract copy and only add DW_AT_inline and
7271 DW_AT_artificial here and call set_decl_origin_self.
7272 If the DIE has an abstract origin don't do anything.
7273 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7274 if we have none yet (Go fails to build one, PR78628).
7275 (variably_modified_type_p): Prevent endless recursion for Ada
7276 cyclic pointer types.
7277 * lto-streamer-in.c: Include debug.h.
7278 (dref_queue): New global.
7279 (lto_read_tree_1): Stream in DIE references.
7280 (lto_input_tree): Register DIE references.
7281 (input_function): Stream DECL_DEBUG_ARGS.
7282 * lto-streamer-out.c: Include debug.h.
7283 (lto_write_tree_1): Output DIE references.
7284 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7285 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7286 (output_function): Stream DECL_DEBUG_ARGS.
7287 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7288 Stream DECL_ABSTRACT_ORIGIN.
7289 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7290 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7291 DECL_CONTEXT for file-scope decls.
7292 * lto-streamer.h (struct dref_entry): Declare.
7293 (dref_queue): Likewise.
7294 * cfgexpand.c (pass_expand::execute): Do not call the
7295 outlining_inline_function hook here.
7296 * lto-wrapper.c (debug_obj): New global.
7297 (tool_cleanup): Unlink it if required.
7298 (debug_objcopy): New function.
7299 (run_gcc): Handle early debug sections in the IL files by
7300 extracting them to separate files, partially linkin them and
7301 feeding the result back as result to the linker.
7302 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7303 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7304 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7305 sections into a separate segment.
7306 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7308 (darwin_asm_dwarf_section): Likewise.
7309 (darwin_asm_output_dwarf_offset): Likewise.
7310 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7312 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7313 Alan Hayward <alan.hayward@arm.com>
7314 David Sherwood <david.sherwood@arm.com>
7316 * read-md.h (md_reader::record_potential_iterator_use): Replace
7317 pointer argument with an rtx and an index.
7318 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7319 (apply_mode_iterator): Likewise.
7320 (apply_code_iterator): Likewise.
7321 (apply_int_iterator): Likewise.
7322 (apply_subst_iterator): Likewise.
7323 (record_iterator_use): Likewise.
7324 (record_attribute_use): Likewise.
7325 (md_reader::record_potential_iterator_use): Likewise. Update calls
7326 to record_iterator_use and apply_iterator.
7327 (iterator_use): Replace ptr with x and index.
7328 (attribute_use): Likewise.
7329 (apply_attribute_uses): Update calls to apply_iterator.
7330 (apply_iterators): Likewise. Update initialization of iterator_use.
7331 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7332 and record_potential_iterator_use.
7333 (rtx_reader::read_rtx_operand): Likewise.
7335 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7336 Alan Hayward <alan.hayward@arm.com>
7337 David Sherwood <david.sherwood@arm.com>
7339 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7342 2017-08-21 Richard Biener <rguenther@suse.de>
7345 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7346 at struct end conservatively when comparing common bases.
7348 2017-08-21 Richard Biener <rguenther@suse.de>
7350 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7351 (mem_ref_in_stmt): Remove.
7352 (determine_max_movement): Use ref index to get at the reference.
7353 (invariantness_dom_walker::before_dom_children): Deal with
7354 lim data already initialized.
7355 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7357 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7359 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7360 (TARGET_ISA_ROUND): Ditto.
7361 (TARGET_ROUND): Ditto.
7362 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7363 * config/i386/i386.md: Ditto.
7364 * config/i386/sse.md: Ditto.
7365 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7366 with OPTION_MASK_ISA_SSE4_1.
7368 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7371 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7372 __builtin_ia32_lzcnt_u16.
7374 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7377 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7378 (rs6000_set_current_function): Rewrite function to use it.
7380 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7383 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7384 and TYPE_WARN_IF_NOT_ALIGN.
7385 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7386 (handle_warn_if_not_align): New.
7387 (place_union_field): Call handle_warn_if_not_align.
7388 (place_field): Call handle_warn_if_not_align.
7389 Copy TYPE_WARN_IF_NOT_ALIGN.
7390 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7391 (layout_type): Likewise.
7392 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7394 (tree_decl_common): Add warn_if_not_align.
7395 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7396 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7397 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7398 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7399 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7400 * doc/extend.texi: Document warn_if_not_aligned attribute.
7401 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7403 2017-08-17 Martin Liska <mliska@suse.cz>
7406 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7407 (get_data_dependence): Use it as pointer type.
7408 (distribute_loop): Likewise.
7410 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7412 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7413 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7414 (p8_vmrgow_<mode>_direct): New define_insn.
7415 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7416 handle endianness for vmrgew and vmrgow permute patterns.
7418 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7420 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7421 * config/rs6000/rs6000-cpus.def: Remove comment.
7422 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7423 (POWERPC_MASKS): Likewise.
7424 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7425 use of TARGET_VSX_TIMODE.
7426 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7427 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7428 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7429 (rs6000_option_override_internal): Remove dead code.
7430 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7431 (rs6000_legitimize_reload_address): Likewise.
7432 (rs6000_legitimate_address_p): Likewise.
7433 (rs6000_opt_masks): Delete "vsx-timode".
7434 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7435 from function comment.
7436 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7437 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7438 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7440 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7441 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7442 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7445 (define_peephole2): Likewise.
7447 2017-08-17 Martin Sebor <msebor@redhat.com>
7450 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7451 past the end of an array.
7452 (test_pp_format): Add test cases.
7454 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7456 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7457 missing ECF_NOTHROW flags.
7459 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7462 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7463 operands residing in integer registers.
7464 (*vsx_le_perm_load_<mode>): Likewise.
7465 (*vsx_le_perm_store_<mode>): Likewise.
7466 (define_peephole2): Add peepholes to optimize the above.
7468 2017-08-17 Marek Polacek <polacek@redhat.com>
7471 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7472 to mimic what shorten_compare did. Change the return type to bool.
7473 (fold_cond_expr_with_comparison): Update call to
7474 operand_equal_for_comparison_p.
7475 (fold_ternary_loc): Likewise.
7477 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7479 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7481 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7483 2017-08-17 Richard Biener <rguenther@suse.de>
7485 * tree-ssa-structalias.c (solve_graph): When propagating
7486 to successors update the graphs succ edges and avoid duplicate work.
7488 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7491 * config/i386/i386.c (ix86_option_override_internal): Save target
7492 specific options after ix86_stack_protector_guard_reg was changed.
7494 2017-08-17 Richard Biener <rguenther@suse.de>
7496 PR tree-optimization/81827
7497 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7498 (new_var_info): Initialize it conservatively.
7499 (get_call_vi): Mark register vars.
7500 (new_scalar_tmp_constraint_exp): Likewise.
7501 (handle_rhs_call): Likewise.
7502 (handle_const_call): Likewise.
7503 (create_function_info_for): Likewise.
7504 (solve_constraints): Sort varinfos to separate register from
7505 non-register vars to pack points-to solution bitmaps during
7508 2017-08-17 Marek Polacek <polacek@redhat.com>
7510 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7512 2017-08-17 Richard Biener <rguenther@suse.de>
7514 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7515 to true when overflow is undefined and we saturated the result.
7517 2017-08-17 Alan Modra <amodra@gmail.com>
7520 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7521 Don't use store multiple if only one reg needs saving.
7522 (interesting_frame_related_regno): New function.
7523 (rs6000_frame_related): Don't emit frame info for regs that
7525 (rs6000_emit_epilogue): Likewise.
7527 2017-08-16 Nathan Sidwell <nathan@acm.org>
7529 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7530 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7531 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7532 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7533 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7534 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7535 (verify_type): Adjust for TYPE_BINFO move.
7536 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7537 process TYPE_BINFO directly.
7538 (hash_tree): Likewise.
7539 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7541 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7544 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7546 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7548 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7551 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7552 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7553 Add ix86_binary_operator_ok to insn constraint.
7554 (*iordi_1_bts): Ditto.
7555 (*xordi_1_btc): Ditto.
7556 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7557 Update corresponding peephole2 pattern.
7561 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7563 PR tree-optimization/81832
7564 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7565 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7567 2017-08-16 Marek Polacek <polacek@redhat.com>
7570 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7571 regarding size_zero_node.
7573 2017-08-16 Martin Liska <mliska@suse.cz>
7576 * config.gcc: Respect previously set extra_objs in case
7579 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7581 PR tree-optimization/81835
7582 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7583 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7584 not depend on the phi.
7586 2017-08-16 Alan Modra <amodra@gmail.com>
7588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7591 2017-08-16 Alan Modra <amodra@gmail.com>
7593 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7594 (save_reg_p): ..into this. Update all callers.
7595 (first_reg_to_save): Simplify.
7597 2017-08-16 Alan Modra <amodra@gmail.com>
7599 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7602 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7606 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7607 constant count if that count is less than 32.
7609 2017-08-15 Nathan Sidwell <nathan@acm.org>
7611 * gcc.c (execute): Emit friendlier message if inferior is killed
7612 by an external cause.
7614 2017-08-15 Richard Biener <rguenther@suse.de>
7616 PR tree-optimization/81790
7617 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7618 CONSTRUCTORs from simplifying and VN.
7620 2017-08-14 Martin Sebor <msebor@redhat.com>
7622 * builtin-attrs.def: Add comments.
7624 2017-08-14 Martin Sebor <msebor@redhat.com>
7627 * doc/extend.texi (attribute nonstring): Document new attribute.
7629 2017-08-14 Martin Sebor <msebor@redhat.com>
7632 * tree-diagnostic.c (default_tree_printer): Handle %G.
7633 * gimple-pretty-print.h (percent_G_format): Declare new function.
7634 * gimple-pretty-print.c (percent_G_format): Define.
7635 * tree-pretty-print.c (percent_K_format): Add argument.
7637 2017-08-14 Martin Sebor <msebor@redhat.com>
7639 PR translation/79998
7640 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7641 Remove a stray space.
7643 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7646 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7647 (*iordi_1_bts): Ditto.
7648 (*xordi_1_btc): Ditto.
7650 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7653 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7654 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7656 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7657 quoted strings, and make more translator-friendly.
7658 (darwin_rs6000_override_options): Likewise.
7659 (rs6000_option_override_internal): Likewise.
7660 (rs6000_return_in_memory): Fix overlong line.
7661 (init_cmulative_args): Use quoted strings, and make more
7662 translator-friendly.
7663 (rs6000_pass_by_reference): Fix overlong line.
7664 (def_builtin): Use quoted strings.
7665 (altivec_expand_predicate_builtin): Use quoted strings, and make
7666 more translator-friendly.
7667 (htm_expand_builtin): Use quoted strings.
7668 (cpu_expand_builtin): Use quoted strings, and make more
7669 translator-friendly.
7670 (altivec_expand_builtin): Likewise.
7671 (paired_expand_predicate_builtin): Likewise.
7672 (rs6000_invalid_builtin): Likewise.
7673 (builtin_function_type): Use quoted strings.
7674 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7675 more translator-friendly.
7676 (rs6000_trampoline_init): Likewise.
7677 (rs6000_handle_altivec_attribute): Likewise.
7678 (rs6000_inner_target_options): Use quoted strings.
7679 (rs6000_disable_incompatible_switches): Likewise.
7680 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7681 strings, and make more translator-friendly.
7682 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7684 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7686 PR tree-optimization/81799
7687 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7688 cond_expr to simple gimple operand.
7690 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7693 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7695 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7699 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7700 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7701 TARGET_GASISR_PROLOGUES.
7702 * config/avr/avr.c (avr_option_override): Same.
7703 (avr_pass_pre_proep::execute): Same.
7705 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7708 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7709 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7711 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7713 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7714 define_insn_and_split. Split to a memory load from 0 in
7715 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7716 using PTR mode iterator.
7717 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7718 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7719 (*add_tp_<mode>): Redefine as define_insn_and_split.
7720 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7721 address space. Merge with *add_tp_x32 using PTR mode iterator.
7722 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7723 Split to an add with a memory load from 0 in
7724 DEFAULT_TLS_SEG_REG address space.
7726 2017-08-12 Andrew Pinski <apinski@cavium.com>
7728 * config/aarch64/aarch64-option-extensions.def (rdma):
7729 Fix feature string to what Linux prints out in /proc/cpuinfo.
7731 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7734 * dwarf2out.c (modified_type_die): For C typedef types that have
7735 an ultimate origin, process the ultimate origin instead of the
7737 (gen_typedef_die): Assert that input DECLs have no ultimate
7739 (gen_type_die_with_usage): For typedef variants that have an
7740 ultimate origin, just call gen_decl_die on the original DECL.
7741 (process_scope_var): Avoid creating DIEs for local typedefs and
7742 concrete static variables.
7744 2017-08-12 Alan Modra <amodra@gmail.com>
7748 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7749 match gnu-user.h startfile.
7750 (ENDFILE_LINUX_SPEC): Similarly.
7752 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7755 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7757 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7759 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7760 * config/aarch64/aarch64.md (mov<mode>): Change.
7761 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7762 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7763 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7765 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7767 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7768 for storage order barriers.
7770 2017-08-11 Martin Liska <mliska@suse.cz>
7773 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7774 variables of void type.
7776 2017-08-11 Martin Liska <mliska@suse.cz>
7778 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7779 TARGET_SUPPORTS_ALIASES.
7780 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7781 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7782 (optimize_weakref): Likewise.
7783 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7784 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7785 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7787 2017-08-11 Martin Liska <mliska@suse.cz>
7790 * config/i386/i386.c (make_resolver_func): Do complete
7791 refactoring of the function.
7793 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7796 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7797 * config/i386/i386.c (ix86_stack_protect_guard): Use
7798 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7799 * doc/invoke.texi (x86 Options): Document
7800 -mstack-protector-guard-symbol= option.
7802 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7804 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7805 * config/i386/i386.c (ix86_split_stack_guard): New function.
7806 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7807 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7808 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7809 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7810 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7811 (split_stack_space_check): Call ix86_split_stack_guard.
7813 2017-08-10 Martin Sebor <msebor@redhat.com>
7815 * print-tree.c (print_node): Print location using the established
7817 Replace spaces with colons.
7818 (debug_raw, debug): Ditto.
7820 2017-08-10 Martin Sebor <msebor@redhat.com>
7823 * pretty-print.c (pp_format): Correct the handling of %s precision.
7825 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7828 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7830 (ix86_finalize_stack_frame_flags): This. Also clear
7831 frame_pointer_needed if -fno-omit-frame-pointer is used without
7833 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7834 with ix86_finalize_stack_frame_flags.
7835 (ix86_expand_epilogue): Likewise.
7836 (ix86_expand_split_stack_prologue): Likewise.
7837 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7839 2017-08-10 Martin Liska <mliska@suse.cz>
7842 * c-attribs.c (handle_target_attribute):
7843 Report warning for an empty string argument of target attribute.
7845 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7848 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7850 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7851 or DECL_NONLOCAL labels.
7852 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7853 or DECL_NONLOCAL labels here.
7855 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7857 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7858 to indicate when early gimple folding has been disabled.
7859 (rs6000_gimple_fold_builtin): Add debug content.
7860 (rs6000_invalid_builtin): Fix whitespace.
7861 (rs6000_expand_builtin): Fix whitespace.
7862 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7864 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7867 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7868 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7870 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7872 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7873 (qdf24xx): Likewise.
7874 * config/aarch64/aarch64.md: Include falkor.md.
7875 * config/aarch64/falkor.md: New.
7877 2017-08-09 Marek Polacek <polacek@redhat.com>
7880 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7881 * diagnostic.c (emit_diagnostic): Add a comment.
7882 (emit_diagnostic_valist): New function.
7884 2017-08-09 Marek Polacek <polacek@redhat.com>
7887 * input.c (make_location): New overload.
7888 * input.h (make_location): Declare.
7890 2017-08-08 Alan Modra <amodra@gmail.com>
7891 H.J. Lu <hongjiu.lu@intel.com>
7894 * gcc.c (NO_PIE_SPEC): Delete.
7895 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7897 (LINK_PIE_SPEC): ..to here.
7898 (LINK_COMMAND_SPEC): Support -no-pie.
7899 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7900 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7902 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7903 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7904 (ENDFILE_CRTEND_SPEC): Similarly.
7906 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7909 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7910 (mstack-protector-guard-offset=): Ditto.
7911 * config/i386/i386.c (ix86_option_override): Handle
7912 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7914 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7915 ix86_stack_protect_guard_offset variables.
7916 (TARGET_STACK_PROTECT_GUARD): Always define.
7917 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7918 and -mstack-protector-guard-offset= options.
7920 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7922 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7923 boundary case for the last candidate.
7925 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7927 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7928 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7930 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7933 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7934 * config/aarch64/aarch64-builtins.c
7935 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7936 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7937 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7939 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7940 Andrew Pinski <pinskia@gmail.com>
7943 * internal-fn.def (XORSIGN): New.
7944 * optabs.def (xorsign_optab): New.
7945 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7946 (convert_expand_mult_copysign): New.
7947 (pass_optimize_widening_mul::execute): Call
7948 convert_expand_mult_copysign.
7950 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7952 PR tree-optimization/81354
7953 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7954 Insert on edges rather than explicitly creating landing pads.
7955 (analyze_candidates_and_replace): Commit edge inserts.
7957 2017-08-08 Richard Biener <rguenther@suse.de>
7960 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7961 (expand_simple_operations): Also look through ADDR_EXPRs with
7962 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7964 2017-08-08 Richard Biener <rguenther@suse.de>
7966 PR tree-optimization/81723
7967 * tree-vect-slp.c (struct bst_traits): New hash traits.
7968 (bst_fail): New global.
7969 (vect_build_slp_tree_2): New worker, split out from ...
7970 (vect_build_slp_tree): ... this now wrapping it with using
7971 bst_fail set to cache SLP tree build fails. Properly handle
7973 (vect_analyze_slp_instance): Allocate and free bst_fail.
7975 2017-08-08 Martin Liska <mliska@suse.cz>
7978 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7979 LABEL_DECLs that can be from a different function.
7981 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7983 PR tree-optimization/81744
7984 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7985 loop's number of iterations.
7987 2017-08-08 Martin Liska <mliska@suse.cz>
7989 * asan.c: Include header files.
7990 * attribs.c (build_decl_attribute_variant): New function moved
7992 (build_type_attribute_qual_variant): Likewise.
7993 (cmp_attrib_identifiers): Likewise.
7994 (simple_cst_list_equal): Likewise.
7995 (omp_declare_simd_clauses_equal): Likewise.
7996 (attribute_value_equal): Likewise.
7997 (comp_type_attributes): Likewise.
7998 (build_type_attribute_variant): Likewise.
7999 (lookup_ident_attribute): Likewise.
8000 (remove_attribute): Likewise.
8001 (merge_attributes): Likewise.
8002 (merge_type_attributes): Likewise.
8003 (merge_decl_attributes): Likewise.
8004 (merge_dllimport_decl_attributes): Likewise.
8005 (handle_dll_attribute): Likewise.
8006 (attribute_list_equal): Likewise.
8007 (attribute_list_contained): Likewise.
8008 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8009 (lookup_attribute_by_prefix): Likewise.
8010 * bb-reorder.c: Include header files.
8011 * builtins.c: Likewise.
8012 * calls.c: Likewise.
8013 * cfgexpand.c: Likewise.
8014 * cgraph.c: Likewise.
8015 * cgraphunit.c: Likewise.
8016 * convert.c: Likewise.
8017 * dwarf2out.c: Likewise.
8018 * final.c: Likewise.
8019 * fold-const.c: Likewise.
8020 * function.c: Likewise.
8021 * gimple-expr.c: Likewise.
8022 * gimple-fold.c: Likewise.
8023 * gimple-pretty-print.c: Likewise.
8024 * gimple.c: Likewise.
8025 * gimplify.c: Likewise.
8026 * hsa-common.c: Likewise.
8027 * hsa-gen.c: Likewise.
8028 * internal-fn.c: Likewise.
8029 * ipa-chkp.c: Likewise.
8030 * ipa-cp.c: Likewise.
8031 * ipa-devirt.c: Likewise.
8032 * ipa-fnsummary.c: Likewise.
8033 * ipa-inline.c: Likewise.
8034 * ipa-visibility.c: Likewise.
8036 * lto-cgraph.c: Likewise.
8037 * omp-expand.c: Likewise.
8038 * omp-general.c: Likewise.
8039 * omp-low.c: Likewise.
8040 * omp-offload.c: Likewise.
8041 * omp-simd-clone.c: Likewise.
8042 * opts-global.c: Likewise.
8043 * passes.c: Likewise.
8044 * predict.c: Likewise.
8045 * sancov.c: Likewise.
8046 * sanopt.c: Likewise.
8047 * symtab.c: Likewise.
8048 * toplev.c: Likewise.
8049 * trans-mem.c: Likewise.
8050 * tree-chkp.c: Likewise.
8051 * tree-eh.c: Likewise.
8052 * tree-into-ssa.c: Likewise.
8053 * tree-object-size.c: Likewise.
8054 * tree-parloops.c: Likewise.
8055 * tree-profile.c: Likewise.
8056 * tree-ssa-ccp.c: Likewise.
8057 * tree-ssa-live.c: Likewise.
8058 * tree-ssa-loop.c: Likewise.
8059 * tree-ssa-sccvn.c: Likewise.
8060 * tree-ssa-structalias.c: Likewise.
8061 * tree-ssa.c: Likewise.
8062 * tree-streamer-in.c: Likewise.
8063 * tree-vectorizer.c: Likewise.
8064 * tree-vrp.c: Likewise.
8066 * ubsan.c: Likewise.
8067 * varasm.c: Likewise.
8068 * varpool.c: Likewise.
8069 * tree.c: Remove functions moved to attribs.[ch].
8071 * config/aarch64/aarch64.c: Add attrs.h header file.
8072 * config/alpha/alpha.c: Likewise.
8073 * config/arc/arc.c: Likewise.
8074 * config/arm/arm.c: Likewise.
8075 * config/avr/avr.c: Likewise.
8076 * config/bfin/bfin.c: Likewise.
8077 * config/c6x/c6x.c: Likewise.
8078 * config/cr16/cr16.c: Likewise.
8079 * config/cris/cris.c: Likewise.
8080 * config/darwin.c: Likewise.
8081 * config/epiphany/epiphany.c: Likewise.
8082 * config/fr30/fr30.c: Likewise.
8083 * config/frv/frv.c: Likewise.
8084 * config/ft32/ft32.c: Likewise.
8085 * config/h8300/h8300.c: Likewise.
8086 * config/i386/winnt.c: Likewise.
8087 * config/ia64/ia64.c: Likewise.
8088 * config/iq2000/iq2000.c: Likewise.
8089 * config/lm32/lm32.c: Likewise.
8090 * config/m32c/m32c.c: Likewise.
8091 * config/m32r/m32r.c: Likewise.
8092 * config/m68k/m68k.c: Likewise.
8093 * config/mcore/mcore.c: Likewise.
8094 * config/microblaze/microblaze.c: Likewise.
8095 * config/mips/mips.c: Likewise.
8096 * config/mmix/mmix.c: Likewise.
8097 * config/mn10300/mn10300.c: Likewise.
8098 * config/moxie/moxie.c: Likewise.
8099 * config/msp430/msp430.c: Likewise.
8100 * config/nds32/nds32-isr.c: Likewise.
8101 * config/nds32/nds32.c: Likewise.
8102 * config/nios2/nios2.c: Likewise.
8103 * config/nvptx/nvptx.c: Likewise.
8104 * config/pa/pa.c: Likewise.
8105 * config/pdp11/pdp11.c: Likewise.
8106 * config/powerpcspe/powerpcspe.c: Likewise.
8107 * config/riscv/riscv.c: Likewise.
8108 * config/rl78/rl78.c: Likewise.
8109 * config/rx/rx.c: Likewise.
8110 * config/s390/s390.c: Likewise.
8111 * config/sh/sh.c: Likewise.
8112 * config/sol2.c: Likewise.
8113 * config/sparc/sparc.c: Likewise.
8114 * config/spu/spu.c: Likewise.
8115 * config/stormy16/stormy16.c: Likewise.
8116 * config/tilegx/tilegx.c: Likewise.
8117 * config/tilepro/tilepro.c: Likewise.
8118 * config/v850/v850.c: Likewise.
8119 * config/vax/vax.c: Likewise.
8120 * config/visium/visium.c: Likewise.
8121 * config/xtensa/xtensa.c: Likewise.
8123 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8126 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8127 constraints since the -mupper-regs-* switches have been
8129 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8130 into a vector from a double word element that was extracted from
8131 another vector, and eliminate extra XXPERMDI instructions.
8132 (vsx_concat_<mode>_2): Likewise.
8133 (vsx_concat_<mode>_3): Likewise.
8134 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8135 concat to allow optimizing inserts from previous extracts.
8137 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8139 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8140 memory reference to a SSP offset in TLS address space.
8141 (ix86_print_operand) <case '@'>: Remove.
8142 (ix86_print_operand_punct_valid_p): Remove '@' code.
8143 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8145 (stack_tls_protect_set_<mode>): Remove.
8146 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8147 (stack_tls_protect_test_<mode>): Remove.
8148 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8150 2017-08-07 Olivier Hainque <hainque@adacore.com>
8153 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8155 2017-08-07 Douglas Rupp <rupp@adacore.com>
8157 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8158 variable was referenced as multidir in command.
8160 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8163 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8166 2017-08-07 Martin Liska <mliska@suse.cz>
8168 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8169 * config/rl78/rl78.c: Add include of attribs.h.
8170 * config/sh/sh.c: Likewise.
8171 * config/v850/v850.c: Likewise.
8173 2017-08-07 Tom de Vries <tom@codesourcery.com>
8176 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8178 2017-08-07 Martin Liska <mliska@suse.cz>
8180 * config/mips/mips.c: Include attribs.h.
8182 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8185 * doc/invoke.texi: Document change in behvaior for -Ofast for
8188 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8190 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8192 (aarch64_pop_regs): Likewise.
8193 (aarch64_gen_load_pair): Likewise.
8194 (aarch64_save_callee_saves): Likewise.
8195 (aarch64_restore_callee_saves): Likewise.
8197 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8199 * config/i386/i386.c: Revert the last change.
8201 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8204 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8206 (ix86_finalize_stack_frame_flags): This. Also clear
8207 frame_pointer_needed if -fno-omit-frame-pointer is used without
8209 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8210 with ix86_finalize_stack_frame_flags.
8211 (ix86_expand_epilogue): Likewise.
8212 (ix86_expand_split_stack_prologue): Likewise.
8214 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8217 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8218 to P_AES for Westmere.
8220 2017-08-07 Jonathan Yong <10walls@gmail.com>
8222 * config/i386/mingw.opt (fset-stack-executable): Removed.
8223 * config/i386/cygming.opt (fset-stack-executable): Moved
8225 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8227 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8229 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8231 2017-08-07 Marek Polacek <polacek@redhat.com>
8234 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8236 2017-08-07 Martin Liska <mliska@suse.cz>
8238 * attribs.h (canonicalize_attr_name): New function.
8239 (cmp_attribs): Move from c-format.c and adjusted.
8240 (is_attribute_p): Moved from tree.h.
8241 * tree-inline.c: Add new includes.
8242 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8243 (private_is_attribute_p): Remove.
8244 (private_lookup_attribute): Likewise.
8245 (private_lookup_attribute_by_prefix): Simplify.
8246 (remove_attribute): Use is_attribute_p.
8247 * tree.h: Remove removed declarations.
8249 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8252 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8253 instead of computing it in the function. Formatting fix.
8254 (expand_case): Don't rely on default_edge being the first edge,
8255 clear it if removing it, pass default_edge to
8256 emit_case_dispatch_table.
8257 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8260 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8262 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8263 insn in the function, emit NOP after the insn.
8265 2017-08-06 Tom de Vries <tom@codesourcery.com>
8267 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8270 2017-08-06 Tom de Vries <tom@codesourcery.com>
8272 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8275 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8277 PR tree-optimization/57371
8278 * match.pd: New pattern.
8280 2017-08-04 Marek Polacek <polacek@redhat.com>
8283 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8284 perform the computation in offset_int.
8286 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8288 PR tree-optimization/81136
8289 * tree-vectorizer.h: Include tree-hash-traits.h.
8290 (vec_base_alignments): New typedef.
8291 (vec_info): Add a base_alignments field.
8292 (vect_record_base_alignments): Declare.
8293 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8295 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8296 (create_data_ref): Add an is_conditional_in_stmt argument.
8297 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8298 the is_conditional_in_stmt field.
8299 (data_ref_loc): Add an is_conditional_in_stmt field.
8300 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8301 (find_data_references_in_stmt): Update call to create_data_ref.
8302 (graphite_find_data_references_in_stmt): Likewise.
8303 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8304 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8305 (vect_record_base_alignment): New function.
8306 (vect_record_base_alignments): Likewise.
8307 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8308 for nested statements even if we fail to compute a misalignment.
8309 Use pooled base alignments for unconditional references.
8310 (vect_find_same_alignment_drs): Compare base addresses instead
8312 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8313 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8315 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8317 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8318 Add an explicit name for the enum. Use auto_vec for slp_instances
8320 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8322 (_bb_vec_info): Add a constructor and destructor.
8323 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8324 (destroy_loop_vec_info): Delete.
8325 (vect_destroy_datarefs): Likewise.
8326 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8327 (vec_info::vec_info): New function.
8328 (vec_info::~vec_info): Likewise.
8329 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8330 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8331 destroy_loop_vec_info.
8332 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8333 (_loop_vec_info::_loop_vec_info): ...this.
8334 (destroy_loop_vec_info): Replace with...
8335 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8336 the stmt_vec_infos. Leave handling of vec_info information to its
8337 destructor. Remove explicit vector releases.
8338 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8339 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8340 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8341 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8342 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8343 (destroy_bb_vec_info): Replace with...
8344 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8345 information to its destructor.
8346 (vect_slp_analyze_bb_1): Use new and delete instead of
8347 new_bb_vec_info and destroy_bb_vec_info.
8348 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8351 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8353 * tree-data-ref.h (subscript): Add access_fn field.
8354 (data_dependence_relation): Add could_be_independent_p.
8355 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8356 (same_access_functions): Move to tree-data-ref.c.
8357 * tree-data-ref.c (ref_contains_union_access_p): New function.
8358 (access_fn_component_p): Likewise.
8359 (access_fn_components_comparable_p): Likewise.
8360 (dr_analyze_indices): Add a reference to access_fn_component_p.
8361 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8363 (constant_access_functions): Likewise.
8364 (add_other_self_distances): Likewise.
8365 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8366 (initialize_data_dependence_relation): Use XCNEW and remove
8367 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8368 of access functions that have the same type. Allow the
8369 subsequence to end with different bases in some circumstances.
8370 Record the chosen access functions in SUB_ACCESS_FN.
8371 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8372 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8373 (subscript_dependence_tester_1): Likewise dra and drb.
8374 (build_classic_dist_vector): Update calls accordingly.
8375 (subscript_dependence_tester): Likewise.
8376 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8377 DDR_COULD_BE_INDEPENDENT_P.
8378 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8379 comp_alias_ddrs instead of may_alias_ddrs.
8380 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8382 (vect_analyze_data_ref_dependence): Use it if
8383 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8384 distance vectors if that fails.
8385 (dependence_distance_ge_vf): New function.
8386 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8387 LOOP_VINFO_MAY_ALIAS_DDRS.
8389 2017-08-04 Richard Biener <rguenther@suse.de>
8392 * fold-const.c (fold_binary_loc): Properly restrict
8393 minus_var0 && minus_var1 case when associating undefined overflow
8396 2017-08-04 Tom de Vries <tom@codesourcery.com>
8398 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8400 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8402 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8403 Don't start diagnostic messages with a capital letter.
8404 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8406 (rs6000_invalid_builtin): Likewise.
8407 (rs6000_trampoline_init): Likewise.
8409 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8412 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8413 after setting changeable df flags.
8415 2017-08-03 Richard Biener <rguenther@suse.de>
8417 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8418 up if the use is in USE - X.
8420 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8422 * toplev.c (dumpfile.h): New include.
8423 (internal_error_reentered): New static function. Use it...
8424 (internal_error_function): ...here to handle reentered internal_error.
8426 2017-08-03 Richard Biener <rguenther@suse.de>
8429 * fold-const.c (split_tree): Add minus_var and minus_con
8430 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8431 here but always use minus_*.
8432 (associate_trees): Assert we never associate with MINUS_EXPR
8433 and NULL first operand. Do not recurse for PLUS_EXPR operands
8434 when associating as MINUS_EXPR either.
8435 (fold_binary_loc): Track minus_var and minus_con.
8437 2017-08-03 Tom de Vries <tom@codesourcery.com>
8440 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8441 ACCEL_COMPILER, apply finish_options on
8442 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8444 2017-08-03 Tom de Vries <tom@codesourcery.com>
8447 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8448 function_entry_patch_area_size > 0.
8450 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8453 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8454 instead of 10??LU, perform unit multiplication in wide_int,
8455 don't change alloc_object_size_limit if the limit is larger
8458 PR tree-optimization/81655
8459 PR tree-optimization/81588
8460 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8461 the case when ranges[i].low and high are 1 for unsigned type with
8465 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8466 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8468 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8470 * tree-vrp.h: Add include guard.
8472 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8475 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8476 (ud2): New insn pattern.
8477 * config/i386/i386.c (ix86_expand_epilogue):
8478 For naked functions, generate ud2 instead of trap insn.
8480 2017-08-02 Marek Polacek <polacek@redhat.com>
8483 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8485 2017-08-02 Tom de Vries <tom@codesourcery.com>
8486 Cesar Philippidis <cesar@codesourcery.com>
8488 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8489 Add missing edge probabilities.
8491 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8493 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8496 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8499 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8500 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8501 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8502 prologue_seq sequences - if any.
8504 2017-08-02 Richard Biener <rguenther@suse.de>
8506 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8507 via vectors if supported, integer extracts via punning if supported
8508 or otherwise vector extracts.
8510 2017-08-02 Richard Biener <rguenther@suse.de>
8512 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8514 (bitmap_insert_into_set): ... this.
8516 2017-08-02 Richard Biener <rguenther@suse.de>
8518 PR tree-optimization/81633
8520 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8522 PR tree-optimization/71752
8523 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8525 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8527 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8528 (machine_function::call_ms2sysv_pad_out): Remove field.
8529 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8530 (ix86_compute_frame_layout): Likewise.
8532 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8535 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8536 attribute with interrupt attribute.
8538 2017-08-01 Andrew Pinski <apinski@cavium.com>
8540 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8541 BIT_INSERT_EXPR's operand 1
8542 to see if the types precision matches.
8544 2017-08-01 Martin Liska <mliska@suse.cz>
8547 * builtins.c (expand_builtin_memcpy_args): Remove.
8548 (expand_builtin_memcpy): Call newly added function
8549 expand_builtin_memory_copy_args.
8550 (expand_builtin_memcpy_with_bounds): Likewise.
8551 (expand_builtin_mempcpy): Remove last argument.
8552 (expand_builtin_mempcpy_with_bounds): Likewise.
8553 (expand_builtin_memory_copy_args): New function created from
8554 expand_builtin_mempcpy_args with small modifications.
8555 (expand_builtin_mempcpy_args): Remove.
8556 (expand_builtin_stpcpy): Remove unused argument.
8557 (expand_builtin): Likewise.
8558 (expand_builtin_with_bounds): Likewise.
8560 2017-08-01 Martin Liska <mliska@suse.cz>
8563 Make mempcpy more optimal (PR middle-end/70140).
8565 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8568 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8569 __builtin_vec_cmpne verify both arguments are compatible vectors
8570 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8571 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8572 move computation of aligned to after checking the argument types.
8576 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8577 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8580 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8581 Doug Rupp <rupp@adacore.com>
8582 Olivier Hainque <hainque@adacore.com>
8584 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8585 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8587 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8588 for TARGET_OS_CPP_BUILTIN.
8589 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8590 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8592 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8593 passing required abi options to the assembler for EABI configurations.
8594 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8595 of .text.hot and .text.unlikely sections for kernel modules when
8596 using ARM style exceptions.
8597 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8598 options. Add EXTRA_CC1_SPEC.
8599 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8601 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8603 (ARM_TARGET2_DWARF_FORMAT): Define.
8604 * config/arm/t-vxworks: Adjust multilib control to removal of the
8605 Diab command line options.
8607 2017-08-01 Martin Liska <mliska@suse.cz>
8609 PR gcov-profile/81561
8610 * gcov.c (unblock): Make unblocking safe as we need to preserve
8611 index correspondence of blocks and block_lists.
8613 2017-08-01 Richard Biener <rguenther@suse.de>
8615 PR tree-optimization/81181
8616 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8617 (compute_antic): ... end of iteration here.
8619 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8621 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8622 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8623 (ftree-slp-vectorize): Likewise.
8624 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8625 can no longer be set independent of flag_tree_loop_vectorize.
8626 * omp-general.c (emp_max_vf): Likewise.
8627 * opts.c (enable_fdo_optimizations): Remove references to
8628 flag_tree_vectorize, these are now implicit.
8629 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8630 and leave it for the options machinery.
8632 2017-08-01 Martin Liska <mliska@suse.cz>
8635 * builtins.c (expand_builtin_memcpy_args): Remove.
8636 (expand_builtin_memcpy): Call newly added function
8637 expand_builtin_memory_copy_args.
8638 (expand_builtin_memcpy_with_bounds): Likewise.
8639 (expand_builtin_mempcpy): Remove last argument.
8640 (expand_builtin_mempcpy_with_bounds): Likewise.
8641 (expand_builtin_memory_copy_args): New function created from
8642 expand_builtin_mempcpy_args with small modifications.
8643 (expand_builtin_mempcpy_args): Remove.
8644 (expand_builtin_stpcpy): Remove unused argument.
8645 (expand_builtin): Likewise.
8646 (expand_builtin_with_bounds): Likewise.
8648 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8651 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8652 print "ds:" only for immediates in generic address space.
8654 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8657 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8658 (ix86_function_ok_for_sibcall): Return false for naked functions.
8660 2017-08-01 Richard Biener <rguenther@suse.de>
8662 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8663 (compute_antic): Seed worklist with exit block predecessors.
8664 * cfganal.c (dfs_find_deadend): For a cycle return the source
8665 of the edge closing it.
8667 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8669 * config/aarch64/aarch64.c
8670 (aarch64_can_const_movi_rtx_p): Move 0 check.
8672 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8674 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8675 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8677 * match.pd: Ditto in address comparison pattern.
8679 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8681 PR tree-optimization/81627
8682 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8683 closed ssa form for store-store chain.
8685 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8687 PR tree-optimization/81620
8688 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8689 for store-store chain.
8691 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8693 PR tree-optimization/81588
8694 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8695 ranges[i].in_p, invert comparison code ccode. For >/>=,
8696 swap rhs1 and rhs2 and comparison code unconditionally,
8697 for </<= don't do that. Don't swap rhs1/rhs2 again if
8698 ranges[i].in_p, instead invert comparison code ccode if
8699 opcode or oe->rank is BIT_IOR_EXPR.
8702 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8703 a direct optab to conversion optab.
8704 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8705 with GET_MODE_INNER as last argument instead of optab_handler.
8706 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8707 vector extraction if possible and optab is available.
8708 * expr.c (store_constructor): Use convert_optab_handler instead
8709 of optab_handler. Use vector initialization from smaller
8710 vectors if possible and optab is available.
8711 * tree-vect-stmts.c (vectorizable_load): Likewise.
8712 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8714 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8715 of vec_init from half-sized vectors with the same element mode.
8716 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8717 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8718 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8719 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8720 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8721 after mode in gen_vec_extract* calls.
8722 (vec_extract<mode>): Renamed to ...
8723 (vec_extract<mode><ssescalarmodelower>): ... this.
8724 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8725 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8726 element mode after mode in gen_vec_init* calls.
8727 (VEC_INIT_HALF_MODE): New mode iterator.
8728 (vec_init<mode>): Renamed to ...
8729 (vec_init<mode><ssescalarmodelower>): ... this.
8730 (vec_init<mode><ssehalfvecmodelower>): New expander.
8731 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8732 (vec_extractv2sfsf): ... this.
8733 (vec_initv2sf): Renamed to ...
8734 (vec_initv2sfsf): ... this.
8735 (vec_extractv2si): Renamed to ...
8736 (vec_extractv2sisi): ... this.
8737 (vec_initv2si): Renamed to ...
8738 (vec_initv2sisi): ... this.
8739 (vec_extractv4hi): Renamed to ...
8740 (vec_extractv4hihi): ... this.
8741 (vec_initv4hi): Renamed to ...
8742 (vec_initv4hihi): ... this.
8743 (vec_extractv8qi): Renamed to ...
8744 (vec_extractv8qiqi): ... this.
8745 (vec_initv8qi): Renamed to ...
8746 (vec_initv8qiqi): ... this.
8747 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8748 (vec_init<mode>): Renamed to ...
8749 (vec_init<mode><VEC_base_l>): ... this.
8750 (vec_extract<mode>): Renamed to ...
8751 (vec_extract<mode><VEC_base_l>): ... this.
8752 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8753 (vec_initv2sfsf): ... this.
8754 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8755 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8756 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8757 element mode after mode in gen_vec_init* calls.
8758 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8759 (vec_init<mode><Vel>): ... this.
8760 (vec_extract<mode>): Renamed to ...
8761 (vec_extract<mode><Vel>): ... this.
8762 * config/aarch64/iterators.md (Vel): New mode attribute.
8763 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8764 Add element mode after mode in gen_vec_extract* calls.
8765 * config/s390/vector.md (non_vec_l): New mode attribute.
8766 (vec_extract<mode>): Renamed to ...
8767 (vec_extract<mode><non_vec_l>): ... this.
8768 (vec_init<mode>): Renamed to ...
8769 (vec_init<mode><non_vec_l>): ... this.
8770 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8771 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8773 * config/arm/iterators.md (V_elem_l): New mode attribute.
8774 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8775 (vec_extract<mode><V_elem_l>): ... this.
8776 (vec_extractv2di): Renamed to ...
8777 (vec_extractv2didi): ... this.
8778 (vec_init<mode>): Renamed to ...
8779 (vec_init<mode><V_elem_l>): ... this.
8780 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8781 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8782 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8783 Add element mode after gen_vec_extract* calls.
8784 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8785 (vec_init<mode><unitmode>): ... this.
8786 (vec_extract<mode>): Renamed to ...
8787 (vec_extract<mode><unitmode>): ... this.
8788 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8789 (vec_init<mode><unitmode>): ... this.
8790 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8791 (vec_initv2sfsf): ... this.
8792 (vec_extractv2sf): Renamed to ...
8793 (vec_extractv2sfsf): ... this.
8794 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8795 Add element mode after gen_vec_extract* calls.
8796 * config/mips/mips.md (unitmode): New mode iterator.
8797 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8798 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8799 * config/spu/spu.md (inner_l): New mode attribute.
8800 (vec_init<mode>): Renamed to ...
8801 (vec_init<mode><inner_l>): ... this.
8802 (vec_extract<mode>): Renamed to ...
8803 (vec_extract<mode><inner_l>): ... this.
8804 * config/sparc/sparc.md (veltmode): New mode iterator.
8805 (vec_init<VMALL:mode>): Renamed to ...
8806 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8807 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8808 (vec_initv2sisi): ... this.
8809 (vec_initv2sf): Renamed to ...
8810 (vec_initv2sfsf): ... this.
8811 (vec_extractv2sf): Renamed to ...
8812 (vec_extractv2sfsf): ... this.
8813 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8814 (vec_init<mode>): Renamed to ...
8815 (vec_init<mode><VEC_base_l>): ... this.
8816 (vec_extract<mode>): Renamed to ...
8817 (vec_extract<mode><VEC_base_l>): ... this.
8818 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8819 (vec_initv2sfsf): ... this.
8820 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8821 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8822 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8823 gen_vec_init* calls.
8825 2017-08-01 Richard Biener <rguenther@suse.de>
8827 PR tree-optimization/81297
8828 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8829 TREE_OVERFLOW from INTEGER_CSTs.
8831 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8833 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8835 2017-07-31 Carl Love <cel@us.ibm.com>
8837 * config/rs6000/rs6000-c: Add support for built-in functions
8838 vector signed char vec_xl_be (signed long long, signed char *);
8839 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8840 vector signed int vec_xl_be (signed long long, signed int *);
8841 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8842 vector signed long long vec_xl_be (signed long long, signed long long *);
8843 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8844 vector signed short vec_xl_be (signed long long, signed short *);
8845 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8846 vector double vec_xl_be (signed long long, double *);
8847 vector float vec_xl_be (signed long long, float *);
8848 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8849 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8850 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8852 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8853 (altivec_expand_builtin): Add switch statement to call
8854 altivec_expand_xl_be for each builtin.
8855 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8856 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8857 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8858 __builtin_vsx_le_be_v16qi.
8859 * doc/extend.texi: Update the built-in documentation file for the
8860 new built-in functions.
8862 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8865 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8867 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8869 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8871 * config.gcc: Add z14.
8872 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8873 CPU model numbers for z13s and z14.
8874 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8876 * config/s390/s390-opts.h (enum processor_type): Rename
8877 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8878 * config/s390/s390.c (processor_table): Add field for CPU name to
8879 be passed to Binutils.
8880 (s390_asm_output_machine_for_arch): Use the new field in
8881 processor_table for Binutils.
8882 (s390_expand_builtin): Replace arch12 with z14.
8883 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8884 (s390_get_sched_attrmask): Likewise.
8885 (s390_get_unit_mask): Likewise.
8886 * config/s390/s390.opt: Add z14 to processor_type enum.
8888 2017-07-31 Martin Jambor <mjambor@suse.cz>
8891 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8892 regardless of optimization level.
8894 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8895 Martin Liska <mliska@suse.cz>
8897 * predict.def: Remove old comment and adjust probability.
8898 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8901 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8904 * config/i386/i386.c (ix86_function_naked): New function.
8905 (ix86_can_use_return_insn_p): Return false for naked functions.
8906 (ix86_expand_prologue): Skip prologue for naked functions.
8907 (ix86_expand_epilogue): Skip epilogue for naked functions
8908 and emit trap instruction.
8909 (ix86_warn_func_return): New function.
8910 (ix86_attribute_table): Add "naked" attribute specification.
8911 (TARGET_WARN_FUNC_RETURN): Define.
8912 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8914 2017-07-31 Martin Liska <mliska@suse.cz>
8916 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8917 (dump_gimple_bb_header): Always dump BB info.
8918 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8920 2017-07-31 Martin Liska <mliska@suse.cz>
8923 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8924 also with current_function_decl non-null equality.
8926 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8929 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8930 change type to the element type, instead add eltype variable and
8931 use it where we are interested in the element type.
8933 PR tree-optimization/81603
8934 * ipa-polymorphic-call.c
8935 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8936 offset arithmetic in offset_int, bail out if the resulting bit offset
8937 doesn't fit into shwi.
8939 2017-07-31 Martin Liska <mliska@suse.cz>
8941 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8942 (gimplify_save_expr): Fix comment.
8944 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8947 * config/i386/i386.c (ix86_function_arg): Update arguments for
8949 (ix86_compute_frame_layout): Set the initial stack offset to
8950 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8951 INCOMING_FRAME_SP_OFFSET.
8952 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8953 stack before exception handler returns.
8954 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8955 the 'ERROR_CODE' for exception handler.
8957 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8959 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8960 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8961 (ASM_OUTPUT_REG_POP): Ditto.
8962 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8963 instead of asm_fprintf to output pure string.
8965 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8967 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8968 to imported_module_or_decl hook.
8969 (debug_nothing_tree_tree_tree_bool): Remove.
8970 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8971 * debug.c (do_nothing_debug_hooks): Use
8972 debug_nothing_tree_tree_tree_bool_bool instead of
8973 debug_nothing_tree_tree_tree_bool.
8974 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8975 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8976 * sdbout.c (sdb_debug_hooks): Likewise.
8977 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8978 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8980 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8981 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8982 attribute, don't add anything.
8984 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8986 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8987 (fold_build2_stat_loc): Likewise.
8988 (fold_build3_stat_loc): Likewise.
8989 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8990 (fold_build1_loc): Remove macro.
8991 (fold_build2_loc): Likewise.
8992 (fold_build3_loc): Likewise.
8994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8996 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8997 (gimple_build_debug_bind_source_stat): Likewise.
8998 * gimple.h (gimple_build_debug_bind): Remove macro.
8999 (gimple_build_debug_bind_source): Likewise.
9001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9003 * bitmap.c (bitmap_alloc): Adjust.
9004 (bitmap_gc_alloc): Likewise.
9005 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9007 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9009 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9010 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9011 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9012 (bitmap_gc_alloc_stat): Likewise.
9013 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9017 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9018 * rtl.h (shallow_copy_rtx): Remove macro.
9020 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9022 * emit-rtl.c (gen_raw_REG): Adjust.
9023 * gengenrtl.c (gendef): Likewise.
9024 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9025 * rtl.h (rtx_alloc): Remove macro.
9027 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9029 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9030 (build_tree_list_stat): Likewise.
9031 * tree.h (build_tree_list): Remove macro.
9032 (build_tree_list_vec): Likewise.
9034 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9036 * tree.c (make_vector_stat): Remove _stat from name.
9037 (build_vector_stat): Likewise.
9038 * tree.h (make_vector_stat): Remove macro.
9039 (build_vector_stat): Likewise.
9041 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9043 * tree.h (build_var_debug_value): Remove prototype.
9045 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9047 * tree.c (tree_cons_stat): Remove _stat from name.
9048 * tree.h (tree_cons): Remove macro.
9050 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9052 * tree.c (build_vl_exp_stat): Remove _stat from name.
9053 * tree.h (build_vl_exp): Remove macro.
9055 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9057 * tree.c (build_decl_stat): Remove _stat from name.
9058 * tree.h (build_decl): Remove macro.
9060 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9062 * gimple.c (gimple_build_with_ops_stat): Adjust.
9063 (gimple_alloc_stat): Remove _stat from name.
9064 * gimple.h (gimple_alloc): Remove macro.
9066 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9068 * tree.c (make_tree_vec_stat): Remove _stat from name.
9069 (grow_tree_vec_stat): Likewise.
9070 * tree.h (make_tree_vec_stat): Adjust prototype.
9071 (grow_tree_vec_stat): Likewise.
9072 (make_tree_vec): Remove macro.
9073 (grow_tree_vec): Likewise.
9075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9077 * fold-const.c (fold_build1_stat_loc): Adjust.
9078 (fold_build2_stat_loc): Likewise.
9079 (fold_build3_stat_loc): Likewise.
9080 * tree.c (build0_stat): Remove _stat from name.
9081 (build1_stat): Likewise.
9082 (build2_stat): Likewise.
9083 (build3_stat): Likewise.
9084 (build4_stat): Likewise.
9085 (build5_stat): Likewise.
9086 * tree.h (build1_loc): Remove macro, and rename _stat function
9088 (build2_loc): Likewise.
9089 (build3_loc): Likewise.
9090 (build4_loc): Likewise.
9091 (build5_loc): Likewise.
9093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9095 * tree.c (make_int_cst_stat): Remove _stat from name.
9096 * tree.h (make_int_cst_stat): Adjust prototype.
9097 (make_int_cst): Remove macro.
9099 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9101 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9102 * tree.h (make_tree_binfo_stat): Adjust prototype.
9103 (make_tree_binfo): Remove.
9105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9107 * tree.c (copy_node_stat): Rename to copy_node.
9108 (build_distinct_type_copy): Adjust.
9109 * tree.h (copy_node_stat): Adjust prototype.
9110 (copy_node): Remove macro.
9112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9114 * tree.c (make_node_stat): rename to make_node.
9115 (build_tree_list_stat): Adjust.
9116 (build0_stat): Likewise.
9117 (build2_stat): Likewise.
9118 (build3_stat): Likewise.
9119 (build4_stat): Likewise.
9120 (build5_stat): Likewise.
9121 (build_decl_stat): Likewise.
9122 * tree.h (make_node_stat): Adjust prototype.
9123 (make_node): remove macro.
9125 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9127 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9128 (PPC_FEATURE2_SCV): Likewise.
9129 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9131 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9133 * config/aarch64/aarch64.c
9134 (aarch64_internal_mov_immediate): Add new special pattern.
9135 * config/aarch64/aarch64.md (*movdi_aarch64):
9136 Add reg/32bit const mov case.
9138 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9139 Richard Sandiford <richard.sandiford@linaro.org>
9141 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9142 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9143 Add integer and movi cases.
9144 (movi-split-hf-df-sf split, fp16): New.
9145 (enabled): Added TARGET_FP_F16INST.
9146 * config/aarch64/iterators.md (GPF_HF): New.
9147 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9149 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9151 * config/aarch64/aarch64.c
9152 (aarch64_simd_container_mode): Add prototype.
9153 (aarch64_expand_mov_immediate): Add HI support.
9154 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9155 (aarch64_can_const_movi_rtx_p): New.
9156 (aarch64_preferred_reload_class):
9157 Remove restrictions of using FP registers for certain SIMD operations.
9158 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9159 (aarch64_valid_floating_const): Add integer move validation.
9160 (aarch64_simd_imm_scalar_p): Remove.
9161 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9162 (aarch64_legitimate_constant_p): Expand list of supported cases.
9163 * config/aarch64/aarch64-protos.h
9164 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9165 (aarch64_reinterpret_float_as_int): New.
9166 (aarch64_simd_imm_scalar_p): Remove.
9167 * config/aarch64/constraints.md (Uvi): New.
9168 (Dd): Split into Ds and new Dd.
9169 * config/aarch64/aarch64.md (*movsi_aarch64):
9171 (*movdi_aarch64): Add SIMD mov case.
9173 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9175 * tree-predcom.c: (struct chain): Handle store-store chain in which
9176 stores for elimination only store loop invariant values.
9177 (execute_pred_commoning_chain): Ditto.
9178 (prepare_initializers_chain_store_elim): Ditto.
9179 (prepare_finalizers): Ditto.
9180 (is_inv_store_elimination_chain): New function.
9181 (initialize_root_vars_store_elim_1): New function.
9183 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9185 * tree-predcom.c: Revise general description of the pass.
9186 (enum chain_type): New enum type for store elimination.
9187 (struct chain): New field supporting store elimination.
9188 (struct component): Ditto.
9189 (dump_chain): Dump store-stores chain.
9190 (release_chain): Release resources.
9191 (split_data_refs_to_components): Compute and create component
9192 contains only stores for elimination.
9193 (get_chain_last_ref_at): New function.
9194 (make_invariant_chain): Initialization.
9195 (make_rooted_chain): Specify chain type in parameter and record it.
9196 (add_looparound_copies): Skip for store-stores chain.
9197 (determine_roots_comp): Compute type of chain and pass it to
9199 (initialize_root_vars_store_elim_2): New function.
9200 (finalize_eliminated_stores): New function.
9201 (remove_stmt): Handle store for elimination.
9202 (execute_pred_commoning_chain): Execute predictive commoning on
9204 (determine_unroll_factor): Skip unroll for store-stores chain.
9205 (prepare_initializers_chain_store_elim): New function.
9206 (prepare_initializers_chain): Hanlde store-store chain.
9207 (prepare_finalizers_chain, prepare_finalizers): New function.
9208 (tree_predictive_commoning_loop): Return integer value indicating
9209 if loop is unrolled or lcssa form is corrupted.
9210 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9212 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9214 * tree-predcom.c (initialize_root): Delete.
9215 (execute_pred_commoning_chain): Initialize root vars and replace
9216 reference of non-combined chain directly, rather than call above
9219 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9221 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9222 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9224 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9226 * tree-predcom.c (struct chain): New field init_seq.
9227 (release_chain): Release init_seq.
9228 (prepare_initializers_chain): Record intialization stmts in above
9230 (insert_init_seqs): New function.
9231 (tree_predictive_commoning_loop): Call insert_init_seqs.
9233 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9235 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9237 2017-07-28 Richard Biener <rguenther@suse.de>
9239 * match.pd: Remove superfluous :c.
9240 * genmatch.c (simplify::id): Add member.
9241 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9243 (current_id): New global.
9244 (dt_node::parent): Move from ...
9245 (dt_operand::parent): ... here. Add for_id member.
9246 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9247 (decision_tree::find_node): Relax order requirement when
9248 merging DT_TRUE nodes to ones inbetween the current simplify
9249 and the one we try to merge with. Add diagnostic whenever
9250 we need to enforce pattern order by not merging.
9251 (decision_tree::insert): Set current_id.
9252 (decision_tree::print_node): Dump parent node and for_id.
9253 (parser::last_id): Add member.
9254 (parser::push_simplify): Assign unique id.
9255 (parser::parser): Initialize last_id.
9257 2017-07-28 Martin Liska <mliska@suse.cz>
9260 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9261 gimple_build_debug_bind.
9263 2017-07-28 Richard Biener <rguenther@suse.de>
9265 PR tree-optimization/81502
9266 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9268 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9270 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9271 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9272 for BIT_FIELD_REF args.
9273 * fold-const.c (make_bit_field_ref): Likewise.
9274 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9276 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9279 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9280 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9281 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9282 Or it into SANITIZER_UNDEFINED.
9283 * ubsan.c: Include gimple-fold.h and varasm.h.
9284 (ubsan_expand_ptr_ifn): New function.
9285 (instrument_pointer_overflow): New function.
9286 (maybe_instrument_pointer_overflow): New function.
9287 (instrument_object_size): Formatting fix.
9288 (pass_ubsan::execute): Call instrument_pointer_overflow
9289 and maybe_instrument_pointer_overflow.
9290 * internal-fn.c (expand_UBSAN_PTR): New function.
9291 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9292 * sanitizer.def (__ubsan_handle_pointer_overflow,
9293 __ubsan_handle_pointer_overflow_abort): New builtins.
9294 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9295 * internal-fn.def (UBSAN_PTR): New internal function.
9296 * opts.c (sanitizer_opts): Add pointer-overflow.
9297 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9298 * fold-const.c (build_range_check): Compute pointer range check in
9299 integral type if pointer arithmetics would be needed. Formatting
9302 2017-07-28 Martin Liska <mliska@suse.cz>
9305 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9306 parameters that are of a variable-length.
9308 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9310 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9312 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9313 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9314 (CRT_CALL_STATIC_FUNCTION): Likewise.
9315 (ASM_DEFAULT_SPEC): New define.
9316 (ASM_SPEC32): Likewise.
9317 (ASM_SPEC64): Likewise.
9318 (ASM_SPEC_COMMON): Likewise.
9319 (ASM_SPEC): Likewise.
9320 (INVALID_64BIT): Likewise.
9321 (LINK_OS_DEFAULT_SPEC): Likewise.
9322 (LINK_OS_SPEC32): Likewise.
9323 (LINK_OS_SPEC64): Likewise.
9324 (POWERPC_LINUX): Likewise.
9325 (PTRDIFF_TYPE): Likewise.
9326 (RESTORE_FP_PREFIX): Likewise.
9327 (RESTORE_FP_SUFFIX): Likewise.
9328 (SAVE_FP_PREFIX): Likewise.
9329 (SAVE_FP_SUFFIX): Likewise.
9330 (SIZE_TYPE): Likewise.
9331 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9332 (TARGET_64BIT): Likewise.
9333 (TARGET_64BIT): Likewise.
9334 (TARGET_AIX): Likewise.
9335 (WCHAR_TYPE_SIZE): Likewise.
9336 (WCHAR_TYPE): Undefine.
9337 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9338 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9339 (CPP_OS_RTEMS_SPEC): Delete.
9340 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9341 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9343 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9345 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9347 PR tree-optimization/81578
9348 * tree-parloops.c (build_new_reduction): Bail out if
9349 reduction_code isn't one of the standard OpenMP reductions.
9350 Move the details printing after that decision.
9352 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9354 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9355 related to reload_in_progress.
9356 (splat_input_operand): Likewise.
9357 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9359 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9361 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9362 (TARGET_INSTANTIATE_DECLS): Likewise.
9363 (legitimate_indexed_address_p): Delete reload_in_progress code.
9364 (rs6000_debug_legitimate_address_p): Likewise.
9365 (rs6000_eliminate_indexed_memrefs): Likewise.
9366 (rs6000_emit_le_vsx_store): Likewise.
9367 (rs6000_emit_move_si_sf_subreg): Likewise.
9368 (rs6000_emit_move): Likewise.
9369 (register_to_reg_type): Likewise.
9370 (rs6000_pre_atomic_barrier): Likewise.
9371 (rs6000_machopic_legitimize_pic_address): Likewise.
9372 (rs6000_allocate_stack_temp): Likewise.
9373 (rs6000_address_for_fpconvert): Likewise.
9374 (rs6000_address_for_altivec): Likewise.
9375 (rs6000_secondary_memory_needed_rtx): Delete function.
9376 (rs6000_check_sdmode): Likewise.
9377 (rs6000_alloc_sdmode_stack_slot): Likewise.
9378 (rs6000_instantiate_decls): Likewise.
9379 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9380 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9381 Delete reload_in_progress.
9382 (*vec_reload_and_plus_<mptrsize>): Likewise.
9383 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9384 (vsx_div_v2di): Likewise.
9385 (vsx_udiv_v2di): Likewise.
9387 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9389 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9390 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9391 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9392 (rs6000_debug_reg_global): Delete print of LRA status.
9393 (rs6000_option_override_internal): Delete dead LRA related code.
9394 (rs6000_lra_p): Delete function.
9395 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9397 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9399 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9400 * config/riscv/rtems.h: New file.
9402 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9403 Sudakshina Das <sudi.das@arm.com>
9405 * config/aarch64/aarch64.md
9406 (define_split for and<mode>3nr_compare): Move
9407 non aarch64_logical_operand to a register.
9408 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9409 register immediate operand to a register.
9410 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9412 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9415 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9417 2017-07-27 Richard Biener <rguenther@suse.de>
9419 PR tree-optimization/81573
9420 PR tree-optimization/81494
9421 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9422 multi defuse cycle case.
9424 2017-07-27 Richard Biener <rguenther@suse.de>
9426 PR tree-optimization/81571
9427 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9430 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9432 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9433 earlier and only if MASK_FPU is set. Adjust formatting.
9435 2017-07-27 Martin Liska <mliska@suse.cz>
9437 * opt-functions.awk: Add validation of value of Init.
9438 * optc-gen.awk: Pass new argument.
9440 2017-07-27 Martin Liska <mliska@suse.cz>
9442 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9443 Fix wrong condition.
9445 2017-07-27 Martin Liska <mliska@suse.cz>
9447 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9448 BBs and edges seen by autoFDO.
9450 2017-07-27 Richard Biener <rguenther@suse.de>
9452 PR tree-optimization/81502
9453 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9454 with incompatible but same sized type.
9455 (execute_update_addresses_taken): Likewise.
9457 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9459 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9460 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9462 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9465 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9466 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9467 Change s_operand to memory_operand.
9469 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9471 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9472 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9473 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9474 Emit instructions rather than returning an expression. Handle TFmode
9475 and KFmode by casting to TImode.
9476 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9477 (rs6000_emit_le_vsx_store): Likewise.
9478 * config/rs6000/vsx.md (VSX_TI): New iterator.
9479 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9480 (*vsx_le_undo_permute_<mode>): Likewise.
9481 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9482 emit the split sequence.
9483 (*vsx_le_perm_store_<mode>): Likewise.
9485 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9487 PR tree-optimization/81555
9488 PR tree-optimization/81556
9489 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9490 if true, force CHANGED for the recursive invocation.
9491 (reassociate_bb): Remember original length of ops array, pass
9492 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9494 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9495 attributes for noipa attribute. For naked attribute use
9496 lookup_attribute first before lookup_attribute_spec.
9497 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9499 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9500 for functions with noipa attribute.
9501 (cgraph_externally_visible_p): Return true for functions with noipa
9503 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9504 for functions with noipa attribute.
9505 * doc/extend.texi: Document noipa function attribute.
9506 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9507 also for functions with noipa attribute.
9508 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9510 2017-07-26 Andrew Pinski <apinski@cavium.com>
9512 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9513 vec_unalign_load_cost and vec_unalign_store_cost.
9515 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9517 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9518 -mvsx-small-integer option.
9519 (ISA_3_0_MASKS_IEEE): Likewise.
9520 (OTHER_VSX_VECTOR_MASKS): Likewise.
9521 (POWERPC_MASKS): Likewise.
9522 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9523 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9524 code, only testing for DImode being allowed in non-VSX floating
9526 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9527 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9529 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9530 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9531 TARGET_P8_VECTOR test.
9532 (rs6000_secondary_reload_simple_move): Likewise.
9533 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9534 since TARGET_P9_VECTOR was already tested.
9535 (rs6000_opt_masks): Remove -mvsx-small-integer.
9536 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9537 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9539 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9540 test for TARGET_VEXTRACTUB was used, and that uses
9542 (p9 extract splitter): Likewise.
9543 (vsx_extract_<mode>_di_p9): Likewise.
9544 (vsx_extract_<mode>_store_p9): Likewise.
9545 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9546 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9547 the elimination of TARGET_VSX_SMALL_INTEGER.
9548 (vsx_extract_<mode>_p8): Likewise.
9549 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9550 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9551 (vsx_set_<mode>_p9): Likewise.
9552 (vsx_set_v4sf_p9): Likewise.
9553 (vsx_set_v4sf_p9_zero): Likewise.
9554 (vsx_insert_extract_v4sf_p9): Likewise.
9555 (vsx_insert_extract_v4sf_p9_2): Likewise.
9556 * config/rs6000/rs6000.md (sign extend splitter): Change
9557 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9558 (floatsi<mode>2_lfiwax_mem): Likewise.
9559 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9560 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9561 since a test for TARGET_P9_VECTOR was used.
9562 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9563 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9564 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9565 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9566 TARGET_P8_VECTOR test.
9567 (fix_trunc<mode>si2_stfiwx): Likewise.
9568 (fix_trunc<mode>si2_internal): Likewise.
9569 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9570 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9572 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9573 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9574 TARGET_P8_VECTOR test.
9575 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9576 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9577 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9579 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9580 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9581 since a test for TARGET_P9_VECTOR was used.
9582 (splitter for loading small constants): Likewise.
9584 2017-07-26 Andrew Pinski <apinski@cavium.com>
9586 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9589 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9592 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9593 (fp_valid_at): Likewise.
9595 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9597 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9598 (qdf24xx_addrcost_table): Likewise.
9599 (cortexa57_tunings): Update to use generic_branch_cost.
9600 (cortexa72_tunings): Likewise.
9601 (cortexa73_tunings): Likewise.
9602 (qdf24xx_tunings): Likewise.
9604 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9606 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9607 (thunderx2t99_branch_cost): Likewise.
9608 (cortexa35_tunings): Update to use generic_branch_cost.
9609 (cortexa53_tunings): Likewise.
9610 (cortexa57_tunings): Likewise.
9611 (cortexa72_tunings): Likewise.
9612 (cortexa73_tunings): Likewise.
9613 (thunderx2t99_tunings): Likewise.
9615 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9617 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9618 (sparc_option_override): Honour MASK_FSMULD.
9619 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9620 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9621 * config/sparc/sparc.opt (mfsmuld): New option.
9622 * doc/invoke.texi (mfsmuld): Document option.
9624 2017-07-26 Marek Polacek <polacek@redhat.com>
9627 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9629 2017-07-26 Richard Biener <rguenther@suse.de>
9631 * gimple-match-head.c (do_valueize): Return OP if valueize
9633 (get_def): New helper to get at the def stmt of a SSA name
9635 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9636 do_valueize to get at the def stmt.
9637 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9639 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9642 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9644 2017-07-26 Martin Liska <mliska@suse.cz>
9647 * function.c (expand_function_start): Make expansion of
9648 nonlocal_goto_save_area after parm_birth_insn.
9650 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9652 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9653 from all CPU target flags enable members.
9655 2017-07-26 Richard Biener <rguenther@suse.de>
9657 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9658 (decision_tree::gen): Make 'type' const.
9659 (write_predicate): Likewise.
9661 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9663 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9664 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9665 (rs6000_option_override_internal): Likewise.
9666 (rs6000_expand_vector_set): Likewise.
9667 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9668 (TARGET_UPPER_REGS_SF): Likewise.
9669 (TARGET_UPPER_REGS_DI): Likewise.
9670 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9671 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9672 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9673 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9674 (Splitters for DI constants in Altivec registers): Likewise.
9675 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9676 (vsx_set_v4sf_p9): Likewise.
9677 (vsx_set_v4sf_p9_zero): Likewise.
9678 (vsx_insert_extract_v4sf_p9): Likewise.
9679 (vsx_insert_extract_v4sf_p9_2): Likewise.
9681 2017-07-25 Carl Love <cel@us.ibm.com>
9683 * doc/extend.texi: Update the built-in documentation file for the
9684 existing built-in functions
9685 vector signed char vec_cnttz (vector signed char);
9686 vector unsigned char vec_cnttz (vector unsigned char);
9687 vector signed short vec_cnttz (vector signed short);
9688 vector unsigned short vec_cnttz (vector unsigned short);
9689 vector signed int vec_cnttz (vector signed int);
9690 vector unsigned int vec_cnttz (vector unsigned int);
9691 vector signed long long vec_cnttz (vector signed long long);
9692 vector unsigned long long vec_cnttz (vector unsigned long long);
9694 2017-07-25 Andrew Pinski <apinski@cavium.com>
9696 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9697 accesses where the use is for the first operand of a BIT_INSERT.
9699 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9702 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9703 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9705 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9707 * config/i386/gstabs.h: Delete.
9708 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9710 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9712 * config/i386/i386.c (ix86_decompose_address): Do not check for
9713 register RTX when looking at index_reg or base_reg.
9714 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9716 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9718 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9719 to update EH info here.
9721 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9723 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9725 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9727 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9729 2017-07-25 Torsten Duwe <duwe@suse.de>
9731 * common.opt: Introduce -fpatchable-function-entry
9732 command line option, and its variables function_entry_patch_area_size
9733 and function_entry_patch_area_start.
9734 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9735 including a two-value parser.
9736 * target.def (print_patchable_function_entry): New target hook.
9737 * targhooks.h (default_print_patchable_function_entry): New function.
9738 * targhooks.c (default_print_patchable_function_entry): Likewise.
9739 * toplev.c (process_options): Switch off IPA-RA if
9740 patchable function entries are being generated.
9741 * varasm.c (assemble_start_function): Look at the
9742 patchable-function-entry command line switch and current
9743 function attributes and maybe generate NOP instructions by
9744 calling the print_patchable_function_entry hook.
9745 * doc/extend.texi: Document patchable_function_entry attribute.
9746 * doc/invoke.texi: Document -fpatchable_function_entry
9747 command line option.
9748 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9750 * doc/tm.texi: Re-generate.
9752 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9755 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9756 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9758 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9760 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9762 2017-07-25 Richard Biener <rguenther@suse.de>
9764 PR tree-optimization/81455
9765 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9766 not walk in cycles when looking for guards.
9768 2017-07-25 Richard Biener <rguenther@suse.de>
9770 PR tree-optimization/81529
9771 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9772 when optimizing backedge uses.
9774 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9776 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9778 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9779 to dl_section_ref. On AIX, append an expression to subtract
9780 the size of the section length to dl_section_ref.
9782 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9784 * configure.ac: If any of the config.* scripts fail, exit 1.
9785 * configure: Regenerate.
9787 2017-07-25 Richard Biener <rguenther@suse.de>
9790 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9791 of immediate uses, be more verbose on errors.
9793 2017-07-25 Richard Biener <rguenther@suse.de>
9795 PR tree-optimization/81510
9796 * tree-vect-loop.c (vect_is_simple_reduction): When the
9797 reduction stmt is not inside the loop bail out.
9799 2017-07-25 Richard Biener <rguenther@suse.de>
9801 PR tree-optimization/81303
9802 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9803 profitability check against LOOP_VINFO_NITERSM1.
9805 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9807 * domwalk.c (cmp_bb_postorder): Simplify.
9808 (sort_bbs_postorder): New function. Use it...
9809 (dom_walker::walk): ...here to optimize common cases.
9811 2017-07-25 Martin Liska <mliska@suse.cz>
9814 * ipa-visibility.c (function_and_variable_visibility): Make the
9815 redirection just on target that supports aliasing.
9816 Fix GNU coding style.
9818 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9821 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9823 * config/arm/rtems.h: Likewise.
9824 * config/bfin/rtems.h: Likewise.
9825 * config/i386/rtemself.h: Likewise.
9826 * config/lm32/rtems.h: Likewise.
9827 * config/m32c/rtems.h: Likewise.
9828 * config/m68k/rtemself.h: Likewise.
9829 * config/microblaze/rtems.h: Likewise.
9830 * config/mips/rtems.h: Likewise.
9831 * config/moxie/rtems.h: Likewise.
9832 * config/nios2/rtems.h: Likewise.
9833 * config/powerpcspe/rtems.h: Likewise.
9834 * config/rs6000/rtems.h: Likewise.
9835 * config/rtems.h: Likewise.
9836 * config/sh/rtems.h: Likewise.
9837 * config/sh/rtemself.h: Likewise.
9838 * config/sparc/rtemself.h: Likewise.
9840 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9843 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9844 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9845 * tree-ssa-structalias.c (alias_get_name): Same.
9847 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9850 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9851 instructions if no du chain is found.
9853 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9855 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9857 2017-07-25 Richard Biener <rguenther@suse.de>
9860 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9863 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9865 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9867 (ISA_2_7_MASKS_SERVER): Likewise.
9868 (ISA_3_0_MASKS_IEEE): Likewise.
9869 (OTHER_P8_VECTOR_MASKS): Likewise.
9870 (OTHER_VSX_VECTOR_MASKS): Likewise.
9871 (POWERPC_MASKS): Likewise.
9872 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9873 duplicate list of options.
9874 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9875 explicit -mupper-regs options.
9876 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9877 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9878 alias for -mupper-regs-df.
9879 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9880 (rs6000_init_hard_regno_mode_ok): Likewise.
9881 (rs6000_option_override_internal): Likewise.
9882 (rs6000_opt_masks): Likewise.
9883 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9884 options in terms of whether -mvsx or -mpower8-vector was used.
9885 (TARGET_UPPER_REGS_DI): Likewise.
9886 (TARGET_UPPER_REGS_SF): Likewise.
9887 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9888 -mupper-regs-* options.
9890 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9892 * passes.c (emergency_dump_function): Print some empty lines and a
9893 header before the RTL dump.
9895 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9897 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9899 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9902 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9903 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9905 2017-07-24 Carl Love <cel@us.ibm.com>
9907 * config/rs6000/rs6000-c.c: Add support for built-in functions
9908 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9909 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9910 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9911 vec_extract_fp_from_shortl): Add defines for the two builtins.
9912 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9913 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9915 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9916 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9917 * doc/extend.texi: Update the built-in documentation file for the
9918 new built-in function.
9920 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9923 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9925 * doc/generic.texi: Likewise.
9926 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9927 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9929 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9931 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9932 (aarch64_mls_elt_merge<mode>): Likewise.
9934 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9936 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9937 having __cxa_atexit.
9939 2017-07-23 Michael Collison <michael.collison@arm.com>
9941 * config/arm/arm.c (arm_option_override): Deprecate
9942 use of -mstructure-size-boundary.
9943 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9944 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9946 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9949 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9950 Reduce cost estimate for direct moves.
9952 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9955 * config/i386/i386.c (ix86_option_override_internal): Disable
9956 BMI, BMI2 and TBM instructions for -m16.
9958 2017-07-21 Carl Love <cel@us.ibm.com>
9960 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9961 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9962 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9963 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9964 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9965 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9966 VMULOSW): New enum "unspec" values.
9967 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9968 altivec_vmulosw): New patterns.
9969 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9970 VMULOSW): Add definitions.
9972 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9974 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9975 (qdf24xx): Likewise.
9976 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9977 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9978 (AARCH64_FL_V8_1): Renumber.
9979 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9980 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9981 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9982 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9983 rdma to feature modifiers list.
9985 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9988 * ipa-visibility (function_and_variable_visibility): Convert
9989 recursive PLT call to direct call if appropriate.
9991 2017-07-21 Andrew Pinski <apinski@cavium.com>
9993 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9994 operand 1 to see if the types precision matches.
9995 * fold-const.c (operand_equal_p): Likewise.
9997 2017-07-21 Richard Biener <rguenther@suse.de>
9999 PR tree-optimization/81303
10000 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10001 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10002 (vect_peeling_hash_get_lowest_cost): Adjust.
10003 (vect_enhance_data_refs_alignment): Likewise. Use
10004 vect_get_peeling_costs_all_drs to compute the penalty for no
10005 peeling to match up costs.
10007 2017-07-21 Richard Biener <rguenther@suse.de>
10009 PR tree-optimization/81500
10010 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10011 we didn't identify a reduction path.
10013 2017-07-21 Tom de Vries <tom@codesourcery.com>
10014 Cesar Philippidis <cesar@codesourcery.com>
10016 PR gcov-profile/81442
10017 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10020 2017-07-21 Tom de Vries <tom@codesourcery.com>
10023 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10025 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10026 nvptx_override_options_after_change.
10028 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10030 * dwarf2out.c (output_file_names): Avoid double testing for
10031 dwarf_version >= 5.
10033 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10035 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10037 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10039 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10041 (try_crossjump_to_edge): Do not punt on partitioned functions.
10043 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10045 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10046 Put all BBs reachable only via paths crossing cold region to cold
10048 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10050 2016-07-21 Richard Biener <rguenther@suse.de>
10052 PR tree-optimization/81303
10053 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10054 into account prologue and epilogue iterations when raising
10055 min_profitable_iters to sth at least covering one vector iteration.
10057 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10059 * config/arm/arm.c (arm_test_cpu_arch_dat):
10062 2017-07-20 Nathan Sidwell <nathan@acm.org>
10064 Remove TYPE_METHODS.
10065 * tree.h (TYPE_METHODS): Delete.
10066 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10067 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10068 (dbxout_type_methods): Scan TYPE_FIELDS.
10069 (dbxout_type): Don't check TYPE_METHODS here.
10070 * function.c (use_register_for_decl): Always ignore register for
10071 class types when not optimizing.
10072 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10073 * tree.c (free_lang_data_in_type): Stitch out member functions and
10074 templates from TYPE_FIELDS.
10075 (build_distinct_type_copy, verify_type_variant,
10076 verify_type): Member fns are on TYPE_FIELDS.
10077 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10078 * tree-pretty-print.c (dump_generic_node): Likewise.
10080 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10083 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10084 V2TImode and V4TImode.
10085 (ix86_expand_vector_extract): Likewise.
10086 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10087 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10088 (ssescalarmode): Handle V4TImode and V2TImode.
10089 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10090 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10091 (VEXTRACTI128_MODE): New mode iterator.
10092 (splitter for *vec_extractv?ti first element): New.
10093 (VEC_INIT_MODE): New mode iterator.
10094 (vec_init<mode>): Consolidate 3 expanders into one using
10095 VEC_INIT_MODE mode iterator.
10097 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10099 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10100 non_spilled_static_chain_regno_p.
10102 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10104 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10106 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10108 * bb-reorder.c (connect_traces): Allow copying of blocks within
10111 2017-07-20 Richard Biener <rguenther@suse.de>
10113 * gimple.h (gimple_phi_result): Add gphi * overload.
10114 (gimple_phi_result_ptr): Likewise.
10115 (gimple_phi_arg): Likewise. Adjust index assert to only
10116 allow actual argument accesses rather than all slots available
10118 (gimple_phi_arg_def): Add gphi * overload.
10119 * tree-phinodes.c (make_phi_node): Initialize only actual
10121 (resize_phi_node): Clear memory not covered by old node,
10122 do not initialize excess argument slots.
10123 (reserve_phi_args_for_new_edge): Initialize new argument slot
10126 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10128 PR tree-optimization/81388
10130 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10132 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10133 by removing computation of may_be_zero.
10135 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10136 Tom de Vries <tom@codesourcery.com>
10138 PR middle-end/81030
10139 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10140 when gimple level profile disagrees with what RTL expander did.
10142 2017-07-20 Richard Biener <rguenther@suse.de>
10144 PR tree-optimization/61171
10145 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10146 (vect_analyze_stmt): Add slp instance parameter.
10147 (vectorizable_reduction): Likewise.
10148 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10149 (vect_is_simple_reduction): Deal with chains not detected
10150 as SLP reduction chain, specifically not properly associated
10151 chains containing a mix of plus/minus.
10152 (get_reduction_op): Remove.
10153 (get_initial_defs_for_reduction): Simplify, pass in whether
10154 this is a reduction chain, pass in the SLP node for the PHIs.
10155 (vect_create_epilog_for_reduction): Get the SLP instance as
10157 (vectorizable_reduction): Get the SLP instance as arg.
10158 During analysis remember the SLP node with the PHIs in the
10159 instance. Simplify getting at the vectorized reduction PHIs.
10160 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10161 through SLP instance.
10162 (vect_slp_analyze_operations): Likewise.
10163 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10164 (vect_transform_stmt): Likewise.
10166 2017-07-20 Tom de Vries <tom@codesourcery.com>
10168 PR tree-optimization/81489
10169 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10170 read of phi arg location to before loop that modifies phi.
10172 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10174 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10177 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10179 PR middle-end/81331
10180 * except.c (execute): Fix ordering issue.
10182 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10184 PR rtl-optimization/81423
10185 * combine.c (make_compound_operation_int): Don't try to optimize
10186 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10188 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10190 PR rtl-optimization/81423
10191 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10192 with a constant that is -1 in the truncated to mode.
10194 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10196 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10197 (determine_unlikely_bbs): ... here.
10198 * predict.h (propagate_unlikely_bbs_forward): Declare.
10199 * cfgexpand.c (pass_expand::execute): Use it.
10200 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10202 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10203 propagate_unlikely_bbs_forward.
10205 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10207 PR middle-end/81331
10208 * except.c (maybe_add_nop_after_section_switch): New function.
10211 2017-07-19 Tom de Vries <tom@codesourcery.com>
10213 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10215 2017-07-19 Tom de Vries <tom@codesourcery.com>
10217 * gimple.h (gimple_phi_arg): Make assert more strict.
10219 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10221 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10222 * config/rs6000/mmintrin.h: New file.
10223 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10225 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10227 PR tree-optimization/81346
10228 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10230 2017-07-19 Tom de Vries <tom@codesourcery.com>
10232 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10234 2017-07-19 Tom de Vries <tom@codesourcery.com>
10236 * config/nvptx/nvptx-modes.def: Add V2DImode.
10237 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10238 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10239 (nvptx_output_mov_insn): Handle lack of mov.b128.
10240 (nvptx_print_operand): Handle 'H' and 'L' codes.
10241 (nvptx_vector_mode_supported): Allow V2DImode.
10242 (nvptx_preferred_simd_mode): New function.
10243 (nvptx_data_alignment): New function.
10244 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10245 nvptx_preferred_simd_mode.
10246 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10248 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10250 2017-07-19 Tom de Vries <tom@codesourcery.com>
10252 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10253 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10254 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10255 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10256 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10257 (mov<VECIM>_insn): New define_insn.
10258 (define_expand "mov<VECIM>): New define_expand.
10260 2017-07-19 Tom de Vries <tom@codesourcery.com>
10262 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10264 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10266 PR tree-optimization/81346
10267 * fold-const.h (fold_div_compare, range_check_type): Declare.
10268 * fold-const.c (range_check_type): New function.
10269 (build_range_check): Use range_check_type.
10270 (fold_div_compare): No longer static, rewritten into
10271 a match.pd helper function.
10272 (fold_comparison): Don't call fold_div_compare here.
10273 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10274 as helper function.
10276 2017-07-19 Nathan Sidwell <nathan@acm.org>
10278 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10279 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10280 * tree.c (find_decls_types_r, verify_type): Use
10281 TYPE_{MIN,MAX}_VALUE_RAW.
10282 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10283 (hash_tree): Likewise.
10284 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10286 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10289 2017-07-18 Tom de Vries <tom@codesourcery.com>
10291 PR middle-end/81464
10292 * omp-expand.c (expand_omp_for_static_chunk): Handle
10293 equal-argument loop exit phi.
10295 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10298 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10299 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10300 operand 2 predicate.
10301 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10302 operand 2 predicate.
10303 (ror,rol -> rorx splitters): Use const_int_operand as
10304 operand 2 predicate.
10306 2017-06-18 Richard Biener <rguenther@suse.de>
10308 PR tree-optimization/81410
10309 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10310 the gap in the ! slp_perm SLP case after each group.
10312 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10314 PR middle-end/81463
10315 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10318 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10320 PR middle-end/81462
10321 * predict.c (set_even_probabilities): Cleanup; do not affect
10322 probabilities that are already known.
10323 (combine_predictions_for_bb): Call even when count is set.
10325 2017-07-18 Nathan Sidwell <nathan@acm.org>
10327 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10330 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10333 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10334 optimization for loop niter analysis.
10336 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10339 * config/avr/avr.c (avr_optimize_casesi): Don't use
10340 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10342 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10344 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10345 body_cost_vec from _vect_peel_extended_info.
10346 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10347 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10350 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10352 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10354 2017-07-18 Richard Biener <rguenther@suse.de>
10356 PR tree-optimization/80620
10357 PR tree-optimization/81403
10358 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10359 info when re-using a VN table entry.
10361 2017-07-18 Richard Biener <rguenther@suse.de>
10363 PR tree-optimization/81418
10364 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10365 vectype_in. Verify that with lane-reducing reduction operations
10366 we have a single def-use cycle.
10368 2017-07-17 Carl Love <cel@us.ibm.com>
10370 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10372 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10373 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10374 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10375 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10376 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10377 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10378 VMULOSW): New enum "unspec" values.
10379 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10380 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10381 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10382 altivec_vmulosw): New patterns.
10383 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10384 VMULOSW): Add definitions.
10385 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10387 * config/alpha/alpha.c: Include predict.h.
10389 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10391 * tree-vrp.c (compare_assert_loc): Fix comparison function
10392 to return predictable results.
10394 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10396 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10398 (subdi3): Likewise.
10399 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10400 * doc/invoke.texi (mexpand-adddi): Update text.
10402 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10404 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10405 that also clobbers the CC register. The old expand code is moved
10407 (*arc_clzsi2): ... here.
10408 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10409 the CC register. The old expand code is moved to ...
10410 (arc_ctzsi2): ... here.
10412 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10414 * config/arc/arc.opt (mindexed-loads): Use initial value
10415 TARGET_INDEXED_LOADS_DEFAULT.
10416 (mauto-modify-reg): Use initial value
10417 TARGET_AUTO_MODIFY_REG_DEFAULT.
10418 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10419 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10420 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10421 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10423 2017-07-17 Martin Liska <mliska@suse.cz>
10426 * opts.c (finish_options): Do not allow -fgnu-tm
10427 w/ -fsanitize={kernel-,}address. Say sorry.
10429 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10432 * tree-loop-distribution.c (classify_partition): Only assert on
10433 numer of iterations.
10434 (merge_dep_scc_partitions): Delete prameter. Update function call.
10435 (distribute_loop): Remove code handling loop with unknown niters.
10436 (pass_loop_distribution::execute): Skip loop with unknown niters.
10438 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10441 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10442 function sort_partitions_by_post_order.
10444 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10446 PR tree-optimization/81374
10447 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10448 the max index of basic blocks, rather than number of basic blocks.
10450 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10452 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10454 (arc_legitimate_pic_operand_p): Likewise.
10455 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10457 (arc_needs_pcl_p): Likewise.
10458 (arc_legitimate_pc_offset_p): Likewise.
10459 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10460 function is also used in constrains.md.
10461 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10462 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10463 PLUS. Only return true/false in known cases, otherwise assert.
10464 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10465 is already called in arc_legitimate_constant_p.
10466 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10468 (LEGITIMATE_PIC_OPERAND_P): Use
10469 arc_raw_symbolic_reference_mentioned_p function.
10470 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10475 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10476 Andrew Burgess <andrew.burgess@embecosm.com>
10478 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10479 (arc_return_address_register): New function.
10480 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10481 (arc_handle_fndecl_attribute): Add naked attribute.
10482 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10483 (TARGET_WARN_FUNC_RETURN): Likewise.
10484 (arc_allocate_stack_slots_for_args): New function.
10485 (arc_warn_func_return): Likewise.
10486 (machine_function): Change type fn_type.
10487 (arc_compute_function_type): Consider new naked function type,
10488 change function return type.
10489 (arc_must_save_register): Adapt to handle new
10490 arc_compute_function_type's return type.
10491 (arc_expand_prologue): Likewise.
10492 (arc_expand_epilogue): Likewise.
10493 (arc_return_address_regs): Delete.
10494 (arc_return_address_register): New function.
10495 (arc_epilogue_uses): Use above function.
10496 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10497 (arc_function_type): Change encoding, add naked type.
10498 (ARC_INTERRUPT_P): Change to handle the new encoding.
10499 (ARC_FAST_INTERRUPT_P): Likewise.
10500 (ARC_NORMAL_P): Define.
10501 (ARC_NAKED_P): Likewise.
10502 (arc_compute_function_type): Delete prototype.
10503 * config/arc/arc.md (in_ret_delay_slot): Use
10504 arc_return_address_register function.
10505 (simple_return): Likewise.
10506 (p_return_i): Likewise.
10508 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10510 PR tree-optimization/81428
10511 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10512 can't be built for those types.
10514 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10516 Remove stuff dead since r239246.
10518 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10519 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10520 (avr_inform_devices): Remove dead stuff.
10522 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10524 * config/arm/arm_neon.h: Fix softp typo.
10526 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10528 PR tree-optimization/81365
10529 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10530 aggregate moves onto bb predecessor edges, make sure there are no
10531 loads that could alias the lhs in between the start of bb and the
10534 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10537 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10538 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10539 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10541 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10543 PR tree-optimization/81396
10544 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10545 (init_symbolic_number): Initialize it to 1.
10546 (perform_symbolic_merge): Add n_ops from both operands into the new
10548 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10549 without base_addr as useless if they need more than one operation.
10550 (bswap_replace): Handle !bswap case for NULL base_addr.
10552 2017-07-17 Tom de Vries <tom@codesourcery.com>
10555 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10558 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10560 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10561 conditional builtin define __FIX_LEON3FT_B2BST.
10563 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10565 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10566 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10569 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10571 PR rtl-optimization/81424
10572 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10573 to remove potential trapping from operands if -fnon-call-exceptions.
10575 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10577 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10578 profile_proability for scalling.
10579 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10581 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10583 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10585 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10587 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10588 fixpoint arithmetics.
10590 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10592 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10593 fixpoint arithmetics.
10595 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10597 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10598 fixpoint arithmetics.
10600 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10602 * profile-count.h (profile_probability::from_reg_br_prob_note,
10603 profile_probability::to_reg_br_prob_note): New functions.
10604 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10605 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10606 * predict.c (probability_reliable_p): Update.
10607 (edge_probability_reliable_p): Update.
10608 (br_prob_note_reliable_p): Update.
10609 (invert_br_probabilities): Update.
10610 (add_reg_br_prob_note): New function.
10611 (combine_predictions_for_insn): Update.
10612 * asan.c (asan_clear_shadow): Update.
10613 * cfgbuild.c (compute_outgoing_frequencies): Update.
10614 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10615 (update_br_prob_note): Update.
10616 (rtl_verify_edges): Update.
10617 (purge_dead_edges): Update.
10618 (fixup_reorder_chain): Update.
10619 * emit-rtl.c (try_split): Update.
10620 * ifcvt.c (cond_exec_process_insns): Update.
10621 (cond_exec_process_if_block): Update.
10622 (dead_or_predicable): Update.
10623 * internal-fn.c (expand_addsub_overflow): Update.
10624 (expand_neg_overflow): Update.
10625 (expand_mul_overflow): Update.
10626 * loop-doloop.c (doloop_modify): Update.
10627 * loop-unroll.c (compare_and_jump_seq): Update.
10628 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10629 * predict.h: Update.
10630 * reorg.c (mostly_true_jump): Update.
10632 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10633 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10634 * config/arc/arc.c: (emit_unlikely_jump): Update.
10635 * config/arm/arm.c: (emit_unlikely_jump): Update.
10636 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10637 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10638 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10639 (ix86_print_operand): Update.
10640 (ix86_split_fp_branch): Update.
10641 (predict_jump): Update.
10642 * config/ia64/ia64.c (ia64_print_operand): Update.
10643 * config/mmix/mmix.c (mmix_print_operand): Update.
10644 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10645 (rs6000_expand_split_stack_prologue): Update.
10646 * config/rs6000/rs6000.c: Update.
10647 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10648 (s390_expand_vec_movstr): Update.
10649 (s390_expand_cs_tdsi): Update.
10650 (s390_expand_split_stack_prologue): Update.
10651 * config/sh/sh.c (sh_print_operand): Update.
10652 (expand_cbranchsi4): Update.
10653 (expand_cbranchdi4): Update.
10654 * config/sparc/sparc.c (output_v9branch): Update.
10655 * config/spu/spu.c (get_branch_target): Update.
10656 (ea_load_store_inline): Update.
10657 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10658 * config/tilepro/tilepro.c: Update.
10660 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10662 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10663 (gimplify_save_expr): Likewise.
10665 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10667 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10669 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10671 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10673 * timevar.def (TV_IPA_FNSUMMARY): Define.
10675 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10677 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10678 to back store errata sensitive sequence from being generated.
10679 (sqrtdf2_fix): Likewise.
10681 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10683 * tree-ssa-threadupdate.c (compute_path_counts,
10684 update_joiner_offpath_counts): Use profile_probability.
10686 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10689 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10691 * config/arm/arm-c.c (arm_cpu_builtins): Define
10692 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10694 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10696 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10697 array entries to represent __ieee128 versions of the
10698 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10699 scalar_extract_sig, and scalar_insert_exp built-in functions.
10700 (altivec_resolve_overloaded_builtin): Add special case handling
10701 for the __builtin_scalar_insert_exp function, as represented by
10702 the P9V_BUILTIN_VEC_VSIEDP constant.
10703 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10704 exponent support for __ieee128 argument.
10705 (VSESQP): Add scalar extract signature support for __ieee128
10707 (VSTDCNQP): Add scalar test negative support for __ieee128
10709 (VSIEQP): Add scalar insert exponent support for __int128 argument
10710 with __ieee128 result.
10711 (VSIEQPF): Add scalar insert exponent support for __ieee128
10712 argument with __ieee128 result.
10713 (VSTDCQP): Add scalar test data class support for __ieee128
10715 (VSTDCNQP): Add overload support for scalar test negative with
10716 __ieee128 argument.
10717 (VSTDCQP): Add overload support for scalar test data class
10718 __ieee128 argument.
10719 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10720 UNSPEC_VSX_SXSIGDP.
10721 (UNSPEC_VSX_SIEXPQP): New constant.
10722 (xsxexpqp): New insn for VSX scalar extract exponent quad
10724 (xsxsigqp): New insn for VSX scalar extract significand quad
10726 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10727 precision with floating point argument.
10728 (xststdcqp): New expand for VSX scalar test data class quad
10730 (xststdcnegqp): New expand for VSX scalar test negative quad
10732 (xststdcqp): New insn to match expansions for VSX scalar test data
10733 class quad precision and VSX scalar test negative quad precision.
10734 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10735 special case operand checking to enforce that second operand of
10736 VSX scalar test data class with quad precision argument is a 7-bit
10738 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10739 prototypes and descriptions of __ieee128 versions of
10740 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10741 scalar_test_data_class, and scalar_test_neg built-in functions.
10743 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10745 PR tree-optimization/81162
10746 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10747 replace a negate with an add.
10749 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10751 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10753 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10755 * config/arm/arm-c.c (arm_cpu_builtins): Define
10756 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10758 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10760 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10761 (armv8-r): Set ARM Cortex-R52 as default CPU.
10762 * config/arm/arm-tables.opt: Regenerate.
10763 * config/arm/arm-tune.md: Regenerate.
10764 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10766 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10767 extension for -mcpu=cortex-r52.
10769 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10771 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10772 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10773 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10774 (fp-armv8): Define it as FP_ARMv8 only.
10775 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10776 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10778 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10779 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10780 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10781 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10782 than TARGET_FPU_ARMV8.
10783 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10784 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10785 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10787 * config/arm/neon.md (neon_vrint): Likewise.
10788 (neon_vcvt): Likewise.
10789 (neon_<fmaxmin_op><mode>): Likewise.
10790 (<fmaxmin><mode>3): Likewise.
10791 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10792 * config/arm/predicates.md (arm_cond_move_operator): Check against
10793 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10795 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10797 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10798 to top of function.
10800 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10802 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10803 loop in comment with memset.
10805 2017-07-14 Martin Liska <mliska@suse.cz>
10807 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10808 * dwarf2out.c (is_java): Remove the function.
10809 (output_pubname): Remove usage of the function.
10810 (lower_bound_default): Remove usage of DW_LANG_Java.
10811 (gen_compile_unit_die): Likewise.
10812 * gcc.c: Remove compiler defaults for .java and .zip files.
10813 * gimple-expr.c (remove_suffix): Change as there's no longer
10814 extension than 4-letter one.
10815 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10816 (gimplify_save_expr): Likewise.
10817 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10818 as it's possible even for other languages than Java.
10819 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10820 * lto-opts.c (lto_write_options): Remove reference to Java.
10821 * opts.c (strip_off_ending): Update file extension handling.
10822 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10823 * tree-eh.c (lower_resx): Likewise.
10824 * tree.c (free_lang_data_in_type): Remove dead code.
10825 (find_decls_types_r): Likewise.
10826 (build_common_builtin_nodes): Remove Java from a comment.
10827 (verify_type): Remove dead code.
10828 * varasm.c (assemble_external): Remove Java from a comment.
10830 2017-07-14 Martin Liska <mliska@suse.cz>
10832 * opts.c (finish_options): Add quotes.
10833 (common_handle_option): Likewise.
10835 2017-07-14 Martin Liska <mliska@suse.cz>
10837 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10838 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10839 Remove N_SO_PASCAL.
10840 * dwarf2out.c (lower_bound_default): Do not handle
10842 (gen_compile_unit_die): Likewise.
10843 * gcc.c: Remove default extension binding for GNU Pascal.
10844 * stmt.c: Remove Pascal language from a comment.
10845 * xcoffout.c: Likewise.
10847 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10850 * diagnostic-show-locus.c (fixit_cmp): New function.
10851 (layout::layout): Sort m_fixit_hints.
10852 (column_range::column_range): Assert that the values are valid.
10853 (struct char_span): New struct.
10854 (correction::overwrite): New method.
10855 (struct source_line): New struct.
10856 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10857 calls in terms of classes source_line and char_span, and
10858 correction::overwrite.
10859 (selftest::test_overlapped_fixit_printing_2): New function.
10860 (selftest::diagnostic_show_locus_c_tests): Call it.
10862 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10864 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10865 early if there is no lhs.
10867 2017-07-13 Martin Liska <mliska@suse.cz>
10869 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10870 (gen_reference_type_die): Likewise.
10871 * stor-layout.c: Remove Pascal-related comment.
10873 2017-07-13 Martin Liska <mliska@suse.cz>
10875 * opts.c (finish_options): Add quotes to error messages.
10876 (parse_sanitizer_options): Likewise.
10878 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10880 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10882 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10884 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10886 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10888 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10890 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10892 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10895 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10896 provides the hardware capability bits, define the macro
10897 __BUILTIN_CPU_SUPPORTS__.
10898 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10899 if GLIBC does not provide the hardware capability bits. Add a
10900 gcc_unreachable call if the built-in cpu function is neither
10901 __builtin_cpu_is nor __builtin_cpu_supports.
10902 (rs6000_get_function_versions_dispatcher): Change the warning
10903 that an old GLIBC is used which does not export the capability
10904 bits to be an error.
10905 * doc/extend.texi (target_clones attribute): Document the
10906 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10907 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10908 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10909 the macros defined by GCC if the newer GLIBC is available.
10911 2017-07-12 Jeff Law <law@redhat.com>
10913 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10914 remaining includes slightly.
10915 * config/riscv/riscv-builtins.c: Include profile-count.h.
10917 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10920 * config/avr/avr.c (avr_set_current_function): In diagnostic
10921 messages: Quote keywords and (parts of) identifiers.
10922 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10925 2017-07-12 Carl Love <cel@us.ibm.com>
10927 * config/rs6000/rs6000-c.c: Add support for built-in functions
10928 vector bool char vec_revb (vector bool char);
10929 vector bool short vec_revb (vector short char);
10930 vector bool int vec_revb (vector bool int);
10931 vector bool long long vec_revb (vector bool long long);
10932 * doc/extend.texi: Update the built-in documentation file for the
10933 new built-in functions.
10935 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10937 * config/s390/s390.md: Remove movcc splitter.
10939 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10941 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10942 load/store on condition.
10944 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10947 * config/avr/avr.c (avr_encode_section_info)
10948 [progmem && !TREE_READONLY]: Error if progmem object needs
10951 2017-07-11 Michael Collison <michael.collison@arm.com>
10953 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10956 2017-07-11 Carl Love <cel@us.ibm.com>
10958 * config/rs6000/rs6000-c.c: Add support for builtins
10959 vector unsigned int vec_parity_lsbb (vector signed int);
10960 vector unsigned int vec_parity_lsbb (vector unsigned int);
10961 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10962 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10963 vector unsigned long long vec_parity_lsbb (vector signed long long);
10964 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10965 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10966 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10967 * doc/extend.texi: Update the built-in documentation file for the
10968 new built-in functions.
10970 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10972 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10973 (layout::m_primary_loc): New field.
10974 (layout::layout): Initialize new field. Move location filtering
10975 logic from here to...
10976 (layout::maybe_add_location_range): ...this new method. Add
10977 support for filtering to just the lines already specified by other
10979 (layout::will_show_line_p): New method.
10980 (gcc_rich_location::add_location_if_nearby): New method.
10981 (selftest::test_add_location_if_nearby): New test function.
10982 (selftest::diagnostic_show_locus_c_tests): Call it.
10983 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10986 2017-07-11 Tom de Vries <tom@codesourcery.com>
10988 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10989 (bb_first_real_insn): New function.
10990 (nvptx_single): Add extra initialization of broadcasted condition
10993 2017-07-11 Nathan Sidwell <nathan@acm.org>
10995 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10997 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10999 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11000 Binutils doc as TEXI will mess them up.
11001 * doc/invoke.texi (AVR Options): Same here.
11003 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11005 * config/sparc/sparc.opt (mfix-ut700): New option.
11006 (mfix-gr712rc): Likewise.
11007 (sparc_fix_b2bst): New variable.
11008 * doc/invoke.texi (SPARC options): Document them.
11009 (ARM options): Fix warnings.
11010 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11011 instructions to prevent sequences that can trigger the store-store
11012 errata for certain LEON3FT processors.
11013 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11014 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11015 * config/sparc/sparc.md (fix_b2bst): New attribute.
11016 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11018 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11021 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11023 (*rsqrtsf2_sse): Ditto.
11025 (div<mode>3): Macroize insn from divdf3 and divsf3
11026 using MODEF mode iterator.
11028 2017-07-10 Martin Sebor <msebor@redhat.com>
11030 PR tree-optimization/80397
11031 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11032 instead of testing for equality to INTEGER_TYPE.
11034 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11036 * config.gcc: Remove uclibc from arc target spec.
11038 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11040 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11042 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11045 * lto-wrapper.c (remove_option): New function.
11046 (merge_and_complain): Merge PIC/PIE options more realistically.
11048 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11050 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11054 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11055 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11056 * config.in: Regenerate.
11057 * configure: Regenerate.
11058 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11059 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11060 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11061 (TARGET_GASISR_PROLOGUES): ...target mask.
11062 * common/config/avr/avr-common.c
11063 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11064 Set -mgas-isr-prologues.
11065 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11066 INSERT_PASS_BEFORE for it.
11067 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11068 * config/avr/avr.c (avr_option_override)
11069 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11070 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11071 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11072 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11073 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11075 (make_avr_pass_pre_proep): New function.
11076 (emit_push_sfr) <treg>: Add argument to function and use it
11077 instead of TMP_REG.
11078 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11079 and set machine->gasisr.yes.
11080 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11081 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11082 __gcc_isr.n_pushed to .L__stack_usage.
11083 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11084 (avr_asm_final_postscan_insn): ...this new static function.
11085 * config/avr/avr.h (machine_function)
11086 <is_no_gccisr, use_L__stack_usage>: New fields.
11087 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11088 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11089 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11090 (gasisr, *gasisr): New expander and insn.
11091 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11092 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11093 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11095 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11097 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11100 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11102 Move jump-tables out of .text again.
11105 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11106 (ASM_OUTPUT_ADDR_VEC): New function.
11107 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11108 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11109 INSN_ADDRESSes as asm comment.
11110 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11111 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11112 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11113 * config/avr/avr.md (*tablejump): Adjust comment.
11114 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11115 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11117 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11118 (avr_output_addr_vec): New proto.
11119 (avr_log_t) <insn_addresses>: New field.
11121 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11124 * config/i386/i386.c (ix86_function_arg_advance): Set
11125 outgoing_args_on_stack to true if there are outgoing arguments
11127 (ix86_function_arg): Likewise.
11128 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11129 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11130 * config/i386/i386.h (machine_function): Add
11131 outgoing_args_on_stack.
11133 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11135 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11136 supporting pthreds.
11137 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11139 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11141 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11142 (REAL_H): Remove $(MACHMODE_H).
11143 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11145 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11146 $(MACHMODE_H) and double-int.h.
11147 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11149 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11152 2017-07-07 Andrew Pinski <apinski@cavium.com>
11154 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11155 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11157 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11159 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11160 Add warning if GCC was not configured to link against a GLIBC that
11161 exports the hardware capability bits.
11162 (make_resolver_func): Make resolver function private and not a
11163 COMDAT function. Create the name with clone_function_name instead
11164 of make_unique_name.
11167 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11168 correct operand in doing the split.
11170 2017-07-07 Carl Love <cel@us.ibm.com>
11172 * config/rs6000/rs6000-c: Add support for built-in function
11173 vector unsigned short vec_pack_to_short_fp32 (vector float,
11175 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11176 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11177 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11178 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11179 (convert_4f32_8i16): Add define_expand.
11180 * doc/extend.texi: Update the built-in documentation file for the
11181 new built-in function.
11183 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11185 * config/sparc/m8.md: New file.
11186 * config/sparc/sparc.md: Include m8.md.
11188 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11190 * config/sparc/sparc.opt: New option -mvis4b.
11191 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11192 (sparc_option_override): Handle VIS4B.
11193 (enum sparc_builtins): Define
11194 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11195 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11196 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11197 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11198 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11199 (check_constant_argument): New function.
11200 (sparc_vis_init_builtins): Define builtins
11201 __builtin_vis_dictunpack{8,16,32},
11202 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11203 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11204 __builtin_vis_fpcmpde{8,16,32}shl and
11205 __builtin_vis_fpcmpur{8,16,32}shl.
11206 (sparc_expand_builtin): Check that the constant operands to
11207 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11208 constant and in range.
11209 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11211 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11212 (SPARC_IMM5_P): Likewise.
11213 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11214 (enabled): Handle vis4b.
11215 (UNSPEC_DICTUNPACK): New unspec.
11216 (UNSPEC_FPCMPSHL): Likewise.
11217 (UNSPEC_FPUCMPSHL): Likewise.
11218 (UNSPEC_FPCMPDESHL): Likewise.
11219 (UNSPEC_FPCMPURSHL): Likewise.
11220 (cpu_feature): New CPU feature `vis4b'.
11221 (dictunpack{8,16,32}): New insns.
11222 (FPCSMODE): New mode iterator.
11223 (fpcscond): New code iterator.
11224 (fpcsucond): Likewise.
11225 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11226 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11227 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11228 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11229 * config/sparc/constraints.md: Define constraints `q' for unsigned
11230 2-bit integer constants and `t' for unsigned 5-bit integer
11232 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11234 (imm5_operand_dictunpack16): Likewise.
11235 (imm5_operand_dictunpack32): Likewise.
11236 (imm2_operand): Likewise.
11237 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11238 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11239 ditunpack* and fpcmp*shl builtins.
11241 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11243 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11244 * config.in: Add HAVE_AS_SPARC6 define.
11245 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11247 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11249 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11250 (CPP_CPU_SPEC): Handle m8.
11251 (ASM_CPU_SPEC): Likewise.
11252 * config/sparc/sparc-opts.h (enum processor_type): Add
11254 * config/sparc/sparc.c (m8_costs): New struct.
11255 (sparc_option_override): Handle TARGET_CPU_m8.
11256 (sparc32_initialize_trampoline): Likewise.
11257 (sparc64_initialize_trampoline): Likewise.
11258 (sparc_issue_rate): Likewise.
11259 (sparc_register_move_cost): Likewise.
11260 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11261 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11262 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11263 (CPP_CPU_SPEC): Handle M8.
11264 (ASM_CPU_SPEC): Likewise.
11265 (AS_M8_FLAG): Define.
11266 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11267 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11268 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11270 * configure: Regenerate.
11271 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11274 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11276 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11278 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11279 ("*movdi_insn_sp32"): Do not set v3pipe.
11280 ("*movsi_insn"): Likewise.
11281 ("*movdi_insn_sp64"): Likewise.
11282 ("*movsf_insn"): Likewise.
11283 ("*movdf_insn_sp32"): Likewise.
11284 ("*movdf_insn_sp64"): Likewise.
11285 ("*zero_extendsidi2_insn_sp64"): Likewise.
11286 ("*sign_extendsidi2_insn"): Likewise.
11287 ("*mov<VM32:mode>_insn"): Likewise.
11288 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11289 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11290 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11291 ("<vlop:code><VL:mode>3"): Likewise.
11292 ("*not_<vlop:code><VL:mode>3"): Likewise.
11293 ("*nand<VL:mode>_vis"): Likewise.
11294 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11295 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11296 ("one_cmpl<VL:mode>2"): Likewise.
11297 ("faligndata<VM64:mode>_vis"): Likewise.
11298 ("alignaddrsi_vis"): Likewise.
11299 ("alignaddrdi_vis"): Likweise.
11300 ("alignaddrlsi_vis"): Likewise.
11301 ("alignaddrldi_vis"): Likewise.
11302 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11303 ("bmaskdi_vis"): Likewise.
11304 ("bmasksi_vis"): Likewise.
11305 ("bshuffle<VM64:mode>_vis"): Likewise.
11306 ("cmask8<P:mode>_vis"): Likewise.
11307 ("cmask16<P:mode>_vis"): Likewise.
11308 ("cmask32<P:mode>_vis"): Likewise.
11309 ("pdistn<P:mode>_vis"): Likewise.
11310 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11312 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11314 * config/sparc/sparc.md ("subtype"): New insn attribute.
11315 ("*wrgsr_sp64"): Set insn subtype.
11316 ("*rdgsr_sp64"): Likewise.
11317 ("alignaddrsi_vis"): Likewise.
11318 ("alignaddrdi_vis"): Likewise.
11319 ("alignaddrlsi_vis"): Likewise.
11320 ("alignaddrldi_vis"): Likewise.
11321 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11322 ("fexpand_vis"): Likewise.
11323 ("fpmerge_vis"): Likewise.
11324 ("faligndata<VM64:mode>_vis"): Likewise.
11325 ("bshuffle<VM64:mode>_vis"): Likewise.
11326 ("cmask8<P:mode>_vis"): Likewise.
11327 ("cmask16<P:mode>_vis"): Likewise.
11328 ("cmask32<P:mode>_vis"): Likewise.
11329 ("fchksm16_vis"): Likewise.
11330 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11331 ("fmean16_vis"): Likewise.
11332 ("fp<plusminus_insn>64_vis"): Likewise.
11333 ("<plusminus_insn>v8qi3"): Likewise.
11334 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11335 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11336 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11337 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11338 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11339 ("*movqi_insn"): Likewise.
11340 ("*movhi_insn"): Likewise.
11341 ("*movsi_insn"): Likewise.
11342 ("movsi_pic_gotdata_op"): Likewise.
11343 ("*movdi_insn_sp32"): Likewise.
11344 ("*movdi_insn_sp64"): Likewise.
11345 ("movdi_pic_gotdata_op"): Likewise.
11346 ("*movsf_insn"): Likewise.
11347 ("*movdf_insn_sp32"): Likewise.
11348 ("*movdf_insn_sp64"): Likewise.
11349 ("*zero_extendhisi2_insn"): Likewise.
11350 ("*zero_extendqihi2_insn"): Likewise.
11351 ("*zero_extendqisi2_insn"): Likewise.
11352 ("*zero_extendqidi2_insn"): Likewise.
11353 ("*zero_extendhidi2_insn"): Likewise.
11354 ("*zero_extendsidi2_insn_sp64"): Likewise.
11355 ("ldfsr"): Likewise.
11356 ("prefetch_64"): Likewise.
11357 ("prefetch_32"): Likewise.
11358 ("tie_ld32"): Likewise.
11359 ("tie_ld64"): Likewise.
11360 ("*tldo_ldub_sp32"): Likewise.
11361 ("*tldo_ldub1_sp32"): Likewise.
11362 ("*tldo_ldub2_sp32"): Likewise.
11363 ("*tldo_ldub_sp64"): Likewise.
11364 ("*tldo_ldub1_sp64"): Likewise.
11365 ("*tldo_ldub2_sp64"): Likewise.
11366 ("*tldo_ldub3_sp64"): Likewise.
11367 ("*tldo_lduh_sp32"): Likewise.
11368 ("*tldo_lduh1_sp32"): Likewise.
11369 ("*tldo_lduh_sp64"): Likewise.
11370 ("*tldo_lduh1_sp64"): Likewise.
11371 ("*tldo_lduh2_sp64"): Likewise.
11372 ("*tldo_lduw_sp32"): Likewise.
11373 ("*tldo_lduw_sp64"): Likewise.
11374 ("*tldo_lduw1_sp64"): Likewise.
11375 ("*tldo_ldx_sp64"): Likewise.
11376 ("*mov<VM32:mode>_insn"): Likewise.
11377 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11378 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11380 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11382 * config/sparc/sparc.md ("type"): New insn type viscmp.
11383 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11385 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11386 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11387 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11388 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11390 ("n7_vis_logical_11cycle"): Likewise.
11391 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11392 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11393 * config/sparc/niagara.md ("niag_vis"): Likewise.
11394 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11395 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11397 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11399 * config/sparc/sparc.md: New instruction type `bmask'.
11400 (bmaskdi_vis): Use the `bmask' type.
11401 (bmasksi_vis): Likewise.
11402 * config/sparc/ultra3.md (us3_array): Likewise.
11403 * config/sparc/niagara7.md (n7_array): Likewise.
11404 * config/sparc/niagara4.md (n4_array): Likewise.
11405 * config/sparc/niagara2.md (niag2_vis): Likewise.
11406 (niag3_vis): Likewise.
11407 * config/sparc/niagara.md (niag_vis): Likewise.
11409 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11411 * ipa-comdats.c: Remove optimize check from gate.
11412 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11413 for functions not optimized.
11414 (ipa_fn_summary_read): Skip optimize check.
11415 (ipa_fn_summary_write): Likewise.
11416 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11418 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11420 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11421 (check_callers): Give up on caller which is not optimized.
11422 (inline_small_functions): Likewise.
11423 (ipa_inline): Do not give up when not optimizing.
11424 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11425 away unoptimizes cdtors.
11426 (whole_program_function_and_variable_visibility): Do
11427 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11428 * ipa.c (process_references): Do not check optimize.
11429 (symbol_table::remove_unreachable_nodes): Update optimize check.
11430 (set_writeonly_bit): Update optimize check.
11431 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11432 (pass_ipa_single_use::gate): Remove.
11434 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11436 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11437 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11438 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11439 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11440 permute_load, permute_store, adjust_extract, adjust_splat,
11441 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11442 replace_swap_with_copy, dump_swap_insn_table,
11443 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11444 recombine_lvx_pattern, recombine_stvx_pattern,
11445 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11446 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11447 to file rs6000-p8swap.c.
11448 * config/rs6000/rs6000-p8swap.c: New file.
11449 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11450 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11451 and rs6000*-*-* targets.
11453 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11455 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11457 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11459 * lto-wrapper.c (merge_and_complain): Do not merge
11460 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11461 fsigned_zeros, ftrapping_math, fwrapv.
11462 (append_compiler_options): Do not track these options.
11463 (append_linker_options): Likewie
11465 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11467 * cgraphunit.c (cgraph_node::finalize_function): When
11468 !flag_toplevel_reorde set no_reorder flag.
11469 (varpool_node::finalize_decl): Likewise.
11470 (symbol_table::compile): Drop no toplevel reorder path.
11472 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11474 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11475 edges; zero probability is not better than uninitialized.
11477 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11479 * asan.h (asan_sanitize_allocas_p): Declare.
11480 * asan.c (asan_sanitize_allocas_p): New function.
11481 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11482 (handle_builtin_alloca): Likewise.
11483 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11484 if !asan_sanitize_allocas_p.
11485 * params.def (asan-instrument-allocas): Add new option.
11486 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11487 * opts.c (common_handle_option): Disable allocas sanitization for
11490 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11492 * asan.c: Include gimple-fold.h.
11493 (get_last_alloca_addr): New function.
11494 (handle_builtin_stackrestore): Likewise.
11495 (handle_builtin_alloca): Likewise.
11496 (asan_emit_allocas_unpoison): Likewise.
11497 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11498 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11499 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11500 (instrument_builtin_call): Pass gimple iterator to
11501 get_mem_refs_of_builtin_call.
11502 (last_alloca_addr): New global.
11503 * asan.h (asan_emit_allocas_unpoison): Declare.
11504 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11505 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11506 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11507 if function calls alloca.
11508 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11509 * gimple-fold.h (replace_call_with_value): Declare.
11510 * internal-fn.c: Include asan.h.
11511 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11512 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11514 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11516 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11517 (C_SELFTEST_FLAGS): New.
11518 (CPP_SELFTEST_FLAGS): New.
11519 (SELFTEST_DEPS): New, from deps of s-selftest.
11520 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11521 (CPP_SELFTEST_DEPS): New.
11522 (selftest): Add dependency on s-selftest-c++.
11523 (s-selftest): Rename to...
11524 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11525 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11526 than SELFTEST_FLAGS.
11527 (selftest-gdb): Rename to...
11528 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11530 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11531 (selftest-valgrind): Rename to...
11532 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11534 (selftest-valgrind): Reintroduce as an alias for
11535 selftest-c-valgrind.
11536 (s-selftest-c++): New.
11537 (selftest-c++-gdb): New.
11538 (selftest-c++-valgrind): New.
11540 2017-07-06 Olivier Hainque <hainque@adacore.com>
11542 * gcc.c (process_command): When deciding if undefined variables
11543 should be ignored when processing specs, accept "gcc -v" as well.
11545 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11547 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11548 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11550 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11552 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11553 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11554 * config/arm/arm-tables.opt: Regenerate.
11555 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11557 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11559 2017-07-06 Carl Love <cel@us.ibm.com>
11561 * ChangeLog: Clean up from mid air collision
11563 2017-07-06 Carl Love <cel@us.ibm.com>
11565 * config/rs6000/rs6000-c.c: Add support for built-in functions
11566 vector signed int vec_subc (vector signed int, vector signed int);
11567 vector signed __int128 vec_subc (vector signed __int128,
11568 vector signed __int128);
11569 vector unsigned __int128 vec_subc (vector unsigned __int128,
11570 vector unsigned __int128);
11571 vector signed int vec_sube (vector signed int, vector signed int,
11572 vector signed int);
11573 vector unsigned int vec_sube (vector unsigned int,
11574 vector unsigned int,
11575 vector unsigned int);
11576 vector signed __int128 vec_sube (vector signed __int128,
11577 vector signed __int128,
11578 vector signed__int128);
11579 vector unsigned __int128 vec_sube (vector unsigned __int128,
11580 vector unsigned __int128,
11581 vector unsigned __int128);
11582 vector signed int vec_subec (vector signed int, vector signed int,
11583 vector signed int);
11584 vector unsigned int vec_subec (vector unsigned int,
11585 vector unsigned int,
11586 vector unsigned int);
11587 vector signed __int128 vec_subec (vector signed __int128,
11588 vector signed __int128,
11589 vector signed__int128);
11590 vector unsigned __int128 vec_subec (vector unsigned __int128,
11591 vector unsigned __int128,
11592 vector unsigned __int128);
11593 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11594 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11595 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11596 BU_ALTIVEC_OVERLOAD_X definitions.
11597 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11598 * doc/extend.texi: Update the built-in documentation file for the new
11599 built-in functions.
11601 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11604 * diagnostic-show-locus.c (layout::layout): Use start and finish
11605 spelling location for the start and finish of each range.
11606 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11607 Add unused aspect param.
11608 * input.c (expand_location_1): Add "aspect" param, and use it
11609 to access the correct part of the location.
11610 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11612 (expand_location_to_spelling_point): Likewise.
11613 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11614 param, and pass it to expand_location_1.
11616 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11618 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11619 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11620 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11621 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11622 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11623 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11624 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11625 _mm_maskz_getmant_ss): New intrinsics.
11626 (__builtin_ia32_getexpss128_mask): Changed to ...
11627 __builtin_ia32_getexpss128_round ... this.
11628 (__builtin_ia32_getexpsd128_mask): Changed to ...
11629 __builtin_ia32_getexpsd128_round ... this.
11630 * config/i386/i386-builtin-types.def
11631 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11632 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11633 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11634 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11635 __builtin_ia32_getmantss_mask_round): New builtins.
11636 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11637 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11638 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11639 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11640 * config/i386/sse.md
11641 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11642 avx512f_sgetexp<mode><mask_scalar_name>
11643 <round_saeonly_scalar_name> ... this.
11644 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11645 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11646 vgetexp<ssescalarmodesuffix>
11647 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11648 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11649 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11650 avx512f_vgetmant<mode><mask_scalar_name>
11651 <round_saeonly_scalar_name> ... this.
11652 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11653 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11654 vgetmant<ssescalarmodesuffix>
11655 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11656 %0<mask_scalar_operand4>, %1, %2
11657 <round_saeonly_scalar_mask_op4>, %3} ... this.
11658 * config/i386/subst.md (mask_scalar_operand4,
11659 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11660 round_saeonly_scalar_nimm_predicate): New subst attributes.
11662 2017-07-06 Julia Koval <julia.koval@intel.com>
11664 * config/i386/i386.c (ix86_erase_embedded_rounding):
11665 Remove code for old rounding pattern.
11667 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11669 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11671 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11673 * doc/sourcebuild.texi (Test Directives, Variants of
11674 dg-require-support): Add documentation for dg-require-stack-check.
11676 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11678 * config/i386/subst.md (mask_scalar, round_scalar,
11679 round_saeonly_scalar): New meta-templates.
11680 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11681 round_scalar_mask_operand3, round_scalar_mask_op3,
11682 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11683 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11684 round_saeonly_scalar_constraint,
11685 round_saeonly_scalar_prefix): New subst attribute.
11686 * config/i386/sse.md
11687 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11688 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11689 <round_scalar_name> ... this.
11690 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11691 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11692 <round_scalar_name> ... this.
11693 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11694 <sse>_vm<code><mode>3<mask_scalar_name>
11695 <round_saeonly_scalar_name> ... this.
11696 (v<plusminus_mnemonic><ssescalarmodesuffix>
11697 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11698 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11699 v<plusminus_mnemonic><ssescalarmodesuffix>
11700 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11701 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11702 (v<multdiv_mnemonic><ssescalarmodesuffix>
11703 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11704 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11705 v<multdiv_mnemonic><ssescalarmodesuffix>
11706 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11707 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11708 (v<maxmin_float><ssescalarmodesuffix>
11709 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11710 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11711 v<maxmin_float><ssescalarmodesuffix>
11712 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11713 %0<mask_scalar_operand3>, %1, %<iptr>2
11714 <round_saeonly_scalar_mask_op3>} ... this.
11716 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11718 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11719 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11721 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11722 Alan Hayward <alan.hayward@arm.com>
11723 David Sherwood <david.sherwood@arm.com>
11725 * combine.c (simplify_if_then_else): Remove "enum" before
11727 * compare-elim.c (can_eliminate_compare): Likewise.
11728 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11730 (aarch64_lookup_simd_builtin_type): Likewise.
11731 (aarch64_simd_builtin_type): Likewise.
11732 (aarch64_init_simd_builtin_types): Likewise.
11733 (aarch64_simd_expand_args): Likewise.
11734 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11736 (aarch64_reverse_mask): Likewise.
11737 (aarch64_simd_emit_reg_reg_move): Likewise.
11738 (aarch64_gen_adjusted_ldpstp): Likewise.
11739 (aarch64_ccmp_mode_to_code): Likewise.
11740 (aarch64_operands_ok_for_ldpstp): Likewise.
11741 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11742 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11744 (aarch64_min_divisions_for_recip_mul): Likewise.
11745 (aarch64_reassociation_width): Likewise.
11746 (aarch64_get_condition_code_1): Likewise.
11747 (aarch64_simd_emit_reg_reg_move): Likewise.
11748 (aarch64_simd_attr_length_rglist): Likewise.
11749 (aarch64_reverse_mask): Likewise.
11750 (aarch64_operands_ok_for_ldpstp): Likewise.
11751 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11752 (aarch64_gen_adjusted_ldpstp): Likewise.
11753 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11755 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11756 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11757 (arm_lookup_simd_builtin_type): Likewise.
11758 (arm_simd_builtin_type): Likewise.
11759 (arm_init_simd_builtin_types): Likewise.
11760 (arm_expand_builtin_args): Likewise.
11761 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11762 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11763 (ft32_setup_incoming_varargs): Likewise.
11764 (ft32_function_arg): Likewise.
11765 (ft32_function_arg_advance): Likewise.
11766 (ft32_pass_by_reference): Likewise.
11767 (ft32_arg_partial_bytes): Likewise.
11768 (ft32_valid_pointer_mode): Likewise.
11769 (ft32_addr_space_pointer_mode): Likewise.
11770 (ft32_addr_space_legitimate_address_p): Likewise.
11771 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11773 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11774 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11775 (iamcu_alignment): Likewise.
11776 (canonicalize_vector_int_perm): Likewise.
11777 (ix86_noce_conversion_profitable_p): Likewise.
11778 (ix86_mpx_bound_mode): Likewise.
11779 (ix86_operands_ok_for_move_multiple): Likewise.
11780 * config/microblaze/microblaze-protos.h
11781 (microblaze_expand_conditional_branch_reg): Likewise.
11782 * config/microblaze/microblaze.c
11783 (microblaze_expand_conditional_branch_reg): Likewise.
11784 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11786 (rs6000_reassociation_width): Likewise.
11787 (rs6000_invalid_binary_op): Likewise.
11788 (fusion_p9_p): Likewise.
11789 (emit_fusion_p9_load): Likewise.
11790 (emit_fusion_p9_store): Likewise.
11791 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11793 (riscv_hard_regno_mode_ok_p): Likewise.
11794 (riscv_address_insns): Likewise.
11795 (riscv_split_symbol): Likewise.
11796 (riscv_legitimize_move): Likewise.
11797 (riscv_function_value): Likewise.
11798 (riscv_hard_regno_nregs): Likewise.
11799 (riscv_expand_builtin): Likewise.
11800 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11801 (riscv_build_integer): Likewise.
11802 (riscv_split_integer): Likewise.
11803 (riscv_legitimate_constant_p): Likewise.
11804 (riscv_cannot_force_const_mem): Likewise.
11805 (riscv_regno_mode_ok_for_base_p): Likewise.
11806 (riscv_valid_base_register_p): Likewise.
11807 (riscv_valid_offset_p): Likewise.
11808 (riscv_valid_lo_sum_p): Likewise.
11809 (riscv_classify_address): Likewise.
11810 (riscv_legitimate_address_p): Likewise.
11811 (riscv_address_insns): Likewise.
11812 (riscv_load_store_insns): Likewise.
11813 (riscv_force_binary): Likewise.
11814 (riscv_split_symbol): Likewise.
11815 (riscv_force_address): Likewise.
11816 (riscv_legitimize_address): Likewise.
11817 (riscv_move_integer): Likewise.
11818 (riscv_legitimize_const_move): Likewise.
11819 (riscv_legitimize_move): Likewise.
11820 (riscv_address_cost): Likewise.
11821 (riscv_subword): Likewise.
11822 (riscv_output_move): Likewise.
11823 (riscv_canonicalize_int_order_test): Likewise.
11824 (riscv_emit_int_order_test): Likewise.
11825 (riscv_function_arg_boundary): Likewise.
11826 (riscv_pass_mode_in_fpr_p): Likewise.
11827 (riscv_pass_fpr_single): Likewise.
11828 (riscv_pass_fpr_pair): Likewise.
11829 (riscv_get_arg_info): Likewise.
11830 (riscv_function_arg): Likewise.
11831 (riscv_function_arg_advance): Likewise.
11832 (riscv_arg_partial_bytes): Likewise.
11833 (riscv_function_value): Likewise.
11834 (riscv_pass_by_reference): Likewise.
11835 (riscv_setup_incoming_varargs): Likewise.
11836 (riscv_print_operand): Likewise.
11837 (riscv_elf_select_rtx_section): Likewise.
11838 (riscv_save_restore_reg): Likewise.
11839 (riscv_for_each_saved_reg): Likewise.
11840 (riscv_register_move_cost): Likewise.
11841 (riscv_hard_regno_mode_ok_p): Likewise.
11842 (riscv_hard_regno_nregs): Likewise.
11843 (riscv_class_max_nregs): Likewise.
11844 (riscv_memory_move_cost): Likewise.
11845 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11846 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11847 (rl78_addr_space_address_mode): Likewise.
11848 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11850 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11851 (rs6000_reassociation_width): Likewise.
11852 (rs6000_invalid_binary_op): Likewise.
11853 (fusion_p9_p): Likewise.
11854 (emit_fusion_p9_load): Likewise.
11855 (emit_fusion_p9_store): Likewise.
11856 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11857 (ok_for_simple_move_operands): Likewise.
11858 (ok_for_simple_move_strict_operands): Likewise.
11859 (ok_for_simple_arith_logic_operands): Likewise.
11860 (visium_legitimize_reload_address): Likewise.
11861 (visium_select_cc_mode): Likewise.
11862 (output_cbranch): Likewise.
11863 (visium_split_double_move): Likewise.
11864 (visium_expand_copysign): Likewise.
11865 (visium_expand_int_cstore): Likewise.
11866 (visium_expand_fp_cstore): Likewise.
11867 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11868 (visium_function_arg): Likewise.
11869 (visium_function_arg_advance): Likewise.
11870 (visium_libcall_value): Likewise.
11871 (visium_setup_incoming_varargs): Likewise.
11872 (visium_legitimate_constant_p): Likewise.
11873 (visium_legitimate_address_p): Likewise.
11874 (visium_legitimize_address): Likewise.
11875 (visium_secondary_reload): Likewise.
11876 (visium_register_move_cost): Likewise.
11877 (visium_memory_move_cost): Likewise.
11878 (prepare_move_operands): Likewise.
11879 (ok_for_simple_move_operands): Likewise.
11880 (ok_for_simple_move_strict_operands): Likewise.
11881 (ok_for_simple_arith_logic_operands): Likewise.
11882 (visium_function_value_1): Likewise.
11883 (rtx_ok_for_offset_p): Likewise.
11884 (visium_legitimize_reload_address): Likewise.
11885 (visium_split_double_move): Likewise.
11886 (visium_expand_copysign): Likewise.
11887 (visium_expand_int_cstore): Likewise.
11888 (visium_expand_fp_cstore): Likewise.
11889 (visium_split_cstore): Likewise.
11890 (visium_select_cc_mode): Likewise.
11891 (visium_split_cbranch): Likewise.
11892 (output_cbranch): Likewise.
11893 (visium_print_operand_address): Likewise.
11894 * expmed.c (flip_storage_order): Likewise.
11895 * expmed.h (emit_cstore): Likewise.
11896 (flip_storage_order): Likewise.
11897 * genrecog.c (validate_pattern): Likewise.
11898 * hsa-gen.c (gen_hsa_addr): Likewise.
11899 * internal-fn.c (expand_arith_overflow): Likewise.
11900 * ira-color.c (allocno_copy_cost_saving): Likewise.
11901 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11902 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11903 (process_invariant_for_inheritance): Likewise.
11904 * lra-eliminations.c (move_plus_up): Likewise.
11905 * omp-low.c (lower_oacc_reductions): Likewise.
11906 * simplify-rtx.c (simplify_subreg): Likewise.
11907 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11908 (TARGET_CHKP_BOUND_MODE): Likewise..
11909 * targhooks.c (default_chkp_bound_mode): Likewise.
11910 (default_setup_incoming_vararg_bounds): Likewise.
11911 * targhooks.h (default_chkp_bound_mode): Likewise.
11912 (default_setup_incoming_vararg_bounds): Likewise.
11913 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11914 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11915 (have_whole_vector_shift): Likewise.
11916 * tree-vect-stmts.c (vectorizable_load): Likewise.
11917 * doc/tm.texi: Regenerate.
11919 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11921 Graceful degrade if Binutils PR21472 is not available.
11924 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11925 .rodata in flash test fails.
11926 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11927 * confgure: Regenerate.
11928 * config.in: Regenerate.
11929 * config/avr/avr.c (avr_asm_named_section)
11930 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11931 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11932 (avr_asm_init_sections): Same.
11934 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11936 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11937 (fma<VH:mode>4_intrinsic): Likewise.
11938 (*fmsub<VCVTF:mode>4): Likewise.
11939 (*fmsub<VH:mode>4_intrinsic): Likewise.
11941 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11944 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11945 Don't depend on "optimize > 0".
11946 (out_movhi_r_mr, out_movqi_mr_r): Same.
11947 (out_movhi_mr_r, out_movqi_r_mr): Same.
11948 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11949 io_address_operand on "optimize > 0".
11951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11953 * tree-loop-distribution.c: Add general explanantion on the pass.
11954 (generate_loops_for_partition): Mark distributed loop.
11955 (pg_add_dependence_edges): New parameter. Handle alias data
11956 dependence specially and record it in the parameter if asked.
11957 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11958 (init_partition_graph_vertices, add_partition_graph_edge): New.
11959 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11960 (free_partition_graph_vdata, build_partition_graph): New.
11961 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11962 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11963 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11964 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11965 (version_for_distribution_p, finalize_partitions): New.
11966 (distribute_loop): Handle alias data dependence specially. Factor
11967 out loop fusion code as functions and call these functions.
11969 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11971 * tree-loop-distribution.c (classify_partition): New parameter and
11972 better handle reduction statement.
11973 (rdg_build_partitions): Revise comment.
11974 (distribute_loop): Compute statements in all partitions and pass it
11975 to classify_partition.
11977 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11979 * tree-loop-distribution.c (enum partition_type): New.
11980 (struct partition): New field type.
11981 (partition_merge_into): Add parameter. Update partition type.
11982 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11983 (build_rdg_partition_for_vertex): Compute partition type.
11984 (rdg_build_partitions): Dump partition type.
11985 (distribute_loop): Update calls to partition_merge_into.
11987 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11989 * tree-loop-distribution.c (struct ddr_hasher): New.
11990 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11992 (classify_partition): Call get_data_dependence.
11993 (pg_add_dependence_edges): Ditto.
11994 (distribute_loop): Release data dependence hash table.
11996 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11998 * tree-loop-distribution.c (ref_base_address): Delete.
11999 (similar_memory_accesses): Rename ...
12000 (share_memory_accesses): ... to this. Check if partitions access
12001 the same memory reference.
12002 (distribute_loop): Call share_memory_accesses.
12004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12006 * tree-loop-distribution.c (struct partition): New field recording
12007 its data reference.
12008 (partition_alloc, partition_free): Init and release data refs.
12009 (partition_merge_into): Merge data refs.
12010 (build_rdg_partition_for_vertex): Collect data refs for partition.
12011 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12013 (distribute_loop): Remve data refs from vertice data of partition
12016 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12018 * tree-loop-distribution.c (params.h): Include header file.
12019 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12020 (datarefs_vec): New global var.
12021 (create_rdg_vertices): Use datarefs_vec directly.
12022 (free_rdg): Don't free data references.
12023 (build_rdg): Update use. Don't free data references.
12024 (distribute_loop): Compute global variable for data references.
12025 Bail out if there are too many data references.
12027 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12029 * tree-loop-distribution.c (loop_nest): New global var.
12030 (build_rdg): Use loop directly, rather than loop nest.
12031 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12033 (distribute_loop): Compute global variable loop nest. Update use.
12035 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12037 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12038 (partition_merge_into): New parameter. Dump reason for fusion.
12039 (distribute_loop): Update use of partition_merge_into.
12041 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12043 * tree-loop-distribution.c (bb_top_order_index): New.
12044 (bb_top_order_index_size, bb_top_order_cmp): New.
12045 (stmts_from_loop): Use topological order.
12046 (pass_loop_distribution::execute): Compute and release topological
12047 order for basic blocks.
12049 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12051 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12054 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12056 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12057 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12058 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12059 * internal-fn.def (LOOP_DIST_ALIAS): New.
12060 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12061 (fold_loop_internal_call): ... this.
12062 (vect_loop_dist_alias_call): New function.
12063 (set_uid_loop_bbs): Call fold_loop_internal_call.
12064 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12067 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12070 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12071 Require dead FLAGS_REG at the beginning of a peephole.
12073 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12076 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12077 arguments in the call to __builtin_ia32_sbb_u32.
12078 (_subborrow_u64): Swap _X and _Y arguments in the call to
12079 __builtin_ia32_sbb_u64.
12081 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12084 * tree-vrp.c (compare_assert_loc): Turn into a function template
12085 with stable template parameter. Only test if a->e is NULL,
12086 !a->e == !b->e has been verified already. Use e == NULL or
12087 e != NULL instead of e or ! e tests. If stable is true, don't use
12088 iterative_hash_expr, on the other side allow a or b or both NULL
12089 and sort the NULLs last.
12090 (process_assert_insertions): Sort using compare_assert_loc<false>
12091 instead of compare_assert_loc, later sort using
12092 compare_assert_loc<true> before calling process_assert_insertions_for
12093 in a loop. Use break instead of continue once seen NULL pointer.
12095 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12097 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12098 Cortex-R7 and Cortex-R8 processors.
12100 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12102 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12103 uninitialized while src is not.
12105 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12107 * common/config/arm/arm-common.c: Adjust include path for
12109 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12110 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12111 (arm-cpu-data.h): Likewise.
12112 (arm-cpu-cdata.h): Likewise.
12113 * config/arm/arm-cpu.h: Delete.
12114 * config/arm/arm-cpu-cdata.h: Delete.
12115 * config/arm/arm-cpu-data.h: Delete.
12117 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12119 * config/arm/arm-cpus.in (cortex-a55): New.
12120 (cortex-a75): Likewise.
12121 (cortex-a75.cortex-a55): Likewise.
12122 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12124 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12125 * config/arm/arm-cpu-cdata.h: Regenerate.
12126 * config/arm/arm-cpu-data.h: Regenerate.
12127 * config/arm/arm-cpu.h: Regenerate.
12128 * config/arm/arm-tables.opt: Regenerate.
12129 * config/arm/arm-tune.md: Regenerate.
12131 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12133 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12135 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12137 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12140 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12142 PR tree-optimization/81292
12143 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12144 full_string_p, also call adjust_related_strinfos if the adjustment
12145 is simple, otherwise invalidate related strinfos.
12147 2017-07-04 Martin Liska <mliska@suse.cz>
12150 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12151 newly created variable as DECL_IGNORED_P.
12153 2017-07-04 Martin Liska <mliska@suse.cz>
12156 * ipa-inline.c (inline_small_functions):
12157 Use xstrdup_for_dump.
12159 2017-07-04 Tom de Vries <tom@codesourcery.com>
12161 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12163 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12166 * config/darwin.c (darwin_function_switched_text_sections):
12169 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12171 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12173 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12175 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12177 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12179 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12180 min_profitable_iters, and th as inclusive lower bounds.
12181 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12182 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12183 for min_profitable_iters and min_profitable_estimate.
12184 (vect_transform_loop): Treat th as an inclusive lower bound.
12185 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12187 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12190 * config/darwin.c (darwin_function_switched_text_sections):
12191 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12192 in two pieces, and suppress the use of buf.
12194 2017-07-03 Nathan Sidwell <nathan@acm.org>
12196 * hash-table.h (hash_table_mod1): Fix indentation.
12198 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12200 PR middle-end/81290
12201 * predict.c (force_edge_cold): Be more careful about propagation
12203 * profile-count.h (profile_probability::guessed,
12204 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12206 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12208 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12210 * doc/invoke.texi (rcpc architecture extension): Document it.
12212 2017-07-03 Richard Biener <rguenther@suse.de>
12214 PR tree-optimization/60510
12215 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12216 the scalar reduction PHI and use it.
12217 (vectorizable_reduction): Properly guard the single_defuse_cycle
12218 path for non-SLP reduction chains where we cannot use it.
12219 Rework reduc_def/index and vector type deduction. Rework
12220 vector operand gathering during reduction op code-gen.
12221 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12222 chains dissolve the chain and leave it to non-SLP reduction
12225 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12227 * tree-data-ref.h (dr_alignment): Declare.
12228 * tree-data-ref.c (dr_alignment): New function.
12229 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12230 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12232 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12234 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12236 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12237 and base_misalignment fields.
12238 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12239 * tree-data-ref.c: Include builtins.h.
12240 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12241 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12242 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12243 * tree-vect-data-refs.c: Include tree-cfg.h.
12244 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12245 fields instead of calculating an alignment here.
12246 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12247 innermost_loop_behavior fields.
12249 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12251 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12253 (DR_STEP_ALIGNMENT): New macro.
12254 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12255 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12256 (create_data_ref): Print it.
12257 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12258 to tell whether the step preserves vector (mis)alignment.
12259 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12260 Move the check for an integer step and generalise to all INTEGER_CST.
12261 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12262 Print the outer step alignment.
12264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12266 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12267 with offset_alignment.
12268 (DR_ALIGNED_TO): Delete.
12269 (DR_OFFSET_ALIGNMENT): New macro.
12270 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12271 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12272 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12273 (create_data_ref): Likewise.
12274 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12275 (vect_analyze_data_refs): Likewise.
12276 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12277 creating dummy innermost behavior.
12279 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12281 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12282 with a "innermost_loop_behavior *" and refeence tree.
12283 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12284 (create_data_ref): Update call accordingly.
12285 * tree-predcom.c (find_looparound_phi): Likewise.
12287 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12289 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12290 fields with dr_wrt_vec_loop.
12291 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12292 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12293 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12294 (vect_dr_behavior): New function.
12295 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12296 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12297 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12298 track whether the step preserves the misalignment.
12299 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12300 Use vect_dr_behavior.
12301 (vect_setup_realignment): Update call accordingly.
12302 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12303 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12304 call to vect_create_addr_base_for_vector_ref.
12305 (vect_create_cond_for_align_checks): Likewise.
12306 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12307 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12308 (vect_recog_mask_conversion_pattern): Likewise.
12309 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12310 (new_stmt_vec_info): Remove redundant zeroing.
12312 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12314 * common/config/arm/arm-common.c (arm_be8_option): New function.
12315 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12316 (ISA_ARMv6): Add isa_bit_be8.
12317 * config/arm/arm.h (arm_be8_option): Add prototype.
12318 (BE8_SPEC_FUNCTION): New define.
12319 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12320 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12321 (mlittle-endian): Similarly.
12322 (mbe8, mbe32): New options.
12323 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12324 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12326 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12328 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12330 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12332 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12333 (cleanup_tree_cfg_bb): Use it.
12334 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12336 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12338 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12341 * loop-doloop.c (add_test): Update profile.
12343 2017-07-03 Martin Liska <mliska@suse.cz>
12346 * sanopt.c (rewrite_usage_of_param): New function.
12347 (sanitize_rewrite_addressable_params): Likewise.
12348 (pass_sanopt::execute): Call rewrite_usage_of_param.
12350 2017-07-03 Richard Biener <rguenther@suse.de>
12352 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12353 back to using VIEW_CONVERT_EXPR.
12355 2017-07-03 Martin Liska <mliska@suse.cz>
12358 * doc/extend.texi: Document when a resolver function is
12359 generated for target_clones.
12361 2017-07-03 Martin Liska <mliska@suse.cz>
12363 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12364 and shadow memory of auto variables which are subject of
12365 use-after-scope sanitization.
12366 (asan_expand_mark_ifn): Add do set only when is_poison.
12368 2016-07-03 Richard Biener <rguenther@suse.de>
12370 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12372 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12373 (vectorizable_reduction): Perform reduction PHI creation when
12374 visiting a reduction PHI and adjust and simplify code generation
12375 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12376 (vect_transform_loop): Visit reduction PHIs.
12377 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12378 defs into the SLP tree.
12379 (vect_build_slp_tree): Reduction defs terminate the recursion.
12380 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12382 (vect_get_vec_defs_for_stmt_copy): Export.
12383 (vect_get_vec_defs): Likewise.
12384 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12386 (vect_get_vec_defs_for_stmt_copy): Declare.
12387 (vect_get_vec_defs): Likewise.
12389 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12391 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12392 parameter with a "loop" parameter and use it instead of the
12393 loop containing DR_STMT. Don't check simple_iv when doing
12394 BB analysis. Describe the two analysis modes in the comment.
12396 2017-07-03 Tom de Vries <tom@codesourcery.com>
12398 PR tree-optimization/69468
12399 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12400 (find_same_succ_bb): Handle ignore_edge_flags.
12402 2017-07-03 Tom de Vries <tom@codesourcery.com>
12404 PR tree-optimization/81192
12405 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12407 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12409 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12411 2017-07-03 Tom de Vries <tom@codesourcery.com>
12413 PR tree-optimization/81192
12414 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12415 BB_SAME_SUCC (bb) == NULL.
12417 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12419 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12422 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12424 * dumpfile.c: Include profile-count.h
12425 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12427 (insert_cond_bb): Update profile.
12428 * tree-cfg.h (insert_cond_bb): Update prototype.
12429 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12430 * tree-dump.c: Do not include tree-cfg.
12432 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12434 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12436 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12438 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12441 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12443 * tree-complex.c (expand_complex_div_wide): update profile.
12445 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12446 Alan Hayward <alan.hayward@arm.com>
12447 David Sherwood <david.sherwood@arm.com>
12449 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12450 (CORETYPES_H): New define.
12451 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12452 (insn-modes-inline.h, s-modes-inline-h): New rules.
12453 (generated_files): Add insn-modes-inline.h.
12454 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12455 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12456 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12457 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12458 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12459 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12460 (build/gencodes.o, build/genconditions.o): Likewise.
12461 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12462 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12463 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12464 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12465 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12466 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12467 * coretypes.h: Include everything up to real.h for generators.
12468 Include insn-modes.h first. Include wide-int-print.h after
12469 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12470 * machmode.h: Don't include insn-modes.h here.
12471 * function-tests.c: Remove includes of signop.h, machmode.h,
12472 double-int.h and wide-int.h.
12474 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12476 * optc-save-gen.awk: Likewise.
12477 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12478 * godump.c: Remove include of wide-int-print.h.
12479 * pretty-print.h: Likewise.
12480 * wide-int-print.cc: Likewise.
12481 * wide-int.cc: Likewise.
12482 * hash-map-tests.c: Remove include of signop.h.
12483 * hash-set-tests.c: Likewise.
12484 * rtl-tests.c: Likewise.
12485 * mkconfig.sh: Remove include of machmode.h.
12486 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12488 (emit_insn_modes_inline_h): ...this new function. Emit the code
12489 into an insn-modes-inline.h header file, adding appropriate
12490 include guards and end comments.
12491 (emit_insn_modes_c_header): Remove include of machmode.h.
12492 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12494 (main): Handle -i flag and call emit_insn_modes_inline_h when
12497 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12499 * tree-ssa-strlen.c (strinfo): Rename the length field to
12500 nonzero_chars. Add a full_string_p field.
12501 (compare_nonzero_chars, zero_length_string_p): New functions.
12502 (get_addr_stridx): Add an offset_out parameter.
12503 Use compare_nonzero_chars.
12504 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12505 (new_strinfo): Update after above changes to strinfo.
12506 (set_endptr_and_length): Set full_string_p.
12507 (get_string_length): Update after above changes to strinfo.
12508 (unshare_strinfo): Update call to new_strinfo.
12509 (maybe_invalidate): Likewise.
12510 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12511 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12512 as a uhwi instead of an shwi. Update after above changes to
12513 strinfo and new_strinfo.
12514 (zero_length_string): Assert that chainsi contains full strings.
12515 Use zero_length_string_p. Update call to new_strinfo.
12516 (adjust_related_strinfos): Update after above changes to strinfo.
12517 Copy full_string_p from origsi.
12518 (adjust_last_stmt): Use zero_length_string_p.
12519 (handle_builtin_strlen): Update after above changes to strinfo and
12520 new_strinfo. Install the lhs as the string length if the previous
12521 entry didn't describe a full string.
12522 (handle_builtin_strchr): Update after above changes to strinfo
12524 (handle_builtin_strcpy): Likewise.
12525 (handle_builtin_strcat): Likewise.
12526 (handle_builtin_malloc): Likewise.
12527 (handle_pointer_plus): Likewise.
12528 (handle_builtin_memcpy): Likewise. Track nonzero characters
12529 that aren't necessarily followed by a nul terminator.
12530 (handle_char_store): Likewise.
12532 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12534 PR tree-optimization/80769
12535 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12536 for malloc and calloc. Document the new invariant that all related
12537 strinfos have delayed lengths or none do.
12538 (verify_related_strinfos): Move earlier in file.
12539 (set_endptr_and_length): New function, split out from...
12540 (get_string_length): ...here. Also set the lengths of related
12542 (zero_length_string): Assert that chainsi has known (rather than
12544 (adjust_related_strinfos): Likewise.
12546 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12548 PR tree-optimization/81136
12549 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12550 assert that two references with the same misalignment have the same
12551 compile-time misalignment if those compile-time misalignments
12554 2017-07-01 Andi Kleen <ak@linux.intel.com>
12556 * print-tree.c (print_node): Print all attributes.
12558 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12560 * cfg.c (scale_bbs_frequencies): New function.
12561 * cfg.h (scale_bbs_frequencies): Declare it.
12562 * cfgloopanal.c (single_likely_exit): Cleanup.
12563 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12565 (scale_loop_profile): Likewise.
12566 (loop_version): Likewise.
12567 (create_empty_loop_on_edge): Update.
12568 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12569 scale_loop_frequencies, scale_loop_profile, loopify,
12570 loop_version): Update prototypes.
12571 * modulo-sched.c (sms_schedule): Update.
12572 * predict.c (unlikely_executed_edge_p): Also check probability.
12573 (probably_never_executed_edge_p): Fix typo.
12574 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12575 * tree-parloops.c (gen_parallel_loop): Update.
12576 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12577 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12578 * tree-ssa-loop-split.c (split_loop): Update.
12579 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12580 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12581 (vect_loop_versioning): Update.
12582 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12584 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12586 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12588 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12590 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12591 to keep profile consistent.
12593 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12595 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12596 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12597 * profile-count.h (max_safe_multiplier): Make unsigned.
12598 (profile_count::guessed_zero): New.
12600 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12602 * bb-reorder.c (fix_up_crossing_landing_pad,
12603 fix_crossing_conditional_branches): Use make_single_succ_edge
12604 to keep profile consistent.
12606 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12608 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12611 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12614 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12615 the right scopes, make sure cond_jump isn't preserved between multiple
12616 iterations. Search for fallthru edge whenever there are 3+ edges and
12617 use find_fallthru_edge for it.
12619 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12621 Patch by Alexander Monakov <amonakov@ispras.ru>
12622 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12623 probabilities consistently.
12625 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12627 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12628 emit_cmp_and_jump_insns.
12630 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12633 * tree-inline.c (expand_call_inline): Combine profile statuses.
12635 2017-06-30 Andrew Pinski <apinski@cavium.com>
12637 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12638 fold_stmt returned true.
12640 2017-06-30 Nathan Sidwell <nathan@acm.org>
12642 * ggc.h (empty_string): Delete.
12643 * cfgexpand.c (expand_asm_stmt): Use plain "".
12644 * optabs.c (expand_asm_memory_barrier): Likewise.
12645 * stringpool.c (empty_string): Delete.
12646 (digit_vector, digit_string): Delete.
12647 (ggc_alloc_string): Use plain "", don't optimize single digit
12648 strings. Use ggc_alloc_atomic.
12650 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12652 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12653 comparison set and one other set, use the cost of the non-comparison
12656 2017-06-30 Nathan Sidwell <nathan@acm.org>
12658 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12661 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12663 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12664 loops. Remove now unneeded calls to gimple_switch_set_label() that
12665 just set removed labels to NULL_TREE.
12667 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12669 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12670 (set_range_info): ...here. Only call set_range_info_raw if domain
12672 (set_nonzero_bits): Call set_range_info_raw.
12673 * tree-ssanames.h (set_range_info_raw): New.
12675 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12678 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12679 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12680 of nonimmediate_operand and <store_mask_constraint> instead of m
12681 for the input operand. For V8FI iterator, always split if input
12682 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12683 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12684 <store_mask_predicate> instead of register_operand and
12685 <store_mask_constraint> instead of v for the input operand. Make
12686 sure both operands aren't MEMs for if not <mask_applied>.
12688 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12690 * lto-wrapper.c (copy_file) Close both file descriptors before
12693 2017-06-30 Martin Liska <mliska@suse.cz>
12696 * multiple_target.c (create_dispatcher_calls): Make ifunc
12697 also for function that don't have calls or are not referenced.
12699 2017-06-30 Richard Biener <rguenther@suse.de>
12701 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12702 analyze the first scalar stmt. Move vector type computation
12703 for the BB case here from ...
12704 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12705 live operation processing in the SLP case properly.
12707 2017-06-30 Richard Biener <rguenther@suse.de>
12709 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12711 2017-06-30 Martin Liska <mliska@suse.cz>
12714 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12715 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12717 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12719 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12722 2017-06-30 Richard Biener <rguenther@suse.de>
12724 PR tree-optimization/81249
12725 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12726 condition reduction result to original scalar type.
12728 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12730 * profile-count.h (enum profile_quality): Fix typos and whitespace
12733 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12735 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12736 type for branch probabilities.
12738 2017-06-29 Julian Brown <julian@codesourcery.com>
12739 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12741 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12742 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12743 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12744 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12746 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12748 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12749 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12750 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12751 CC usage from generic code to here.
12752 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12753 CC usage into the target macros.
12755 2017-06-29 Maya Rashish <coypu@sdf.org>
12757 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12760 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12762 * arm/arm-builtins.c: Include profile-count.h
12763 * except.c (sjlj_emit_function_enter): Use
12764 profile_probability::unlikely.
12766 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12768 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12769 and tocrel_offset be pointer args rather than implicitly using
12771 (legitimate_constant_pool_address_p, rs6000_emit_move,
12772 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12773 tocrel_offset and use in toc_relative_expr_p call.
12774 (print_operand, print_operand_address): Use static tocrel_base_oac
12775 and tocrel_offset_oac.
12776 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12779 2017-06-29 Maya Rashish <coypu@sdf.org>
12781 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12783 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12785 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12786 objects, take into account only the alignment of 'op0' and 'mode1' if
12789 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12791 * ccmp.c (ccmp_tree_comparison_p): New function.
12792 (ccmp_candidate_p): Update to use above function.
12793 (get_compare_parts): New function.
12794 (expand_ccmp_next): Update to use new functions.
12795 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12797 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12798 take mode as argument.
12799 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12800 * expr.c (expand_expr_real_1): Pass mode as argument.
12802 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12804 * combine.c (combine_instructions): Print insns to dump_file, together
12807 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12809 * asan.c (asan_emit_stack_protection): Update.
12810 (create_cond_insert_point): Update.
12811 * auto-profile.c (afdo_propagate_circuit): Update.
12812 * basic-block.h (struct edge_def): Turn probability to
12813 profile_probability.
12814 (EDGE_FREQUENCY): Update.
12815 * bb-reorder.c (find_traces_1_round): Update.
12816 (better_edge_p): Update.
12817 (sanitize_hot_paths): Update.
12818 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12819 (make_single_succ_edge): Update.
12820 (check_bb_profile): Update.
12821 (dump_edge_info): Update.
12822 (update_bb_profile_for_threading): Update.
12823 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12824 probabilitycount to 0.
12825 * cfgbuild.c (compute_outgoing_frequencies): Update.
12826 * cfgcleanup.c (try_forward_edges): Update.
12827 (outgoing_edges_match): Update.
12828 (try_crossjump_to_edge): Update.
12829 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12830 (expand_gimple_tailcall): Update.
12831 (construct_init_block): Use make_single_succ_edge.
12832 (construct_exit_block): Use make_single_succ_edge.
12833 * cfghooks.c (verify_flow_info): Update.
12834 (redirect_edge_succ_nodup): Update.
12835 (split_edge): Update.
12836 (account_profile_record): Update.
12837 * cfgloopanal.c (single_likely_exit): Update.
12838 * cfgloopmanip.c (scale_loop_profile): Update.
12839 (set_zero_probability): Remove.
12840 (duplicate_loop_to_header_edge): Update.
12841 * cfgloopmanip.h (loop_version): Update prototype.
12842 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12843 (force_nonfallthru_and_redirect): Update.
12844 (update_br_prob_note): Update.
12845 (rtl_verify_edges): Update.
12846 (purge_dead_edges): Update.
12847 (rtl_lv_add_condition_to_bb): Update.
12848 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12849 * cgraphunit.c (init_lowered_empty_function): Update.
12850 (cgraph_node::expand_thunk): Update.
12851 * cilk-common.c: Include profile-count.h
12852 * dojump.c (inv): Remove.
12853 (jumpifnot): Update.
12854 (jumpifnot_1): Update.
12855 (do_jump_1): Update.
12857 (do_jump_by_parts_greater_rtx): Update.
12858 (do_compare_rtx_and_jump): Update.
12859 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12860 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12861 * dwarf2cfi.c: Include profile-count.h
12862 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12863 (sjlj_emit_dispatch_table): Likewise.
12864 * explow.c: Include profile-count.h
12865 * expmed.c (emit_store_flag_force): Update.
12866 (do_cmp_and_jump): Update.
12867 * expr.c (compare_by_pieces_d::generate): Update.
12868 (compare_by_pieces_d::finish_mode): Update.
12869 (emit_block_move_via_loop): Update.
12870 (store_expr_with_bounds): Update.
12871 (store_constructor): Update.
12872 (expand_expr_real_2): Update.
12873 (expand_expr_real_1): Update.
12874 * expr.h (try_casesi, try_tablejump): Update prototypes.
12875 * gimple-pretty-print.c (dump_probability): Update.
12876 (dump_profile): New.
12877 (dump_gimple_label): Update.
12878 (dump_gimple_bb_header): Update.
12879 * graph.c (draw_cfg_node_succ_edges): Update.
12880 * hsa-gen.c (convert_switch_statements): Update.
12881 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12882 (find_if_case_1): Update.
12883 (find_if_case_2): Update.
12884 * internal-fn.c (expand_arith_overflow_result_store): Update.
12885 (expand_addsub_overflow): Update.
12886 (expand_neg_overflow): Update.
12887 (expand_mul_overflow): Update.
12888 (expand_vector_ubsan_overflow): Update.
12889 * ipa-cp.c (good_cloning_opportunity_p): Update.
12890 * ipa-split.c (split_function): Use make_single_succ_edge.
12891 * ipa-utils.c (ipa_merge_profiles): Update.
12892 * loop-doloop.c (add_test): Update.
12893 (doloop_modify): Update.
12894 * loop-unroll.c (compare_and_jump_seq): Update.
12895 (unroll_loop_runtime_iterations): Update.
12896 * lra-constraints.c (lra_inheritance): Update.
12897 * lto-streamer-in.c (input_cfg): Update.
12898 * lto-streamer-out.c (output_cfg): Update.
12899 * mcf.c (adjust_cfg_counts): Update.
12900 * modulo-sched.c (sms_schedule): Update.
12901 * omp-expand.c (expand_omp_for_init_counts): Update.
12902 (extract_omp_for_update_vars): Update.
12903 (expand_omp_ordered_sink): Update.
12904 (expand_omp_for_ordered_loops): Update.
12905 (expand_omp_for_generic): Update.
12906 (expand_omp_for_static_nochunk): Update.
12907 (expand_omp_for_static_chunk): Update.
12908 (expand_cilk_for): Update.
12909 (expand_omp_simd): Update.
12910 (expand_omp_taskloop_for_outer): Update.
12911 (expand_omp_taskloop_for_inner): Update.
12912 * omp-simd-clone.c (simd_clone_adjust): Update.
12913 * optabs.c (expand_doubleword_shift): Update.
12914 (expand_abs): Update.
12915 (emit_cmp_and_jump_insn_1): Update.
12916 (expand_compare_and_swap_loop): Update.
12917 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12918 * predict.c (predictable_edge_p): Update.
12919 (edge_probability_reliable_p): Update.
12920 (set_even_probabilities): Update.
12921 (combine_predictions_for_insn): Update.
12922 (combine_predictions_for_bb): Update.
12923 (propagate_freq): Update.
12924 (estimate_bb_frequencies): Update.
12925 (force_edge_cold): Update.
12926 * profile-count.c (profile_count::dump): Add missing space into dump.
12927 (profile_count::debug): Add newline.
12928 (profile_count::differs_from_p): Explicitly convert to unsigned.
12929 (profile_count::stream_in): Update.
12930 (profile_probability::dump): New member function.
12931 (profile_probability::debug): New member function.
12932 (profile_probability::differs_from_p): New member function.
12933 (profile_probability::differs_lot_from_p): New member function.
12934 (profile_probability::stream_in): New member function.
12935 (profile_probability::stream_out): New member function.
12936 * profile-count.h (profile_count_quality): Rename to ...
12937 (profile_quality): ... this one.
12938 (profile_probability): New.
12939 (profile_count): Update.
12940 * profile.c (compute_branch_probabilities): Update.
12941 * recog.c (peep2_attempt): Update.
12942 * sched-ebb.c (schedule_ebbs): Update.
12943 * sched-rgn.c (find_single_block_region): Update.
12944 (compute_dom_prob_ps): Update.
12945 (schedule_region): Update.
12946 * sel-sched-ir.c (compute_succs_info): Update.
12947 * stmt.c (struct case_node): Update.
12948 (do_jump_if_equal): Update.
12949 (get_outgoing_edge_probs): Update.
12950 (conditional_probability): Update.
12951 (emit_case_dispatch_table): Update.
12952 (expand_case): Update.
12953 (expand_sjlj_dispatch_table): Update.
12954 (emit_case_nodes): Update.
12955 * targhooks.c: Update.
12956 * tracer.c (better_p): Update.
12957 (find_best_successor): Update.
12958 * trans-mem.c (expand_transaction): Update.
12959 * tree-call-cdce.c: Update.
12960 * tree-cfg.c (gimple_split_edge): Upate.
12961 (move_sese_region_to_fn): Upate.
12962 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12963 * tree-eh.c (lower_resx): Upate.
12964 (cleanup_empty_eh_move_lp): Upate.
12965 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12966 * tree-inline.c (copy_edges_for_bb): Update.
12967 (copy_cfg_body): Update.
12968 * tree-parloops.c (gen_parallel_loop): Update.
12969 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12970 (gimple_gen_time_profiler): Update.
12971 * tree-ssa-dce.c (remove_dead_stmt): Update.
12972 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12973 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12974 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12975 (unloop_loops): Update.
12976 (try_peel_loop): Update.
12977 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12978 * tree-ssa-loop-split.c (connect_loops): Update.
12979 (split_loop): Update.
12980 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12981 (hoist_guard): Update.
12982 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12983 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12984 (value_replacement): Update.
12985 * tree-ssa-reassoc.c (branch_fixup): Update.
12986 * tree-ssa-tail-merge.c (replace_block_by): Update.
12987 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12988 (create_edge_and_update_destination_phis): Update.
12989 (compute_path_counts): Update.
12990 (recompute_probabilities): Update.
12991 (update_joiner_offpath_counts): Update.
12992 (freqs_to_counts_path): Update.
12993 (duplicate_thread_path): Update.
12994 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12995 (struct switch_conv_info): Update.
12996 (gen_inbound_check): Update.
12997 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12998 (vect_do_peeling): Update.
12999 (vect_loop_versioning): Update.
13000 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13001 (optimize_mask_stores): Update.
13002 * ubsan.c (ubsan_expand_null_ifn): Update.
13003 * value-prof.c (gimple_divmod_fixed_value): Update.
13004 (gimple_divmod_fixed_value_transform): Update.
13005 (gimple_mod_pow2): Update.
13006 (gimple_mod_pow2_value_transform): Update.
13007 (gimple_mod_subtract): Update.
13008 (gimple_mod_subtract_transform): Update.
13009 (gimple_ic): Update.
13010 (gimple_stringop_fixed_value): Update.
13011 (gimple_stringops_transform): Update.
13012 * value-prof.h: Update.
13014 2017-06-29 Carl Love <cel@us.ibm.com>
13016 * config/rs6000/rs6000-c.c: Add support for built-in functions
13017 vector signed int vec_signed (vector float);
13018 vector signed long long vec_signed (vector double);
13019 vector signed int vec_signed2 (vector double, vector double);
13020 vector signed int vec_signede (vector double);
13021 vector signed int vec_signedo (vector double);
13022 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13023 instruction generator.
13024 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13025 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13026 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13028 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13029 vunsignede_v2df): Add define_expands.
13030 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13031 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13032 VEC_UNSIGNEDO): Add definitions.
13033 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13034 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13035 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13036 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13037 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13038 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13039 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13040 vec_unsignede, vec_unsignedo): Add builtin defines.
13041 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13043 * doc/extend.texi: Update the built-in documentation file for the
13044 new built-in functions.
13046 2017-06-29 Richard Biener <rguenther@suse.de>
13048 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13049 reduction chains to LOOP_VINFO_REDUCTIONS.
13050 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13051 SLP reductions after processing reduction chains.
13053 2017-06-29 Nathan Sidwell <nathan@acm.org>
13055 * builtins.c (fold_builtin_FUNCTION): Use
13056 lang_hooks.decl_printable_name.
13058 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13060 PR middle-end/81194
13061 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13062 with only one label.
13063 * stmt.c (expand_case): Assert NCASES is greater than one.
13065 2017-06-29 Richard Biener <rguenther@suse.de>
13067 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13069 (group_case_labels): Likewise.
13070 (find_taken_edge): Push sanity checking on val to workers...
13071 (find_taken_edge_cond_expr): ... here
13072 (find_taken_edge_switch_expr): ... and here, handle cases
13073 with just a default label.
13074 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13075 (group_case_labels): Likewise.
13076 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13077 group_case_labels does anything cleanup the CFG again.
13079 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13081 PR tree-optimization/81196
13082 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13083 exit condition comparing two IVs.
13085 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13087 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13088 profile to the dummy entry at the end of the list of architectures.
13089 * config/arm/arm-cpu-cdata.h: Regenerated.
13091 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13092 Michael Collison <michael.collison@arm.com>
13095 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13097 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13098 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13099 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13100 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13101 of shift when the shift amount is masked with constant equal to
13102 the size of the mode.
13103 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13106 2017-06-29 Martin Liska <mliska@suse.cz>
13108 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13110 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13113 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13116 (edge_growth_cache_entry::edge_growth_cache_entry): New
13118 (reset_edge_growth_cache): Update to use constructor.
13120 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13122 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13123 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13124 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13126 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13128 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13129 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13131 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13133 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13134 (*-linux-uclibc*): Add t-uclibc tmake_file.
13135 * config/t-musl: New.
13136 * config/t-uclibc: New.
13138 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13140 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13142 (gen_comm_data): Emit architectural setting of arch_prof.
13143 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13145 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13146 (armv8-m.base, armv8-m.main): Likewise.
13147 * arm-protos.h (arm_build_target): Add profile field.
13148 (arch_option): Likewise.
13149 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13151 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13152 arm_active_target.profile.
13154 2017-06-28 Richard Biener <rguenther@suse.de>
13156 PR middle-end/81227
13157 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13158 TYPE_OVERFLOW_WRAPS.
13159 * match.pd (negate_expr_p): Likewise.
13160 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13161 fold_build2, not fold_binary.
13163 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13165 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13166 Convert memory address to Pmode.
13167 (aarch64_print_operand): Assert MEM operands are always Pmode.
13169 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13172 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13173 Remove redundant if.
13174 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13175 * config/arm/aarch-common-protos.h
13176 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13177 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13179 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13182 * multiple_target.c (create_dispatcher_calls): Set the default
13183 clone to be static, not public.
13185 2017-06-28 Richard Biener <rguenther@suse.de>
13187 * tree-vect-loop.c (vectorizable_reduction): Move special
13188 cond reduction IV var creation ...
13189 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13190 parameter. Use STMT_VINFO_VECTYPE.
13191 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13194 2017-06-28 Martin Liska <mliska@suse.cz>
13197 * ipa-visibility.c (non_local_p): Handle visibility.
13199 2017-06-28 Martin Liska <mliska@suse.cz>
13202 * common.opt: Add IntegerRange to various options.
13203 * opt-functions.awk (integer_range_info): New function.
13204 * optc-gen.awk: Add integer_range_info to cl_options struct.
13205 * opts-common.c (decode_cmdline_option): Handle
13206 CL_ERR_INT_RANGE_ARG.
13207 (cmdline_handle_error): Likewise.
13208 * opts.c (print_filtered_help): Show valid interval in
13209 when --help is provided.
13210 * opts.h (struct cl_option): Add range_min and range_max fields.
13211 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13213 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13215 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13216 (x * C EQ/NE y * C): New transformation.
13218 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13220 * genmultilib (combination_space): Accept '+' in option names.
13222 2017-06-28 Martin Liska <mliska@suse.cz>
13225 * asan.c (instrument_derefs): Bail out inner references
13226 that are hard register variables.
13228 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13231 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13232 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13234 2017-06-28 Richard Biener <rguenther@suse.de>
13236 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13237 (vect_get_slp_defs): Adjust.
13238 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13240 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13242 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13243 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13244 (vectorizable_reduction): Adjust.
13245 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13247 (vect_get_slp_defs): Likewise.
13248 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13249 (vectorizable_bswap): Adjust.
13250 (vectorizable_call): Likewise.
13251 (vectorizable_conversion): Likewise.
13252 (vectorizable_assignment): Likewise.
13253 (vectorizable_shift): Likewise.
13254 (vectorizable_operation): Likewise.
13255 (vectorizable_store): Likewise.
13256 (vectorizable_condition): Likewise.
13257 (vectorizable_comparison): Likewise.
13259 2017-06-28 Michael Collison <michael.collison@arm.com>
13262 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13264 (arm_gen_movmemqi): Removed unused variable 'i'.
13265 Convert 'for' loop into 'while' loop.
13266 (arm_expand_prologue): Remove last unnecessary set of insn.
13267 (thumb_pop): Remove unused variable 'pushed_words'.
13268 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13270 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13272 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13273 * config/s390/s390-protos.h: Add prototype of
13274 s390_rel_address_ok_p.
13275 * config/s390/s390.c (s390_got_symbol): New function.
13276 (s390_rel_address_ok_p): New function.
13277 (legitimize_pic_address): Use s390_rel_address_ok_p.
13278 (s390_load_got): Use s390_got_symbol.
13279 (s390_option_override): Issue error if
13280 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13281 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13283 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13285 2017-06-27 Andrew Pinski <apinski@cavium.com>
13287 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13288 (X * copysign (1.0, X)): New pattern.
13289 (X * copysign (1.0, -X)): New pattern.
13290 (copysign (-1.0, CST)): New pattern.
13292 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13294 * genmultilib (combination_space): Remove variable.
13295 Validate reuse rules against regular expression for any sequence
13296 of multilib options in any order.
13298 2017-06-27 Michael Collison <michael.collison@arm.com>
13300 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13301 call aarch64_split_simd_combine.
13302 * (aarch64_combine_internal<mode>): Delete pattern.
13303 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13304 Allow register and subreg operands.
13306 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13308 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13309 specific need, just fallback on defaults.
13310 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13312 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13313 Olivier Hainque <hainque@adacore.com>
13315 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13317 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13318 targets. Pick a default if no particular attempt applied.
13319 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13322 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13324 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13325 (x86_64-wrs-vxworks7): Likewise.
13327 2017-06-27 Marek Polacek <polacek@redhat.com>
13330 * ubsan.c (instrument_null): Check get_base_address's result for null.
13332 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13334 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13336 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13338 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13339 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13340 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13341 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13342 New function types.
13343 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13344 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13345 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13346 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13347 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13348 BUILT_IN_FEUPDATEENV): New builtins.
13349 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13350 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13351 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13352 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13354 (builtin_structptr_types): Adjust size.
13355 * tree.c (builtin_structptr_types): Add four entries.
13357 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13358 Olivier Hainque <hainque@adacore.com>
13360 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13361 (TLS_SYM): New local macro, forcing reference to __tls__ on
13362 link command lines for VxWorks 7 RTPs, triggering initialization
13364 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13365 OS features TLS support, true for RTPs on VxWorks 7.
13366 * config/vxworks.c (vxworks_override_options): Setup emutls
13369 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13371 * predict.c (test_prediction_value_range): Use -1U instead of -1
13372 to avoid narrowing conversion warning.
13373 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13374 to avoid narrowing conversion warning.
13375 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13377 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13379 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13381 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13382 64bit configurations.
13383 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13384 (SIZE_TYPE): Likewise.
13385 * config/vxworks.c (vxworks_emutls_var_fields): Use
13386 long_unsigned_type_node instead of unsigned_type_node as the offset
13387 field type, which is "pointer" mode in emutls.c.
13389 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13392 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13394 PR middle-end/81207
13395 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13396 gimple_vuse copying separately from gimple_vdef copying.
13398 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13400 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13403 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13404 Olivier Hainque <hainque@adacore.com>
13406 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13407 for all vxworks7 targets.
13408 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13409 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13410 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13411 variations for VX6/VX7 and 32/64bits later on in ...
13412 (VXWORKS_LIB_SPEC): Leverage new macros.
13413 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13414 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13416 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13418 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13419 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13421 2017-06-26 Carl Love <cel@us.ibm.com>
13423 * config/rs6000/rs6000-c.c: Add support for built-in functions
13424 vector bool char vec_reve (vector bool char);
13425 vector signed char vec_reve (vector signed char);
13426 vector unsigned char vec_reve (vector unsigned char);
13427 vector bool int vec_reve (vector bool int);
13428 vector signed int vec_reve (vector signed int);
13429 vector unsigned int vec_reve (vector unsigned int);
13430 vector bool long long vec_reve (vector bool long long);
13431 vector signed long long vec_reve (vector signed long long);
13432 vector unsigned long long vec_reve (vector unsigned long long);
13433 vector bool short vec_reve (vector bool short);
13434 vector signed short vec_reve (vector signed short);
13435 vector double vec_reve (vector double);
13436 vector float vec_reve (vector float);
13437 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13438 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13439 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13440 (altivec_vreve): New pattern.
13441 * config/rs6000/altivec.h (vec_reve): New define.
13442 * doc/extend.texi (vec_rev): Update the built-in documentation file
13443 for the new built-in functions.
13445 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13447 PR tree-optimization/71815
13448 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13450 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13452 (slsr_process_phi): Likewise.
13453 (replace_uncond_cands_and_profitable_phis): Don't replace a
13454 multiply candidate with a stride of 1 (copy or cast).
13455 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13457 (lowest_cost_path): Likewise.
13458 (total_savings): Likewise.
13460 2017-06-26 Richard Biener <rguenther@suse.de>
13463 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13464 Use def_builtin_pure for all gather builtins.
13466 2017-06-26 Richard Biener <rguenther@suse.de>
13468 PR tree-optimization/81203
13469 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13470 non-dominating BBs.
13472 2017-06-26 Marek Polacek <polacek@redhat.com>
13475 * doc/invoke.texi: Document -Wmultistatement-macros.
13477 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13479 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13480 arm_neon_ok_no_float_abi effective target.
13482 2017-06-26 Richard Biener <rguenther@suse.de>
13484 PR tree-optimization/80928
13485 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13486 (copy_bbs): Set BB_DUPLICATED flag early.
13487 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13489 (execute_on_shrinking_pred): Likewise.
13490 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13491 BB_DUPLICATED blocks.
13492 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13493 iterate over all PHIs considering removal of *gsi.
13495 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13497 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13500 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13502 * config/rs6000/rs6000-string.c: (expand_block_clear,
13503 do_load_for_compare, select_block_compare_mode,
13504 compute_current_alignment, expand_block_compare,
13505 expand_strncmp_align_check, expand_strn_compare,
13506 expand_block_move, rs6000_output_load_multiple)
13507 Move functions related to string/block move/compare
13508 to a separate file.
13509 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13510 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13511 for this function which is now used in two files.
13512 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13513 * config.gcc: Add rs6000-string.o to extra_objs for
13514 targets powerpc*-*-* and rs6000*-*-*.
13516 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13519 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13520 32-bit, since indexed is not valid for DImode.
13521 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13522 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13523 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13524 (define_peephole2 for Altivec d-form store): Likewise.
13527 * multiple_target.c (create_dispatcher_calls): Only create the
13528 dispatcher call if the function is the default clone of a
13529 versioned function.
13531 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13533 PR middle-end/80902
13534 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13535 a call, force the call to not be a tail call.
13537 2017-06-23 Jeff Law <law@redhat.com>
13539 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13542 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13544 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13545 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13546 handling for early expansion of vector shifts (sl,sr,sra,rl).
13547 (builtin_function_type): Add vector shift right instructions
13548 to the unsigned argument list.
13550 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13552 rtl-optimizatoin/79286
13553 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13554 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13555 trap. PIC register plus a const unspec without offset can never trap.
13557 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13559 * tree.h (builtin_structptr_type): New type.
13560 (builtin_structptr_types): Declare new array.
13561 * tree.c (builtin_structptr_types): New array.
13562 (free_lang_data, build_common_tree_nodes): Use it.
13564 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13567 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13570 2017-06-22 Matt Turner <mattst88@gmail.com>
13572 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13573 Lake models to skylake case. Assume skylake for unknown
13574 models with clflushopt.
13576 2017-06-22 Jeff Law <law@redhat.com>
13578 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13579 frame sizes that do not satisfy aarch64_uimm12_shift.
13581 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13583 * profile-count.h (apply_probability,
13584 apply_scale, probability_in): Fix checks for zero.
13586 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13588 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13589 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13591 2016-06-22 Richard Biener <rguenther@suse.de>
13593 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13594 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13595 REDUC_MAX_EXPR support.
13596 (vectorizable_reduction): Likewise.
13597 (vect_create_epilog_for_reduction): Likewise.
13599 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13601 * match.pd (A / (1 << B) -> A >> B): New.
13602 * generic-match-head.c: Include optabs-tree.h.
13603 * gimple-match-head.c: Likewise.
13604 * optabs-tree.h (target_supports_op_p): New.
13605 * optabs-tree.c (target_supports_op_p): New.
13607 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13609 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13610 $gcc_cv_ld --help output.
13611 (gcc_cv_ld_demangle): Likewise.
13612 (gcc_cv_ld_eh_frame_hdr): Likewise.
13613 (gcc_cv_ld_pie): Likewise.
13614 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13615 (gcc_cv_ld_buildid): Likewise.
13616 (gcc_cv_ld_sysroot): Likewise.
13617 (ld_bndplt_support): Likewise.
13618 (ld_pushpopstate_support): Likewise.
13619 * configure: Regenerate.
13620 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13622 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13625 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13626 operands indexes to avoid gap between operands and match_dups.
13628 2017-06-21 Andrew Pinski <apinski@cavium.com>
13630 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13631 Increment Arith_shift and Arith_shift_reg by 1.
13632 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13634 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13635 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13636 (aarch64_strip_extend): Add new argument and test for it.
13637 (aarch64_cheap_mult_shift_p): New function.
13638 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13639 add a cost if it is true.
13640 Update calls to aarch64_strip_extend.
13641 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13643 2017-06-21 Andrew Pinski <apinski@cavium.com>
13645 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13647 (thunderxt88): Likewise.
13648 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13649 (thunderx_prefetch_tune): New variable.
13650 (thunderx2t99_prefetch_tune): Update for the correct values.
13651 (thunderxt88_tunings): New variable.
13652 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13653 generic_prefetch_tune.
13654 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13656 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13658 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13659 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13660 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13661 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13662 (aarch64_atomic_cas<mode>, GPI): Likewise.
13664 2017-06-21 Martin Liska <mliska@suse.cz>
13666 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13667 statements on cold and hot labels.
13668 * predict.c (tree_estimate_probability_bb): Remove the
13669 prediction from this place.
13671 2017-06-21 Martin Liska <mliska@suse.cz>
13673 PR tree-optimization/79489
13674 * gimplify.c (maybe_add_early_return_predict_stmt): New
13676 (gimplify_return_expr): Call the function.
13677 * predict.c (tree_estimate_probability_bb): Remove handling
13679 * predict.def: Update comment about early return predictor.
13680 * gimple-predict.h (is_gimple_predict): New function.
13681 * predict.def: Change default value of early return to 66.
13682 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13684 * passes.def: Put pass_strip_predict_hints to the beginning of
13687 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13689 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13690 FUNCTION_DECL declarations.
13691 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13693 (dwaf2out_decl): Likewise.
13694 * godump.c (go_early_global_decl): Skip call to the real debug hook
13695 for FUNCTION_DECL declarations.
13696 * passes.c (rest_of_decl_compilation): Skip call to the
13697 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13698 -fdump-go-spec is passed.
13700 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13702 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13703 Reorder for compactness.
13704 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13705 (def_builtin_pure, def_builtin_pure2): New functions.
13706 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13708 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13710 * match.pd (nop_convert): New predicate.
13711 ((A +- CST1) +- CST2): Allow some NOP conversions.
13713 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13716 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13717 with ctors/dtors if GOVD_SHARED is set.
13719 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13721 * config/aarch64/aarch64.md (movti_aarch64):
13722 Emit mov rather than orr.
13723 (movtf_aarch64): Likewise.
13724 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13725 Emit mov rather than orr.
13727 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13729 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13730 Swap alternatives, make integer dup more expensive.
13732 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13734 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13735 Return true for non-tls symbols.
13737 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13739 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13740 (cortex-a75): Likewise.
13741 (cortex-a75.cortex-a55): Likewise.
13742 * config/aarch64/aarch64-tune.md: Regenerate.
13743 * doc/invoke.texi (-mtune): Document new values for -mtune.
13745 2017-06-21 Tom de Vries <tom@codesourcery.com>
13747 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13748 stack_size feature.
13749 (Effective-Target Keywords, Other attributes): Suggest using
13750 dg-add-options stack_size feature to get stack limit in stack_size
13751 effective target documentation.
13753 2017-06-21 Julian Brown <julian@codesourcery.com>
13754 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13756 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13757 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13758 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13760 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13761 attribute type list for neon_multiply.
13762 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13763 attribute type list for neon_multiply.
13764 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13765 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13766 attribute type list for neon_multiply.
13767 * config/arm/types.md (crypto_pmull): Add.
13768 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13769 attribute type list.
13771 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13773 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13776 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13778 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13779 to make sure not to dereference a NULL cost_classes_ptr pointer.
13781 2017-06-20 Carl Love <cel@us.ibm.com>
13783 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13784 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13785 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13786 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13787 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13788 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13789 VMULOSW): New enum "unspec" values.
13790 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13791 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13792 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13793 altivec_vmulosw): New patterns.
13794 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13795 VMULOSW): Add definitions.
13797 2017-06-20 Julia Koval <julia.koval@intel.com>
13799 * config/i386/i386.c: Fix rounding expand for new pattern.
13800 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13802 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13804 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13805 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13807 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13809 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13812 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13814 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13815 architecture, then by implementer ID.
13816 * config/aarch64/aarch64-tune.md: Regenerate.
13818 2017-06-20 Richard Biener <rguenther@suse.de>
13820 PR middle-end/81097
13821 * fold-const.c (split_tree): Fold to type before negating.
13823 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13825 * diagnostic-show-locus.c
13826 (selftest::test_fixit_deletion_affecting_newline): New function.
13827 (selftest::diagnostic_show_locus_c_tests): Call it.
13829 2017-06-20 Andreas Schwab <schwab@suse.de>
13832 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13835 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13837 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13838 __ARM_FEATURE_COPROC according to support.
13840 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13842 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13843 Rewritten to avoid overflow for > 32-bit pointers.
13846 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13847 by removing enum keyword.
13848 (ubsan_type_descriptor): Likewise. Formatting fix.
13851 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13852 splitter): Require TARGET_SSE2 in the condition.
13854 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13857 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13858 for doing vector set of SFmode on ISA 3.0.
13859 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13860 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13862 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13863 SFmode value into a V4SF variable that was extracted from another
13864 V4SF variable without converting the element to double precision
13865 and back to single precision vector format.
13866 (vsx_insert_extract_v4sf_p9_2): Likewise.
13868 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13870 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13871 in UWHI to avoid undefined overflow.
13874 * ubsan.h (enum ubsan_encode_value_phase): New.
13875 (ubsan_encode_value): Change second argument to
13876 enum ubsan_encode_value_phase with default value of
13877 UBSAN_ENCODE_VALUE_GENERIC.
13878 * ubsan.c (ubsan_encode_value): Change second argument to
13879 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13880 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13881 create_tmp_var_raw instead of create_tmp_var and use a
13883 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13884 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13885 ubsan_encode_value callers.
13888 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13889 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13890 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13892 2017-06-19 Richard Biener <rguenther@suse.de>
13894 PR middle-end/81118
13895 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13896 estimates if we changed anything.
13898 2017-06-19 Richard Biener <rguenther@suse.de>
13900 PR tree-optimization/80887
13901 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13902 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13903 simplified lookups, then reset mprts_hook.
13904 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13906 (try_to_simplify): Likewise.
13908 2017-06-19 Martin Liska <mliska@suse.cz>
13911 * gimplify.c (gimplify_switch_expr):
13912 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13914 2017-06-19 Martin Liska <mliska@suse.cz>
13916 * doc/install.texi: Document that PGO runs in 4 stages.
13918 2017-06-19 Martin Liska <mliska@suse.cz>
13921 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13922 to dispatcher function name.
13923 * multiple_target.c (replace_function_decl): New function.
13924 (create_dispatcher_calls): Redirect both edges and references.
13926 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13928 * profile-count.c (profile_count::dump): Dump quality.
13929 (profile_count::differs_from_p): Update for unsigned val.
13930 * profile-count.h (profile_count_quality): New enum.
13931 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13933 2017-06-19 Richard Biener <rguenther@suse.de>
13935 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13936 struct function as arg.
13937 (estimate_numbers_of_iterations): Export overload with loop arg.
13938 (free_numbers_of_iterations_estimates_loop): Use an overload of
13939 free_numbers_of_iterations_estimates instead.
13940 * tree-cfg.c (remove_bb): Adjust.
13941 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13942 * tree-parloops.c (gen_parallel_loop): Likewise.
13943 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13945 (tree_unroll_loops_completely): Likewise.
13946 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13947 Use an overload instead and export.
13948 (estimated_loop_iterations): Adjust.
13949 (max_loop_iterations): Likewise.
13950 (likely_max_loop_iterations): Likewise.
13951 (estimate_numbers_of_iterations): Take struct function as arg
13953 (loop_exits_before_overflow): Adjust.
13954 (free_numbers_of_iterations_estimates_loop): Use an overload.
13955 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13956 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13958 2017-06-19 Richard Biener <rguenther@suse.de>
13961 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13962 RANGE_EXPR conservatively.
13964 2017-06-16 Carl Love <cel@us.ibm.com>
13966 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13967 definitions for vec_float, vec_float2, vec_floato,
13968 vec_floate built-ins.
13969 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13970 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13972 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13973 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13974 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13975 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13976 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13977 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13978 vec_floato): Add builtin defines.
13979 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13980 Update the built-in documentation file for the new built-in
13983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13985 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13986 (mthumb): Mark as the negative of -marm.
13988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13990 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13992 (ARM Options, -mtune): Document that this accepts the same
13993 extension options as -mcpu.
13994 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13998 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13999 permitted extensions.
14001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14003 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14004 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14005 (armv8-m.main): Add option +nodsp.
14006 * config/arm/arm-cpu-cdata.h: Regenerated.
14008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14010 * config/arm/t-fuchsia: New file.
14011 * config.gcc (arm*-*-fuchsia*): Use it.
14013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14015 * config/arm/t-symbian: Rewrite for new option infrastructure.
14017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14019 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14020 (MULTILIB_REQUIRED): Likewise.
14022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14024 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14025 (MULTILIB_RESUE): Likewise.
14026 (MULTILIB_MATCHES): Likewise.
14027 (MULTLIB_REQUIRED): Likewise.
14029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14031 * config/arm/t-rtems: Rewrite for new option framework.
14033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14035 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14036 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14037 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14038 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14039 * config/arm/t-multilib: ... here.
14040 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14041 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14042 armv7-a and armv8*-a when A-profile libraries have not been built.
14043 * config/arm/t-rmprofile: Rewrite.
14045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14047 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14048 with a backslash. Remove the backslash after substituting unescaped
14050 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14054 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14055 the driver to pass through the default setting of -mfloat-abi.
14056 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14058 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14059 (all_feat_combs): New rule.
14060 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14062 * config/arm/t-aprofile: Rewrite.
14064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14066 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14067 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14068 fpu is not specified by the user/command-line.
14069 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14070 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14071 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14072 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14073 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14074 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14078 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14079 * config/arm/t-arm-elf: Rewritten.
14081 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14083 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14084 have some floating-point instructions.
14085 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14086 (TARGET_MAYBE_HARD_FLOAT): New macro.
14087 * config/arm/arm-builtins.c (arm_init_builtins): Use
14088 TARGET_MAYBE_HARD_FLOAT.
14089 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14093 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14094 (configargs.h): Include it.
14095 (arm_print_hint_for_fpu_option): New function.
14096 (arm_parse_fpu_option): New function.
14097 (candidate_extension): New class.
14098 (arm_canon_for_multilib): New function.
14099 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14100 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14101 (ARCH_CANONICAL_SPECS): New macro.
14102 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14106 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14107 are set after handling multilib fragments. Set target_cpu_default2
14110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14112 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14114 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14115 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14117 (check_arch): Likewise.
14118 * config/arm/arm.c (arm_configure_build_target): Handle
14119 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14120 options in the default.
14122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14124 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14125 when an option is an alias of another.
14126 * config/arm/parsecpu.awk (optalias): New parser token.
14127 (gen_comm_data): Mark non-alias options as such. Emit entries
14128 for extension aliases.
14129 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14130 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14131 (armv6kz, armv6zk, armv6t2): Likewise.
14132 (armv7): Make vfpv3-d16 an alias.
14133 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14135 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14136 Sort in canonical order.
14137 (armv8-a): Sort in canonical order.
14138 (armv8.1-a, armv8.2-a): Likewise.
14139 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14141 (cortex-a9): Sort in canonical order.
14142 * config/arm/arm.c (selftests.h): Include it.
14143 (arm_test_cpu_arch_data): New function.
14144 (arm_run_self_tests): New function.
14145 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14146 (targetm): Move declaration to the end of the file.
14147 * arm-cpu-cdata.h: Regenerated.
14149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14151 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14152 call to target_mode_check describing the type of option passed.
14153 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14154 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14155 arm_parse_cpu_option_name to match parameters against list of
14157 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14158 arm_arch_core_flags data structure.
14159 * config/arm/arm-cpu_cdata.h: Regenerated.
14161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14163 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14165 (arm_print_hint_for_cpu_option): Likewise.
14166 (arm_print_hint_for_arch_option): Likewise.
14167 (arm_parse_cpu_option_name): Likewise.
14168 (arm_parse_arch_option_name): Likewise.
14169 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14170 of entries in the all_fpus list.
14171 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14172 (arm_parse_cpu_option_name): Declare.
14173 (arm_parse_arch_option_name): Declare.
14174 (arm_parse_option_features): Declare.
14175 (arm_intialize_isa): Declare.
14176 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14178 (gen_comm_data): ... here. Make definitions non-static.
14179 * config/arm/arm-cpu-data.h: Regenerated.
14180 * config/arm/arm-cpu-cdata.h: Regenerated.
14182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14184 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14185 (cpu_arch_extension): New structure.
14186 (cpu_arch_option, arch_option, cpu_option): New structures.
14187 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14188 architecture types.
14189 (gen_data): Generate new format data tables.
14190 * config/arm/arm.c (cpu_tune): New structure.
14191 (cpu_option, processors): Delete.
14192 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14193 (arm_print_hint_for_cpu_option): ... this and ...
14194 (arm_print_hint_for_arch_option): ... this.
14195 (arm_parse_arch_cpu_name): Delete. Replace with ...
14196 (arm_parse_cpu_option_name): ... this and ...
14197 (arm_parse_arch_option_name): ... this.
14198 (arm_unrecognized_feature): Change type of target parameter to
14200 (arm_parse_arch_cpu_features): Delete. Replace with ...
14201 (arm_parse_option_features): ... this.
14202 (arm_configure_build_target): Rework to use new configuration data
14204 (arm_print_tune_info): Rework for new configuration data tables.
14205 * config/arm/arm-cpu-data.h: Regenerated.
14206 * config/arm/arm-cpu.h: Regenerated.
14208 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14210 * Makefile.in (OBJS): Move sbitmap.o from here ...
14211 (OBJS-libcommon): ... to here.
14213 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14215 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14216 (ISA_ALL_CRYPTO): New macro.
14217 (ISA_ALL_SIMD): New macro
14218 (ISA_ALL_FP): New macro.
14219 * config/arm/arm.c (fpu_bitlist): Update initializer.
14220 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14222 (arm9e): Add fpu. Add option for nofp
14223 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14224 (arm926ej-s, arm1026ej-s): Likewise.
14225 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14226 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14227 neon-fp16, neon-vfpv4, nofp and nosimd.
14228 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14229 (cortex-a8): Add fpu. Add option for nofp.
14230 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14231 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14232 (cortex-r4f): Add fpu.
14233 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14234 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14236 (cortex-r8): Likewise.
14237 (cortex-m4): Add fpu. Add option for nofp.
14238 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14239 (cortex-a17.cortex-a7): Likewise.
14240 (cortex-a32): Add fpu. Add options for crypto and nofp.
14241 (cortex-a35, cortex-a53): Likewise.
14242 (cortex-a57): Add fpu. Add option for crypto.
14243 (cortex-a72, cortex-a73): Likewise.
14244 (exynos-m1): Likewise.
14245 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14246 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14247 (cortex-m33): Add fpu. Add option for nofp.
14248 * config/arm/arm-cpu-cdata.h: Regenerated
14249 * config/arm/arm-cpu-data.h: Regenerated.
14251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14253 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14254 (armv5te, armv5tej): Likewise.
14255 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14256 (armv7): Add options fp and vfpv3-d16.
14257 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14258 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14260 (armv7ve): Likewise.
14261 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14262 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14263 (armv8-a): Add nocrypto option.
14264 (armv8.1-a, armv8.2-a): Likewise.
14265 (armv8-m.main): add options fp, fp.dp and nofp.
14267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14269 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14271 (armv8-a+crc): Delete.
14272 (armv8.1-a): Add options simd, crypto and nofp.
14273 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14274 (armv8.2-a+fp16): Delete.
14275 (armv8-m.main): Add option dsp.
14276 (armv8-m.main+dsp): Delete.
14277 (cortex-a8): Add fpu. Add nofp option.
14278 (cortex-a9): Add fpu. Add nofp and nosimd options.
14279 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14280 link to main cpu and architecture data structures.
14281 (gen_comm_data): Only put isa attributes from the main architecture
14283 (option): New statement for architecture and CPU entries.
14284 * arm.c (struct cpu_option): New structure.
14285 (struct processors): Add entry for options.
14286 (arm_unrecognized_feature): New function.
14287 (arm_parse_arch_cpu_name): Ignore any characters after the first
14289 (arm_parse_arch_cpu_feature): New function.
14290 (arm_configure_build_target): Separate out any CPU and architecture
14291 features and parse separately. Don't error out if -mfpu=auto is
14292 used with only an architecture string.
14293 (arm_print_asm_arch_directives): New function.
14294 (arm_file_start): Call it.
14295 * config/arm/arm-cpu-cdata.h: Regenerated.
14296 * config/arm/arm-cpu-data.h: Likewise.
14297 * config/arm/arm-tables.opt: Likewise.
14299 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14301 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14302 assembler when it is not -mfpu=auto.
14304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14306 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14307 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14308 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14309 (ASM_CPU_SPEC): Rewrite.
14310 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14311 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14312 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14313 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14314 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14315 copied string is NUL-terminated. Also strip any characters prefixed
14317 (arm_rewrite_selected_arch): New function.
14318 (arm_rewrite_march): New function.
14320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14322 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14323 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14324 (march, mcpu, mtune): Convert to string-based options.
14325 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14326 (arm_parse_arch_cpu_name): New function.
14327 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14328 identify selected architecture or CPU.
14329 (arm_option_save): New function.
14330 (TARGET_OPTION_SAVE): Redefine.
14331 (arm_option_restore): Restore string options.
14332 (arm_option_print): Print string options.
14334 2017-06-16 Martin Sebor <msebor@redhat.com>
14336 PR tree-optimization/80933
14337 PR tree-optimization/80934
14338 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14339 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14340 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14341 (gimple_fold_builtin): Call them.
14343 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14345 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14346 as unlikely; update profile.
14348 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14350 * predict.c (force_edge_cold): Handle declaring edges impossible
14353 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14355 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14357 (try_unroll_loop_completely): Fix reporting.
14359 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14361 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14363 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14366 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14367 if given a non-constant argument for an intrinsic which requires a
14370 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14372 * profile.c (compare_freqs): New function.
14373 (branch_prob): Sort edge list.
14374 (find_spanning_tree): Assume that the list is priority sorted.
14376 2017-06-16 Richard Biener <rguenther@suse.de>
14378 PR tree-optimization/81090
14379 * passes.def (pass_record_bounds): Remove.
14380 * tree-pass.h (make_pass_record_bounds): Likewise.
14381 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14382 make_pass_record_bounds): Likewise.
14383 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14384 not free niter estimates at the beginning but at the end.
14385 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14387 2017-06-16 Richard Biener <rguenther@suse.de>
14389 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14390 initializer to workaround ICE in host GCC 4.8.
14392 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14394 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14396 (clone_inlined_nodes): Update.
14398 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14400 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14401 prefetch settings, and enable prefetching by default at -O3.
14403 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14405 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14406 Set flag_prefetch_loop_arrays according to tuning data.
14408 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14410 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14411 New tune structure.
14412 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14413 [Unrelated to main purpose of the patch] Place the pointer field last
14414 to enable type checking errors when tune structure are wrongly merged.
14415 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14416 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14417 (thunderx2t99_prefetch_tune): New tune constants.
14418 (tune_params *_tunings): Update all tunings (no functional change).
14419 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14420 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14421 from tunings structures.
14423 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14426 * ubsan.c (instrument_null): Add T argument, use it instead
14427 of computing it based on IS_LHS.
14428 (instrument_object_size): Likewise.
14429 (pass_ubsan::execute): Adjust instrument_null and
14430 instrument_object_size callers to pass gimple_get_lhs or
14431 gimple_assign_rhs1 result to it. Use instrument_null instead of
14432 calling get_base_address and instrument_mem_ref. Handle
14433 aggregate call arguments for object-size sanitization.
14435 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14437 PR tree-optimization/81089
14438 * tree-vrp.c (is_masked_range_test): Validate operands of
14441 2017-06-15 Martin Sebor <msebor@redhat.com>
14444 * dumpfile.c (dump_register): Avoid calling memset to initialize
14445 a class with a default ctor.
14446 * gcc.c (struct compiler): Remove const qualification.
14447 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14448 * hash-table.h: Ditto.
14449 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14451 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14452 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14454 * params.h (struct param_info): Make struct members non-const.
14455 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14456 with default initialization.
14457 * vec.h (vec_copy_construct, vec_default_construct): New helper
14459 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14460 with vec_copy_construct.
14461 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14462 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14463 * doc/invoke.texi (-Wclass-memaccess): Document.
14465 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14467 * emit-rtl.h (is_leaf): Update comment about local
14468 register allocator.
14470 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14473 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14474 for a variable to have a section before checking if the section has a
14476 Set section to.persistent if persistent attribute is set.
14477 Warn if .persistent attribute is used on an automatic variable.
14479 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14481 PR rtl-optimization/80474
14482 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14484 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14486 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14489 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14491 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14492 (casesi): Emit bounds checking as RTL.
14493 (casesi_internal_mips16_<mode>): Remove bounds checking.
14495 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14497 * config/xtensa/xtensa.c (xtensa_option_override): Append
14498 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14499 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14500 xtensa_doloop_hooks): Define unconditionally.
14501 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14503 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14504 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14505 for it in xtensa_option_override.
14506 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14507 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14509 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14511 * doc/cppopts.texi: Document '-' special value to -MF.
14513 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14515 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14516 (cortex_a53_fconst): Likewise.
14517 (cortex_a53_fpmul): Likewise.
14518 (cortex_a53_f_load_64): Likewise.
14519 (cortex_a53_f_load_many): Likewise.
14520 (cortex_a53_advsimd_alu): Likewise.
14521 (cortex_a53_advsimd_alu_q): Likewise.
14522 (cortex_a53_advsimd_mul): Likewise.
14523 (cortex_a53_advsimd_mul_q): Likewise.
14524 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14525 Add missing fmul, r2f_cvt and fconst cases.
14527 2017-06-14 Richard Biener <rguenther@suse.de>
14529 PR middle-end/81088
14530 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14532 (fold_binary_loc): When associating do not treat pre-existing
14533 TREE_OVERFLOW on literal constants as a reason to allow
14534 TREE_OVERFLOW on associated literal constants.
14536 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14538 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14539 (MASK_FEATURES): New macro.
14540 * config/sparc/sparc.c (sparc_option_override): Remove the special
14541 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14543 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14545 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14546 a division of 0 if non-call exceptions are enabled.
14548 2017-06-14 Andrew Pinski <apinski@cavium.com>
14549 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14552 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14553 Improve vector initialization code gen for only variable case.
14555 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14557 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14559 2017-06-14 Richard Biener <rguenther@suse.de>
14561 PR tree-optimization/81083
14562 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14565 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14567 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14568 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14569 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14570 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14571 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14572 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14574 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14576 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14577 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14579 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14581 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14583 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14585 * config/rs6000/t-rtems: Don't handle SPE.
14587 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14589 * config/rs6000/t-linux: Don't handle SPE.
14591 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14593 * config/rs6000/eabispe.h: Delete file.
14595 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14597 * config/rs6000/t-spe: Delete file.
14599 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14601 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14602 (rs6000_legitimate_offset_address_p): Return false for anything in
14603 V2SImode or V2SFmode.
14605 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14607 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14608 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14609 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14611 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14612 (rs6000_legitimate_offset_address_p): Ditto.
14613 (rs6000_emit_move): Ditto.
14614 (rs6000_init_builtins): Remove V4HI_type_node.
14616 2017-06-13 Martin Liska <mliska@suse.cz>
14619 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14620 (gate_asan): Likewise.
14621 * asan.h (asan_no_sanitize_address_p): Remove the function.
14622 (sanitize_flags_p): New function.
14623 * builtins.def: Fix coding style.
14624 * common.opt: Use renamed enum value.
14625 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14626 * doc/extend.texi: Document no_sanitize attribute.
14627 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14628 to SANITIZE_UNDEFINED_NONDEFAULT.
14629 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14630 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14631 Use sanitize_flags_p.
14632 * gimplify.c (gimplify_function_tree): Likewise.
14633 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14634 * opts.c (parse_no_sanitize_attribute): New function.
14635 (common_handle_option): Use renamed enum value.
14636 * opts.h (parse_no_sanitize_attribute): Declare.
14637 * tree.c (sanitize_flags_p): New function.
14638 * tree.h: Declared here.
14639 * tsan.c: Use sanitize_flags_p.
14640 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14641 (instrument_mem_ref): Likewise.
14642 (instrument_bool_enum_load): Likewise.
14643 (do_ubsan_in_current_function): Remove the function.
14644 (pass_ubsan::execute): Use sanitize_flags_p.
14645 * ubsan.h: Remove do_ubsan_in_current_function
14646 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14647 (dump_function_to_file): Use it here.
14649 2017-06-13 Martin Jambor <mjambor@suse.cz>
14651 PR tree-optimization/80803
14652 PR tree-optimization/81063
14653 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14654 (propagate_subaccesses_across_link): Enqueue subtree whenever
14655 necessary instead of relying on the caller.
14657 2017-06-13 Martin Jambor <mjambor@suse.cz>
14659 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14660 that have a first_link.
14661 (sort_and_splice_var_accesses): Do not check first_link before
14663 (subtree_mark_written_and_enqueue): Likewise.
14664 (propagate_all_subaccesses): Likewise and do not stop at first
14665 parent with a first_link.
14667 2017-06-13 Martin Jambor <mjambor@suse.cz>
14669 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14672 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14674 * match.pd: New pattern.
14676 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14678 * tree-vrp.c (is_masked_range_test): New function.
14679 (register_edge_assert_for): Determine ranges for
14682 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14684 PR tree-optimization/67328
14685 * fold-const.c (maskable_range_p): New function.
14686 (build_range_check): Generate bittests if possible.
14688 2017-06-13 Martin Liska <mliska@suse.cz>
14690 * gimple-pretty-print.c (dump_probability): Add new argument.
14691 (dump_edge_probability): Dump both probability and count.
14692 (dump_gimple_label): Likewise.
14693 (dump_gimple_bb_header): Likewise.
14695 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14698 * config/avr/avr-devices.c: Fix indentation.
14699 * config/avr/gen-avr-mmcu-specs.c: Dito.
14701 2017-06-13 Richard Biener <rguenther@suse.de>
14703 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14704 instead get vector type from stmt_info.
14705 (vectorizable_reduction): Adjust. Remove dead code.
14707 2017-06-13 Richard Biener <rguenther@suse.de>
14709 PR middle-end/81065
14710 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14711 case of C * (x * C2 + C3).
14712 (fold_addr_of_array_ref_difference): Properly fold index difference.
14714 2017-06-12 David S. Miller <davem@davemloft.net>
14717 * config/sparc/sparc.md (return expander): Emit frame blockage if
14718 function uses alloca.
14720 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14722 * combine.c (make_field_assignment): Check len rather than the mode
14723 precision when calling force_to_mode.
14725 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14727 Support multilibs and devices that see flash in RAM address range.
14730 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14731 (avr_mcu_t) <flash_pm_offset>: New field.
14732 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14733 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14734 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14735 (AVR_TINY_PM_OFFSET): Remove macro.
14736 * config/avr/avr.opt (-mshort-calls): New option.
14737 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14738 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14739 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14740 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14741 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14742 instead of avr_arch->have_jmp_call.
14743 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14744 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14745 avr_arch->flash_pm_offset to define.
14746 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14747 new field flash_pm_offset. Add entry for avrxmega3.
14748 (avr_texinfo): Add entry for avrxmega3.
14749 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14750 attiny212, attiny214,
14751 attiny412, attiny414, attiny416, attiny417,
14752 attiny814, attiny816, attiny817,
14753 attiny1614, attiny1616, attiny1617,
14754 attiny3214, attiny3216, attiny3217.
14755 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14756 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14757 (avr_print_operand_address) [AVR_TINY]: Same.
14758 (avr_asm_init_sections) <readonly_data_section>: Only patch
14759 callback if avr_arch->flash_pm_offset = 0.
14760 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14761 for rodata if avr_arch->flash_pm_offset != 0.
14762 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14763 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14764 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14765 (m_options): Append opt_rcall.
14766 (m_dirnames): Append dir_rcall.
14767 * config/avr/t-multilib: Regenerate.
14769 * configure.ac [target=avr]: Check whether avrxmega3 default
14770 linker description file works as needed.
14771 * configure: Regenerate.
14772 * doc/avr-mmcu.texi: Regenerate.
14773 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14774 <__AVR_ARCH__>: Document avrxmega3 and 103.
14775 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14776 <__AVR_SHORT_CALLS__>: Document it.
14777 <__AVR_PM_BASE_ADDRESS__>: Document it.
14778 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14779 (AVR Variable Attributes) <progmem>: Document this is
14780 not needed for avrxmega3.
14781 (AVR Named Address Spaces) <__flash>: Dito.
14783 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14785 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14787 2017-06-12 Doug Rupp <rupp@adacore.com>
14789 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14790 Append vxworks-stdint.h to the tm_file list.
14791 * config/vxworks-stdint.h: New file.
14793 2017-06-12 Martin Liska <mliska@suse.cz>
14795 PR tree-optimization/81041
14796 * tree-profile.c (gimple_gen_ic_func_profiler):
14797 Create an extra BB in profile-generate
14798 (gimple_gen_time_profiler): Likewise.
14800 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14802 PR tree-optimization/81003
14803 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14804 (update_range_test): Use it instead of force_gimple_operand_gsi.
14806 2017-06-12 Richard Biener <rguenther@suse.de>
14808 PR tree-optimization/81053
14809 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14810 with backedge value not defined in loop. Simplify def stmt
14813 2017-06-11 Tom de Vries <tom@codesourcery.com>
14816 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14818 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14819 nvptx_cannot_force_const_mem.
14821 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14823 * opts.c (finish_options): Move test for flag_split_stack after
14824 it has been initialized.
14826 2017-06-11 Jason Merrill <jason@redhat.com>
14828 * tree.h (id_equal): New.
14829 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14830 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14831 instead of strcmp of IDENTIFIER_POINTER.
14833 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14835 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14836 (mark_all_inlined_calls_cdtor): Fix formating.
14837 (inline_transform): Rescale profile before inlining.
14839 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14841 * cgraph.h (cgraph_edge::clone): Update prototype.
14842 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14843 (cgraph_node::create_clone): Update.
14844 (cgraph_node::create_version_clone): Update.
14845 * tree-inline.c (copy_bb): Update.
14846 (expand_call_inline): Update.
14848 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14850 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14851 factored out from ...
14852 (rs6000_emit_prologue): ... here.
14854 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14856 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14857 factored out from ...
14858 (rs6000_emit_prologue): ... here.
14860 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14862 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14864 (handle_missing_profiles): Fix computation of tp_first_run.
14865 (counts_to_freqs): Do not touch freqs when count is 0.
14867 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14869 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14872 2017-06-10 Tom de Vries <tom@codesourcery.com>
14874 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14875 attributes): Document signal effective target.
14877 2017-06-10 Tom de Vries <tom@codesourcery.com>
14879 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14880 Document effective target stack_size.
14882 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14884 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14885 to the edit_context if they can be auto-applied.
14887 2017-06-9 Ian Lance Taylor <iant@golang.org>
14889 * opts.c (finish_options): If -fsplit-stack, disable implicit
14890 -forder-blocks-and-partition.
14891 * doc/invoke.texi (Optimize Options): Document that when using
14892 -fsplit-stack -forder-blocks-and-partition is not implicitly
14895 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14897 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14898 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14899 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14900 * builtins.def (abort, trap, unreachable): Declare cold.
14901 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14902 * tree-core.h (ECF_COLD): New.
14903 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14904 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14906 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14908 * predict.c (unlikely_executed_stmt_p): Cleanup.
14910 2017-06-09 Richard Biener <rguenther@suse.de>
14912 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14913 model if the ref is always written to.
14915 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14917 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14919 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14921 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14924 2017-06-09 Tom de Vries <tom@codesourcery.com>
14927 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14928 "target cannot support label values" when encountering LABEL_REF.
14930 2017-06-09 Martin Liska <mliska@suse.cz>
14932 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14933 (gimple_gen_ic_func_profiler): Emit direct comparison
14934 of __gcov_indirect_call_callee with NULL.
14935 (gimple_gen_time_profiler): Change probability from
14936 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14938 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14940 * profile.c (edge_gcov_counts): Turn to pointer.
14941 (compute_branch_probabilities, compute_branch_probabilities): Update.
14942 (branch_prob): Do not clear edge_gcov_count.
14943 * profile.h (edge_gcov_counts): Turn to pointer.
14944 (edge_gcov_count): Update.
14946 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14948 * gimple.h (gimple_check_failed): Mark cold.
14950 2017-06-09 Richard Biener <rguenther@suse.de>
14952 PR tree-optimization/66623
14953 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14954 refactor check_reduction into two parts, properly computing
14955 whether we have to check reduction validity for outer loop
14958 2017-06-09 Richard Biener <rguenther@suse.de>
14960 PR tree-optimization/79483
14961 * graphite-scop-detection.c (order): New global.
14962 (get_order): Compute bb to order mapping that satisfies code
14963 generation constraints.
14964 (cmp_pbbs): New helper.
14965 (build_scops): Start domwalk at entry block, sort generated
14968 2017-06-09 Richard Biener <rguenther@suse.de>
14970 PR middle-end/81007
14971 * ipa-polymorphic-call.c
14972 (ipa_polymorphic_call_context::restrict_to_inner_class):
14973 Skip FIELD_DECLs with error_mark_node type.
14974 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14977 2017-06-09 Martin Liska <mliska@suse.cz>
14979 * predict.c (struct branch_predictor): New struct.
14980 (test_prediction_value_range): New test.
14981 (predict_c_tests): New function.
14982 * selftest-run-tests.c (selftest::run_tests): Run the function.
14983 * selftest.h: Declare new tests.
14985 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14988 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14989 gen_add3_insn did not fail.
14990 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14991 r0, construct that number in a temporary reg and add that reg to r0.
14992 If asked to put the result in r0 as well, fail.
14994 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14996 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14997 for early expansion of vec_eqv.
14999 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15001 PR middle-end/81005
15002 * ubsan.c (instrument_null): Avoid pointless code temporary.
15003 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15005 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15009 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15012 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15015 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15017 * predict.c (unlikely_executed_edge_p): Move ahead.
15018 (probably_never_executed_edge_p): Use it.
15020 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15022 PR middle-end/79988
15023 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15024 gimple_call_builtin_p call.
15026 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15028 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15029 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15030 rtl_check_failed_type2, rtl_check_failed_code1,
15031 rtl_check_failed_code2, rtl_check_failed_code_mode,
15032 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15033 rtvec_check_failed_bounds, rtl_check_failed_flag,
15034 _fatal_insn_not_found, _fatal_insn): Likewise.
15035 * tree.h (tree_contains_struct_check_failed,
15036 tree_check_failed, tree_not_check_failed,
15037 tree_class_check_failed, tree_range_check_failed,
15038 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15039 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15040 tree_operand_check_failed, omp_clause_check_failed,
15041 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15044 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15046 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15047 flag_branch_probabilities.
15048 * ipa-inline.c (edge_badness): Likewise.
15049 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15050 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15051 * predict.c (maybe_hot_frequency_p): Likewise.
15052 (probably_never_executed): Likewise.
15053 * sched-ebb.c (schedule_ebbs): Likewise.
15054 * sched-rgn.c (find_single_block_region): Likewise.
15055 * tracer.c (tail_duplicate): Likewise.
15057 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15059 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15060 longer requires x_flag_profile_use.
15062 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15064 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15065 instead of flag_reorder_blocks_and_partition.
15066 * dbxout.c (dbxout_function_end): Likewise.
15067 * dwarf2out.c (gen_subprogram_die): Likewise.
15068 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15069 * hw-doloop.c (reorg_loops): Likewise.
15070 * varasm.c (assemble_start_function,
15071 assemble_end_function): Likewise.
15072 (decide_function_section): Do not check for
15073 flag_reorder_blocks_and_partition.
15075 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15077 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15079 (chkp_get_hard_register_fake_addr_expr): Ditto.
15080 (chkp_build_addr_expr): Add check for hard reg case.
15081 (chkp_parse_array_and_component_ref): Ditto.
15082 (chkp_find_bounds_1): Ditto.
15083 (chkp_process_stmt): Don't generate bounds store for
15086 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15088 * predict.c (maybe_hot_bb_p): Do not check profile status.
15089 (maybe_hot_edge_p): Likewise.
15090 (probably_never_executed): Check for zero counts even if profile
15092 (unlikely_executed_edge_p): New function.
15093 (unlikely_executed_stmt_p): New function.
15094 (unlikely_executed_bb_p): New function.
15095 (set_even_probabilities): Use unlikely predicates.
15096 (combine_predictions_for_bb): Likewise.
15097 (predict_paths_for_bb): Likewise.
15098 (predict_paths_leading_to_edge): Likewise.
15099 (determine_unlikely_bbs): New function.
15100 (estimate_bb_frequencies): Use it.
15101 (compute_function_frequency): Use zero counts even if profile is
15103 * profile-count.h: Fix typo.
15105 2017-08-08 Julia Koval <julia.koval@intel.com>
15107 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15108 _mm512_mask_cvtsepi16_storeu_epi8,
15109 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15110 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15111 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15112 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15113 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15114 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15115 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15116 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15117 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15118 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15119 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15120 __builtin_ia32_pmovuswb256mem_mask,
15121 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15122 __builtin_ia32_pmovwb512mem_mask): New builtins.
15124 2017-08-08 Julia Koval <julia.koval@intel.com>
15126 PR target/73350,80862
15127 * config/i386/subst.md (round): Fix round pattern.
15128 * config/i386/i386.c (ix86_erase_embedded_rounding):
15129 Fix erasing rounding for the fixed pattern.
15131 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15133 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15135 2017-06-08 Martin Liska <mliska@suse.cz>
15137 PR gcov-profile/80911
15138 * gcov.c (block_info::block_info): New constructor.
15140 2017-06-07 Carl Love <cel@us.ibm.com>
15142 * config/rs6000/rs6000-c: The return type of the following
15143 built-in functions was implemented as int not long long. Fix sign
15144 of return value for the unsigned version of vec_mulo and vec_mule.
15145 vector unsigned long long vec_bperm (vector unsigned long long,
15146 vector unsigned char)
15147 vector signed long long vec_mule (vector signed int,
15149 vector unsigned long long vec_mule (vector unsigned int,
15150 vector unsigned int)
15151 vector signed long long vec_mulo (vector signed int,
15153 vector unsigned long long vec_mulo (vector unsigned int,
15154 vector unsigned int)
15155 * doc/extend.texi: Fix the documentation for the built-in
15158 2017-06-07 Carl Love <cel@us.ibm.com>
15161 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15164 2017-06-07 Carl Love <cel@us.ibm.com>
15166 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15167 support, Generate doublehv for signed int/float for BE case only.
15169 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15171 * doc/invoke.texi (mcx16): Rewrite.
15173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15175 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15176 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15177 *mov<mode>_softfloat, and an anonymous splitter): Use
15178 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15180 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15182 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15184 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15185 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15186 (rs6000_debug_reg_global): Adjust.
15187 (rs6000_init_hard_regno_mode_ok): Adjust.
15188 (rs6000_dbx_register_number): Adjust.
15189 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15190 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15191 Remove SPE_ACC and SPEFSCR.
15192 (REG_ALLOC_ORDER): Ditto.
15193 (FRAME_POINTER_REGNUM): Change to 111.
15194 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15195 (REG_CLASS_NAMES): Ditto.
15196 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15197 (REGISTER_NAMES): Ditto.
15198 (ADDITIONAL_REG_NAMES): Ditto.
15199 (rs6000_reg_names): Ditto.
15200 * config/rs6000/rs6000.md: Renumber some register number
15203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15205 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15207 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15208 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15210 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15211 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15212 Delete the SPE high registers.
15213 (REG_ALLOC_ORDER): Ditto.
15214 (enum reg_class): Remove SPE_HIGH_REGS.
15215 (REG_CLASS_NAMES): Ditto.
15216 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15217 (REGISTER_NAMES): Ditto.
15218 (rs6000_reg_names): Ditto.
15219 * doc/tm.texi.in: Remove SPE as example.
15220 * doc/tm.texi: Regenerate.
15222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15224 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15225 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15226 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15227 * config/rs6000/rs6000.md (type): Remove "brinc".
15229 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15231 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15232 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15233 * config/rs6000/linuxspe.h: Delete file.
15234 * config/rs6000/rs6000.md: Don't include spe.md.
15235 * config/rs6000/spe.h: Delete file.
15236 * config/rs6000/spe.md: Delete file.
15237 * config/rs6000/t-rs6000: Remove spe.md.
15239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15241 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15242 (reg_or_none500mem_operand): Delete.
15243 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15244 instead of reg_or_none500mem_operand.
15246 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15249 handling of SPE flags.
15250 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15252 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15254 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15256 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15257 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15258 paired_divv2sf3): Similar.
15259 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15260 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15261 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15263 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15264 Rename the paired_* instruction patterns.
15265 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15267 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15268 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15269 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15270 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15271 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15272 PAIRED_VECTOR_MODE.
15273 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15274 (spe_func_has_64bit_regs_p): Delete.
15275 (spe_expand_predicate_builtin): Delete.
15276 (spe_expand_evsel_builtin): Delete.
15277 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15278 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15279 (invalid_e500_subreg): Delete.
15280 (rs6000_legitimize_address): Always force_reg op2 as well, for
15281 paired single memory accesses.
15282 (rs6000_member_type_forces_blk): Delete.
15283 (rs6000_spe_function_arg): Delete.
15284 (rs6000_expand_unop_builtin): Delete SPE handling.
15285 (rs6000_expand_binop_builtin): Ditto.
15286 (spe_expand_stv_builtin): Delete.
15287 (bdesc_2arg_spe): Delete.
15288 (spe_expand_builtin): Delete.
15289 (spe_expand_predicate_builtin): Delete.
15290 (spe_expand_evsel_builtin): Delete.
15291 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15292 (spe_init_builtins): Delete.
15293 (spe_func_has_64bit_regs_p): Delete.
15294 (savres_routine_name): Delete "info" parameter. Adjust callers.
15295 (rs6000_emit_stack_reset): Ditto.
15296 (rs6000_dwarf_register_span): Delete.
15297 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15298 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15299 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15301 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15303 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15304 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15305 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15306 mulv2sf3, divv2sf3): Delete expanders.
15308 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15310 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15312 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15314 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15315 * config/rs6000/rs6000.c: Ditto.
15317 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15319 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15320 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15321 comparison_operator.
15323 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15325 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15326 * config/rs6000/rs6000.opt: Ditto.
15327 * config/rs6000/t-rtems: Ditto.
15329 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15331 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15332 TARGET_E500_SINGLE by 0, simplify.
15333 * config/rs6000/rs6000.c: Ditto.
15334 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15335 (spe_build_register_parallel): Delete.
15336 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15337 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15338 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15339 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15340 (E500_CONVERT): Delete.
15341 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15343 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15345 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15346 * config/rs6000/dfp.md: Ditto.
15347 (negdd2, *negdd2_fpr): Merge.
15348 (absdd2, *absdd2_fpr): Merge.
15349 (negtd2, *negtd2_fpr): Merge.
15350 (abstd2, *abstd2_fpr): Merge.
15351 * config/rs6000/e500.h: Delete file.
15352 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15353 TARGET_FPRS by 1 and simplify.
15354 * config/rs6000/rs6000-c.c: Ditto.
15355 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15356 TARGET_DF_SPE by 0.
15357 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15359 * config/rs6000/rs6000.md: Ditto.
15360 (floatdidf2, *floatdidf2_fpr): Merge.
15361 (move_from_CR_gt_bit): Delete.
15362 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15363 (E500_CR_IOR_COMPARE): Delete.
15364 (All patterns that require !TARGET_FPRS): Delete.
15365 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15367 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15369 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15371 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15373 * graphds.c (add_edge): Intitialize edge's attached data.
15374 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15375 pointer parameter. Call pointed function on each edge during
15376 graph traversing. Skip traversing the edge when the function
15378 (graphds_dfs, graphds_scc): Ditto.
15379 (for_each_edge): New parameter. Pass the new parameter to callback
15381 * graphds.h (skip_edge_callback): New function pointer type.
15382 (graphds_dfs, graphds_scc): New function pointer parameter.
15383 (graphds_edge_callback, for_each_edge): New parameter.
15385 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15387 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15388 out code checking if runtime alias check is possible to below ...
15389 Call the new function.
15390 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15391 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15393 2017-06-07 Marek Polacek <polacek@redhat.com>
15396 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15397 TYPE_OVERFLOW_WRAPS check.
15399 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15401 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15402 if versioning is required.
15403 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15404 peeling with the check for versioning.
15406 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15408 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15409 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15410 Set true to new parameter if new ssa variable is defined.
15411 (vect_gen_vector_loop_niters): Refactor. Set range information
15412 for the new vector loop bound variable.
15413 (vect_do_peeling): Ditto.
15415 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15417 * tree-affine.c (ssa.h): Include header file.
15418 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15419 has wrapping overflow behavior.
15421 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15423 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15425 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15427 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15428 (tree_to_aff_combination): ... here.
15430 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15432 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15433 reg_pressure model function.
15434 (ivopts_global_cost_for_size): Delete.
15435 (determine_set_costs, iv_ca_recount_cost): Call new model function
15436 ivopts_estimate_reg_pressure.
15438 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15440 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15441 expensive than udiv. Remove floating point cases from mod.
15443 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15445 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15446 Increase idiv cost.
15448 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15450 * config/aarch64/aarch64.md
15451 (copysignsf3): Fix mask generation.
15453 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15455 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15457 (class dump_manager): Add register_dumps method.
15458 * dumpfile.c: Include langhooks.h.
15459 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15460 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15461 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15462 (dump_manager::dump_register): Start with 512 entries instead of 32.
15463 (dump_manager::register_dumps): New method.
15464 * toplev.c (general_init): Instead of invoking register_dumps
15465 langhook, invoke register_dumps method on the dump manager.
15466 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15469 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15471 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15472 Say that # requires an associated define_split to exist, and that
15473 the define_split must be suitable for use after register allocation.
15475 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15477 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15478 (compute_outgoing_frequencies): Also initialize zero counts.
15479 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15480 around loops; preserve more of profile when nothing changes.
15482 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15484 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15486 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15487 * config/arm/arm-cpu-cdata.h: Regenerate.
15488 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15489 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15490 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15491 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15492 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15494 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15496 * config/arm/t-rmprofile: Likewise.
15497 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15499 2017-06-06 David S. Miller <davem@davemloft.net>
15502 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15503 blockage if function uses alloca.
15505 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15507 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15508 New "uid" fields to hold pretty-print IDs of group and ref.
15509 Memory references are now identified as <group_id>:<ref_id>
15510 instead of using [random] addresses.
15511 (dump_mem_details): Simplify, no functional change.
15512 (dump_mem_ref): Simplify and make output more concise.
15513 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15514 (find_or_create_group): Initialize group uid.
15515 (record_ref): Initialize ref uid. Improve debug output.
15516 (prune_group_by_reuse, should_issue_prefetch_p,)
15517 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15518 (mark_nontemporal_store, determine_loop_nest_reuse):
15519 Improve debug output.
15521 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15523 * dbgcnt.def (prefetch): New debug counter.
15524 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15525 (schedule_prefetches): Stop issueing prefetches if debug counter
15528 2017-06-06 Tom de Vries <tom@codesourcery.com>
15530 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15531 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15533 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15535 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15536 Use aarch64_reg_or_zero predicate for operand 4.
15537 (aarch64_compare_and_swap<mode> define_insn_and_split):
15538 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15539 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15541 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15543 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15544 (arm_compute_save_core_reg_mask): This.
15545 (thumb1_compute_save_reg_mask): Rename into ...
15546 (thumb1_compute_save_core_reg_mask): This.
15547 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15548 (arm_compute_frame_layout): Likewise.
15550 2017-06-06 Richard Biener <rguenther@suse.de>
15552 PR tree-optimization/80974
15553 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15554 keep or clear leaders SSA info.
15556 2017-06-06 Tom de Vries <tom@codesourcery.com>
15558 * config/nvptx/nvptx.c (split_mode_p): New function.
15559 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15561 2017-06-06 Tom de Vries <tom@codesourcery.com>
15563 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15565 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15568 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15571 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15573 * shrink-wrap.c (handle_simple_exit): Update profile.
15574 (try_shrink_wrapping): Upate profile.
15576 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15578 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15579 (tree_guess_outgoing_edge_probabilities): New.
15580 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15581 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15583 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15585 * ipa-split.c (split_function): Initialize return bb profile.
15587 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15589 * profile.c (compute_branch_probabilities): Also initialize
15590 EXIT_BLOCK profile.
15592 2017-06-06 Richard Biener <rguenther@suse.de>
15594 PR tree-optimization/80928
15595 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15596 (vect_analyze_loop_operations): Properly guard analysis for
15598 (vect_transform_loop): Likewise.
15599 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15600 (vect_model_induction_cost): Do not cost for pure SLP.
15601 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15602 of induction in inner loop vectorization.
15603 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15604 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15605 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15607 (vect_analyze_slp_cost_1): Cost induction.
15608 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15609 (vect_get_slp_vect_defs): Likewise.
15610 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15611 (vect_transform_stmt): Handle SLP reductions.
15612 * tree-vectorizer.h (vectorizable_induction): Adjust.
15614 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15616 * config/rs6000/rs6000.c (make_resolver_func): Update
15617 init_lowered_empty_function call.
15619 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15621 * doc/invoke.texi: Document the -fprofile-abs-path option.
15622 * common.opt (fprofile-abs-path): New option.
15623 * gcov-io.h (gcov_write_filename): Declare.
15624 * gcov-io.c (gcov_write_filename): New function.
15625 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15626 * profile.c (output_location): Likewise.
15628 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15630 * shring-wrap.c: Revert accidental commit.
15632 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15634 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15636 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15638 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15640 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15641 profile in callgraph edge.
15642 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15643 (apply_scale): Likewise.
15644 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15645 Also scale profile when inlining function with zero profile.
15646 (initialize_cfun): Update exit block profile even when it is zero.
15647 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15648 when profile is read.
15650 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15652 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15653 (CLONE_*): New constants to define the processors we can generate
15654 code for with the target_clone attribute.
15655 (rs6000_clone_map): New array to identify which clone processors
15656 the current program is running on.
15657 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15658 target_clone attribute.
15659 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15660 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15661 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15662 (cpu_expand_builtin): Add support for target_clone attribute.
15663 (rs6000_valid_attribute_p): Allow "default" attribute.
15664 (get_decl_name): New debug function to simplify printing the
15665 current function name in debugging statements.
15666 (rs6000_clone_priority): New functions to support the target_clone
15667 attribute, and be able to generate code to switch between ISA 2.05
15668 through ISA 3.0 (power6 through power9).
15669 (rs6000_compare_version_priority): Likewise.
15670 (rs6000_get_function_versions_dispatcher): Likewise.
15671 (make_resolver_func): Likewise.
15672 (add_condition_to_bb): Likewise.
15673 (dispatch_function_versions): Likewise.
15674 (rs6000_generate_version_dispatcher_body): Likewise.
15675 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15676 (fusion_gpr_load_p): Fix a spacing issue.
15677 * doc/extend.texi (Common Function Attributes): Document that the
15678 PowerPC supports the target_clone attribute.
15680 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15682 * config/arm/arm.h: explain F symbol found in description of ARM
15683 register allocation in its legend.
15685 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15687 * config/mips/frame-header-opt.c: Include profile-count.h.
15688 * config/riscv/riscv.c: Include profile-count.h
15690 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15692 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15694 (sm_set_flag_if_changed): Add bbs field.
15695 (execute_sm_if_changed_flag_set): Pass BBS.
15696 (execute_sm): Update.
15698 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15700 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15703 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15705 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15706 (peephole2): New peephole2 to emit the above.
15707 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15709 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15711 * config/aarch64/aarch64.c (define_peephole2 above
15712 *sub_<shift>_<mode>): New peephole.
15714 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15716 * config/i386/i386.c (make_resolver_func): Update.
15717 * Makefile.in: Add profile-count.h and profile-count.o
15718 * auto-profile.c (afdo_indirect_call): Update to new API.
15719 (afdo_set_bb_count): Update.
15720 (afdo_propagate_edge): Update.
15721 (afdo_propagate_circuit): Update.
15722 (afdo_calculate_branch_prob): Update.
15723 (afdo_annotate_cfg): Update.
15724 * basic-block.h: Include profile-count.h
15725 (struct edge_def): Turn count to profile_count.
15726 (struct basic_block_def): Likewie.
15727 (REG_BR_PROB_BASE): Move to profile-count.h
15728 (RDIV): Move to profile-count.h
15729 * bb-reorder.c (max_entry_count): Turn to profile_count.
15730 (find_traces): Update.
15731 (rotate_loop):Update.
15732 (connect_traces):Update.
15733 (sanitize_hot_paths):Update.
15734 * bt-load.c (migrate_btr_defs): Update.
15735 * cfg.c (RDIV): Remove.
15736 (init_flow): Use alloc_block.
15737 (alloc_block): Uninitialize count.
15738 (unchecked_make_edge): Uninitialize count.
15739 (check_bb_profile): Update.
15740 (dump_edge_info): Update.
15741 (dump_bb_info): Update.
15742 (update_bb_profile_for_threading): Update.
15743 (scale_bbs_frequencies_int): Update.
15744 (scale_bbs_frequencies_gcov_type): Update.
15745 (scale_bbs_frequencies_profile_count): New.
15746 * cfg.h (update_bb_profile_for_threading): Update.
15747 (scale_bbs_frequencies_profile_count): Declare.
15748 * cfgbuild.c (compute_outgoing_frequencies): Update.
15749 (find_many_sub_basic_blocks): Update.
15750 * cfgcleanup.c (try_forward_edges): Update.
15751 (try_crossjump_to_edge): Update.
15752 * cfgexpand.c (expand_gimple_tailcall): Update.
15753 (construct_exit_block): Update.
15754 * cfghooks.c (verify_flow_info): Update.
15755 (dump_bb_for_graph): Update.
15756 (split_edge): Update.
15757 (make_forwarder_block): Update.
15758 (duplicate_block): Update.
15759 (account_profile_record): Update.
15760 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15761 (get_estimated_loop_iterations): Update.
15762 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15763 (single_likely_exit): Update.
15764 * cfgloopmanip.c (scale_loop_profile): Update.
15766 (set_zero_probability): Update.
15767 (lv_adjust_loop_entry_edge): Update.
15768 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15769 (purge_dead_edges): Update.
15770 (rtl_account_profile_record): Update.
15771 * cgraph.c (cgraph_node::create): Uninitialize count.
15772 (symbol_table::create_edge): Uninitialize count.
15773 (cgraph_update_edges_for_call_stmt_node): Update.
15774 (cgraph_edge::dump_edge_flags): Update.
15775 (cgraph_node::dump): Update.
15776 (cgraph_edge::maybe_hot_p): Update.
15777 * cgraph.h: Include profile-count.h
15778 (create_clone), create_edge, create_indirect_edge): Update.
15779 (cgraph_node): Turn count to profile_count.
15780 (cgraph_edge0: Likewise.
15781 (make_speculative, clone): Update.
15782 (create_edge): Update.
15783 (init_lowered_empty_function): Update.
15784 * cgraphclones.c (cgraph_edge::clone): Update.
15785 (duplicate_thunk_for_node): Update.
15786 (cgraph_node::create_clone): Update.
15787 * cgraphunit.c (cgraph_node::analyze): Update.
15788 (cgraph_node::expand_thunk): Update.
15789 * final.c (dump_basic_block_info): Update.
15790 * gimple-streamer-in.c (input_bb): Update.
15791 * gimple-streamer-out.c (output_bb): Update.
15792 * graphite.c (print_global_statistics): Update.
15793 (print_graphite_scop_statistics): Update.
15794 * hsa-brig.c: Include basic-block.h.
15795 * hsa-dump.c: Include basic-block.h.
15796 * hsa-gen.c (T sum_slice): Update.
15797 (convert_switch_statements):Update.
15798 * hsa-regalloc.c: Include basic-block.h.
15799 * ipa-chkp.c (chkp_produce_thunks): Update.
15800 * ipa-cp.c (struct caller_statistics): Update.
15801 (init_caller_stats): Update.
15802 (gather_caller_stats): Update.
15803 (ipcp_cloning_candidate_p): Update.
15804 (good_cloning_opportunity_p): Update.
15805 (get_info_about_necessary_edges): Update.
15806 (dump_profile_updates): Update.
15807 (update_profiling_info): Update.
15808 (update_specialized_profile): Update.
15809 (perhaps_add_new_callers): Update.
15810 (decide_about_value): Update.
15811 (ipa_cp_c_finalize): Update.
15812 * ipa-devirt.c (struct odr_type_warn_count): Update.
15813 (struct decl_warn_count): Update.
15814 (struct final_warning_record): Update.
15815 (possible_polymorphic_call_targets): Update.
15816 (ipa_devirt): Update.
15817 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15818 * ipa-icf.c (sem_function::merge): Update.
15819 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15820 * ipa-inline.c (compute_uninlined_call_time): Update.
15821 (compute_inlined_call_time): Update.
15822 (want_inline_small_function_p): Update.
15823 (want_inline_self_recursive_call_p): Update.
15824 (edge_badness): Update.
15825 (lookup_recursive_calls): Update.
15826 (recursive_inlining): Update.
15827 (inline_small_functions): Update.
15828 (dump_overall_stats): Update.
15829 (dump_inline_stats): Update.
15830 * ipa-profile.c (ipa_profile_generate_summary): Update.
15831 (ipa_propagate_frequency): Update.
15832 (ipa_profile): Update.
15833 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15834 * ipa-utils.c (ipa_merge_profiles): Update.
15835 * loop-doloop.c (doloop_modify): Update.
15836 * loop-unroll.c (report_unroll): Update.
15837 (unroll_loop_runtime_iterations): Update.
15838 * lto-cgraph.c (lto_output_edge): Update.
15839 (lto_output_node): Update.
15840 (input_node): Update.
15841 (input_edge): Update.
15842 (merge_profile_summaries): Update.
15843 * lto-streamer-in.c (input_cfg): Update.
15844 * lto-streamer-out.c (output_cfg): Update.
15845 * mcf.c (create_fixup_graph): Update.
15846 (adjust_cfg_counts): Update.
15847 (sum_edge_counts): Update.
15848 * modulo-sched.c (sms_schedule): Update.
15849 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15850 * predict.c (maybe_hot_count_p): Update.
15851 (probably_never_executed): Update.
15852 (dump_prediction): Update.
15853 (combine_predictions_for_bb): Update.
15854 (propagate_freq): Update.
15855 (handle_missing_profiles): Update.
15856 (counts_to_freqs): Update.
15857 (rebuild_frequencies): Update.
15858 (force_edge_cold): Update.
15859 * predict.h: Include profile-count.h
15860 (maybe_hot_count_p, counts_to_freqs): UPdate.
15861 * print-rtl-function.c: Do not include cfg.h
15862 * print-rtl.c: Include basic-block.h
15863 * profile-count.c: New file.
15864 * profile-count.h: New file.
15865 * profile.c (is_edge_inconsistent): Update.
15866 (correct_negative_edge_counts): Update.
15867 (is_inconsistent): Update.
15868 (set_bb_counts): Update.
15869 (read_profile_edge_counts): Update.
15870 (compute_frequency_overlap): Update.
15871 (compute_branch_probabilities): Update; Initialize and deinitialize
15873 (branch_prob): Update.
15874 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15875 (edge_gcov_count): New.
15876 (bb_gcov_count): New.
15877 * shrink-wrap.c (try_shrink_wrapping): Update.
15878 * tracer.c (better_p): Update.
15879 * trans-mem.c (expand_transaction): Update.
15880 (ipa_tm_insert_irr_call): Update.
15881 (ipa_tm_insert_gettmclone_call): Update.
15882 * tree-call-cdce.c: Update.
15883 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15884 (gimple_duplicate_sese_tail): Update.
15885 (gimple_account_profile_record): Update.
15886 (execute_fixup_cfg): Update.
15887 * tree-inline.c (copy_bb): Update.
15888 (copy_edges_for_bb): Update.
15889 (initialize_cfun): Update.
15890 (freqs_to_counts): Update.
15891 (copy_cfg_body): Update.
15892 (expand_call_inline): Update.
15893 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15894 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15895 (try_unroll_loop_completely): Update.
15896 (try_peel_loop): Update.
15897 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15898 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15899 * tree-ssa-loop-split.c (connect_loops): Update.
15900 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15901 * tree-ssa-reassoc.c (branch_fixup): Update.
15902 * tree-ssa-tail-merge.c (replace_block_by): Update.
15903 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15904 (compute_path_counts): Update.
15905 (update_profile): Update.
15906 (recompute_probabilities): Update.
15907 (update_joiner_offpath_counts): Update.
15908 (estimated_freqs_path): Update.
15909 (freqs_to_counts_path): Update.
15910 (clear_counts_path): Update.
15911 (ssa_fix_duplicate_block_edges): Update.
15912 (duplicate_thread_path): Update.
15913 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15914 (struct switch_conv_info): Update.
15915 * tree-tailcall.c (decrease_profile): Update.
15916 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15917 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15918 * value-prof.c (check_counter): Update.
15919 (gimple_divmod_fixed_value): Update.
15920 (gimple_mod_pow2): Update.
15921 (gimple_mod_subtract): Update.
15922 (gimple_ic_transform): Update.
15923 (gimple_stringop_fixed_value): Update.
15924 * value-prof.h (gimple_ic): Update.
15926 2017-06-02 Carl Love <cel@us.ibm.com>
15928 * config/rs6000/rs6000-c: Add support for built-in functions
15929 vector double vec_doublee (vector signed int);
15930 vector double vec_doublee (vector unsigned int);
15931 vector double vec_doublee (vector float);
15932 vector double vec_doubleh (vector signed int);
15933 vector double vec_doubleh (vector unsigned int);
15934 vector double vec_doubleh (vector float);
15935 vector double vec_doublel (vector signed int);
15936 vector double vec_doublel (vector unsigned int);
15937 vector double vec_doublel (vector float);
15938 vector double vec_doubleo (vector signed int);
15939 vector double vec_doubleo (vector unsigned int);
15940 vector double vec_doubleo (vector float);.
15941 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15942 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15944 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15945 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15946 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15948 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15949 vec_doublel, vec_doubleh.
15950 * doc/extend.texi: Update the built-in documentation file for the
15951 new built-in functions.
15953 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15956 * ipa-inline-analysis.c (free_growth_caches): Set
15957 edge_removal_hook_holder to NULL after removing it.
15959 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15961 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15962 comparision with zero.
15964 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15965 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15966 for early expansion of vec_min and vec_max builtins.
15967 (builtin_function_type): Add min/max unsigned variants to those
15968 identified as having unsigned arguments.
15970 2017-06-02 Olivier Hainque <hainque@adacore.com>
15972 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15974 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15976 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15977 Use VALL_F16 iterator rather than VALL.
15979 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15981 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15982 Emit CBNZ inside loop when doing a strong exchange and comparing
15983 against zero. Generate the CC flags after the loop.
15985 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15987 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15988 (dl_section_ref): New.
15989 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15990 On AIX, append an expression to subtract the size of the
15991 section length to dl_section_ref.
15993 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15995 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15996 for early expansion of vector absolute builtins.
15998 2017-06-02 Richard Biener <rguenther@suse.de>
16000 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16001 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16003 2017-06-02 Richard Biener <rguenther@suse.de>
16005 PR tree-optimization/80948
16006 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16007 stmt order as well.
16009 2017-06-02 Richard Biener <rguenther@suse.de>
16011 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16013 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16014 for inductions as relevant.
16016 2017-06-02 Richard Biener <rguenther@suse.de>
16018 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16019 (vectorizable_induction): ... this. Remove dead code.
16021 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16023 * builtins. (expand_builtin_alloca): Remove second parameter and
16024 infer its value from the first parameter instead.
16025 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16027 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16029 PR rtl-optimization/80903
16030 * loop-doloop.c (add_test): Unshare sequence.
16032 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16034 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16036 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16038 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16040 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16041 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16042 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16043 (xlogue_layout::get_stub_rtx): Make static.
16044 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16045 (xlogue_layout::compute_stub_managed_regs): Rename to...
16046 (xlogue_layout::count_stub_managed_regs): ...this.
16047 (xlogue_layout::is_stub_managed_reg): New function.
16048 (xlogue_layout::m_stub_names): Rename to...
16049 (xlogue_layout::s_stub_names): ...this, make static.
16050 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16051 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16052 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16053 xlogue_layout::s_stub_names): Instantiate statics.
16054 (stub_managed_regs): Remove.
16055 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16056 (disable_call_ms2sysv_xlogues): Rename to...
16057 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16058 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16060 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16061 change after reload_completed.
16062 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16064 (ix86_expand_prologue): Likewise.
16065 (ix86_expand_epilogue): Likewise.
16066 (ix86_expand_split_stack_prologue): Likewise.
16067 (ix86_compute_frame_layout): Remove frame parameter ...
16068 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16069 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16071 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16072 (ix86_frame): Move from here ...
16073 * config/i386/i386.h (ix86_frame): ... to here.
16074 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16075 complete ix86_frame data structure instead. Remove some_ld_name.
16077 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16079 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16080 symbols that hold a DECL_VALUE_EXPR.
16082 2017-06-01 Martin Jambor <mjambor@suse.cz>
16084 PR tree-optimization/80898
16085 * tree-sra.c (process_subtree_disqualification): Removed.
16086 (disqualify_candidate): Do not acll
16087 process_subtree_disqualification.
16088 (subtree_mark_written_and_enqueue): New function.
16089 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16090 RHS has been disqualified and re-queue LHS if necessary. Apart
16091 from that, ignore disqualified RHS.
16093 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16095 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16096 address fetch for z10 or later.
16098 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16100 * config/arc/arc.md (tst_movb): Add guard when splitting.
16102 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16104 * config/arc/arc.c (arc_can_eliminate): Test against
16105 arc_frame_pointer_needed.
16107 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16109 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16110 to prevent store reordering.
16111 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16112 (type): Add block type.
16113 (stack_tie): Define special instruction to be used in
16116 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16118 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16119 constraint. It is not valid for the pattern.
16120 (noncommutative_binary_comparison): Likewise.
16122 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16124 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16127 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16129 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16130 be used by the reg-alloc.
16132 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16134 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16135 reg-alloc when having mul64 or mul32x16 instructions.
16136 (mulsidi3): Likewise.
16137 (umulsidi3): Likewise.
16138 (mulsi32x16): New pattern.
16139 (mulsi64): Likewise.
16140 (mulsidi64): Likewise.
16141 (umulsidi64): Likewise.
16142 (MUL32x16_REG): Define.
16143 (mul64_600): Use MUL32x16_REG.
16144 (mac64_600): Likewise.
16145 (umul64_600): Likewise.
16146 (umac64_600): Likewise.
16148 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16150 * config/arc/arc.md (mulsi3_700): Make it commutative.
16152 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16154 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16156 (*sign_extendsidi2_insn): Likewise for movstosw.
16158 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16160 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16161 the type of the input discriminant value. Convert the
16162 discriminant value of signedness vary.
16164 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16166 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16167 Add to -Wall section.
16169 2017-06-01 Richard Biener <rguenther@suse.de>
16171 PR middle-end/66313
16172 * fold-const.c (fold_plusminus_mult_expr): If the factored
16173 factor may be zero use a wrapping type for the inner operation.
16174 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16175 and handle moved defs.
16176 (process_assignment): Properly guard the unary op case. Return a
16177 tri-state indicating that moving the stmt before the call may allow
16178 to continue. Pass through to_move.
16179 (find_tail_calls): Handle moving unrelated defs before
16182 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16185 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16186 splitter result in the canonical way.
16188 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16190 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16191 also for 32bit target. Update insn attributes.
16192 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16194 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16196 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16197 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16198 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16199 (_mm_maskz_min_ss): New intrinsics.
16201 2017-05-31 Martin Liska <mliska@suse.cz>
16203 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16204 Change comment style to one we normally use.
16205 (vectorizable_reduction): Likewise.
16206 (vectorizable_induction): Likewise.
16207 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16208 (vectorizable_call): Likewise.
16209 (vectorizable_simd_clone_call): Likewise.
16210 (vectorizable_conversion): Likewise.
16211 (vectorizable_assignment): Likewise.
16212 (vectorizable_shift): Likewise.
16213 (vectorizable_operation): Likewise.
16214 (vectorizable_store): Likewise.
16215 (vectorizable_load): Likewise.
16216 * tree-vectorizer.h: Likewise.
16218 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16220 * passes.c (emergency_dump_function): New.
16221 * tree-pass.h (emergency_dump_function): Declare.
16222 * plugin.c (plugins_internal_error_function): Remove.
16223 * plugin.h (plugins_internal_error_function): Remove declaration.
16224 * toplev.c (internal_error_function): New static function. Use it...
16225 (general_init): ...here.
16227 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16229 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16230 (arc_rtx_costs): Add costs for new patterns.
16231 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16232 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16234 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16236 * tree-ssa-strlen.c (get_next_strinfo): New function.
16237 (get_stridx_plus_constant): Use it.
16238 (zero_length_string): Likewise.
16239 (adjust_related_strinfos): Likewise.
16240 (adjust_last_stmt): Likewise.
16242 2017-05-31 Richard Biener <rguenther@suse.de>
16245 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16246 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16248 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16250 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16251 loop_vinfo argument and use of dependence distance vectors.
16252 Check instead whether the two references differ only in their
16253 initial value and assume that they have the same alignment if the
16254 difference is a multiple of the vector alignment.
16255 (vect_analyze_data_refs_alignment): Update call accordingly.
16257 2017-05-31 Martin Liska <mliska@suse.cz>
16260 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16262 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16264 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16265 (create_intersect_range_checks): Move from ...
16266 * tree-data-ref.c (create_intersect_range_checks_index)
16267 (create_intersect_range_checks): ... to here.
16268 (create_runtime_alias_checks): New function factored from ...
16269 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16270 here. Call above function.
16271 * tree-data-ref.h (create_runtime_alias_checks): New function.
16273 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16275 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16276 segment length for dr_b and compute it in wide_int.
16278 2017-05-31 Richard Biener <rguenther@suse.de>
16280 PR tree-optimization/80906
16281 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16282 and pass through iv_map.
16283 (copy_bb_and_scalar_dependences): Adjust.
16284 (translate_pending_phi_nodes): Likewise.
16285 (copy_loop_close_phi_args): Handle code-generating IVs instead
16288 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16290 * diagnostic-color.c (color_dict): Add "type-diff".
16291 (parse_gcc_colors): Update comment.
16292 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16293 -fdiagnostics-show-template-tree and -fno-elide-type.
16294 (GCC_COLORS): Add type-diff to example.
16296 (-fdiagnostics-show-template-tree): New.
16297 (-fno-elide-type): New.
16298 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16299 the pp_format_decoder callback. Call any m_format_postprocessor's
16301 (pretty_printer::pretty_printer): Initialize
16302 m_format_postprocessor.
16303 (pretty_printer::~pretty_printer): Delete any
16304 m_format_postprocessor.
16305 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16306 (class format_postprocessor): New class.
16307 (struct pretty_printer::format_decoder): Document the new parameters.
16308 (struct pretty_printer::m_format_postprocessor): New field.
16309 * tree-diagnostic.c (default_tree_printer): Update for new
16310 bool and const char ** params.
16311 * tree-diagnostic.h (default_tree_printer): Likewise.
16313 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16315 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16316 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16317 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16318 rs6000_gen_cell_microcode code.
16319 (rs6000_final_prescan_insn): Delete.
16320 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16321 "warn-cell-microcode" entries.
16322 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16323 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16324 throughout. Change cc_reg_not_micro_cr0_operand to
16325 cc_reg_not_cr0_operand throughout.
16326 (*extendhi<mode>2_noload): Delete.
16327 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16328 (mwarn-cell-microcode): Delete.
16329 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16330 -mgen-cell-microcode and -mwarn-cell-microcode.
16332 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16335 * config/i386/constraints.md (Yd): New constraint.
16337 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16338 and (?Yd, r) alternatives. Update insn attributes.
16339 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16340 and (?*Yd, r) alternatives. Update insn attributes.
16341 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16343 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16345 * gimplify.c (gimplify_modify_expr): Don't create a
16346 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16349 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16351 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16353 2017-05-30 Richard Biener <rguenther@suse.de>
16355 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16356 and reduc_def fields.
16357 (STMT_VINFO_REDUC_TYPE): New define.
16358 (STMT_VINFO_REDUC_DEF): Likewise.
16359 (vect_force_simple_reduction): Adjust prototype.
16360 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16361 (vect_is_simple_reduction): Remove check_reduction argument.
16362 (vect_force_simple_reduction): Adjust and set
16363 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16364 (vectorizable_reduction): Do not re-do reduction analysis
16365 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16366 * tree-parloops.c (gather_scalar_reductions): Adjust.
16368 2017-05-30 Richard Biener <rguenther@suse.de>
16370 PR middle-end/80901
16371 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16374 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16376 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16377 Introduce unknown_misalignment parameter and remove vf.
16378 (vect_peeling_hash_get_lowest_cost):
16379 Pass unknown_misalignment parameter.
16380 (vect_enhance_data_refs_alignment):
16381 Fix unsupportable data ref treatment.
16383 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16385 * tree-vect-data-refs.c (vect_get_data_access_cost):
16386 Workaround for SLP handling.
16387 (vect_enhance_data_refs_alignment):
16388 Compute costs for doing no peeling at all, compare to the best
16389 peeling costs so far and avoid peeling if cheaper.
16391 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16393 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16394 Return peeling info and set costs to zero for unlimited cost
16396 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16397 with unknown misalignment. Compute and costs for unknown
16398 misalignment, compare them to the costs for known misalignment
16399 and choose the cheapest for peeling.
16401 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16403 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16404 (vect_get_peeling_costs_all_drs): Create function.
16405 (vect_peeling_hash_get_lowest_cost):
16406 Use vect_get_peeling_costs_all_drs.
16407 (vect_peeling_supportable): Create function.
16408 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16410 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16412 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16413 DR_HAS_NEGATIVE_STEP.
16414 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16415 (vect_enhance_data_refs_alignment): Use.
16416 (vect_duplicate_ssa_name_ptr_info): Use.
16417 * tree-vectorizer.h (dr_misalignment): Use.
16418 (known_alignment_for_access_p): Use.
16420 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16423 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16425 (has_section_name): New function.
16427 2017-05-30 Martin Liska <mliska@suse.cz>
16430 * auto-profile.c (get_function_decl_from_block): Fix
16433 2017-05-30 Richard Biener <rguenther@suse.de>
16435 PR middle-end/80876
16436 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16438 2017-05-30 Martin Liska <mliska@suse.cz>
16440 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16441 * dumpfile.h (struct dump_file_info): Remove ctors.
16443 2017-05-30 Martin Liska <mliska@suse.cz>
16445 * predict.def: Fix GNU coding style.
16447 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16449 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16450 Mark 'to' argument with ATTRIBUTE_UNUSED.
16452 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16454 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16455 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16456 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16459 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16461 * doc/install.texi (Options specification): Restore entry of
16462 --enable-sjlj-exceptions.
16464 2017-05-27 Michael Eager <eager@eagercon.com>
16467 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16469 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16471 * config/microblaze/microblaze.h
16472 (FIXED_REGISTERS): Update in macro.
16473 (CALL_USED_REGISTERS): Update in macro.
16475 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16477 * doc/install.texi: Add links to macOS binary distributions.
16479 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16483 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16485 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16487 2017-05-26 Martin Liska <mliska@suse.cz>
16489 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16491 2017-05-26 Martin Liska <mliska@suse.cz>
16493 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16494 always leading ';; '.
16495 (dump_bb_info): Likewise.
16496 (brief_dump_cfg): Likewise.
16497 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16498 * dumpfile.c: Remove usage of TDF_VERBOSE.
16499 * dumpfile.h (enum dump_kind): Likewise.
16500 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16501 * print-tree.c (debug_verbose): Remove.
16502 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16503 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16504 * tree-diagnostic.c (default_tree_printer): Replace
16505 TDF_DIAGNOSTIC with TDF_SLIM.
16507 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16509 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16510 in parameter loop, rather than loop_vinfo.
16511 (create_intersect_range_checks): Ditto.
16512 (vect_create_cond_for_alias_checks): Update call to above functions.
16514 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16516 PR tree-optimization/80815
16517 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16518 for merging runtime alias checks. Handle negative DR_STEPs.
16520 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16522 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16524 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16525 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16526 out code pruning runtime alias checks.
16527 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16528 factored out from above.
16529 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16531 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16533 (prune_runtime_alias_test_list): New decalaration.
16535 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16537 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16538 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16539 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16540 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16541 (operator==, comp_dr_with_seg_len_pair): Ditto.
16542 (vect_prune_runtime_alias_test_list): Ditto.
16544 2017-05-26 Martin Liska <mliska@suse.cz>
16547 * params.def: Bound partial-inlining-entry-probability param.
16549 2017-05-26 Marek Polacek <polacek@redhat.com>
16552 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16555 2017-05-26 Richard Biener <rguenther@suse.de>
16557 PR tree-optimization/80842
16558 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16561 2017-05-26 Richard Biener <rguenther@suse.de>
16563 PR tree-optimization/80844
16564 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16566 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16568 * doc/md.texi (Machine Constraints): Update x86 family
16569 machine constraints section to match 'config/i386/constraints.md'.
16571 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16573 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16575 2017-05-25 Nathan Sidwell <nathan@acm.org>
16577 * doc/invoke.texi (--enable-languages): Update documentation.
16579 2017-05-25 Martin Liska <mliska@suse.cz>
16581 * dumpfile.c: Add TDF_FOLDING.
16582 * dumpfile.h (enum dump_kind): Likewise.
16583 * genmatch.c (dt_simplify::gen_1): Use it.
16585 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16587 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16589 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16591 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16592 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16594 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16596 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16597 * match.pd (X == C): Rewrite it here.
16598 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16599 with_certain_nonzero_bits2): New predicates.
16600 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16602 2017-05-24 Nathan Sidwell <nathan@acm.org>
16604 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16607 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16610 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16612 * config/powerpcspe: New port. Files are copied from the rs6000
16613 port, with "rs6000" in filenames replaced by "powerpcspe".
16615 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16617 PR rtl-optimization/80754
16618 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16620 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16622 * config/sparc/sparc.md (length): Return the correct value for -mflat
16623 sibcalls to match output_sibcall.
16625 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16629 * config/rs6000/rs6000.c (struct machine_function): Add new field
16631 (rs6000_get_separate_components): Init that field, use it.
16632 (rs6000_components_for_bb): Use the field.
16634 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16636 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16638 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16640 PR middle-end/80823
16641 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16643 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16646 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16647 address against address_operand predicate.
16648 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16650 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16652 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16653 parameters passed indirectly.
16655 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16657 * config/i386/i386.md (*movdi_internal): Remove SSE4
16658 alternative 18 (?r, *v). Update insn attributes.
16659 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16660 Update insn attributes.
16661 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16662 Update insn attributes.
16663 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16664 alternative 1 (r, v). Remove isa attribute.
16665 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16666 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16667 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16669 2017-05-23 Tom de Vries <tom@codesourcery.com>
16671 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16674 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16676 * cgraphunit.c (symbol_table::process_new_functions): Update.
16677 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16678 (inline_generate_summary): Rename to ...
16679 (ipa_fn_summary_generate): ... this one.
16680 (inline_read_summary): Rename to ...
16681 (ipa_fn_summary_read): ... this one.
16682 (inline_write_summary): Rename to ...
16683 (ipa_fn_summary_write): ... this one.
16684 (inline_free_summary): Rename to ...
16685 (ipa_free_fn_summary): ... this one.
16686 (pass_data_local_fn_summary, pass_local_fn_summary,
16687 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16688 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16689 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16690 make_pass_ipa_fn_summary): New.
16691 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16692 inline_write_summary, inline_free_summary): Remove.
16693 (ipa_free_fn_summary) : New.
16694 * ipa-inline.c (ipa_inline): Update.
16695 (pass_ipa_inline): Do not generate summaries.
16696 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16698 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16699 and add pass_ipa_fn_summary.
16700 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16702 (make_pass_inline_parameters): Remove.
16704 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16706 * omp-low.c (struct omp_context): Remove "default_kind" member.
16709 * omp-offload.c (execute_oacc_device_lower): Remove the
16710 parallelism dimensions function attributes for unparallelized
16711 OpenACC kernels constructs.
16713 2017-05-23 Martin Liska <mliska@suse.cz>
16715 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16717 (cgraph_edge::make_speculative): Likewise.
16718 (cgraph_edge::resolve_speculation): Likewise.
16719 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16720 (cgraph_node::dump): Likewise.
16721 * cgraph.h: Likewise.
16722 * cgraphunit.c (analyze_functions): Likewise.
16723 (symbol_table::compile): Likewise.
16724 * ipa-cp.c (print_all_lattices): Likewise.
16725 (determine_versionability): Likewise.
16726 (initialize_node_lattices): Likewise.
16727 (ipcp_verify_propagated_values): Likewise.
16728 (estimate_local_effects): Likewise.
16729 (update_profiling_info): Likewise.
16730 (create_specialized_node): Likewise.
16731 (perhaps_add_new_callers): Likewise.
16732 (decide_about_value): Likewise.
16733 (decide_whether_version_node): Likewise.
16734 (identify_dead_nodes): Likewise.
16735 (ipcp_store_bits_results): Likewise.
16736 * ipa-devirt.c (dump_targets): Likewise.
16737 (ipa_devirt): Likewise.
16738 * ipa-icf.c (sem_item::dump): Likewise.
16739 (sem_function::equals): Likewise.
16740 (sem_variable::equals): Likewise.
16741 (sem_item_optimizer::read_section): Likewise.
16742 (sem_item_optimizer::execute): Likewise.
16743 (congruence_class::dump): Likewise.
16744 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16745 (dump_inline_summary): Likewise.
16746 (estimate_node_size_and_time): Likewise.
16747 (inline_analyze_function): Likewise.
16748 * ipa-inline-transform.c (inline_call): Likewise.
16749 * ipa-inline.c (report_inline_failed_reason): Likewise.
16750 (want_early_inline_function_p): Likewise.
16751 (edge_badness): Likewise.
16752 (update_edge_key): Likewise.
16753 (inline_small_functions): Likewise.
16754 * ipa-profile.c (ipa_profile): Likewise.
16755 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16756 (ipa_make_edge_direct_to_target): Likewise.
16757 (remove_described_reference): Likewise.
16758 (ipa_impossible_devirt_target): Likewise.
16759 (propagate_controlled_uses): Likewise.
16760 (ipa_print_node_params): Likewise.
16761 (ipcp_transform_function): Likewise.
16762 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16763 (propagate_pure_const): Likewise.
16764 * ipa-reference.c (generate_summary): Likewise.
16765 (read_write_all_from_decl): Likewise.
16766 (propagate): Likewise.
16767 (ipa_reference_read_optimization_summary): Likewise.
16768 * ipa-utils.c (ipa_merge_profiles): Likewise.
16769 * ipa.c (walk_polymorphic_call_targets): Likewise.
16770 (symbol_table::remove_unreachable_nodes): Likewise.
16771 (ipa_single_use): Likewise.
16772 * passes.c (execute_todo): Likewise.
16773 * predict.c (drop_profile): Likewise.
16774 * symtab.c (symtab_node::get_dump_name): New function.
16775 (symtab_node::dump_name): Likewise.
16776 (symtab_node::dump_asm_name): Likewise.
16777 (symtab_node::dump_references): Likewise.
16778 (symtab_node::dump_referring): Likewise.
16779 (symtab_node::dump_base): Likewise.
16780 (symtab_node::debug_symtab): Likewise.
16781 * tree-sra.c (convert_callers_for_node): Likewise.
16782 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16783 * value-prof.c (init_node_map): Likewise.
16785 2017-05-23 Martin Liska <mliska@suse.cz>
16787 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16788 and symtab_node::debug_symtab to symbol_table::debug.
16789 * cgraphunit.c (analyze_functions): Use the renamed function.
16790 (symbol_table::compile): Likewise.
16791 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16792 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16793 * passes.c (execute_todo): Likewise.
16794 * symtab.c (symbol_table::dump): New function.
16795 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16797 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16799 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16800 that nonconst implies exec.
16802 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16804 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16805 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16806 (inline_edge_summary_vec): Turn into ...
16807 (ipa_call_summaries): ... this one.
16808 (redirect_to_unreachable, edge_set_predicate,
16809 evaluate_properties_for_edge, inline_summary_alloc,
16810 reset_ipa_call_summary, reset_inline_summary,
16811 inline_summary_t::duplicate): Update.
16812 (inline_edge_duplication_hook): Turn to ...
16813 (ipa_call_summary_t::duplicate): ... this one.
16814 (inline_edge_removal_hook): Turn to ...
16815 (ipa_call_summary_t::remove): ... this one.
16816 (dump_inline_edge_summary): Turn to ...
16817 (dump_ipa_call_summary): ... this one.
16818 (estimate_function_body_sizes): Update.
16819 (inline_update_callee_summaries): Update.
16820 (remap_edge_change_prob): Update.
16821 (remap_edge_summaries): Update.
16822 (inline_merge_summary): Update.
16823 (do_estimate_edge_time): Update.
16824 (inline_generate_summary): Update.
16825 (inline_read_section): Update.
16826 (inline_read_summary): Update.
16827 (inline_free_summary): Update.
16828 * ipa-inline.c (can_inline_edge_p): Update.
16829 (compute_inlined_call_time): Update.
16830 (want_inline_small_function_p): Update.
16831 (edge_badness): Update.
16832 (early_inliner): Update.
16833 * ipa-inline.h (inline_edge_summary): Turn to ...
16834 (ipa_call_summary): ... this one.
16835 (ipa_call_summary_t): New class.
16836 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16837 (ipa_call_summaries): New.
16838 (inline_edge_summary): Remove.
16839 (estimate_edge_growth): Update.
16840 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16841 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16842 * ipa-split.c (execute_split_functions): Update.
16843 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16845 2017-05-23 Tom de Vries <tom@codesourcery.com>
16847 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16848 attributes): Document rdrand effective target.
16850 2017-05-23 Tom de Vries <tom@codesourcery.com>
16852 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16853 attributes): Sort alphabetically.
16855 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16857 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16859 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16862 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16863 V2DF/V2DI splat into two separate patterns, one that handles
16864 registers, and the other that only handles memory. Drop support
16865 for splatting from a GPR on ISA 2.07 and then splitting the
16866 splat into direct move and splat.
16867 (vsx_splat_<mode>_reg): Likewise.
16868 (vsx_splat_<mode>_mem): Likewise.
16870 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16872 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16874 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16876 PR middle-end/80809
16877 * omp-low.c (finish_taskreg_remap): New function.
16878 (finish_taskreg_scan): If unit size of ctx->record_type
16879 is non-constant, unshare the size expression and replace
16880 decls in it with possible outer var refs.
16882 PR middle-end/80809
16883 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16884 GOVD_SHARED rather than GOVD_PRIVATE with it.
16885 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16886 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16888 PR middle-end/80853
16889 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16890 as last argument to build_outer_var_ref for pointer bases of array
16891 section reductions.
16893 2017-05-19 Martin Sebor <msebor@redhat.com>
16895 * print-tree.c (print_node): Print DECL_READ_P flag.
16897 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16899 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16900 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16901 * cgraph.c: Likewise.
16902 * cgraphunit.c: Likewise.
16903 * gengtype.c: Likewise.
16904 * ipa-cp.c: Likewise.
16905 * ipa-devirt.c: Likewise.
16906 * ipa-icf.c: Likewise.
16907 * ipa-predicate.c: Likewise.
16908 * ipa-profile.c: Likewise.
16909 * ipa-prop.c: Likewise.
16910 * ipa-split.c: Likewise.
16912 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16913 edge_predicate_pool, dump_inline_hints,
16914 inline_summary::account_size_time, redirect_to_unreachable,
16915 edge_set_predicate, set_hint_predicate,
16916 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16917 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16918 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16919 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16920 ipa_call_summary_t::remove, initialize_growth_caches,
16921 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16922 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16923 mark_modified, unmodified_parm_1, unmodified_parm,
16924 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16925 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16926 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16927 will_be_nonconstant_predicate, record_modified_bb_info,
16928 get_minimal_bb, record_modified, param_change_prob,
16929 phi_result_unknown_predicate, predicate_for_phi_result,
16930 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16931 estimate_function_body_sizes, compute_inline_parameters,
16932 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16933 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16934 inline_update_callee_summaries, remap_edge_change_prob,
16935 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16936 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16937 inline_analyze_function, inline_summary_t::insert,
16938 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16939 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16940 inline_free_summary): Move to ipa-fnsummary.h
16941 (predicate_t): Remove.
16942 * ipa-fnsummary.c: New file.
16943 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16944 (enum inline_hints_vals, inline_hints, agg_position_info,
16945 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16946 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16947 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16948 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16949 inline_read_summary, inline_write_summary, inline_free_summary,
16950 inline_analyze_function, initialize_inline_failed,
16951 inline_merge_summary, inline_update_overall_summary,
16952 compute_inline_parameters): Move to ipa-fnsummary.h
16953 * ipa-fnsummary.h: New file.
16954 * ipa-inline-transform.h: Include ipa-inline.h.
16955 * ipa-inline.c: LIkewise.
16957 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16959 * ipa-inline.c (edge_badness): Use inlined_time instead of
16960 inline_summaries->get.
16962 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16964 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16966 2017-05-22 Nathan Sidwell <nathan@acm.org>
16968 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16969 (fdump-lang): Document 'raw' option.
16970 * dumpfile.h (TDI_tu): Delete.
16971 * dumpfile.c (dump_files): Remove translation-unit.
16972 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16974 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16976 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16977 command option from $(AWK) call.
16978 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16979 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16980 [FORMAT]: Remove handling of variable.
16981 * config/avr/t-multilib: Regenerate.
16983 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16985 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16987 (dump_inline_summary): Do not print self_time.
16988 (estimate_function_body_sizes): Do not set self_time.
16989 (compute_inline_parameters): Likewise.
16990 (inline_read_section, inline_write_summary): Do not stream self_time.
16991 * ipa-inline.h (inline_summary): Drop self_time.
16993 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16995 * ipa-inline-analysis.c (account_size_time): Rename to ...
16996 (inline_summary::account_size_time): ... this one.
16997 (reset_ipa_call_summary): Turn to ...
16998 (ipa_call_summary::reset): ... this one.
16999 (reset_inline_summary): Turn to ...
17000 (inline_summary::reset): ... this one.
17001 (inline_summary_t::remove): Update.
17002 (inline_summary_t::duplicate): Update.
17003 (ipa_call_summary_t::remove): Update.
17004 (dump_inline_summary): Update.
17005 (estimate_function_body_sizes): Update.
17006 (compute_inline_parameters): Update.
17007 (estimate_node_size_and_time): Update.
17008 (inline_merge_summary): Update.
17009 (inline_update_overall_summary): Update.
17010 (inline_read_section): Update.
17011 (inline_write_summary): Update.
17012 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17013 add account_size_time and reset member functions.
17014 (ipa_call_summary): Add reset function.
17015 * ipa-predicate.h (predicate::operator &): Constify.
17017 2017-05-22 Richard Biener <rguenther@suse.de>
17019 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17021 2017-05-19 Jason Merrill <jason@redhat.com>
17023 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17025 2017-05-19 Marek Polacek <polacek@redhat.com>
17028 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17029 TYPE_OVERFLOW_WRAPS checks.
17031 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17033 * tree-core.h (enum omp_clause_default_kind): Add
17034 "OMP_CLAUSE_DEFAULT_PRESENT".
17035 * tree-pretty-print.c (dump_omp_clause): Handle it.
17036 * gimplify.c (enum gimplify_omp_var_data): Add
17037 "GOVD_MAP_FORCE_PRESENT".
17038 (gimplify_adjust_omp_clauses_1): Map it to
17039 "GOMP_MAP_FORCE_PRESENT".
17040 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17042 * gimplify.c (oacc_default_clause): Clarify.
17044 2017-05-19 Nathan Sidwell <nathan@acm.org>
17046 LANG_HOOK_REGISTER_DUMPS
17047 * toplev.c (general_init): Call register dump lang hook.
17048 * doc/invoke.texi: Document -fdump-lang option family.
17049 * dumpfile.c (dump_files): Remove class dump here.
17050 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17051 * dumpfile.h (tree_dump_index): Remove TDI_class.
17052 * langhooks-def.h (lhd_register_dumps): Declare.
17053 (LANG_HOOKS_REGISTER_DUMPS): Define.
17054 (LANG_HOOKS_INITIALIZER): Add it.
17055 * langhooks.c (lhd_register_dumps): Define.
17056 * langhooks.h (struct lang_hooks): Add register_dumps.
17058 2017-05-19 Nathan Sidwell <nathan@acm.org>
17060 * context.h (context::set_passes): New.
17061 * context.c (context::context): Do not create pass manager.
17062 * toplev.c (general_init): Create pass manager here.
17064 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17066 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17067 use this splitter if two add or or instructions would also work for
17068 the constant we want to generate.
17070 2017-05-19 Richard Biener <rguenther@suse.de>
17073 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17074 predicate evaluation.
17076 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17078 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17080 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17083 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17085 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17086 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17087 (inline_edge_summary_vec): Turn into ...
17088 (ipa_call_summaries): ... this one.
17089 (redirect_to_unreachable, edge_set_predicate,
17090 evaluate_properties_for_edge, inline_summary_alloc,
17091 reset_ipa_call_summary, reset_inline_summary,
17092 inline_summary_t::duplicate): Update.
17093 (inline_edge_duplication_hook): Turn to ...
17094 (ipa_call_summary_t::duplicate): ... this one.
17095 (inline_edge_removal_hook): Turn to ...
17096 (ipa_call_summary_t::remove): ... this one.
17097 (dump_inline_edge_summary): Turn to ...
17098 (dump_ipa_call_summary): ... this one.
17099 (estimate_function_body_sizes): Update.
17100 (inline_update_callee_summaries): Update.
17101 (remap_edge_change_prob): Update.
17102 (remap_edge_summaries): Update.
17103 (inline_merge_summary): Update.
17104 (do_estimate_edge_time): Update.
17105 (inline_generate_summary): Update.
17106 (inline_read_section): Update.
17107 (inline_read_summary): Update.
17108 (inline_free_summary): Update.
17109 * ipa-inline.c (can_inline_edge_p): Update.
17110 (compute_inlined_call_time): Update.
17111 (want_inline_small_function_p): Update.
17112 (edge_badness): Update.
17113 (early_inliner): Update.
17114 * ipa-inline.h (inline_edge_summary): Turn to ...
17115 (ipa_call_summary): ... this one.
17116 (ipa_call_summary_t): New class.
17117 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17118 (ipa_call_summaries): New.
17119 (inline_edge_summary): Remove.
17120 (estimate_edge_growth): Update.
17121 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17122 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17123 * ipa-split.c (execute_split_functions): Update.
17124 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17126 2017-05-19 Richard Biener <rguenther@suse.de>
17128 PR middle-end/80764
17129 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17131 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17133 * config/rs6000/rs6000.c (struct machine_function): Add field
17134 fpr_is_wrapped_separately.
17135 (rs6000_get_separate_components): Use 64 components. Handle the
17136 new FPR components.
17137 (rs6000_components_for_bb): Handle the FPR components.
17138 (rs6000_emit_prologue_components): Handle the FPR components.
17139 (rs6000_emit_epilogue_components): Handle the FPR components.
17140 (rs6000_set_handled_components): Handle the FPR components.
17141 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17142 that are already separately shrink-wrapped.
17143 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17144 that are already separately shrink-wrapped.
17146 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17149 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17152 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17153 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17154 cases where the register allocator uses a move and an offsettable
17155 memory operation to/from a FPR register on ISA 2.06/2.07.
17156 (define_peephole2 for Altivec d-form store): Likewise.
17158 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17161 * config/i386/mmx.md (*mov<mode>_internal): Enable
17162 alternatives 11, 12, 13 and 14 also for 32bit targets.
17163 Remove alternatives 15, 16, 17 and 18.
17164 * config/i386/sse.md (vec_concatv2di): Change
17165 alternative (!x, *y) to (x, ?!*Yn).
17167 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17169 * dumpfile.h (enum dump_kind): Remove stray comma.
17171 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17173 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17174 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17175 predicate::num_conditions
17176 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17177 (CHANGED): turn into predicate::changed.
17178 (agg_position_info): Move to ipa-predicate.h
17179 (add_condition, predicate::add_clause, predicate::operator &=,
17180 predicate::or_with, predicate::evaluate, predicate::probability,
17181 dump_condition, dump_clause, predicate::dump,
17182 predicate::remap_after_duplication, predicate::remap_after_inlining,
17183 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17184 (evaluate_conditions_for_known_args): Update.
17185 (set_cond_stmt_execution_predicate): Update.
17186 * ipa-inline.h: Include ipa-predicate.h
17187 (condition, inline_param_summary, conditions, agg_position_info,
17188 predicate): Move to ipa-predicate.h
17189 * ipa-predicate.c: New file.
17190 * ipa-predicate.h: New file.
17192 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17194 * final.c (leaf_function_p): Check we are not in a sequence.
17196 2017-05-18 Martin Liska <mliska@suse.cz>
17198 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17199 * dumpfile.c (dump_register): Use new enum dump_kind.
17200 (get_dump_file_name): Likewise.
17201 (dump_enable_all): Likewise.
17202 (dump_switch_p_1): Likewise.
17203 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17204 * dumpfile.h (enum dump_kind): New enum type.
17205 (struct dump_file_info): Create constructor and
17206 format fields and comments.
17207 * passes.c (pass_manager::register_one_dump_file):
17209 * statistics.c (statistics_early_init): Likewise.
17210 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17211 TDF_TREE with TDF_SLIM.
17212 (gather_memory_references_ref): Likewise.
17214 2017-05-18 Martin Liska <mliska@suse.cz>
17216 * vec.h (struct vnull): Use it.
17218 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17220 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17221 (true_predicate, false_predicate, true_predicate_p,
17222 false_predicate_p): Remove.
17223 (single_cond_predicate, not_inlined_predicate): Turn to member function
17225 (add_condition): Update.
17226 (add_clause): Turn to...
17227 (predicate::add_clause): ... this one; update; allow passing NULL
17229 (and_predicates): Turn to ...
17230 (predicate::operator &=): ... this one.
17231 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17232 (or_predicates): Turn to ...
17233 (predicate::or_with): ... this one.
17234 (evaluate_predicate): Turn to ...
17235 (predicate::evaluate): ... this one.
17236 (predicate_probability): Turn to ...
17237 (predicate::probability): ... this one.
17238 (dump_condition): Update.
17239 (dump_predicate): Turn to ...
17240 (predicate::dump): ... this one.
17241 (account_size_time): Update.
17242 (edge_set_predicate): Update.
17243 (set_hint_predicate): UPdate.
17244 (evaluate_conditions_for_known_args): Update.
17245 (evaluate_properties_for_edge): Update.
17246 (remap_predicate_after_duplication): Turn to...
17247 (predicate::remap_after_duplication): ... this one.
17248 (remap_hint_predicate_after_duplication): Update.
17249 (inline_summary_t::duplicate): UPdate.
17250 (dump_inline_edge_summary): Update.
17251 (dump_inline_summary): Update.
17252 (set_cond_stmt_execution_predicate): Update.
17253 (set_switch_stmt_execution_predicate): Update.
17254 (compute_bb_predicates): Update.
17255 (will_be_nonconstant_expr_predicate): Update.
17256 (will_be_nonconstant_predicate): Update.
17257 (phi_result_unknown_predicate): Update.
17258 (predicate_for_phi_result): Update.
17259 (array_index_predicate): Update.
17260 (estimate_function_body_sizes): Update.
17261 (estimate_node_size_and_time): Update.
17262 (estimate_ipcp_clone_size_and_time): Update.
17263 (remap_predicate): Rename to ...
17264 (predicate::remap_after_inlining): ... this one.
17265 (remap_hint_predicate): Update.
17266 (inline_merge_summary): Update.
17267 (inline_update_overall_summary): Update.
17268 (estimate_size_after_inlining): Update.
17269 (read_predicate): Rename to ...
17270 (predicate::stream_in): ... this one.
17271 (read_inline_edge_summary): Update.
17272 (write_predicate): Rename to ...
17273 (predicate::stream_out): ... this one.
17274 (write_inline_edge_summary): Update.
17275 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17276 (clause_t): Turn to uint32_t
17277 (predicate): Turn to class; implement constructor and operators
17279 (size_time_entry): Update.
17280 (inline_summary): Update.
17281 (inline_edge_summary): Update.
17283 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17285 * fold-const.c (fold_binary_loc): Move transformation...
17286 * match.pd (C - X CMP X): ... here.
17288 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17290 * config/sparc/sparc.c (sparc_option_override): Set function
17291 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17292 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17294 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17296 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17298 2017-05-18 Marek Polacek <polacek@redhat.com>
17301 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17302 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17303 gimple_assign_load_p.
17305 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17307 PR middle-end/80692
17308 * real.c (do_compare): Give decimal_do_compare preference over
17309 comparing just the signs.
17311 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17313 * doc/md.texi (Canonicalization of Instructions): Describe the
17314 canonical form of instructions that inherently set a condition
17317 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17319 PR middle-end/80775
17320 * tree-cfg.c: Move deletion of unreachable case statements to after
17321 the merging of consecutive case labels.
17323 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17325 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17326 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17327 restoring of callee-saved registers.
17329 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17331 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17332 * config/visium/visium.c (single_set_and_flags): Likewise.
17333 * config/visium/visium.md (Substitutions): Likewise.
17335 2017-05-17 Martin Liska <mliska@suse.cz>
17337 * cfg.c: Introduce dump_flags_t type and
17338 use it instead of int type.
17340 * cfghooks.c: Likewise.
17341 * cfghooks.h (struct cfg_hooks): Likewise.
17342 * cfgrtl.c: Likewise.
17343 * cfgrtl.h: Likewise.
17344 * cgraph.c (cgraph_node::get_body): Likewise.
17345 * coretypes.h: Likewise.
17346 * domwalk.c: Likewise.
17347 * domwalk.h: Likewise.
17348 * dumpfile.c (struct dump_option_value_info): Likewise.
17349 (dump_enable_all): Likewise.
17350 (dump_switch_p_1): Likewise.
17351 (opt_info_switch_p): Likewise.
17352 * dumpfile.h (enum tree_dump_index): Likewise.
17353 (struct dump_file_info): Likewise.
17354 * genemit.c: Likewise.
17355 * generic-match-head.c: Likewise.
17356 * gengtype.c (open_base_files): Likewise.
17357 * gimple-pretty-print.c: Likewise.
17358 * gimple-pretty-print.h: Likewise.
17359 * graph.c (print_graph_cfg): Likewise.
17360 * graphite-scop-detection.c (dot_all_sese): Likewise.
17361 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17362 * loop-unroll.c (report_unroll): Likewise.
17363 * passes.c (pass_manager::register_one_dump_file): Likewise.
17364 * print-tree.c: Likewise.
17365 * statistics.c: Likewise.
17366 * tree-cfg.c: Likewise.
17367 * tree-cfg.h: Likewise.
17368 * tree-dfa.c: Likewise.
17369 * tree-dfa.h: Likewise.
17370 * tree-dump.c (dump_function): Likewise.
17371 * tree-dump.h (struct dump_info): Likewise.
17372 * tree-pretty-print.c: Likewise.
17373 * tree-pretty-print.h: Likewise.
17374 * tree-ssa-live.c: Likewise.
17375 * tree-ssa-live.h: Likewise.
17376 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17377 * tree-vect-loop.c: Likewise.
17378 * tree-vect-slp.c: Likewise.
17380 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17381 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17383 PR tree-optimization/80457
17384 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17385 of all arguments to a statement as scalar_to_vec operations.
17386 (vectorizable_call): Adjust call to vect_model_simple_cost for
17388 (vectorizable_conversion): Likewise.
17389 (vectorizable_assignment): Likewise.
17390 (vectorizable_shift): Likewise.
17391 (vectorizable_operation): Likewise.
17392 (vectorizable_comparison): Likewise.
17393 (vect_is_simple_cond): Record the def types for operands.
17394 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17395 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17396 for statement argument count.
17398 2017-05-16 Carl Love <cel@us.ibm.com>
17400 * config/rs6000/rs6000-c: Add support for built-in functions
17401 vector unsigned long long vec_bperm (vector unsigned long long,
17402 vector unsigned char)
17403 vector signed long long vec_mule (vector signed int,
17405 vector unsigned long long vec_mule (vector unsigned int,
17406 vector unsigned int)
17407 vector signed long long vec_mulo (vector signed int,
17409 vector unsigned long long vec_mulo (vector unsigned int,
17410 vector unsigned int)
17411 vector signed char vec_sldw (vector signed char,
17412 vector signed char,
17414 vector unsigned char vec_sldw (vector unsigned char,
17415 vector unsigned char,
17417 vector signed short vec_sldw (vector signed short,
17418 vector signed short,
17420 vector unsigned short vec_sldw (vector unsigned short,
17421 vector unsigned short,
17423 vector signed int vec_sldw (vector signed int,
17426 vector unsigned int vec_sldw (vector unsigned int,
17427 vector unsigned int,
17429 vector signed long long vec_sldw (vector signed long long,
17430 vector signed long long,
17432 vector unsigned long long vec_sldw (vector unsigned long long,
17433 vector unsigned long long,
17435 * config/rs6000/rs6000-c: Add support for built-in functions
17436 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17437 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17438 * doc/extend.texi: Update the built-in documentation for the
17439 new built-in functions.
17441 2017-05-16 Marek Polacek <polacek@redhat.com>
17445 * tree.c (save_expr): Don't fold the expression.
17447 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17449 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17450 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17451 and (?*y,m). Update insn attributes.
17453 2017-05-16 Martin Liska <mliska@suse.cz>
17455 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17456 flags argument of print_gimple_stmt, print_gimple_expr,
17457 print_generic_stmt and print_generic_expr.
17458 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17459 * coretypes.h: Likewise.
17460 * except.c (dump_eh_tree): Likewise.
17461 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17462 * gimple-pretty-print.h: Likewise.
17463 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17464 (backprop::push_to_worklist): Likewise.
17465 (backprop::pop_from_worklist): Likewise.
17466 (backprop::process_use): Likewise.
17467 (backprop::intersect_uses): Likewise.
17468 (note_replacement): Likewise.
17469 * gimple-ssa-store-merging.c
17470 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17471 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17472 (pass_store_merging::execute): Likewise.
17473 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17474 (ssa_base_cand_dump_callback): Likewise.
17475 (dump_incr_vec): Likewise.
17476 (replace_refs): Likewise.
17477 (replace_mult_candidate): Likewise.
17478 (create_add_on_incoming_edge): Likewise.
17479 (create_phi_basis): Likewise.
17480 (insert_initializers): Likewise.
17481 (all_phi_incrs_profitable): Likewise.
17482 (introduce_cast_before_cand): Likewise.
17483 (replace_one_candidate): Likewise.
17484 * gimplify.c (gimplify_expr): Likewise.
17485 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17486 (set_rename): Likewise.
17487 (rename_uses): Likewise.
17488 (copy_loop_phi_nodes): Likewise.
17489 (add_close_phis_to_merge_points): Likewise.
17490 (copy_loop_close_phi_args): Likewise.
17491 (copy_cond_phi_args): Likewise.
17492 (graphite_copy_stmts_from_block): Likewise.
17493 (translate_pending_phi_nodes): Likewise.
17494 * graphite-poly.c (print_pdr): Likewise.
17495 (dump_gbb_cases): Likewise.
17496 (dump_gbb_conditions): Likewise.
17497 (print_scop_params): Likewise.
17498 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17499 (build_cross_bb_scalars_use): Likewise.
17500 (gather_bbs::before_dom_children): Likewise.
17501 * hsa-dump.c (dump_hsa_immed): Likewise.
17502 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17503 (get_replacement_map): Likewise.
17504 * ipa-inline-analysis.c (dump_condition): Likewise.
17505 (estimate_function_body_sizes): Likewise.
17506 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17507 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17508 * ipa-prop.c (ipa_dump_param): Likewise.
17509 (ipa_print_node_jump_functions_for_edge): Likewise.
17510 (ipa_modify_call_arguments): Likewise.
17511 (ipa_modify_expr): Likewise.
17512 (ipa_dump_param_adjustments): Likewise.
17513 (ipa_dump_agg_replacement_values): Likewise.
17514 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17515 * ipa-pure-const.c (check_stmt): Likewise.
17516 (pass_nothrow::execute): Likewise.
17517 * ipa-split.c (execute_split_functions): Likewise.
17518 * omp-offload.c (dump_oacc_loop_part): Likewise.
17519 (dump_oacc_loop): Likewise.
17520 * trans-mem.c (tm_log_emit): Likewise.
17521 (tm_memopt_accumulate_memops): Likewise.
17522 (dump_tm_memopt_set): Likewise.
17523 (dump_tm_memopt_transform): Likewise.
17524 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17525 (print_loop): Likewise.
17526 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17527 (chkp_gather_checks_info): Likewise.
17528 (chkp_get_check_result): Likewise.
17529 (chkp_remove_check_if_pass): Likewise.
17530 (chkp_use_outer_bounds_if_possible): Likewise.
17531 (chkp_reduce_bounds_lifetime): Likewise.
17532 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17533 (chkp_mark_completed_bounds): Likewise.
17534 (chkp_register_incomplete_bounds): Likewise.
17535 (chkp_mark_invalid_bounds): Likewise.
17536 (chkp_maybe_copy_and_register_bounds): Likewise.
17537 (chkp_build_returned_bound): Likewise.
17538 (chkp_get_bound_for_parm): Likewise.
17539 (chkp_build_bndldx): Likewise.
17540 (chkp_get_bounds_by_definition): Likewise.
17541 (chkp_generate_extern_var_bounds): Likewise.
17542 (chkp_get_bounds_for_decl_addr): Likewise.
17543 * tree-chrec.c (chrec_apply): Likewise.
17544 * tree-data-ref.c (dump_data_reference): Likewise.
17545 (dump_subscript): Likewise.
17546 (dump_data_dependence_relation): Likewise.
17547 (analyze_overlapping_iterations): Likewise.
17548 * tree-inline.c (expand_call_inline): Likewise.
17549 (tree_function_versioning): Likewise.
17550 * tree-into-ssa.c (dump_defs_stack): Likewise.
17551 (dump_currdefs): Likewise.
17552 (dump_names_replaced_by): Likewise.
17553 (dump_update_ssa): Likewise.
17554 (update_ssa): Likewise.
17555 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17556 * tree-parloops.c (build_new_reduction): Likewise.
17557 (try_create_reduction_list): Likewise.
17558 (ref_conflicts_with_region): Likewise.
17559 (oacc_entry_exit_ok_1): Likewise.
17560 (oacc_entry_exit_single_gang): Likewise.
17561 * tree-pretty-print.h: Likewise.
17562 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17563 (get_scalar_evolution): Likewise.
17564 (add_to_evolution): Likewise.
17565 (get_loop_exit_condition): Likewise.
17566 (analyze_evolution_in_loop): Likewise.
17567 (analyze_initial_condition): Likewise.
17568 (analyze_scalar_evolution): Likewise.
17569 (instantiate_scev): Likewise.
17570 (number_of_latch_executions): Likewise.
17571 (gather_chrec_stats): Likewise.
17572 (final_value_replacement_loop): Likewise.
17573 (scev_const_prop): Likewise.
17574 * tree-sra.c (dump_access): Likewise.
17575 (disqualify_candidate): Likewise.
17576 (create_access): Likewise.
17577 (reject): Likewise.
17578 (maybe_add_sra_candidate): Likewise.
17579 (create_access_replacement): Likewise.
17580 (analyze_access_subtree): Likewise.
17581 (analyze_all_variable_accesses): Likewise.
17582 (sra_modify_assign): Likewise.
17583 (initialize_constant_pool_replacements): Likewise.
17584 (find_param_candidates): Likewise.
17585 (decide_one_param_reduction): Likewise.
17586 (replace_removed_params_ssa_names): Likewise.
17587 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17588 * tree-ssa-copy.c (dump_copy_of): Likewise.
17589 (copy_prop_visit_cond_stmt): Likewise.
17590 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17591 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17592 (record_equivalences_from_stmt): Likewise.
17593 * tree-ssa-dse.c (compute_trims): Likewise.
17594 (delete_dead_call): Likewise.
17595 (delete_dead_assignment): Likewise.
17596 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17597 (forward_propagate_into_cond): Likewise.
17598 (pass_forwprop::execute): Likewise.
17599 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17600 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17602 (move_computations_worker): Likewise.
17603 (execute_sm): Likewise.
17604 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17605 (remove_exits_and_undefined_stmts): Likewise.
17606 (remove_redundant_iv_tests): Likewise.
17607 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17608 (adjust_iv_update_pos): Likewise.
17609 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17610 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17611 (value_replacement): Likewise.
17612 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17613 * tree-ssa-pre.c (print_pre_expr): Likewise.
17614 (get_representative_for): Likewise.
17615 (create_expression_by_pieces): Likewise.
17616 (insert_into_preds_of_block): Likewise.
17617 (eliminate_insert): Likewise.
17618 (eliminate_dom_walker::before_dom_children): Likewise.
17619 (eliminate): Likewise.
17620 (remove_dead_inserted_code): Likewise.
17621 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17622 * tree-ssa-reassoc.c (get_rank): Likewise.
17623 (eliminate_duplicate_pair): Likewise.
17624 (eliminate_plus_minus_pair): Likewise.
17625 (eliminate_not_pairs): Likewise.
17626 (undistribute_ops_list): Likewise.
17627 (eliminate_redundant_comparison): Likewise.
17628 (update_range_test): Likewise.
17629 (optimize_range_tests_var_bound): Likewise.
17630 (optimize_vec_cond_expr): Likewise.
17631 (rewrite_expr_tree): Likewise.
17632 (rewrite_expr_tree_parallel): Likewise.
17633 (linearize_expr): Likewise.
17634 (break_up_subtract): Likewise.
17635 (linearize_expr_tree): Likewise.
17636 (attempt_builtin_powi): Likewise.
17637 (attempt_builtin_copysign): Likewise.
17638 (transform_stmt_to_copy): Likewise.
17639 (transform_stmt_to_multiply): Likewise.
17640 (dump_ops_vector): Likewise.
17641 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17642 (print_scc): Likewise.
17643 (set_ssa_val_to): Likewise.
17644 (visit_reference_op_store): Likewise.
17645 (visit_use): Likewise.
17646 (sccvn_dom_walker::before_dom_children): Likewise.
17647 (run_scc_vn): Likewise.
17648 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17650 (expr_hash_elt::print): Likewise.
17651 (const_and_copies::pop_to_marker): Likewise.
17652 (const_and_copies::record_const_or_copy_raw): Likewise.
17653 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17654 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17655 (dump_predicates): Likewise.
17656 (find_uninit_use): Likewise.
17657 (warn_uninitialized_phi): Likewise.
17658 (pass_late_warn_uninitialized::execute): Likewise.
17659 * tree-ssa.c (verify_vssa): Likewise.
17660 (verify_ssa): Likewise.
17661 (maybe_optimize_var): Likewise.
17662 * tree-vrp.c (dump_value_range): Likewise.
17663 (dump_all_value_ranges): Likewise.
17664 (dump_asserts_for): Likewise.
17665 (register_edge_assert_for_2): Likewise.
17666 (vrp_visit_cond_stmt): Likewise.
17667 (vrp_visit_switch_stmt): Likewise.
17668 (vrp_visit_stmt): Likewise.
17669 (vrp_visit_phi_node): Likewise.
17670 (simplify_cond_using_ranges_1): Likewise.
17671 (fold_predicate_in): Likewise.
17672 (evrp_dom_walker::before_dom_children): Likewise.
17673 (evrp_dom_walker::push_value_range): Likewise.
17674 (evrp_dom_walker::pop_value_range): Likewise.
17675 (execute_early_vrp): Likewise.
17677 2017-05-16 Richard Biener <rguenther@suse.de>
17679 * dwarf2out.c (loc_list_from_tree_1): Do not create
17680 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17682 2017-05-16 Richard Biener <rguenther@suse.de>
17684 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17686 (note_variable_value_in_expr): If we resolved the decl ref
17687 do not push to the stack.
17689 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17691 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17692 operations in fast-math mode.
17693 (vaddq_f16): Likewise.
17694 (vmul_f16): Likewise.
17695 (vmulq_f16): Likewise.
17696 (vsub_f16): Likewise.
17697 (vsubq_f16): Likewise.
17698 * config/arm/neon.md (add<mode>3): New.
17700 (fma:<VH:mode>3): New. Also remove outdated comment.
17703 2017-05-16 Martin Liska <mliska@suse.cz>
17707 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17708 (odr_types_equivalent_p): Likewise.
17710 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17712 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17714 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17717 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17718 non-interunit SSE move alternatives with '?'.
17719 (zero-extendsidi peephole2): New peephole to skip intermediate
17720 general register in SSE zero-extend sequence.
17722 2017-05-15 Jeff Law <law@redhat.com>
17724 * reorg.c (relax_delay_slots): Create a new variable to hold
17725 the temporary target rather than clobbering TARGET_LABEL.
17727 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17728 missing argument to extract_bit_field call.
17729 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17731 2017-05-15 Martin Liska <mliska@suse.cz>
17734 * gcc.c (process_command): Do not allow empty argument of -o option.
17736 2017-05-15 Renlin Li <renlin.li@arm.com>
17738 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17739 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17740 * config/aarch64/constraints.md (Usf): Add long call check.
17741 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17742 (call_value): Likewise.
17743 (sibcall): Likewise.
17744 (sibcall_value): Likewise.
17746 (call_value_insn): New.
17747 (sibcall_insn): Update rtx pattern.
17748 (sibcall_value_insn): Likewise.
17749 (call_internal): Remove.
17750 (call_value_internal): Likewise.
17751 (sibcall_internal): Likewise.
17752 (sibcall_value_internal): Likewise.
17753 (call_reg): Likewise.
17754 (call_symbol): Likewise.
17755 (call_value_reg): Likewise.
17756 (call_value_symbol): Likewise.
17758 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17761 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17763 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17765 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17766 compatible with CCGOCmode and with CCZmode.
17768 2017-05-14 Martin Sebor <msebor@redhat.com>
17770 PR middle-end/77671
17771 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17772 (gimple_fold_builtin_snprintf): Same.
17773 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17774 (gimple_fold_builtin_snprintf): Same.
17775 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17776 of character types.
17777 (is_call_safe): New function.
17778 (try_substitute_return_value): Call it.
17779 (try_simplify_call): New function.
17780 (pass_sprintf_length::handle_gimple_call): Call it.
17782 2017-05-14 Martin Sebor <msebor@redhat.com>
17784 PR middle-end/80669
17785 * builtins.c (expand_builtin_stpncpy): Simplify.
17787 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17789 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17790 * config/i386/i386.h
17791 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17792 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17793 (struct machine_function): Add new members call_ms2sysv,
17794 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17795 (struct machine_frame_state): New fields sp_realigned and
17796 sp_realigned_offset.
17797 * config/i386/i386.c
17798 (enum xlogue_stub): New enum.
17799 (enum xlogue_stub_sets): New enum.
17800 (class xlogue_layout): New class.
17801 (struct ix86_frame): New fields stack_realign_allocate_offset,
17802 stack_realign_offset and outlined_save_offset. Modify comments to
17803 detail stack layout when using out-of-line stubs.
17804 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17805 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17806 -mcall-ms2sysv-xlogues.
17807 (stub_managed_regs): New static variable.
17808 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17809 registers managed by out-of-line stub.
17810 (disable_call_ms2sysv_xlogues): New function.
17811 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17812 m->call_ms2sysv when appropriate and compute frame layout for
17814 (sp_valid_at, fp_valid_at): New inline functions.
17815 (choose_basereg): New function.
17816 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17818 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17819 Use align parameter of choose_baseaddr to generated aligned SSE movs
17821 (pro_epilogue_adjust_stack): Modify to track
17822 machine_frame_state::sp_realigned.
17823 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17824 (ix86_nsaved_sseregs): Likewise.
17825 (ix86_emit_save_regs): Likewise.
17826 (ix86_emit_save_regs_using_mov): Likewise.
17827 (ix86_emit_save_sse_regs_using_mov): Likewise.
17828 (get_scratch_register_on_entry): Likewise.
17829 (gen_frame_set): New function.
17830 (gen_frame_load): Likewise.
17831 (gen_frame_store): Likewise.
17832 (emit_outlined_ms2sysv_save): Likewise.
17833 (emit_outlined_ms2sysv_restore): Likewise.
17834 (ix86_expand_prologue): Modify stack re-alignment code and call
17835 emit_outlined_ms2sysv_save when appropriate.
17836 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17837 parameter rtx_insn *insn, which allows the function to be used to only
17838 generate the notes.
17839 (ix86_expand_epilogue): Modify validity checks of frame and stack
17840 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17841 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17842 * config/i386/predicates.md
17843 (save_multiple): New predicate.
17844 (restore_multiple): Likewise.
17845 * config/i386/sse.md
17846 (save_multiple<mode>): New pattern.
17847 (save_multiple_realign<mode>): Likewise.
17848 (restore_multiple<mode>): Likewise.
17849 (restore_multiple_and_return<mode>): Likewise.
17850 (restore_multiple_leave_return<mode>): Likewise.
17851 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17853 2017-05-14 Julia Koval <julia.koval@intel.com>
17855 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17856 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17857 (__builtin_ia32_xsetbv): New builtins.
17858 * config/i386/i386.c (ix86_expand_special_args_builtin):
17860 (ix86_expand_builtin): Special expand for new intrinsics.
17861 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17862 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17863 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17865 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17867 * cfganal.c (inverted_post_order_compute): Change argument type
17869 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17870 * df-core.c (rest_of_handle_df_initialize): Adjust.
17871 (rest_of_handle_df_finish): Likewise.
17872 (df_analyze_1): Likewise.
17873 (df_analyze): Likewise.
17874 (loop_inverted_post_order_compute): Change argument to be a vec *.
17875 (df_analyze_loop): Adjust.
17876 (df_get_n_blocks): Likewise.
17877 (df_get_postorder): Likewise.
17878 * df.h (struct df_d): Change field to be a vec.
17879 * lcm.c (compute_laterin): Adjust.
17880 (compute_available): Likewise.
17881 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17882 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17883 * tree-ssa-pre.c (compute_antic): Likewise.
17885 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17887 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17888 (depth_first_search::depth_first_search): Change structure init
17889 function to this constructor.
17890 (depth_first_search::add_bb): Rename function to this member.
17891 (depth_first_search::execute): Likewise.
17892 (flow_dfs_compute_reverse_finish): Adjust.
17894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17896 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17897 (longest_simple_path): Likewise.
17898 * shrink-wrap.c (spread_components): Likewise.
17899 (disqualify_problematic_components): Likewise.
17900 (emit_common_heads_for_components): Likewise.
17901 (emit_common_tails_for_components): Likewise.
17902 (insert_prologue_epilogue_for_components): Likewise.
17904 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17906 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17909 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17911 * df-core.c (df_set_blocks): Start using auto_bitmap.
17912 (df_compact_blocks): Likewise.
17913 * df-problems.c (df_rd_confluence_n): Likewise.
17914 * df-scan.c (df_insn_rescan_all): Likewise.
17915 (df_process_deferred_rescans): Likewise.
17916 (df_update_entry_block_defs): Likewise.
17917 (df_update_exit_block_uses): Likewise.
17918 (df_entry_block_bitmap_verify): Likewise.
17919 (df_exit_block_bitmap_verify): Likewise.
17920 (df_scan_verify): Likewise.
17921 * lra-constraints.c (lra_constraints): Likewise.
17922 (undo_optional_reloads): Likewise.
17923 (lra_undo_inheritance): Likewise.
17924 * lra-remat.c (calculate_gen_cands): Likewise.
17925 (do_remat): Likewise.
17926 * lra-spills.c (assign_spill_hard_regs): Likewise.
17927 (spill_pseudos): Likewise.
17928 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17929 (bitmap_set_subtract_values): Likewise.
17931 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17933 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17934 management with auto_bitmap.
17935 (fix_inter_tick): Likewise.
17936 (fix_recovery_deps): Likewise.
17937 * ira.c (add_store_equivs): Likewise.
17938 (find_moveable_pseudos): Likewise.
17939 (split_live_ranges_for_shrink_wrap): Likewise.
17940 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17941 (rtx_reuse_manager::seen_def_p): Likewise.
17942 (rtx_reuse_manager::set_seen_def): Likewise.
17943 * print-rtl.h (class rtx_reuse_manager): Likewise.
17945 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17947 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17949 (migrate_btr_def): Likewise.
17950 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17951 * df-core.c (loop_post_order_compute): Likewise.
17952 (loop_inverted_post_order_compute): Likewise.
17953 * hsa-common.h: Likewise.
17954 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17955 * init-regs.c (initialize_uninitialized_regs): Likewise.
17956 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17957 (inline_small_functions): Likewise.
17958 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17959 * ira.c (combine_and_move_insns): Likewise.
17960 (build_insn_chain): Likewise.
17961 * loop-invariant.c (find_invariants): Likewise.
17962 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17963 * predict.c (tree_predict_by_opcode): Likewise.
17964 (predict_paths_leading_to): Likewise.
17965 (predict_paths_leading_to_edge): Likewise.
17966 (estimate_loops_at_level): Likewise.
17967 (estimate_loops): Likewise.
17968 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17969 (spread_components): Likewise.
17970 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17971 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17972 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17973 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17974 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17975 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17976 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17977 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17978 (mark_threaded_blocks): Likewise.
17979 (thread_through_all_blocks): Likewise.
17980 * tree-ssa.c (verify_ssa): Likewise.
17981 (execute_update_addresses_taken): Likewise.
17982 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17986 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17988 (post_order_compute): Likewise.
17989 (inverted_post_order_compute): Likewise.
17990 (pre_and_rev_post_order_compute_fn): Likewise.
17992 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17994 * genrecog.c (int_set::int_set): Explicitly construct our
17995 auto_vec base class.
17996 * vec.h (auto_vec::auto_vec): New constructor.
17998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18000 * bitmap.h (class auto_bitmap): New constructor taking
18001 bitmap_obstack * argument.
18003 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18005 * bitmap.h (class auto_bitmap): Change type of m_bits to
18006 bitmap_head, and adjust ctor / dtor and member operators.
18008 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18010 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18011 when returned register mode doesn't match original mode.
18013 2017-05-12 Jeff Law <law@redhat.com>
18014 Jakub Jelinek <jakub@redhat.com>
18016 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18017 we look for cc setter after the compare-elim changes.
18018 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18019 within the vector to match what compare-elim now expects.
18020 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18021 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18023 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18024 after the compare-elim changes.
18025 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18026 the vector to match what compare-elim now expects.
18027 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18028 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18029 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18030 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18031 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18033 * config/visium/visium.c (single_set_and_flags): Fix where
18034 we look for cc setter after the compare-elim changes.
18035 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18036 with the vector to match what compare-elim now expects.
18037 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18038 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18039 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18040 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18041 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18042 (neg<mode>2_insn_set_overflow): Likewise.
18044 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18046 PR middle-end/79794
18047 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18048 maybe_expand_insn call, set ops[0].target. If still set after call,
18049 set alt_rtl. Add extra arg to recursive calls.
18050 (extract_bit_field): Add alt_rtl argument. Pass to
18052 * expmed.h (extract_bit_field): Fix prototype.
18053 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18054 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18055 to extract_bit_field_calls.
18056 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18057 Pass alt_rtl to extract_bit_field calls.
18058 * calls.c (store_unaligned_arguments_into_psuedos)
18059 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18060 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18062 * optabs.h (struct expand_operand): Add target bitfield.
18064 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18066 * compare-elim.c (try_eliminate_compare): Canonicalize
18067 operation with embedded compare to
18068 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18069 (set (reg) (operation)].
18071 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18073 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18076 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18077 cost of adding a carry flag for ADC instruction.
18078 [case MINUS]: Ignore the cost of subtracting a carry flag
18079 for SBB instruction.
18081 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18083 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18085 * config/rs6000/bmiintrin.h: New file.
18086 * config/rs6000/bmi2intrin.h: New file.
18087 * config/rs6000/x86intrin.h: New file.
18089 2017-05-12 Jeff Law <law@redhat.com>
18091 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18094 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18096 PR middle-end/80707
18097 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18099 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18102 early expansion of vector divide builtins.
18103 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18104 builtins identified as having unsigned arguments.
18106 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18108 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18109 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18110 expansion of vector logical operations (and, andc, or, xor,
18113 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18115 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18116 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18118 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18120 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18121 early GIMPLE expansion of vector multiplies.
18123 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18125 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18126 TARGET_HAVE_MOVT conditional.
18127 (movt splitter): Likewise.
18129 2017-05-12 Richard Biener <rguenther@suse.de>
18131 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18132 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18133 Fold all stmts not inplace.
18135 2017-05-12 Richard Biener <rguenther@suse.de>
18137 PR tree-optimization/80713
18138 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18139 inserted_exprs bit for not removed stmts.
18141 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18143 PR middle-end/69921
18144 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18145 parallelized" attribute for parallelized OpenACC kernels.
18146 * omp-offload.c (execute_oacc_device_lower): Use it.
18148 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18149 Set "oacc kernels" attribute.
18150 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18151 parameter. Adjust all users.
18152 (oacc_fn_attrib_kernels_p): Remove function.
18153 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18154 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18155 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18156 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18157 assert "oacc kernels" attribute is set.
18159 2017-05-11 Carl Love <cel@us.ibm.com>
18161 * config/rs6000/rs6000-c: Add support for built-in functions
18162 vector unsigned char vec_popcnt (vector signed char)
18163 vector unsigned char vec_popcnt (vector unsigned char)
18164 vector unsigned short vec_popcnt (vector signed short)
18165 vector unsigned short vec_popcnt (vector unsigned short)
18166 vector unsigned int vec_popcnt (vector signed int)
18167 vector unsigned int vec_popcnt (vector unsigned int)
18168 vector unsigned long long vec_popcnt (vector signed long long)
18169 vector unsigned long long vec_popcnt (vector unsigned long long)
18170 vector signed long long vec_slo (vector signed long long,
18171 vector signed char)
18172 vector signed long long vec_slo (vector signed long long,
18173 vector unsigned char)
18174 vector unsigned long long vec_slo (vector unsigned long long,
18175 vector signed char)
18176 vector unsigned long long vec_slo (vector unsigned long long,
18177 vector unsigned char)
18178 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18179 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18180 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18181 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18182 * doc/extend.texi: Update the built-in documentation file for the
18183 new built-in functions.
18185 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18187 * attribs.h (sorted_attr_string): Move machine independent
18188 functions for target clone support from the i386 port to common
18189 code. Rename ix86_function_versions to common_function_versions.
18190 Rename make_name to make_unique_name.
18191 (common_function_versions): Likewise.
18192 (make_unique_name): Likewise.
18193 (make_dispatcher_decl): Likewise.
18194 (is_function_default_version): Likewise.
18195 * attribs.c (attr_strcmp): Likewise.
18196 (sorted_attr_string): Likewise.
18197 (common_function_versions): Likewise.
18198 (make_unique_name): Likewise.
18199 (make_dispatcher_decl): Likewise.
18200 (is_function_default_version): Likewise.
18201 * config/i386/i386.c (attr_strcmp): Likewise.
18202 (sorted_attr_string): Likewise.
18203 (ix86_function_versions): Likewise.
18204 (make_name): Likewise.
18205 (make_dispatcher_decl): Likewise.
18206 (is_function_default_version): Likewise.
18207 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18209 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18212 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18213 Account for direct move costs for vec_construct of integer
18216 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18219 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18220 (UNSPEC_STX_ATOMIC): Ditto.
18221 (loaddi_via_sse): New insn.
18222 (storedi_via_sse): Ditto.
18223 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18224 Update corresponding peephole2 patterns.
18225 (atomic_storedi_fpu): Ditto.
18227 2017-05-11 Julia Koval <julia.koval@intel.com>
18229 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18230 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18232 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18233 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18234 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18236 2017-05-11 Nathan Sidwell <nathan@acm.org>
18238 * graphite-poly.c: Include dumpfile.h.
18240 * dumpfle.h (dump_function): Declare here ...
18241 * tree-dump.h (dump_function): ... not here.
18242 * dumpfile.c: #include tree-cfg.h.
18243 (dump_function): Move here from ...
18244 * tree-dump.c (dump_function): ... here.
18245 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18246 * graphite-poly.c: Don't include tree-dump.h.
18247 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18248 * print-tree.c: Likewise.
18249 * stor-layout.c: Likewise.
18250 * tree-nested.c: Likewise.
18252 * dumpfile.c (dump_start): Use TDF_FLAGS.
18253 (dump_enable_all): Fix TDF_KIND check thinko.
18255 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18257 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18258 array entries to represent two legal parameterizations of the
18259 overloaded __builtin_cmpb function, as represented by the
18260 P6_OV_BUILTIN_CMPB constant.
18261 (altivec_resolve_overloaded_builtin): Add special case handling
18262 for the __builtin_cmpb function, as represented by the
18263 P6_OV_BUILTIN_CMPB constant.
18264 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18265 (BU_P6_64BIT_2): New macro.
18266 (BU_P6_OVERLOAD_2): New macro
18267 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18268 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18269 (CMPB): Add overload support to represent both 32-bit and 64-bit
18270 compare-bytes function.
18271 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18272 support for TARGET_CMPB.
18273 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18274 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18275 documentation of the __builtin_cmpb overloaded built-in function.
18277 2017-05-11 Richard Biener <rguenther@suse.de>
18279 PR tree-optimization/80705
18280 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18281 bases are not vectorizable.
18283 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18285 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18286 when counting register pressure.
18288 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18290 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18291 (iv_ca_more_deps): Renamed to ...
18292 (iv_ca_compare_deps): ... this.
18293 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18297 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18299 (determine_group_iv_costs): ... here.
18300 (find_inv_vars_cb): Record inv var if it's not recorded before.
18302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18304 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18305 (get_shiftadd_cost): Ditto.
18307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18309 * tree-ssa-address.c: Include header file.
18310 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18312 (add_to_parts): Refactor.
18313 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18314 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18319 PR tree-optimization/53090
18320 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18322 (extract_cond_operands): Detect condition with IV on both sides
18323 and return COMP_IV_EXPR_2.
18324 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18325 (rewrite_use_compare): Simplify by removing call to function
18326 extract_cond_operands.
18328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18330 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18331 (extract_cond_operands): Detect condition comparing against non-
18332 invariant bound and return appropriate enum value.
18333 (find_interesting_uses_cond): Update use of extract_cond_operands.
18334 Handle its return value accordingly.
18335 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18337 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18339 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18340 nonlinear iv_use computation in loop invariant sensitive way.
18342 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18344 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18345 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18347 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18349 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18350 (dump_cand): Support iv_cand.inv_exprs.
18351 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18353 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18356 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18358 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18360 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18361 as local function. Include necessary header files.
18362 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18364 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18366 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18368 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18370 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18371 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18372 RSHIFT_EXPR and BIT_NOT_EXPR.
18374 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18376 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18377 (adjust_setup_cost): New parameter supporting round up adjustment.
18378 (struct address_cost_data): Delete.
18379 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18380 (split_address_cost, ptr_difference_cost): Delete.
18381 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18382 (struct ainc_cost_data): New struct.
18383 (get_address_cost_ainc): New function.
18384 (get_address_cost, get_computation_cost): Reimplement.
18385 (determine_group_iv_cost_address): Record inv_expr for all uses of
18387 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18388 (iv_ca_has_deps): Reimplemented to ...
18389 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18391 (iv_ca_extend): Call iv_ca_more_deps.
18393 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18395 * tree-ssa-address.c (struct mem_address): Move to header file.
18396 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18397 * tree-ssa-address.h (struct mem_address): Move from C file.
18398 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18402 * tree-affine.h (aff_combination_type): New interface.
18403 (aff_combination_zero_p): Remove static.
18404 (aff_combination_const_p): New interface.
18405 (aff_combination_singleton_var_p): New interfaces.
18407 2017-05-11 Richard Biener <rguenther@suse.de>
18409 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18410 Skip unreachable blocks and destinations.
18411 (eliminate): Move stmt removal and fixup ...
18412 (fini_eliminate): ... here. Skip inserted exprs.
18413 (pass_pre::execute): Move fini_pre after fini_eliminate.
18414 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18415 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18416 PRE to get rid of dead code that has invalid SSA form and
18417 split critical edges again.
18419 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18421 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18423 2017-05-11 Richard Biener <rguenther@suse.de>
18425 * passes.c (execute_function_todo): Verify loops if they are
18426 said to be up-to-date.
18427 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18428 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18430 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18433 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18434 handle calling assemble_external ourself.
18437 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18438 modes with zero size. Enhance comment.
18440 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18442 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18443 built-ins for vec_xl and vec_xst with short and char pointer
18446 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18448 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18449 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18450 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18451 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18452 (_mm_maskz_min_round_ss): New intrinsics.
18453 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18454 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18455 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18456 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18457 (__builtin_ia32_minss_mask_round): New builtins.
18458 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18459 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18460 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18462 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18463 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18465 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18468 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18470 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18471 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18472 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18473 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18474 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18475 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18476 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18477 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18478 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18479 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18480 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18481 (__builtin_ia32_mulss_mask_round): New builtins.
18482 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18483 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18484 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18486 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18487 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18489 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18492 2017-05-10 Julia Koval <julia.koval@intel.com>
18494 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18495 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18496 (_mm256_setr_m128i): New intrinsics.
18498 2017-05-10 Julia Koval <julia.koval@intel.com>
18500 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18501 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18502 (_mm_maskz_rcp14_ss): New intrinsics.
18503 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18504 (__builtin_ia32_rcp14ss_mask): New builtins.
18505 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18507 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18509 PR tree-optimization/51513
18510 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18511 (assert_unreachable_fallthru_edge_p): Use it.
18512 (group_case_labels_stmt): Likewise.
18513 * tree-cfg.h: Prototype it.
18514 * stmt.c: Include cfghooks.h and tree-cfg.h.
18515 (emit_case_dispatch_table) <gap_label>: New local variable.
18516 Use it to fill dispatch table gaps.
18517 Test for default_label before updating probabilities.
18518 (expand_case) <default_label>: Remove unneeded initialization.
18519 Test for unreachable default case statement and remove its edge.
18520 Set default_label accordingly.
18521 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18523 2017-05-10 Carl Love <cel@us.ibm.com>
18525 * config/rs6000/rs6000-c: Add support for built-in functions
18526 vector signed char vec_neg (vector signed char)
18527 vector signed short int vec_neg (vector short int)
18528 vector signed int vec_neg (vector signed int)
18529 vector signed long long vec_neg (vector signed long long)
18530 vector float vec_neg (vector float)
18531 vector double vec_neg (vector double)
18532 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18534 * config/rs6000/altivec.h: Add define for vec_neg
18535 * doc/extend.texi: Update the built-in documentation for the
18536 new built-in functions.
18538 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18540 PR tree-optimization/77644
18541 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18543 2017-05-10 Nathan Sidwell <nathan@acm.org>
18545 * dumpfile.h (TDI_lang_all): New.
18546 (TDF_KIND): New. Renumber others
18547 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18549 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18551 (get_dump_file_name): Adjust suffix generation.
18552 (dump_enable_all): Use TDF_KIND.
18553 * doc/invoke.texi (-fdump-lang-all): Document.
18555 * dumpfile.h: Tabify.
18557 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18560 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18561 Move member access before delete.
18563 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18565 * tree-inline.c (expand_call_inline): Split block at stmt
18568 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18571 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18572 are now unused after splitting mov{sf,sd}_hardfloat.
18573 (f32_lr2): Likewise.
18574 (f32_lm): Likewise.
18575 (f32_lm2): Likewise.
18576 (f32_li): Likewise.
18577 (f32_li2): Likewise.
18578 (f32_lv): Likewise.
18579 (f32_sr): Likewise.
18580 (f32_sr2): Likewise.
18581 (f32_sm): Likewise.
18582 (f32_sm2): Likewise.
18583 (f32_si): Likewise.
18584 (f32_si2): Likewise.
18585 (f32_sv): Likewise.
18586 (f32_dm): Likewise.
18587 (f32_vsx): Likewise.
18588 (f32_av): Likewise.
18589 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18590 For movsf, order stores so the VSX stores occur before the GPR
18591 store which encourages the register allocator to use a traditional
18592 FPR instead of a GPR. For movsd, order the stores so that the GPR
18593 store comes before the VSX stores to allow the power6 to work.
18594 This is due to the power6 not having a 32-bit integer store
18595 instruction from a FPR.
18596 (movsf_hardfloat): Likewise.
18597 (movsd_hardfloat): Likewise.
18599 2017-05-09 Martin Sebor <msebor@redhat.com>
18601 PR translation/80280
18602 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18605 PR translation/80280
18606 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18607 data member added in r247778.
18608 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18610 2017-05-09 Nathan Sidwell <nathan@acm.org>
18612 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18614 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18617 2017-05-09 Marek Polacek <polacek@redhat.com>
18619 * doc/invoke.texi: Fix typo.
18621 2017-05-09 Richard Biener <rguenther@suse.de>
18623 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18624 (vrp_val_is_min): Likewise.
18625 (set_value_range_to_value): Likewise.
18626 (set_value_range_to_nonnegative): Likewise.
18627 (gimple_assign_nonzero_p): Likewise.
18628 (gimple_stmt_nonzero_p): Likewise.
18629 (vrp_int_const_binop): Likewise. Remove unreachable case.
18630 (adjust_range_with_scev): Adjust comments.
18631 (compare_range_with_value): Likewise.
18632 (extract_range_from_phi_node): Likewise.
18633 (test_for_singularity): Likewise.
18635 2017-05-09 Richard Biener <rguenther@suse.de>
18637 * tree-vrp.c (get_single_symbol): Add assert that we don't
18638 get overflowed constants as invariant part.
18639 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18640 checks. Use wi::cmp instead of recursing for integer constants.
18641 (compare_values): Just ignore whether we assumed undefined
18642 overflow instead of failing the compare.
18643 (extract_range_for_var_from_comparison_expr): Add comment before the
18644 TREE_NO_WARNING sets.
18645 (test_for_singularity): Likewise.
18646 (extract_range_from_comparison): Do not disable optimization
18647 when we assumed undefined overflow.
18648 (extract_range_basic): Remove init of unused var.
18650 2017-05-09 Richard Biener <rguenther@suse.de>
18652 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18653 (extract_range_from_multiplicative_op_1): Adjust.
18654 (extract_range_from_binary_expr_1): Use int_const_binop.
18656 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18659 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18660 rs6000_store_data_bypass_p in seven define_bypass directives and
18661 in several comments.
18662 * config/rs6000/rs6000-protos.h: Add prototype for
18663 rs6000_store_data_bypass_p function.
18664 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18665 function implements slightly different (rs6000-specific) semantics
18666 than store_data_bypass_p, returning false rather than aborting
18667 with assertion error when arguments do not satisfy the
18668 requirements of store data bypass.
18669 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18670 rs6000_store_data_bypass_p.
18672 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18674 * config/xtensa/xtensa-protos.h
18675 (xtensa_initial_elimination_offset): New declaration.
18676 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18677 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18678 macro definition, add case for FRAME_POINTER_REGNUM when
18679 FRAME_GROWS_DOWNWARD.
18680 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18681 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18682 xtensa_initial_elimination_offset.
18684 2017-05-08 Nathan Sidwell <nathan@acm.org>
18686 * doc/invoke.texi: Alphabetize -fdump options.
18688 2017-05-08 Martin Sebor <msebor@redhat.com>
18690 PR translation/80280
18691 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18693 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18695 * target.def (compute_frame_layout): New optional target hook.
18696 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18697 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18698 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18700 * reload1.c (verify_initial_elim_offsets): Likewise.
18701 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18702 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18703 (arm_get_frame_offsets): Split up into this ...
18704 (arm_compute_frame_layout): ... and this function.
18706 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18708 * config/aarch64/constraints.md (Usa): New constraint.
18709 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18711 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18713 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18714 with_multilib_list after it has been checked.
18716 2017-05-08 Richard Biener <rguenther@suse.de>
18718 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18719 (bitmap_set_subtract_values): Likewise.
18721 2017-05-08 Richard Biener <rguenther@suse.de>
18723 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18724 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18726 (gimple_stmt_nonzero_warnv_p): Rename to ...
18727 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18729 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18730 (extract_range_basic): Adjust, do not disable propagation on
18731 strict overflow sensitive simplification.
18732 (vrp_visit_cond_stmt): Likewise.
18734 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18736 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18737 body size unconditionally.
18739 2017-05-07 Jeff Law <law@redhat.com>
18742 2017-05-06 Jeff Law <law@redhat.com>
18743 PR tree-optimization/78496
18744 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18747 PR tree-optimization/78496
18748 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18749 (simplify_stmt_using_ranges): Call it.
18750 (vrp_dom_walker::before_dom_children): Extract equivalences
18751 from an ASSERT_EXPR with an equality comparison against a
18754 2017-05-06 Jeff Law <law@redhat.com>
18756 PR tree-optimization/78496
18757 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18760 PR tree-optimization/78496
18761 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18762 (simplify_stmt_using_ranges): Call it.
18763 (vrp_dom_walker::before_dom_children): Extract equivalences
18764 from an ASSERT_EXPR with an equality comparison against a
18767 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18769 * lra-constraints.c (lra_copy_reg_equiv): New function.
18770 (split_reg): Use it to copy equivalence information from the
18771 original register to the spill register.
18773 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18775 PR rtl-optimization/75964
18776 * simplify-rtx.c (simplify_const_relational_operation): Remove
18777 invalid handling of comparisons of integer ABS.
18779 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18781 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18782 initialize to zero.
18783 (init_regs): Remove declaration.
18784 (function_arg_advance_32): Initialize error_p as boolean variable.
18786 2017-05-05 Nathan Sidwell <nathan@acm.org>
18788 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18789 lines. Use for (;;).
18791 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18793 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18794 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18795 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18796 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18797 VF=2 that require versioning.
18799 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18801 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18804 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18806 * diagnostic.h (diagnostic_override_option_index): Convert from
18807 macro to inline function.
18809 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18811 * diagnostic.c (last_module_changed_p): New function.
18812 (set_last_module): New function.
18813 (diagnostic_report_current_module): Convert macro usage to
18814 the above functions.
18815 * diagnostic.h (diagnostic_context::last_module): Strengthen
18816 from const line_map * to const line_map_ordinary *.
18817 (diagnostic_last_module_changed): Delete macro.
18818 (diagnostic_set_last_module): Delete macro.
18820 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18822 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18823 with diagnostic_report_diagnostic.
18824 (diagnostic_n_impl_richloc): Likewise.
18825 * diagnostic.h (report_diagnostic): Delete macro.
18826 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18827 with diagnostic_report_diagnostic.
18828 * substring-locations.c (format_warning_va): Likewise.
18830 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18832 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18833 save/restor of format_spec. Move option-printing code to...
18834 (print_option_information): ...this new function, and
18835 reimplement by simply printing to the pretty_printer,
18836 rather than appending to the format string.
18838 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18840 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18841 handling logic into...
18842 (update_effective_level_from_pragmas): ...this new function.
18844 2017-05-04 Andrew Waterman <andrew@sifive.com>
18846 * config/riscv/riscv.opt (mstrict-align): New option.
18847 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18848 (SLOW_UNALIGNED_ACCESS): Define.
18849 (riscv_slow_unaligned_access): Declare.
18850 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18852 (riscv_slow_unaligned_access): New variable.
18853 (rocket_tune_info): Set slow_unaligned_access to true.
18854 (optimize_size_tune_info): Set slow_unaligned_access to false.
18855 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18856 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18857 (riscv_option_override): Set riscv_slow_unaligned_access.
18858 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18860 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18862 * config/riscv/riscv.md: Unify indentation.
18864 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18869 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18871 (extendsi<mode>2): Add support for doing sign extension via
18872 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18873 don't have ISA 3.0 instructions.
18874 (extendsi<mode>2 splitter): Likewise.
18875 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18876 generate the normal insns since SImode can now go in vector
18877 registers. Disallow the special UNSPECs needed for previous
18878 machines to hide SImode being used. Add new insns
18879 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18880 (fix_trunc<mode>si2_stfiwx): Likewise.
18881 (fix_trunc<mode>si2_internal): Likewise.
18882 (fixuns_trunc<mode>si2): Likewise.
18883 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18884 (fctiw<u>z_<mode>_smallint): Likewise.
18885 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18886 of floating point to 32-bit integer from doing a direct move to
18887 the GPR registers to do a store.
18888 (fctiwz_<mode>): Break long line.
18890 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18892 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18893 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18894 (addr_list, addr_offset_valid_p): New.
18895 (split_address_groups): Check offset validity with above function.
18896 (gt-tree-ssa-loop-ivopts.h): Include header file.
18898 2017-05-05 Nathan Sidwell <nathan@acm.org>
18900 * config.gcc (arm*-*-*): Add missing 'fi'.
18902 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18904 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18905 included in -fopt-info does not matter.
18906 * doc/optinfo.texi (-fopt-info): Fix description of default
18907 behavour. Explicitly say order of options included in -fopt-info
18910 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18912 * config.gcc: Allow combinations of aprofile and rmprofile values for
18913 --with-multilib-list.
18914 * config/arm/t-multilib: New file.
18915 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18916 variables. Remove setting of ISA and floating-point ABI in
18917 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18918 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18919 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18920 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18922 * config/arm/t-rmprofile: Likewise except for the matches changes.
18923 * doc/install.texi (--with-multilib-list): Document the combination of
18924 aprofile and rmprofile values and warn about pitfalls in doing that.
18926 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18928 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18929 (movdi_aarch64): Likewise.
18931 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18933 PR tree-optimization/80632
18934 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18936 (build_arrays): Initialize it for virtual phis.
18937 (fix_phi_nodes): Use it for virtual phis.
18939 PR tree-optimization/80558
18940 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18941 [x, y] op z into [x op, y op z] for op & or | if conditions
18944 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18945 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18948 * config/arm/arm.md (use_literal_pool): Remove.
18949 (64-bit immediate split): No longer takes cost into consideration
18950 if arm_disable_literal_pool is enabled.
18951 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18952 used when arm_disable_literal_pool is enabled.
18953 (arm_max_const_double_inline_cost): Remove use of
18954 arm_disable_literal_pool.
18955 (push_minipool_fix): Add assert.
18956 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18957 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18958 (no_literal_pool_sf_immediate): New.
18960 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18962 PR tree-optimization/80613
18963 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18964 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18966 2017-05-05 Richard Biener <rguenther@suse.de>
18968 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18970 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18972 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18973 of this flag from insn conditions due to removal from r247495.
18975 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18977 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18979 (arm_early_store_addr_dep_ptr): Likewise.
18980 * config/arm/aarch-common-protos.h
18981 (arm_early_load_addr_dep_ptr): Add prototype.
18982 (arm_early_store_addr_dep_ptr): Likewise.
18983 * config/arm/cortex-a53.md: Add new bypasses.
18985 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18987 * tree.c (next_type_uid): Change type to unsigned.
18988 (type_hash_canon): Decrement back next_type_uid if
18989 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18990 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18993 2017-05-04 Martin Sebor <msebor@redhat.com>
18995 * builtins.c: Fix a trivial typo in a comment.
18997 PR middle-end/79234
18998 * builtins.c (check_sizes): Adjust to handle reading past the end.
18999 Avoid printing excessive upper bound of ranges. Use %E to print
19000 tree nodes instead of converting them to %wu.
19001 (expand_builtin_memchr): New function.
19002 (compute_dest_size): Rename...
19003 (compute_objsize): ...to this.
19004 (expand_builtin_memcpy): Adjust.
19005 (expand_builtin_mempcpy): Adjust.
19006 (expand_builtin_strcat): Adjust.
19007 (expand_builtin_strcpy): Adjust.
19008 (check_strncat_sizes): Adjust.
19009 (expand_builtin_strncat): Adjust.
19010 (expand_builtin_strncpy): Adjust and simplify.
19011 (expand_builtin_memset): Adjust.
19012 (expand_builtin_bzero): Adjust.
19013 (expand_builtin_memcmp): Adjust.
19014 (expand_builtin): Handle memcmp.
19015 (maybe_emit_chk_warning): Check strncat just once.
19017 2017-05-04 Martin Sebor <msebor@redhat.com>
19019 PR preprocessor/79214
19020 PR middle-end/79222
19021 PR middle-end/79223
19022 * builtins.c (check_sizes): Add inlining context and issue
19023 warnings even when -Wno-system-headers is set.
19024 (check_strncat_sizes): Same.
19025 (expand_builtin_strncat): Same.
19026 (expand_builtin_memmove): New function.
19027 (expand_builtin_stpncpy): Same.
19028 (expand_builtin): Handle memmove and stpncpy.
19030 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19032 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19033 which is not used any more.
19035 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19037 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19039 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19041 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19042 (cortexa53_tunings): Likewise.
19043 (cortexa57_tunings): Likewise.
19044 (cortexa72_tunings): Likewise.
19045 (cortexa73_tunings): Likewise.
19047 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19049 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19050 Set loop alignment to 8.
19052 2017-05-04 Martin Sebor <msebor@redhat.com>
19054 PR translation/80280
19055 * builtins.c (expand_builtin_object_size): Add missing quoting to
19056 %D and like directives.
19057 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19058 (hsa_type_for_tree_type): Same.
19059 (verify_function_arguments): Same.
19060 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19061 * varasm.c (get_section): Same.
19064 2017-05-04 Martin Sebor <msebor@redhat.com>
19066 PR translation/80280
19067 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19069 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19071 * config/aarch64/aarch64.c (generic_addrcost_table):
19072 Change HI/TI mode setting.
19074 2017-05-04 Martin Jambor <mjambor@suse.cz>
19076 PR tree-optimization/80622
19077 * tree-sra.c (comes_initialized_p): New function.
19078 (build_accesses_from_assign): Only set write lazily when
19079 comes_initialized_p is false.
19080 (analyze_access_subtree): Use comes_initialized_p.
19081 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19082 instead of testing for PARM_DECL.
19084 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19086 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19087 constraint on operand 0 to allow more general addressing modes.
19088 Adjust output template.
19089 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19091 * config/aarch64/aarch64-protos.h
19092 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19093 * config/aarch64/constraints.md (Dp): New address constraint.
19094 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19097 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19099 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19100 update use of estimate_ipcp_clone_size_and_time.
19101 (estimate_local_effects): Update use of
19102 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19103 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19104 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19105 Return nonspecialized time.
19107 2017-05-04 Richard Biener <rguenther@suse.de>
19109 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19110 for the last VUSE which def dominates the PHI. Directly call
19112 (get_continuation_for_phi_1): Remove.
19114 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19116 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19117 to explain the use of truncating division. Cap the number of
19118 iterations to the maximum given by nb_iterations_upper_bound,
19121 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19123 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19124 * configure: Regenerate.
19125 * config.in: Regenerate.
19126 * config/i386/driver-mingw32.c: new file.
19127 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19128 * config.host: Link driver-mingw32.o on MinGW host.
19129 * doc/install.texi: Document new --enable-mingw-wildcard configure
19132 2017-05-04 Marek Polacek <polacek@redhat.com>
19134 PR tree-optimization/80612
19135 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19137 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19138 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19140 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19141 (movt splitter): Likewise.
19142 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19143 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19144 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19145 block for Thumb-1 with MOVT.
19146 (thumb2_legitimate_address_p): Move code block ...
19147 (can_avoid_literal_pool_for_label_p): ... into this new function.
19148 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19150 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19151 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19152 "M-profile targets with the MOVT instruction".
19154 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19156 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19157 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19158 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19160 2017-05-04 Martin Liska <mliska@suse.cz>
19162 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19163 variable cond_code.
19165 2017-05-04 Richard Biener <rguenther@suse.de>
19167 * tree.c (array_at_struct_end_p): Handle arrays at struct
19168 end with flexarrays more conservatively. Refactor and treat
19169 arrays of arrays or aggregates more strict. Fix
19170 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19171 * tree.c (array_at_struct_end_p): Adjust prototype.
19172 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19173 * gimple-fold.c (get_range_strlen): Likewise.
19174 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19176 2017-05-04 Richard Biener <rguenther@suse.de>
19178 PR tree-optimization/31130
19179 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19181 (supports_overflow_infinity): Likewise.
19182 (is_negative_overflow_infinity): Likewise.
19183 (is_positive_overflow_infinity): Likewise.
19184 (is_overflow_infinity): Likewise.
19185 (stmt_overflow_infinity): Likewise.
19186 (overflow_infinity_range_p): Likewise.
19187 (usable_range_p): Remove as always returning true.
19188 (make_overflow_infinity): Remove.
19189 (negative_overflow_infinity): Likewise.
19190 (positive_overflow_infinity): Likewise.
19191 (avoid_overflow_infinity): Likewise.
19192 (set_value_range): Adjust accordingly.
19193 (set_value_range_to_nonnegative): Likewise, remove now unused
19194 overflow_infinity arg.
19195 (vrp_operand_equal_p): Adjust.
19196 (update_value_range): Likewise.
19197 (range_int_cst_singleton_p): Likewise.
19198 (operand_less_p): Likewise.
19199 (compare_values_warnv): Likewise.
19200 (extract_range_for_var_from_comparison_expr): Likewise.
19201 (vrp_int_const_binop): Likewise.
19202 (zero_nonzero_bits_from_vr): Likewise.
19203 (extract_range_from_multiplicative_op_1): Likewise.
19204 (extract_range_from_binary_expr_1): Likewise.
19205 (extract_range_from_unary_expr): Likewise.
19206 (extract_range_from_comparison): Likewise.
19207 (extract_range_basic): Likewise.
19208 (adjust_range_with_scev): Likewise.
19209 (compare_ranges): Likewise.
19210 (compare_range_with_value): Likewise.
19211 (dump_value_range): Likewise.
19212 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19214 (simplify_cond_using_ranges): Adjust.
19216 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19218 * brig-builtins.def: Added a builtin for class_f64.
19219 * builtin-types.def: Added a builtin type needed by class_f64.
19221 2017-05-03 Jason Merrill <jason@redhat.com>
19223 * timevar.def: Add TV_CONSTEXPR.
19225 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19227 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19229 2017-05-03 Martin Jambor <mjambor@suse.cz>
19231 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19232 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19233 * ipa-cp.c (ipcp_propagate_stage): Do not call
19234 ipa_update_after_lto_read.
19235 * ipa-inline.c (ipa_inline): Likewise.
19237 2017-05-03 Martin Jambor <mjambor@suse.cz>
19239 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19240 tag. Added a default constructor and a destructor.
19241 (ipa_edge_args_sum_t): New class;
19242 (ipa_edge_args_sum): Declare.
19243 (ipa_edge_args_vector): Remove declaration.
19244 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19245 (ipa_free_edge_args_substructures): Remove declaration.
19246 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19247 (ipa_edge_args_info_available_for_edge_p): Likewise.
19248 * ipa-prop.c (ipa_edge_args_vector): Removed.
19249 (edge_removal_hook_holder): Likewise.
19250 (edge_duplication_hook_holder): Likewise.
19251 (ipa_edge_args_sum): New variable.
19252 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19253 ipa_edge_args_vector.
19254 (ipa_free_edge_args_substructures): Likewise.
19255 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19256 ipa_edge_args_vector.
19257 (ipa_edge_removal_hook): Turned into method
19258 ipa_edge_args_sum_t::remove.
19259 (ipa_edge_duplication_hook): Turned into method
19260 ipa_edge_args_sum_t::duplicate.
19261 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19262 registering edge hooks.
19263 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19264 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19265 ipa_edge_args_sum instead of ipa_edge_args_vector.
19266 * ipa-profile.c (ipa_profile): Likewise.
19268 2017-05-03 Martin Jambor <mjambor@suse.cz>
19270 * symbol-summary.h (function_summary): New method exists.
19271 (function_summary::symtab_removal): Deallocate through release.
19272 (call_summary): New class.
19273 (gt_ggc_mx): New overload.
19274 (gt_pch_nx): Likewise.
19275 (gt_pch_nx): Likewise.
19277 2017-05-03 Jeff Law <law@redhat.com>
19279 PR tree-optimization/78496
19280 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19281 from simplify_cond_using_ranges. Split off code to walk
19282 backwards through casts into ...
19283 (simplify_cond_using_ranges_2): New function.
19284 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19285 (execute_vrp): After identifying jump threads, call
19286 simplify_cond_using_ranges_2.
19288 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19291 * ipa-inline.h (inline_summary): Add ctor.
19292 (create_ggc): Do not use ggc_cleared_alloc.
19294 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19295 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19297 * gcc.c (handle_braces): Support escaping in switch matching
19299 * doc/invoke.texi (Spec Files): Document it.
19300 Remove superfluous @code markup in items.
19302 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19304 * diagnostic-show-locus.c (struct column_range): New struct.
19305 (get_affected_columns): New function.
19306 (get_printed_columns): New function.
19307 (struct correction): New struct.
19308 (correction::ensure_capacity): New function.
19309 (correction::ensure_terminated): New function.
19310 (struct line_corrections): New struct.
19311 (line_corrections::~line_corrections): New dtor.
19312 (line_corrections::add_hint): New function.
19313 (layout::print_trailing_fixits): Reimplement in terms of the new
19315 (selftest::test_overlapped_fixit_printing): New function.
19316 (selftest::diagnostic_show_locus_c_tests): Call it.
19318 2017-05-03 Nathan Sidwell <nathan@acm.org>
19320 Canonicalize canonical type hashing
19321 * tree.h (type_hash_canon_hash): Declare.
19322 * tree.c (type_hash_list, attribute_hash_list): Move into
19323 type_hash_canon_hash.
19324 (build_type_attribute_qual_variant): Break out hash code calc into
19325 type_hash_canon_hash.
19326 (type_hash_canon_hash): New. Generic type hash computation.
19327 (build_range_type_1, build_array_type_1, build_function_type,
19328 build_method_type_directly, build_offset_type, build_complex_type,
19329 make_vector_type): Call it.
19331 2017-05-03 Richard Biener <rguenther@suse.de>
19333 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19334 When all DRs have unknown misaligned do not always peel
19335 when there is a store but apply the same costing model as if
19336 there were only loads.
19338 2017-05-03 Richard Biener <rguenther@suse.de>
19341 PR tree-optimization/80492
19342 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19343 compare_base_decls returning dont-know properly.
19345 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19347 * config/arm/iterators.md (CCSI): New mode iterator.
19348 (arch): New mode attribute.
19349 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19350 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19351 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19352 code iterator for success result mode.
19353 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19354 the corresponding new insn generators.
19356 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19359 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19360 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19362 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19364 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19365 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19366 (DDR_A): Wrap DDR argument in brackets.
19367 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19368 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19369 (DDR_REVERSED_P): Likewise.
19371 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19373 PR tree-optimization/79472
19374 * tree-switch-conversion.c (struct switch_conv_info): Add
19375 contiguous_range and default_case_nonstandard fields.
19376 (collect_switch_conv_info): Compute contiguous_range and
19377 default_case_nonstandard fields, don't clear final_bb if
19378 contiguous_range and only the default case doesn't have the required
19380 (check_all_empty_except_final): Set default_case_nonstandard instead
19381 of failing if contiguous_range and the default case doesn't have empty
19383 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19384 and only the default case doesn't have the required constants. Skip
19386 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19387 if default_case_nonstandard.
19388 (build_constructors): Build constant 1 just once. Assert that default
19389 values aren't inserted in between cases if contiguous_range. Skip
19391 (build_arrays): Skip virtual phis.
19392 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19393 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19394 Handle virtual phis.
19395 (gen_inbound_check): Handle default_case_nonstandard case.
19396 (process_switch): Adjust check_final_bb caller. Call
19397 gather_default_values with the first non-default case instead of
19398 default case if default_case_nonstandard.
19400 2017-05-02 Nathan Sidwell <nathan@acm.org>
19402 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19403 check. Fix formatting.
19405 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19407 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19408 errors when comparing specialized and unspecialized times.
19410 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19412 * diagnostic-show-locus.c
19413 (layout::should_print_annotation_line_p): Make private.
19414 (layout::print_annotation_line): Make private.
19415 (layout::annotation_line_showed_range_p): Make private.
19416 (layout::show_ruler): Make private.
19417 (layout::print_source_line): Make private. Pass in line and
19418 line_width, rather than calling location_get_source_line. Drop
19420 (layout::print_leading_fixits): New method.
19421 (layout::print_any_fixits): Rename to...
19422 (layout::print_trailing_fixits): ...this, and make private.
19423 Don't print newline fixits.
19424 (diagnostic_show_locus): Move logic for printing one row into...
19425 (layout::print_line): ...this new function. Move the
19426 location_get_source_line call and error-handling from
19427 print_source_line to here. Call print_leading_fixits, and rename
19428 print_any_fixits to print_trailing_fixits.
19429 (selftest::test_fixit_insert_containing_newline): Update now that
19430 newlines are partially supported.
19431 (selftest::test_fixit_insert_containing_newline_2): New test.
19432 (selftest::test_fixit_replace_containing_newline): Update comments.
19433 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19434 * edit-context.c (class added_line): New class.
19435 (class edited_line): Describe newline handling in comment.
19436 (edited_line::actually_edited_p): New method.
19437 (edited_line::print_content): Delete redundant decl.
19438 (edited_line::m_predecessors): New field.
19439 (edited_file::print_content): Call edited_line::print_content.
19440 (edited_file::print_diff): Update to support newlines.
19441 (edited_file::print_diff_hunk): Likewise.
19442 (edited_file::print_run_of_changed_lines): New function.
19443 (edited_file::print_diff_line): Convert to...
19444 (print_diff_line): ...this.
19445 (edited_file::get_effective_line_count): New function.
19446 (edited_line::edited_line): Initialize new field m_predecessors.
19447 (edited_line::~edited_line): Clean up m_predecessors.
19448 (edited_line::apply_fixit): Handle newlines.
19449 (edited_line::get_effective_line_count): New function.
19450 (edited_line::print_content): New function.
19451 (edited_line::print_diff_lines): New function.
19452 (selftest::test_applying_fixits_insert_containing_newline): New
19454 (selftest::test_applying_fixits_replace_containing_newline): New
19456 (selftest::insert_line): New function.
19457 (selftest::test_applying_fixits_multiple_lines): Add example of
19459 (selftest::edit_context_c_tests): Call the new tests.
19461 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19463 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19464 parameter cand. Update dump information.
19465 (get_computation_cost): Update uses.
19467 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19469 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19470 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19471 (get_computation_at, rewrite_use_address): Update use of
19472 get_computation_aff.
19474 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19476 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19477 (get_computation): Delete.
19478 (get_computation_cost): Implement like get_computation_cost_at.
19479 Use get_computation_at.
19480 (get_computation_cost_at): Delete.
19481 (rewrite_use_nonlinear_expr): Use get_computation_at.
19482 (rewrite_use_compare, remove_unused_ivs): Ditto.
19484 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19486 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19488 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19490 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19491 (ivopts_global_cost_for_size): Rename parameter and update uses.
19492 (iv_ca_recount_cost): Update uses.
19493 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19494 candidates seperately in n_invs and n_cands.
19495 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19497 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19499 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19500 (find_inv_vars_cb): New.
19501 (find_depends): Renamed to ...
19502 (find_inv_vars): ... this.
19503 (add_candidate_1, force_var_cost): Call find_inv_vars.
19504 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19506 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19508 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19509 inv_vars. Add inv_exprs.
19510 (struct iv_cand): Rename depends_on to inv_vars.
19511 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19512 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19513 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19514 (dump_cand): Dump inv_vars.
19515 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19516 (record_invariant, find_depends, add_candidate_1): Ditto.
19517 (set_group_iv_cost, force_var_cost): Ditto.
19518 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19519 (get_computation_cost_at, get_computation_cost): Ditto.
19520 (determine_group_iv_cost_generic): Ditto.
19521 (determine_group_iv_cost_address): Ditto.
19522 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19523 (determine_group_iv_costs): Ditto.
19524 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19525 (iv_ca_set_remove_invariants): Renamed to ...
19526 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19527 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19528 (iv_ca_set_add_invariants): Renamed to ...
19529 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19530 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19531 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19532 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19533 (create_new_ivs): Remove useless dump.
19535 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19537 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19539 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19540 (iv_ca_set_no_cp, create_new_iv): Ditto.
19542 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19544 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19546 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19548 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19549 function tree_check2.
19551 2017-05-02 Martin Liska <mliska@suse.cz>
19553 * doc/gcov.texi: Add missing preposition.
19554 * gcov.c (function_info::function_info): Properly fill up
19555 all member variables.
19557 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19559 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19561 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19563 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19565 2017-05-02 Martin Liska <mliska@suse.cz>
19568 * lto-streamer-in.c (lto_read_tree_1): Remove
19569 LTO_STREAMER_DEBUG.
19570 * lto-streamer.c (struct tree_hash_entry): Likewise.
19571 (struct tree_entry_hasher): Likewise.
19572 (tree_entry_hasher::hash): Likewise.
19573 (tree_entry_hasher::equal): Likewise.
19574 (lto_streamer_init): Likewise.
19575 (lto_orig_address_map): Likewise.
19576 (lto_orig_address_get): Likewise.
19577 (lto_orig_address_remove): Likewise.
19578 * lto-streamer.h: Likewise.
19579 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19580 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19582 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19584 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19585 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19586 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19587 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19588 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19589 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19590 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19591 (mm_maskz_sub_ss): New intrinsics.
19592 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19593 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19594 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19595 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19596 (__builtin_ia32_subss_mask_round): New builtins.
19597 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19598 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19599 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19601 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19602 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19604 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19607 2017-05-02 Martin Jambor <mjambor@suse.cz>
19609 PR tree-optimization/78687
19610 * tree-sra.c (access): New field parent.
19611 (process_subtree_disqualification): New function.
19612 (disqualify_candidate): Call it.
19613 (build_accesses_from_assign): Reset write flag if creating an
19615 (build_access_subtree): Fill in parent field and also prpagate
19616 down grp_write flag.
19617 (create_artificial_child_access): New parameter set_grp_write, set
19618 grp_write to its value.
19619 (propagate_subaccesses_across_link): Also propagate grp_write flag
19621 (propagate_all_subaccesses): Push the closest parent back to work
19622 queue if add_access_to_work_queue returned true.
19624 2017-05-02 Richard Biener <rguenther@suse.de>
19626 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19627 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19628 -fstrict-overflow documentation.
19629 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19630 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19631 flag_strict_overflow.
19632 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19633 * lto-opts.c (lto_write_options): Do not stream it.
19634 * lto-wrapper.c (merge_and_complain): Do not handle it.
19635 * opts.c (default_options_table): Do not set -fstrict-overflow.
19636 (finish_options): Likewise do not clear it when sanitizing.
19637 * simplify-rtx.c (simplify_const_relational_operation): Do not
19638 test flag_strict_overflow.
19640 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19642 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19643 using enabled attribute.
19644 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19645 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19646 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19647 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19648 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19649 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19650 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19651 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19652 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19653 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19655 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19657 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19659 2017-05-02 Richard Biener <rguenther@suse.de>
19661 PR tree-optimization/80591
19663 2017-04-10 Richard Biener <rguenther@suse.de>
19665 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19668 2017-05-02 Richard Biener <rguenther@suse.de>
19670 PR tree-optimization/80549
19671 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19672 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19673 headers if they do not have a preheader.
19675 2017-05-02 Martin Liska <mliska@suse.cz>
19678 * common.opt: Fix typo.
19679 * doc/invoke.texi: Likewise.
19681 2017-05-01 Jan Beulich <jbeulich@suse.com>
19683 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19684 swapping, add (x,x,m,x,n) alternative.
19686 2017-05-01 Nathan Sidwell <nathan@acm.org>
19688 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19689 unnecessary unadjusted_alignment check.
19691 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19694 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19696 * gimplify.c (gimplify_cilk_detach): New function.
19697 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19698 * tree-core.h: Document EXPR_CILK_SPAWN.
19699 * tree.h (EXPR_CILK_SPAWN): Define.
19701 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19703 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19704 to use new fixit_hint representation, using the "replace" logic.
19705 (get_line_span_for_fixit_hint): Likewise.
19706 (layout::print_any_fixits): Likewise.
19707 (selftest::test_one_liner_many_fixits): Rename to...
19708 (selftest::test_one_liner_many_fixits_1): ...this, and update
19709 comment and expected output to reflect that the multiple fix-it
19710 hints are now consolidated into one insertion.
19711 (selftest::test_one_liner_many_fixits_2): New test.
19712 (selftest::test_diagnostic_show_locus_one_liner): Update for
19714 (selftest::test_fixit_consolidation): Update for fix-it API
19716 * diagnostic.c (print_parseable_fixits): Likewise.
19717 * edit-context.c (edited_line::m_line_events): Convert from
19718 auto_vec <line_event *> to auto_vec <line_event>.
19719 (class line_event): Convert from abstract base class to a concrete
19720 class, taking over the role of replace_event.
19721 (class insert_event): Delete.
19722 (class replace_event): Rename to class line_event. Convert to
19724 (edit_context::add_fixits): Reimplement.
19725 (edit_context::apply_insert): Delete.
19726 (edit_context::apply_replace): Rename to...
19727 (edit_context::apply_fixit): ...this. Convert to half-open range.
19728 (edited_file::apply_insert): Delete.
19729 (edited_file::apply_replace): Rename to...
19730 (edited_file::apply_fixit): ...this.
19731 (edited_line::~edited_line): Drop deletion of events.
19732 (edited_line::apply_insert): Delete.
19733 (edited_line::apply_replace): Rename to...
19734 (edited_line::apply_fixit): ...this. Convert to half-open range.
19735 Update for change to type of m_line_events.
19736 * edit-context.h (edit_context::apply_insert): Delete.
19737 (edit_context::apply_replace): Rename to...
19738 (edit_context::apply_fixit): ...this.
19740 2017-05-01 Martin Sebor <msebor@redhat.com>
19742 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19745 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19748 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19749 __get_cpuid_max returns 0.
19750 (__get_cpuid_count): Ditto.
19752 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19754 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19755 replacement expression is another instance of one of its arguments.
19757 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19760 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19761 check for stack push/pop autoinc.
19762 * config/i386/i386.c (ix86_agi_dependent): Return false
19763 if the only reason why modified_in_p returned true is that
19764 addr is SP based and set_insn is a push or pop.
19766 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19768 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19771 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19774 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19775 (account_size_time): Use two predicates - exec_pred and
19777 (evaluate_conditions_for_known_args): Compute both clause and
19779 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19780 (inline_summary_t::duplicate): Update.
19781 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19783 (compute_inline_parameters): Likewise.
19784 (estimate_edge_size_and_time): Update caluclation of time.
19785 (estimate_node_size_and_time): Compute both time and nonspecialized
19787 (estimate_ipcp_clone_size_and_time): Update.
19788 (inline_merge_summary): Update.
19789 (do_estimate_edge_time): Update.
19790 (do_estimate_edge_size): Update.
19791 (do_estimate_edge_hints): Update.
19792 (inline_read_section, inline_write_summary): Stream both new predicates.
19793 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19795 (compute_inlined_call_time): Cleanup.
19796 (big_speedup_p): Update.
19797 (edge_badness): Update.
19798 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19799 (size_time_entry): Replace predicate by exec_predicate and
19800 nonconst_predicate.
19801 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19802 (estimate_edge_time): Return also nonspec_time.
19803 (reset_edge_growth_cache): Update.
19805 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19807 PR rtl-optimization/80491
19808 * ifcvt.c (noce_process_if_block): When looking for x setter
19809 with missing else_bb, don't check only the insn right before
19810 cond_earliest, but look for the last insn that x is modified in
19811 within the same bb.
19813 PR rtl-optimization/80491
19814 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19816 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19818 PR tree-optimization/80487
19819 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19821 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19823 PR tree-optimization/79697
19824 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19825 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19826 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19828 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19829 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19831 2017-04-28 Martin Sebor <msebor@redhat.com>
19833 PR tree-optimization/80523
19834 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19835 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19837 (maybe_warn, format_directive, parse_directive): Use new functions.
19838 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19840 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19842 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19844 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19846 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19847 target_header_dir): Set correctly.
19848 * configure: Regenerated.
19849 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19850 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19851 instead of SYSTEM_HEADER_DIR.
19853 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19855 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19856 (estimate_local_effects): Likewise.
19857 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19858 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19859 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19860 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19861 do_estimate_edge_time, estimate_edge_time): Likewise.
19862 * ipa-inline-analysis.c (estimate_node_size_and_time,
19863 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19864 (estimate_time_after_inlining): Remove.
19866 2017-04-28 Martin Liska <mliska@suse.cz>
19868 * doc/gcov.texi: Enhance documentation of gcov.
19870 2017-04-28 Martin Liska <mliska@suse.cz>
19872 * doc/gcov.texi: Sort options in alphabetic order.
19873 * doc/gcov-dump.texi: Likewise.
19874 * doc/gcov-tool.texi: Likewise.
19875 * gcov.c (print_usage): Likewise.
19876 * gcov-dump.c (print_usage): Likewise.
19877 * gcov-tool.c (print_merge_usage_message): Likewise.
19878 (print_rewrite_usage_message): Likewise.
19879 (print_overlap_usage_message): Likewise.
19881 2017-04-28 Martin Liska <mliska@suse.cz>
19883 PR gcov-profile/53915
19884 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19886 2017-04-28 Martin Liska <mliska@suse.cz>
19888 PR gcov-profile/79891
19889 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19890 is marked by compiler as living on a line.
19891 (get_cycles_count): Remove usage of the union.
19892 (output_intermediate_file): Likewise.
19893 (find_source): Fix GNU coding style.
19894 (accumulate_line_counts): Remove old non-all block mode.
19895 (output_lines): Remove usage of the union.
19896 * profile.c (output_location): Include all BBs, even if
19897 belonging to a same line (and file) as a previous BB.
19899 2017-04-28 Martin Liska <mliska@suse.cz>
19901 * gcov.c (process_args): Handle new argument 'w'.
19902 (read_graph_file): Assign ID to BBs.
19903 (output_branch_count): Display BB # if verbose flag is set.
19904 (output_lines): Likewise for arcs.
19905 (print_usage): Add '--verbose' option help.
19906 * doc/gcov.texi: Document --verbose (-w) option.
19908 2017-04-28 Martin Liska <mliska@suse.cz>
19910 * gcov.c (struct block_location_info): New struct.
19911 (process_file): Fill up the new structure.
19912 (read_graph_file): Replace usage of encoding by the newly added
19914 (add_line_counts): Likewise.
19915 (accumulate_line_counts): Remove usage of the union.
19916 (function_info::function_info): New function.
19917 (function_info::~function_info): Likewise.
19918 (process_file): Call delete instead of release_function.
19919 (release_function): Release the function.
19920 (release_structures): Call delete instead of release_function.
19921 (solve_flow_graph): Replace usage of num_blocks.
19922 (find_exception_blocks): Likewise.
19923 (output_lines): Fix GNU coding style.
19925 2017-04-28 Martin Liska <mliska@suse.cz>
19928 * coverage.c (coverage_remove_note_file): New function.
19929 * coverage.h: Declare the function.
19930 * toplev.c (finalize): Clean if an error has been seen.
19932 2017-04-28 Martin Liska <mliska@suse.cz>
19934 PR gcov-profile/80031
19935 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19936 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19937 * gcov.c (read_graph_file): Read just number of blocks.
19938 * profile.c (branch_prob): Do not stream 0 flags per a basic
19941 2017-04-28 Martin Liska <mliska@suse.cz>
19943 * gcov-dump.c (tag_*): Add new argument to declarations.
19944 (dump_gcov_file): Likewise.
19945 (tag_blocks): Add and use new argument depth.
19946 (tag_arcs): Likewise.
19947 (tag_lines): Likewise.
19948 (tag_counters): Likewise.
19949 (tag_summary): Likewise.
19950 (dump_working_sets): Use depth to do a proper indentation.
19952 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19955 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19956 * symtab.c (symtab_node::debug_symtab): Move definition here.
19958 2017-04-28 Richard Biener <rguenther@suse.de>
19960 * lto-streamer.h (LTO_major_version): Bump to 7.
19962 2017-04-28 Richard Biener <rguenther@suse.de>
19964 * tree-vrp.c (assert_info): New struct.
19965 (add_assert_info): New helper.
19966 (register_edge_assert_for_2): Refactor to add asserts to a vector
19968 (register_edge_assert_for_1): Likewise.
19969 (register_edge_assert_for): Likewise.
19970 (finish_register_edge_assert_for): New helper actually registering
19971 asserts where live on edge.
19972 (find_conditional_asserts): Adjust.
19973 (find_switch_asserts): Likewise.
19974 (evrp_dom_walker::try_find_new_range): Generalize.
19975 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19977 2017-04-27 Marek Polacek <polacek@redhat.com>
19980 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19981 arg10 and arg11 to itype.
19983 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19985 * doc/extend.texi (Object Size Checking): Improve grammar.
19987 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19990 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19991 that the logic for permitting reciprocal estimates matches that
19994 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19997 * tree.c (type_cache_hasher::equal): Only compare
19998 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19999 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20000 non-aggregate element types.
20001 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20002 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20004 2017-04-27 Richard Biener <rguenther@suse.de>
20006 PR middle-end/80533
20007 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20008 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20009 keeping a reference to a trailing array.
20011 2017-04-27 Richard Biener <rguenther@suse.de>
20013 PR middle-end/80539
20014 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20015 being in loop-closed SSA form conservatively.
20016 (chrec_fold_multiply_poly_poly): Likewise.
20018 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20020 PR middle-end/79665
20021 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20022 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20024 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20027 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20028 (aarch64_function_arg_alignment): Return unsigned int again, but still
20029 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20030 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20031 Don't emit -Wpsabi note.
20032 (aarch64_function_arg_boundary): Likewise.
20033 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20036 2017-04-26 Nathan Sidwell <nathan@acm.org>
20038 * tree.h (crc32_unsigned_n): Declare.
20039 (crc32_unsigned, crc32_unsigned): Make inline.
20040 * tree.c (crc32_unsigned_bits): Replace with ...
20041 (crc32_unsigned_n): ... this.
20042 (crc32_unsigned, crc32_byte): Remove.
20043 (crc32_string): Remove unnecessary braces.
20045 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20047 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20048 * ipa-inline-analysis.c (MAX_TIME): Remove.
20049 (account_size_time): Use sreal for time.
20050 (dump_inline_summary): Update.
20051 (estimate_function_body_sizes): Update.
20052 (estimate_edge_size_and_time): Update.
20053 (estimate_calls_size_and_time): Update.
20054 (estimate_node_size_and_time): Update.
20055 (inline_merge_summary): Update.
20056 (inline_update_overall_summary): Update.
20057 (estimate_time_after_inlining): Update.
20058 (inline_read_section): Update.
20059 (inline_write_summary): Update.
20060 * ipa-inline.c (compute_uninlined_call_time): Update.
20061 (compute_inlined_call_time): Update.
20062 (recursive_inlining): Update.
20063 (inline_small_functions): Update.
20064 (dump_overall_stats): Update.
20065 * ipa-inline.h: Include sreal.h.
20066 (size_time_entry): Turn time to sreal.
20067 (inline_summary): Turn self_time nad time to sreal.
20069 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20071 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20073 (sreal::stream_out, sreal::stream_in): New.
20074 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20076 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20078 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20081 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20084 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20086 (dimode_scalar_chain::compute_convert_gain): Ditto.
20087 (dimode_scalar_chain::make_vector_copies): Ditto.
20088 (dimode_scalar_chain::convert_reg): Ditto.
20089 (dimode_scalar_chain::convert_insn): Ditto.
20090 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20091 (VI248_AVX512BW_1): New mode iterator.
20092 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20093 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20096 2017-04-25 Martin Sebor <msebor@redhat.com>
20098 PR tree-optimization/80497
20099 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20100 constants are representable in HOST_WIDE_INT.
20101 (parse_directive): Ditto.
20103 2017-04-25 Martin Sebor <msebor@redhat.com>
20106 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20107 (new_zero_array): Adjust signature.
20108 (dom_info::dom_init): Used unsigned rather that size_t.
20109 (dom_info::dom_info): Same.
20111 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20112 Jakub Jelinek <jakub@redhat.com>
20115 * config/arm/arm.c: Include gimple.h.
20116 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20117 returns negative, increment ncrn only if it returned positive.
20118 (arm_needs_doubleword_align): Return int instead of bool,
20119 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20120 members, but if there is any such non-FIELD_DECL
20121 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20122 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20123 returns negative, increment nregs only if it returned positive.
20124 (arm_setup_incoming_varargs): Likewise.
20125 (arm_function_arg_boundary): Emit -Wpsabi note if
20126 arm_needs_doubleword_align returns negative, return
20127 DOUBLEWORD_ALIGNMENT only if it returned positive.
20129 2017-04-25 Marek Polacek <polacek@redhat.com>
20132 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20133 first argument to type.
20135 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20139 type checks to test for compatibility instead of equality.
20141 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20142 Jakub Jelinek <jakub@redhat.com>
20145 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20147 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20148 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20149 the alignment computation, but return their maximum in warn_alignment.
20150 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20151 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20153 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20154 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20157 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20159 * config/arc/simdext.md (dmpyh): Fix typo.
20161 2017-04-25 Richard Biener <rguenther@suse.de>
20163 PR tree-optimization/80492
20164 * alias.c (compare_base_decls): Handle registers with asm
20165 specification conservatively.
20166 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20167 compare_base_decls returning dont-know properly.
20169 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20171 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20172 (legitimate_offset_address_p): New function.
20173 (arc_legitimate_address_p): Use above function.
20175 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20177 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20179 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20181 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20182 ACCH registers whenever they are available.
20184 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20186 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20187 double regs fix when not used.
20189 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20191 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20193 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20194 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20196 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20198 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20199 when emitting short ADD instructions.
20201 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20203 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20205 (cmpsi_cc_c_insn): Likewise.
20206 (cbranchsi4_scratch): Compute proper instruction length using
20207 compact_hreg_operand.
20208 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20210 2017-04-25 Richard Biener <rguenther@suse.de>
20212 PR middle-end/80509
20213 * passes.c (pass_manager::pass_manager): Initialize
20214 m_name_to_pass_map.
20216 2017-04-25 Richard Biener <rguenther@suse.de>
20218 PR tree-optimization/79201
20219 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20221 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20224 * config/s390/vector.md: Split MEM->GPR vector moves for
20225 non-s_operand addresses.
20227 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20230 * config/s390/predicates.md (reload_const_wide_int_operand): New
20232 * config/s390/s390.md ("movti"): Remove d/P alternative.
20233 ("movti_bigconst"): New pattern definition.
20235 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20238 * s390-protos.h (s390_expand_cs_hqi): Removed.
20239 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20240 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20241 modes as well as CCZ1mode and CCZmode.
20242 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20243 signature of s390_emit_compare_and_swap.
20244 (s390_expand_cs_hqi): Likewise, make static.
20245 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20246 compare-and-swap, in some cases.
20247 (s390_expand_cs): Wrapper function.
20248 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20250 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20251 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20252 patterns for small and large integers. Forbid symref memory operands.
20253 Move expander to s390.c. Require cc register.
20254 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20255 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20256 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20257 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20258 symref memory operands. Remove CC mode and call s390_match_ccmode
20260 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20262 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20264 * config/s390/s390.md (define_peephole2): New peephole to help
20265 combining the load-and-test pattern with volatile memory.
20267 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20269 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20270 with CCZmode for TARGET_Z196.
20272 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20274 PR rtl-optimization/80501
20275 * combine.c (make_compound_operation_int): Set subreg_code to SET
20276 even for AND with mask of the sign bit of mode.
20278 PR rtl-optimization/80500
20279 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20280 sum's initial value.
20282 2017-04-25 Julian Brown <julian@codesourcery.com>
20283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20285 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20287 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20289 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20291 2017-04-25 Julian Brown <julian@codesourcery.com>
20292 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20294 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20295 (thunderx2t99_sha): New Reservation.
20297 2017-04-25 Julian Brown <julian@codesourcery.com>
20298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20300 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20301 type for 1-element load.
20303 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20305 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20307 2017-04-24 Martin Jambor <mjambor@suse.cz>
20309 PR tree-optimization/80293
20310 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20311 char arrays not totally scalarizable if it is false.
20312 (analyze_all_variable_accesses): Pass correct value in the new
20313 parameter. Add a statistics counter.
20315 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20317 PR middle-end/79931
20318 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20320 2017-04-24 Richard Biener <rguenther@suse.de>
20322 PR tree-optimization/80494
20323 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20324 out for complex types.
20326 2017-04-24 Richard Biener <rguenther@suse.de>
20328 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20329 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20330 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20331 (DFS): Adjust and never fail.
20332 (sccvn_dom_walker::fail): Remove.
20333 (sccvn_dom_walker::before_dom_children): Adjust.
20334 (run_scc_vn): Likewise and never fail.
20335 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20336 (pass_fre::execute): Likewise.
20338 2017-04-24 Richard Biener <rguenther@suse.de>
20340 PR tree-optimization/79725
20341 * tree-ssa-sink.c (statement_sink_location): Return whether
20342 failure reason was zero uses. Move that check later.
20343 (sink_code_in_bb): Deal with zero uses by removing the stmt
20346 2017-04-24 Richard Biener <rguenther@suse.de>
20349 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20350 pointer-based references.
20352 2017-04-24 Richard Biener <rguenther@suse.de>
20355 * pass_manager.h (pass_manager::operator new): Remove.
20356 (pass_manager::operator delete): Likewise.
20357 * passes.c (pass_manager::operator new): Remove.
20358 (pass_manager::operator delete): Likewise.
20359 (pass_manager::pass_manager): Zero individual pass members.
20361 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20364 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20365 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20366 Check "XEXP (src, 1)" operand here.
20367 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20368 Check "XEXP (src, 1)" operand here.
20369 (dimode_scalar_chain::make_vector_copies): Detect count register
20370 of a shift instruction. Zero extend count register from QImode
20371 to DImode to satisfy vector shift pattern count operand predicate.
20372 Substitute vector shift count operand with a DImode copy.
20373 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20376 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20378 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20379 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20380 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20381 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20382 (UNSPEC_NOREX_MEM): Remove definition.
20384 2017-04-21 Richard Biener <rguenther@suse.de>
20386 PR tree-optimization/79547
20387 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20388 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20389 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20390 without any constraints.
20392 2017-04-21 Richard Biener <rguenther@suse.de>
20394 PR tree-optimization/78847
20395 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20397 2017-04-21 Richard Biener <rguenther@suse.de>
20399 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20400 (build_distinct_type_copy): Likewise.
20401 (build_variant_type_copy): Likewise.
20402 * tree.c (build_qualified_type): Pass down mem-stat info.
20403 (build_distinct_type_copy): Likewise.
20404 (build_variant_type_copy): Likewise.
20406 2017-04-21 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/80237
20409 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20411 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20412 for a simplified result.
20414 2016-04-21 Richard Biener <rguenther@suse.de>
20416 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20417 sth as strict as a simple_iv but a chrec without symbols and an
20418 operand defined in the loop we are peeling (and not some subloop).
20419 (propagate_constants_for_unrolling): Propagate all constants.
20421 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20424 * config/i386/i386.c (print_reg): Remove assert for disalowed
20425 regno values, call output_operand_lossage instead.
20427 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20430 * config/i386/constraints.md (Yc): New register constraint.
20431 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20432 Use Yc constraint for alternative 2 of operand 0. Remove
20433 preferred_for_speed attribute.
20435 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20437 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20438 lastprivate clauses in SIMT case.
20440 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20442 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20444 2017-04-20 Richard Biener <rguenther@suse.de>
20446 PR tree-optimization/57796
20447 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20448 as N scalar stores.
20449 (vect_model_load_cost): Cost gathers as N scalar loads.
20451 2017-04-20 Richard Biener <rguenther@suse.de>
20453 * ggc-page.c (ggc_allocated_p): Rename to ...
20454 (safe_lookup_page_table_entry): ... this and return the lookup
20456 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20458 2017-04-20 Richard Biener <rguenther@suse.de>
20460 PR tree-optimization/80453
20461 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20462 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20463 from the conditions.
20464 (vn_phi_eq): Pass them down.
20465 (vn_phi_lookup): Record them.
20466 (vn_phi_insert): Likewise.
20468 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20470 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20471 uninitialized variable warning to avoid buffer overrun.
20473 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20476 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20477 is suppressed for '{ 0 }' in C.
20479 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20481 * BASE-VER: Set to 8.0.0.
20483 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20485 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20486 priority .init_array and .fini_array section with SECTION_NOTYPE
20489 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20491 PR middle-end/80423
20492 * tree.h (build_array_type): Add typeless_storage default argument.
20493 * tree.c (type_cache_hasher::equal): Also compare
20494 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20495 (build_array_type): Add typeless_storage argument, set
20496 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20498 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20499 (build_array_type): Likewise. Add typeless_storage argument.
20501 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20502 Jakub Jelinek <jakub@redhat.com>
20504 PR tree-optimization/80426
20505 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20506 operation on symbolic operands, also compute the overflow for the
20507 invariant part when the operation degenerates into a negation.
20509 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20512 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20513 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20516 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20518 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20521 * config/avr/avr.c (tree.h): Include it.
20522 (cgraph.h): Include it.
20523 (avr_encode_section_info): Don't warn for uninitialized progmem
20524 variable if it's just an alias.
20526 2017-04-19 Richard Biener <rguenther@suse.de>
20529 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20530 when needed by AutoPGO.
20532 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20535 * doc/lto.texi: Remove an extra 'that'.
20537 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20539 PR rtl-optimization/80429
20540 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20541 are only used in debug insns.
20543 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20544 Vladimir Makarov <vmakarov@redhat.com>
20546 * config/sparc/predicates.md (input_operand): Add comment. Return
20547 true for any memory operand when LRA is in progress.
20548 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20550 2017-04-18 Jeff Law <law@redhat.com>
20553 * mips.md ({return,simple_return}_internal): Do not overwrite
20556 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20558 PR tree-optimization/80443
20559 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20560 instead of adding 1, subtract -1 and similarly instead of subtracting
20563 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20565 PR rtl-optimization/80357
20566 * haifa-sched.c (tmp_bitmap): New variable.
20567 (model_recompute): Handle duplicate use records.
20568 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20569 (free_global_sched_pressure_data): Free it.
20571 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20574 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20575 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20576 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20577 instead of SYSTEM_HEADER_DIR.
20579 2017-04-18 Jeff Law <law@redhat.com>
20581 PR middle-end/80422
20582 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20583 predecessors after walking up the insn chain.
20585 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20588 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20589 sizetype type into debug info.
20591 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20594 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20595 unneeded test for TARGET_UPPER_REGS_SF.
20596 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20598 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20601 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20602 instead of gsi_after_labels.
20604 2017-04-18 Jeff Law <law@redhat.com>
20606 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20610 2017-04-13 Jeff Law <law@redhat.com>
20611 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20612 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20614 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20617 * config/avr/avr.c (intl.h): Include it.
20618 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20620 2017-04-18 Martin Liska <mliska@suse.cz>
20622 PR gcov-profile/78783
20623 * gcov-tool.c (gcov_output_files): Validate that destination
20624 file is either removed by the tool or by a user.
20626 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20627 Guy Benyei <guybe@mellanox.com>
20629 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20630 block, and do not negate it, the stored id is already negative.
20632 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20634 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20636 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20639 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20640 masks of options that should be turned off if the VSX vector
20641 options are turned off.
20642 (OTHER_P8_VECTOR_MASKS): Likewise.
20643 (OTHER_VSX_VECTOR_MASKS): Likewise.
20644 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20645 rs6000_disable_incompatible_switches to validate no type switches
20647 (rs6000_incompatible_switch): New function to disallow turning on
20648 other vector options if -mno-vsx, -mno-power8-vector, or
20649 -mno-power9-vector are specified.
20651 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20653 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20655 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20657 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20658 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20659 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20660 (ARG_POINTER_CFA_OFFSET): Likewise.
20662 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20664 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20665 conditions to take advantage of various optimizations.
20667 2017-04-13 Jeff Law <law@redhat.com>
20669 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20670 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20671 (zero_extendsidi2_dext): Likewise.
20673 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20676 * fold-const.c (fold_ternary_loc): Revert
20677 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20680 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20682 PR rtl-optimization/80343
20683 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20684 new scratch pseudo.
20686 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20689 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20690 to ubsan_encode_value.
20692 2017-04-13 Jeff Law <law@redhat.com>
20694 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20695 appearing in DEBUG_INSNs.
20697 2017-04-13 Martin Liska <mliska@suse.cz>
20699 PR gcov-profile/80413
20700 * gcov-io.c (gcov_write_string): Copy to buffer just when
20701 allocated size is greater than zero.
20703 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20706 * dwarf2out.c (decls_for_scope): Ignore declarations of
20707 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20709 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20712 * ipa-visibility.c (non_local_p): Fix typos.
20713 (localize_node): When localizing symbol in same comdat group,
20714 dissolve the group only when we know external symbols are going
20716 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20718 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20720 PR tree-optimization/79390
20721 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20722 order does not result in usable sequence, retry with reversed operand
20728 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20729 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20730 op0 instead of fold_convert_loc (loc, type, arg0).
20732 2017-04-12 Jeff Law <law@redhat.com>
20734 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20735 has a delay slot in the generated code.
20737 * config/cris/cris.md (cris_preferred_reload_class): Return
20738 GENNONACR_REGS rather than GENERAL_REGS.
20740 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20743 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20744 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20745 signedness of the result type.
20747 2017-04-12 Richard Biener <rguenther@suse.de>
20748 Jeff Law <law@redhat.com>
20750 PR tree-optimization/80359
20751 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20752 trim stores to TARGET_MEM_REFs.
20754 2017-04-12 Richard Biener <rguenther@suse.de>
20756 PR tree-optimization/79390
20757 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20758 threading case even more.
20760 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20763 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20764 for quad_address_p for TImode, instead of just not indexed_address.
20766 2017-04-12 Richard Biener <rguenther@suse.de>
20767 Bernd Edlinger <bernd.edlinger@hotmail.de>
20769 PR middle-end/79671
20770 * alias.c (component_uses_parent_alias_set_from): Handle
20771 TYPE_TYPELESS_STORAGE.
20772 (get_alias_set): Likewise.
20773 * tree-core.h (tree_type_common): Add typeless_storage flag.
20774 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20775 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20776 for types containing members with TYPE_TYPELESS_STORAGE.
20777 (place_field): Likewise.
20778 (layout_type): Likewise for ARRAY_TYPE.
20779 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20780 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20781 TYPE_TYPELESS_STORAGE.
20782 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20784 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20787 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20788 first argument to type.
20790 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20794 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20795 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20796 (rs6000_expand_binop_builtin): Likewise.
20797 (rs6000_expand_ternop_builtin): Likewise; also add missing
20798 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20799 vshasigma built-ins.
20800 * doc/extend.texi: Document that vec_xxpermdi's third argument
20801 must be a constant.
20803 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20805 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20806 Use shift_const cost parameter when calculating gain of STV shifts.
20808 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20810 PR rtl-optimization/70478
20811 * lra-constraints.c (process_alt_operands): Check memory for
20812 disfavoring memory insn operand.
20814 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20816 PR middle-end/80100
20817 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20818 left shift in unsigned HOST_WIDE_INT type.
20820 PR rtl-optimization/80385
20821 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20822 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20825 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20826 if they have any depend clauses.
20828 2017-04-11 Martin Liska <mliska@suse.cz>
20831 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20832 * ipa-split.c (split_function): Create a local comdat symbol
20833 if caller is in a comdat group.
20835 2017-04-11 Martin Liska <mliska@suse.cz>
20838 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20841 2017-04-11 Martin Sebor <msebor@redhat.com>
20843 PR middle-end/80364
20844 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20845 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20847 (directive::set_width, directive::set_precision, format_character):
20849 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20852 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20855 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20856 conflict, set target->arch_name instead of target->cpu_name.
20858 2017-04-11 Richard Biener <rguenther@suse.de>
20860 PR tree-optimization/80374
20861 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20862 build_zero_cst, remove fold_convertible_p check again.
20864 2017-04-11 Martin Liska <mliska@suse.cz>
20867 * ubsan.c (instrument_object_size): Do not instrument register
20870 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20873 * config/i386/i386-builtin-types.def
20874 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20875 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20876 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20877 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20878 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20879 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20880 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20881 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20882 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20883 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20884 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20885 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20886 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20887 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20888 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20889 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20890 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20891 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20892 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20893 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20894 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20895 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20896 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20897 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20898 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20899 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20900 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20901 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20902 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20903 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20904 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20905 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20906 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20907 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20908 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20909 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20910 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20911 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20912 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20913 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20914 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20915 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20916 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20917 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20918 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20919 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20921 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20922 flag to second_arg_count, handle 4 argument function type _COUNT
20923 aliases, handle second_arg_count on second argument rather than last.
20925 2017-04-10 Jeff Law <law@redhat.com>
20927 PR tree-optimization/80374
20928 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20929 record anything if we can not convert integer_zero_node to the
20932 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20935 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20936 Enhance special handling given to the TARGET_P9_MINMAX option in
20937 relation to certain other options.
20939 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20941 PR tree-optimization/80153
20942 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20943 remove POINTER_PLUS_EXPR's base part directly, rather than through
20946 2017-04-10 Richard Biener <rguenther@suse.de>
20947 Bin Cheng <bin.cheng@arm.com>
20949 PR tree-optimization/80153
20950 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20951 the first element of pointer type aff_tree. Build result expr in
20953 (add_elt_to_tree): Convert to type unconditionally. Remove other
20954 fold_convert calls.
20955 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20956 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20958 2017-04-10 Richard Biener <rguenther@suse.de>
20960 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20963 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20965 PR rtl-optimization/70478
20966 * lra-constraints.c (curr_small_class_check): New.
20967 (update_and_check_small_class_inputs): New.
20968 (process_alt_operands): Update curr_small_class_check. Disfavor
20969 alternative insn memory operands. Check available regs for small
20972 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20975 * config/mips/mips.opt (-mvirt): Update description.
20976 * doc/invoke.texi (-mvirt): Likewise.
20978 2017-04-10 Richard Biener <rguenther@suse.de>
20980 PR middle-end/80362
20981 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20982 looking for NEGATE_EXPR in -A / -B to A / B folding.
20984 2017-04-10 Martin Liska <mliska@suse.cz>
20986 PR gcov-profile/80224
20987 * gcov.c (print_usage): Fix usage string.
20988 (get_gcov_intermediate_filename): Remove.
20989 (output_gcov_file): Use both for normal and intermediate format.
20990 (generate_results): Do not initialize special file for
20991 intermediate format.
20993 2017-04-10 Richard Biener <rguenther@suse.de>
20995 PR tree-optimization/80304
20996 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20999 2017-04-10 Nathan Sidwell <nathan@acm.org>
21002 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21003 (rs6000_init_builtins): Use it.
21005 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21007 * config/arm/arm.md (<mrc>): Add mode to SET source.
21008 (<mrrc>): Likewise.
21010 2017-04-10 Richard Biener <rguenther@suse.de>
21012 PR middle-end/80344
21013 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21015 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21018 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21019 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21020 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21021 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21022 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21023 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21024 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21025 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21026 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21027 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21028 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21029 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21030 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21031 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21032 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21033 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21034 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21035 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21036 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21037 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21038 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21039 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21040 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21042 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21044 PR rtl-optimization/70478
21045 * lra-constraints.c: Reverse the last patch.
21047 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21049 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21050 Add comment for WCHAR_T.
21052 2017-04-08 Martin Liska <mliska@suse.cz>
21055 2017-04-07 Martin Liska <mliska@suse.cz>
21058 * ipa-split.c (split_function): Add function part to a same comdat
21061 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21064 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21066 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21068 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21069 (vec_pairing): ...to this.
21070 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21071 (rs6000_sched_init): Adjust for name change.
21072 (struct rs6000_sched_context): Likewise.
21073 (rs6000_init_sched_context): Likewise.
21074 (rs6000_set_sched_context): Likewise.
21076 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21082 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21084 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21085 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21086 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21088 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21090 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21092 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21094 PR rtl-optimization/70703
21095 * ira-color.c (update_conflict_hard_regno_costs): Use
21096 int64_t instead of HOST_WIDE_INT.
21098 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21100 PR rtl-optimization/70478
21101 * lra-constraints.c (process_alt_operands): Disfavor alternative
21102 insn memory operands.
21104 2017-04-07 Jeff Law <law@redhat.com>
21106 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21107 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21109 2017-04-07 Martin Liska <mliska@suse.cz>
21112 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21113 Show error message instead of an ICE.
21115 2017-04-07 Martin Liska <mliska@suse.cz>
21118 * ipa-split.c (split_function): Add function part to a same comdat
21121 2017-04-07 Richard Biener <rguenther@suse.de>
21123 PR middle-end/80341
21124 * tree.c (get_unwidened): Also handle ! for_type case for
21126 * convert.c (do_narrow): Split out from ...
21127 (convert_to_integer_1): ... here. Do not pass final truncation
21128 type to get_unwidened for TRUNC_DIV_EXPR.
21130 2017-04-07 Richard Biener <rguenther@suse.de>
21132 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21134 (aff_combination_const): Adjust.
21135 (aff_combination_scale): Likewise.
21136 (aff_combination_add_elt): Likewise.
21137 (aff_combination_add_cst): Likewise.
21138 (aff_combination_convert): Likewise.
21139 (add_elt_to_tree): Likewise. Remove unused argument.
21140 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21142 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21144 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21146 * config/arm/arm.c (arm_default_short_enums): Use
21147 ARM_DEFAULT_SHORT_ENUMS.
21148 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21150 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21153 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21154 members with redundant out-of-class redeclaration.
21156 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21159 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21160 * config/i386/i386.md (*zero_extendsidi2):
21161 Add (?*x,*x) and (?*v,*v) alternatives.
21163 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21166 * config/i386/i386.c (ix86_expand_builtin)
21167 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21168 mode from insn data. Convert operands to insn operand mode.
21169 Copy operands that don't satisfy insn predicate to a register.
21171 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21173 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21176 2017-04-06 Richard Biener <rguenther@suse.de>
21178 PR tree-optimization/80334
21179 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21180 preserve alignment of accesses.
21182 2017-04-06 Richard Biener <rguenther@suse.de>
21184 PR tree-optimization/80262
21185 * tree-sra.c (build_ref_for_offset): Preserve address-space
21187 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21188 Drop useless address-space information on MEM_REF offsets.
21190 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21192 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21194 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21196 PR rtl-optimization/70703
21197 * ira-color.c (update_conflict_hard_regno_costs): Use
21198 HOST_WIDE_INT instead of long.
21200 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21203 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21204 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21206 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21207 for x86_64 target. Handle -m3dnowa option.
21209 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21211 PR rtl-optimization/70703
21212 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21213 (update_conflict_hard_regno_costs): Use long instead of unsigned
21214 arithmetic for cost calculation.
21216 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21217 Bernd Edlinger <bernd.edlinger@hotmail.de>
21220 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21223 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21226 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21227 ptr_mode with Pmode throughout.
21228 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21229 into probe_stack_range and use DImode.
21231 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21234 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21235 call_eh_return is true.
21237 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21239 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21240 Initialize last_match_fntype_index.
21242 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21245 * tree-nvr.c: Include internal-fn.h.
21246 (pass_return_slot::execute): Ignore internal calls without
21249 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21250 Richard Biener <rguenther@suse.de>
21253 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21254 captures used multiple times, except for the last use.
21255 * generic-match-head.c: Include gimplify.h.
21257 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21259 PR tree-optimization/79390
21260 * target.h (struct noce_if_info): Declare.
21261 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21262 * target.def (noce_conversion_profitable_p): New target hook.
21263 * ifcvt.h (struct noce_if_info): New type, moved from ...
21264 * ifcvt.c (struct noce_if_info): ... here.
21265 (noce_conversion_profitable_p): Renamed to ...
21266 (default_noce_conversion_profitable_p): ... this. No longer
21268 (noce_try_store_flag_constants, noce_try_addcc,
21269 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21270 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21271 instead of noce_conversion_profitable_p.
21272 * config/i386/i386.c: Include ifcvt.h.
21273 (ix86_option_override_internal): Don't override
21274 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21275 (ix86_noce_conversion_profitable_p): New function.
21276 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21277 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21278 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21279 * doc/tm.texi: Regenerated.
21281 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21283 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21286 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21289 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21290 instructions for small multiply cores.
21292 2017-04-04 Jeff Law <law@redhat.com>
21294 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21296 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21297 that are not set by the loop over the elements.
21299 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21302 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21303 int mode, convert_modes it to mode as unsigned, otherwise use
21304 lowpart_subreg to mode rather than SImode.
21305 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21306 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21307 Use DImode instead of SImode for the shift count operand.
21308 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21311 2017-04-04 Richard Biener <rguenther@suse.de>
21313 PR middle-end/80281
21314 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21315 arithmetic done for the negate or the plus. Simplify.
21316 (A - (-B) -> A + B): Likewise.
21317 * fold-const.c (split_tree): Make sure to not negate pointers.
21319 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21321 PR rtl-optimization/60818
21322 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21323 a compare of comparisons with the thing compared if this results
21324 in a different machine mode.
21326 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21328 * alias.c (base_alias_check): Fix typo in comment.
21329 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21330 * cgraphunit.c (symbol_table::compile): Likewise.
21331 * collect2.c (maybe_run_lto_and_relink): Likewise.
21332 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21333 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21334 * config/avr/avr.c (avr_map_op_t): Likewise.
21335 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21336 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21337 * config/epiphany/epiphany.md (movcc): Likewise.
21338 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21339 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21341 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21342 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21343 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21344 * config/sh/sh.c (sh_rtx_costs): Likewise.
21345 * fold-const.c (fold_truth_andor): Likewise.
21346 * genautomata.c (collapse_flag): Likewise.
21347 * gengtype.h (struct type::u::s): Likewise.
21348 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21349 * input.c (FORMAT_AMOUNT): Likewise.
21350 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21351 (known_aggs_to_agg_replacement_list): Likewise.
21352 * ipa-inline-analysis.c: Likewise.
21353 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21354 * ipa-polymorphic-call.c
21355 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21356 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21357 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21359 * modulo-sched.c (apply_reg_moves): Likewise.
21360 * omp-expand.c (build_omp_regions_1): Likewise.
21361 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21362 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21363 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21364 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21365 * value-prof.c: Likewise.
21366 * var-tracking.c (val_reset): Likewise.
21368 2017-04-03 Richard Biener <rguenther@suse.de>
21370 PR tree-optimization/80275
21371 * fold-const.c (split_address_to_core_and_offset): Handle
21374 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21376 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21377 descriptors is at least equal to that of functions.
21379 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21381 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21383 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21386 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21387 (mov<IMOD4:mode>): New expander.
21388 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21390 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21392 PR rtl-optimization/79405
21393 * fwprop.c (propagations_left): New variable.
21394 (forward_propagate_into): Decrement it.
21395 (fwprop_init): Initialize it.
21396 (fw_prop): If the variable has reached zero, stop propagating.
21397 (fwprop_addr): Ditto.
21399 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21402 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21403 a FUNCTION_DECL, pass it as decl instead of origin to
21406 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21408 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21411 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21414 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21415 TARGET_VSX_SMALL_INTEGER.
21417 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21419 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21420 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21422 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21424 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21425 extraction from odd-numbered MSA register.
21427 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21429 PR middle-end/80173
21430 * expmed.c (store_bit_field_1): Don't attempt to create
21431 a word subreg out of hard registers wider than word if they
21432 have HARD_REGNO_NREGS of 1 for their mode.
21434 PR middle-end/80163
21435 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21436 conversions to integer types wider than word and pointer.
21439 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21440 (rtx_equal_for_cselib_p): Pass 0 to it.
21441 * cselib.c (cselib_hasher::equal): Likewise.
21442 (rtx_equal_for_cselib_1): Add depth argument. If depth
21443 is 128, don't look up VALUE locs and punt. Increment
21444 depth in recursive calls when walking VALUE locs.
21446 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21448 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21449 (make_gcov_file_name): Use the canonical path name for generating
21451 (read_line): Fix handling of files with ascii null bytes.
21453 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21455 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21456 to initialise a vector register instead
21457 of using a const_int.
21459 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21461 PR translation/80189
21462 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21463 diagnostic messages.
21465 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21468 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21469 (dfp_diex_<mode>): Update mode of operand 1.
21470 * doc/extend.texi (dxex, dxexq): Document change to return type.
21471 (diex, diexq): Document change to argument type.
21473 2017-03-30 Martin Jambor <mjambor@suse.cz>
21476 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21477 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21478 it reflects the signature changes performed at the callee side.
21479 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21480 to cgraph_build_function_type_skip_args.
21481 (build_function_decl_skip_args): Adjust call to the above function.
21483 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21486 * config/i386/sse.md
21487 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21488 register as dest whenever it is a MEM not rtx_equal_p to the
21489 corresponding dup operand, and when forcing into reg move the
21490 reg into the memory afterwards.
21491 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21492 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21493 for the force_reg mode.
21494 (avx512vl_vextractf128<mode>): Use register as dest either
21495 always when a MEM, or when it is a MEM not rtx_equal_p to the
21496 corresponding dup operand, or even not when it is a CONST_VECTOR
21497 depending on the mode and lo vs. hi.
21498 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21500 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21501 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21502 Likewise. Require that operands[2] is even.
21503 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21504 Remove extraneous parens. Require that operands[2] is a multiple
21506 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21507 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21508 disallow memory then.
21510 2017-03-30 Richard Biener <rguenther@suse.de>
21512 PR tree-optimization/77498
21513 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21514 to non-constants over backedges.
21516 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21518 PR rtl-optimization/80233
21519 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21520 as last_combined_insn. Do not test for BARRIER_P separately.
21522 2017-03-29 Andreas Schwab <schwab@suse.de>
21525 * calls.c (prepare_call_address): Convert funexp to Pmode before
21526 copying to temp reg.
21528 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21530 PR tree-optimization/80158
21531 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21532 Handle possible future case of more than one alternate
21534 (replace_rhs_if_not_dup): Likewise.
21535 (replace_one_candidate): Likewise.
21537 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21539 PR rtl-optimization/80193
21540 * ira.c (ira): Do not check allocation for LRA.
21542 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21544 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21545 (nvptx_output_simt_exit): Declare.
21546 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21547 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21548 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21549 (nvptx_declare_function_name): ...here. Emit declaration of local
21550 memory space buffer for omp_simt_enter insn.
21551 (nvptx_output_unisimt_switch): New.
21552 (nvptx_output_softstack_switch): New.
21553 (nvptx_output_simt_enter): New.
21554 (nvptx_output_simt_exit): New.
21555 * config/nvptx/nvptx.h (struct machine_function): New fields
21556 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21557 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21558 (UNSPECV_SIMT_EXIT): Ditto.
21559 (omp_simt_enter_insn): New insn.
21560 (omp_simt_enter): New expansion.
21561 (omp_simt_exit): New insn.
21562 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21564 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21565 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21566 (expand_GOMP_SIMT_EXIT): New.
21567 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21568 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21569 (GOMP_SIMT_EXIT): Ditto.
21570 * target-insns.def (omp_simt_enter): New insn.
21571 (omp_simt_exit): Ditto.
21572 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21574 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21575 (lower_rec_input_clauses): Likewise.
21576 (lower_lastprivate_clauses): Handle SIMT privatization.
21578 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21579 (ompdevlow_adjust_simt_enter): New.
21580 (find_simtpriv_var_op): New.
21581 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21582 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21584 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21585 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21586 (copy_decl_for_dup_finish): Ditto.
21588 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21590 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21593 * config/i386/i386.c (ix86_option_override_internal): Always
21594 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21596 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21598 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21600 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21602 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21603 mark new edge's irreducible flag accordign to it.
21604 (vect_do_peeling): Check loop preheader edge's irreducible flag
21605 and pass it to function slpeel_add_loop_guard.
21607 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21609 PR tree-optimization/80218
21610 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21611 Update block frequencies and counts.
21613 2017-03-28 Richard Biener <rguenther@suse.de>
21615 PR tree-optimization/78644
21616 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21617 of a simplification result we may not use it at all.
21619 2017-03-28 Richard Biener <rguenther@suse.de>
21622 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21623 without arguments, generate default definition of a SSA name.
21625 2017-03-28 Richard Biener <rguenther@suse.de>
21627 PR middle-end/80222
21628 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21629 TYPE_REF_CAN_ALIAS_ALL references.
21630 * fold-const.c (fold_indirect_ref_1): Likewise.
21632 2017-03-28 Martin Liska <mliska@suse.cz>
21635 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21636 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21638 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21639 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21641 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21642 (EXTRA_SPECS): Define.
21643 (SUBTARGET_EXTRA_SPECS): Likewise.
21644 (SUBTARGET_CPP_SPEC): Likewise.
21645 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21646 SUBTARGET_EXTRA_SPECS.
21647 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21649 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21651 * config/arc/simdext.md (vst64_insn): Update pattern.
21652 (vld32wh_insn): Likewise.
21653 (vld32wl_insn): Likewise.
21654 (vld64_insn): Likewise.
21655 (vld32_insn): Likewise.
21657 2017-03-28 Marek Polacek <polacek@redhat.com>
21660 * fold-const.c (fold_comparison): Use protected_set_expr_location
21661 instead of SET_EXPR_LOCATION.
21663 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21665 * tree.c (add_expr): Avoid name lookup warning.
21667 2017-03-27 Jeff Law <law@redhat.com>
21669 PR tree-optimization/80216
21670 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21671 function name. Limit recursion depth.
21672 (record_temporary_equivalences): Corresponding changes.
21674 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21676 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21679 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21682 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21684 * cfgcleanup.c (reg_note_cfa_p): New array.
21685 (insns_have_identical_cfa_notes): New function.
21686 (old_insns_match_p): Don't cross-jump in between /f
21687 and non-/f instructions. If both i1 and i2 are frame related,
21688 verify all CFA notes, their order and content.
21690 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21693 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21694 HImode and SImode with zero extend to DImode to one insn.
21695 (bswap<mode>2_extenddi): Likewise.
21696 (bswapsi2_extenddi): Likewise.
21697 (bswaphi2_extendsi): Likewise.
21698 (bswaphi2): Combine bswap HImode and SImode into one insn.
21699 Separate memory insns from swapping register.
21700 (bswapsi2): Likewise.
21701 (bswap<mode>2): Likewise.
21702 (bswaphi2_internal): Delete, no longer used.
21703 (bswapsi2_internal): Likewise.
21704 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21705 store, and gpr<-gpr swap insns.
21706 (bswap<mode>2_store): Likewise.
21707 (bswaphi2_reg): Register only splitter, combine with the splitter.
21708 (bswaphi2 splitter): Likewise.
21709 (bswapsi2_reg): Likewise.
21710 (bswapsi2 splitter): Likewise.
21711 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21712 the insns into load, store, and register/register insns.
21713 (bswapdi2_ldbrx): Likewise.
21714 (bswapdi2_load): Likewise.
21715 (bswapdi2_store): Likewise.
21716 (bswapdi2_reg): Likewise.
21718 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21720 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21721 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21723 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21726 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21728 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21729 special handling for target option conflicts between dform
21730 options (-mpower9-dform, -mpower9-dform-vector,
21731 -mpower9-dform-scalar) and -mno-direct-move.
21733 2017-03-27 Richard Biener <rguenther@suse.de>
21735 PR tree-optimization/80181
21736 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21738 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21740 * config/arc/predicates.md (move_double_src_operand): Replace the
21741 call to move_double_src_operand with a call to address_operand.
21743 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21745 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21746 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21747 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21749 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21751 * config/arc/predicates.md (long_immediate_loadstore_operand):
21752 Consider scaled addresses cases.
21754 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21756 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21757 restored when in interrupt.
21758 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21759 doesn't have delay slot.
21761 2017-03-27 Richard Biener <rguenther@suse.de>
21764 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21765 inlined thunk clones.
21767 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21770 * asan.c (instrument_derefs): Copy over last operand from
21771 original COMPONENT_REF to the new COMPONENT_REF with
21772 DECL_BIT_FIELD_REPRESENTATIVE.
21773 * ubsan.c (instrument_object_size): Likewise.
21775 2017-03-27 Richard Biener <rguenther@suse.de>
21777 PR tree-optimization/80170
21778 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21779 sure DR/SCEV didnt fold in constants we do not see when looking
21780 at the reference base alignment.
21782 2017-03-27 Richard Biener <rguenther@suse.de>
21784 PR middle-end/80171
21785 * gimple-fold.c (fold_ctor_reference): Properly guard against
21786 NULL return value from canonicalize_constructor_val.
21788 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21791 * config/i386/i386.c (ix86_expand_builtin)
21792 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21793 flags reg setting and flags reg using instructions.
21794 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21795 clobbering instructions to zero extend op2.
21797 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21799 * doc/install.texi (Configuration) <--with-aix-soname>:
21800 Update link to AIX ld.
21802 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21804 PR rtl-optimization/80160
21805 PR rtl-optimization/80159
21806 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21807 reg_alternate_class into account.
21809 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21812 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21813 to consider in curr_insn_transform.
21815 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21817 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21818 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21819 and emit_mode_inner.
21821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21823 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21824 argument to the overloaded builtin variants. Use the new flag to
21825 deprecate certain builtin variants.
21826 * config/s390/s390-builtin-types.def: Add new builtin types.
21827 * config/s390/s390-builtins.h: Support new flags field for
21828 overloaded builtins.
21829 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21830 (s390_macro_to_expand): Enable vector float data type.
21831 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21832 builtins by incrementing the __VEC__ version number.
21833 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21835 (s390_resolve_overloaded_builtin): Emit error messages depending
21836 on the builtin flags.
21837 * config/s390/s390.c (s390_expand_builtin): Support additional
21838 flags argument. Change error message to match the messages
21839 emitted in s390-c.c.
21840 * config/s390/s390.md: New UNSPEC_* constants.
21841 (op_type): Add new instruction types.
21842 * config/s390/vecintrin.h: Add new builtins and test data class
21844 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21845 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21846 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21847 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21848 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21849 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21851 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21852 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21853 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21854 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21856 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21857 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21858 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21859 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21861 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21862 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21863 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21864 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21865 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21866 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21867 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21869 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21870 ("vec_scatter_element<V_HW_4:mode>_DI")
21871 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21872 ("vec_fpint<mode>", "vflls")
21873 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21874 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21875 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21876 ("*vec_cmphe<mode>_cc"): ... these.
21878 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21879 mode constant instead of magic value.
21881 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21883 * config/s390/s390.c (s390_expand_vec_compare): Support other
21884 vector floating point modes than just V2DF.
21885 (s390_expand_vcond): Likewise.
21886 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21887 (s390_cannot_change_mode_class): Prevent mode changes between TF
21888 and V1TF in vector registers.
21889 * config/s390/s390.md (DF, SF): New mode attributes.
21890 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21891 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21892 SFmode support for VRs.
21893 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21895 (VFT, VF_HW): New mode iterators.
21896 (vw, sdx): New mode attributes.
21897 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21898 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21899 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21900 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21901 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21902 also the new vector floating point modes. Renaming to ...
21904 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21905 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21906 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21907 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21908 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21909 ("vec_unordered<mode>"): ... these.
21911 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21912 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21913 ("*vec_extendv2df"): New insn definitions.
21915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21917 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21918 ("mulditi3_2", "*muldi3_sign"): New patterns.
21919 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21920 rename the pattern definition.
21922 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21924 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21926 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21928 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21930 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21931 instruction if possible.
21932 * config/s390/vector.md (vec_halfnumelts): New mode
21934 ("*vec_vllezlf<mode>"): New pattern.
21936 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21938 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21939 ("popcountv4si2", "popcountv2di2"): Rename to ...
21940 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21941 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21943 ("popcount<mode>2_vxe"): New pattern.
21945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21947 * common/config/s390/s390-common.c (processor_flags_table): Add
21949 * config.gcc: Add arch12.
21950 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21951 Default to arch12 for unknown CPU model numbers.
21952 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21953 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21954 PROCESSOR_max sanity check.
21955 * config/s390/s390-opts.h (enum processor_type): Add
21957 * config/s390/s390.c (processor_table): Add arch12.
21958 (s390_expand_builtin): Add check for B_VXE flag.
21959 (s390_issue_rate): Add PROCESSOR_ARCH12.
21960 (s390_get_sched_attrmask): Likewise.
21961 (s390_get_unit_mask): Likewise.
21962 (s390_sched_score): Enable z13 scheduling for arch12.
21963 (s390_sched_reorder): Likewise.
21964 (s390_sched_variable_issue): Likewise.
21965 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21967 (s390_tune_attr): Use z13 scheduling also for arch12.
21968 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21969 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21970 (TARGET_VXE_P): New macros.
21971 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21972 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21973 * config/s390/s390.opt: Add arch12 as processor_type.
21975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21977 * config/s390/s390.md
21978 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21979 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21980 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21982 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21983 Rename expanders to ...
21985 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21986 ("fixuns_truncdddi2_emu"): ... these.
21988 ("fixuns_trunc<mode>si2_emu"): New expander.
21990 ("*fixuns_truncdfdi2_z13"): Rename to ...
21991 ("*fixuns_truncdfdi2_vx"): ... this.
21993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21995 * config/s390/2964.md: Remove the single element vector compare
21996 instructions which are no longer used.
21997 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21999 (s390_canonicalize_comparison): Remove handling of DFmode
22001 (s390_expand_vec_compare_scalar): Remove function.
22002 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22003 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22005 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22007 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22009 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22011 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22012 will anyway by matched by mov<mode>_64dfp.
22014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22016 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22017 vlef/vstef. Add missing operand to vleif.
22019 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22021 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22022 pair for all vector types with 64 bit elements.
22023 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22024 * config/s390/vector.md (V_HW_64): ... here.
22025 (V_128_NOSINGLE): New mode iterator.
22026 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22027 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22028 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22029 ("*vec_load_pairv2di"): Change to ...
22030 ("*vec_load_pair<mode>"): ... this one.
22032 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22034 * config/s390/constraints.md: Add comments.
22035 (jKK): Reject element sizes > 8 bytes.
22036 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22038 * config/s390/s390.md: Add the s_operand checks formerly in
22039 s390_split_ok_p to various splitters where they are still
22041 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22042 for 128 bit vectors. Plus two splitters.
22044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22046 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22049 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22052 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22053 error if the boundary argument is not constant.
22055 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22057 PR rtl-optimization/80112
22058 * loop-doloop.c (doloop_condition_get): Don't check condition
22059 if cmp isn't SET with IF_THEN_ELSE src.
22061 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22063 PR tree-optimization/80158
22064 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22065 replacing a candidate statement, also replace it for the
22066 candidate's alternate interpretation.
22067 (replace_rhs_if_not_dup): Likewise.
22068 (replace_one_candidate): Likewise.
22070 2017-03-24 Richard Biener <rguenther@suse.de>
22072 PR tree-optimization/80167
22073 * graphite-isl-ast-to-gimple.c
22074 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22076 (translate_isl_ast_to_gimple::get_rename): Likewise.
22078 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22080 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22081 handling of certain combinations of target options, including the
22082 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22083 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22085 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22088 * config/arm/arm.md (*load_multiple): Add reload_completed to
22089 matching condition.
22091 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22092 Richard Biener <rguenth@suse.de>
22094 PR tree-optimization/79908
22095 PR tree-optimization/80136
22096 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22097 been cast away, gimplify_and_add suffices.
22099 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22101 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22103 2017-03-23 Richard Biener <rguenther@suse.de>
22105 PR tree-optimization/80032
22106 * gimplify.c (gimple_push_cleanup): Forced unconditional
22107 cleanups still have to go to the conditional_cleanups
22110 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22112 PR tree-optimization/80072
22113 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22115 (next_operand_entry_id): Change type to unsigned int.
22116 (sort_by_operand_rank): Make sure to return the right return value
22117 even if unsigned fields are bigger than INT_MAX.
22118 (struct oecount): Change cnt and id type to unsigned int.
22119 (oecount_hasher::equal): Formatting fix.
22120 (oecount_cmp): Make sure to return the right return value
22121 even if unsigned fields are bigger than INT_MAX.
22122 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22125 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22126 TREE_READONLY on result if writing it more than once.
22129 * doc/invoke.texi (-fsanitize=thread): Document that with
22130 -fnon-call-exceptions atomics are not able to throw
22134 * tsan.c: Include tree-eh.h.
22135 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22136 maybe_clean_or_replace_eh_stmt where needed.
22137 (instrument_memory_accesses): Add cfg_changed argument.
22138 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22139 if it returned true.
22140 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22142 PR rtl-optimization/63191
22143 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22144 wrapper function, moved the whole old content into ...
22145 (ix86_delegitimize_address_1): ... this. New inline function.
22146 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22147 true as last argument instead of ix86_delegitimize_address.
22149 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22151 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22152 cortexa57_branch_cost.
22154 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22156 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22158 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22161 * doc/md.texi (Constraints): Document wA constraint.
22162 * config/rs6000/constraints.md (wA): New.
22163 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22164 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22165 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22166 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22168 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22171 * gimplify.c (is_oacc_declared): New function.
22172 (oacc_default_clause): Use it to set default flags for acc declared
22173 variables inside parallel regions.
22174 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22175 declared variables.
22176 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22177 declare attribute to any decl as necessary.
22179 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22182 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22183 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22184 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22185 (arm_arch_lpae): This.
22186 * config/arm/arm.c (arm_arch7ve): Rename into ...
22187 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22188 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22191 2017-03-22 Martin Liska <mliska@suse.cz>
22194 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22195 error message instead of an ICE.
22197 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22199 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22201 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22203 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22205 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22208 2017-03-21 Martin Sebor <msebor@redhat.com>
22210 * doc/extend.texi: Use "cannot" instead of "can't."
22211 * doc/hostconfig.texi: Same.
22212 * doc/install.texi: Same.
22213 * doc/invoke.texi: Same.
22214 * doc/loop.texi: Same.
22215 * doc/md.texi: Same.
22216 * doc/objc.texi: Same.
22217 * doc/rtl.texi: Same.
22218 * doc/tm.texi: Same.
22219 * doc/tm.texi.in: Same.
22220 * doc/trouble.texi: Same.
22222 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22225 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22226 (collect_checksum_attributes): Set it.
22227 (die_checksum_ordered): Use it.
22229 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22231 PR tree-optimization/79908
22232 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22233 change: For a VA_ARG whose LHS has been cast away, use
22234 force_gimple_operand to construct the side effects.
22236 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22238 PR translation/80001
22239 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22240 more amenable to translation.
22241 (oacc_loop_auto_partitions): Likewise.
22243 2017-03-21 Marek Polacek <polacek@redhat.com>
22244 Martin Sebor <msebor@redhat.com>
22246 PR tree-optimization/80109
22247 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22248 on INTEGRAL_TYPE_P.
22250 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22251 Segher Boessenkool <segher@kernel.crashing.org>
22254 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22255 check reg_used_between_p between insn and one of succ or succ2
22256 depending on if succ is artificial insn not inserted into insn
22259 2017-03-21 Martin Liska <mliska@suse.cz>
22261 PR gcov-profile/80081
22262 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22263 * doc/gcc.texi: Include gcov-dump stuff.
22264 * doc/gcov-dump.texi: New file.
22266 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22268 PR rtl-optimization/79150
22269 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22270 conditional jump, if the jump is the last insn of the loop.
22272 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22273 Richard Biener <rguenth@suse.de>
22275 PR tree-optimization/79908
22276 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22277 been cast away, use force_gimple_operand to construct the side
22280 2017-03-21 Martin Liska <mliska@suse.cz>
22282 PR libfortran/79956
22283 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22286 2017-03-21 Brad Spengler <spender@grsecurity.net>
22289 * plugin.c (htab_hash_plugin): New function.
22290 (add_new_plugin): Use it and adjust.
22291 (parse_plugin_arg_opt): Adjust.
22292 (init_one_plugin): Likewise.
22294 2017-03-21 Richard Biener <rguenther@suse.de>
22296 PR tree-optimization/80032
22297 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22298 if set force the cleanup to happen unconditionally.
22299 (gimplify_target_expr): Push inserted clobbers with force_uncond
22300 to avoid them being removed by control-dependent DCE.
22302 2017-03-21 Richard Biener <rguenther@suse.de>
22304 PR tree-optimization/80122
22305 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22306 va_arg_pack_len when the inlined call stmt requires pack
22308 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22310 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22313 * tsan.c (instrument_builtin_call): If the memory model argument
22314 is not a constant, assume it is valid.
22317 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22320 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22322 PR rtl-optimization/79910
22323 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22324 if its dest is used by an insn before I2 (other than the combined
22325 insns themselves, which are properly handled already).
22327 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22330 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22332 * combine.c (record_used_regs): New static function.
22333 (try_combine): Handle situations where there is an additional
22334 instruction between I2 and I3 which needs to have a LOG_LINK
22338 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22340 * combine.c (try_combine): Delete redundant i1 test. Call
22341 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22343 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22346 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22347 alternatives 13/14.
22349 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22351 PR tree-optimization/80054
22352 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22353 the optimization if a PHI or any of its arguments is not dominated
22354 by the candidate's basis. Use gphi* rather than gimple* as
22356 (replace_profitable_candidates): Clean up a gimple* variable that
22357 should be a gphi* variable.
22359 2017-03-20 Martin Sebor <msebor@redhat.com>
22362 * doc/extend.texi (attribute constructor): Document present limitation.
22364 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22367 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22368 __POWER9_VECTOR__ #ifdef control, change template definition to
22369 use Power9-specific built-in function.
22370 (vec_any_eq): Likewise.
22371 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22372 to control outcomes from this test.
22373 (vector_ae_<mode>p): For VEC_F modes, likewise.
22375 2017-03-20 Ian Lance Taylor <iant@google.com>
22377 * config/i386/i386.c (ix86_function_regparm): Save an extra
22378 register for -fsplit-stack with DECL_STATIC_CHAIN.
22380 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22383 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22384 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22386 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22388 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22390 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22393 2017-03-20 Marek Polacek <polacek@redhat.com>
22396 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22398 2017-03-20 Richard Biener <rguenther@suse.de>
22400 PR tree-optimization/80113
22401 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22402 allocate extra SSA name for PHI def.
22403 (add_close_phis_to_outer_loops): Likewise.
22404 (add_close_phis_to_merge_points): Likewise.
22405 (copy_loop_close_phi_args): Likewise.
22406 (copy_cond_phi_nodes): Likewise.
22408 2017-03-20 Martin Liska <mliska@suse.cz>
22410 PR middle-end/79753
22411 * tree-chkp.c (chkp_build_returned_bound): Do not build
22412 returned bounds for a LHS that's not a BOUNDED_P type.
22414 2017-03-20 Martin Liska <mliska@suse.cz>
22418 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22419 COMPLEX_CST and VECTOR_CST.
22421 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22424 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22425 target operand. A new splitter adds the clobber statement in case
22426 the target operand is dead anyway.
22428 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22430 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22431 to age-old versions of binutils and glibc.
22433 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22435 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22437 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22439 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22441 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22443 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22444 requirement for binutils 2.13.
22446 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22448 * combine.c (try_combine): Delete redundant i1 test. Call
22449 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22451 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22453 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22454 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22456 <riscv64-*-elf>: Re-arrange section
22457 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22458 <riscv32-*-linux>: Likewise.
22459 <riscv64-*-elf>: Likewise
22460 <riscv64-*-linux>: Likewise.
22462 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22465 * aarch64.opt(verbose-cost-dump): Fix typo.
22467 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22470 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22471 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22473 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22475 * reload.c (find_reloads): When reloading a nonoffsettable address,
22476 use RELOAD_OTHER for it and its address reloads.
22478 PR rtl-optimization/79910
22479 * combine.c (record_used_regs): New static function.
22480 (try_combine): Handle situations where there is an additional
22481 instruction between I2 and I3 which needs to have a LOG_LINK
22484 2017-03-17 Jeff Law <law@redhat.com>
22486 PR tree-optimization/71437
22487 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22488 conditional in the hash table first.
22489 (vrp_dom_walker::before_dom_children): Extract condition from
22490 ASSERT_EXPR. Record condition, its inverion and any implied
22491 conditions as well.
22493 2017-03-17 Marek Polacek <polacek@redhat.com>
22494 Markus Trippelsdorf <markus@trippelsdorf.de>
22496 PR tree-optimization/80079
22497 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22500 2017-03-17 Richard Biener <rguenther@suse.de>
22502 PR middle-end/80075
22503 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22504 Properly verify the LHS before the RHS possibly claims to be
22506 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22509 2017-03-17 Martin Jambor <mjambor@suse.cz>
22511 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22512 (List of -O2 options): Likewise.
22513 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22516 2017-03-17 Tom de Vries <tom@codesourcery.com>
22518 * gcov-dump.c (print_usage): Print bug_report_url.
22520 2017-03-17 Richard Biener <rguenther@suse.de>
22522 PR middle-end/80050
22523 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22524 (parser::peek): Likewise.
22526 2017-03-17 Richard Biener <rguenther@suse.de>
22528 PR tree-optimization/80048
22529 * sese.c (free_sese_info): Properly release rename_map and
22530 copied_bb_map elements.
22532 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22534 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22535 Add linked-list forward and backlinks. Insert on
22536 construction, remove on destruction.
22537 (class pass_store_merging): Add m_stores_head field.
22538 (pass_store_merging::terminate_and_process_all_chains):
22539 Iterate over m_stores_head list.
22540 (pass_store_merging::terminate_all_aliasing_chains):
22542 (pass_store_merging::execute): Check for debug stmts first.
22543 Push new chains onto the m_stores_head stack.
22545 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22548 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22549 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22550 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22552 2017-03-16 Jeff Law <law@redhat.com>
22554 PR tree-optimization/71437
22555 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22556 member function. Implementation moved into after_dom_children
22557 member function and into the threader's thread_outgoing_edges
22559 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22560 some code into new thread_outgoing_edges.
22561 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22562 definition. Simplify marker handling (do it here). Assume we always
22563 have the available expression and the const/copies tables.
22564 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22566 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22567 * tree-vrp.c (equiv_stack): No longer file scoped.
22568 (vrp_dom_walker): New class.
22569 (vrp_dom_walker::before_dom_children): New member function.
22570 (vrp_dom_walker::after_dom_children): Likewise.
22571 (identify_jump_threads): Setup domwalker. Use it rather than
22572 walking edges in a random order by hand. Simplify setup/finalization.
22573 (finalize_jump_threads): Remove.
22574 (vrp_finalize): Do not call identify_jump_threads here.
22575 (execute_vrp): Do it here instead and call thread_through_all_blocks
22578 PR tree-optimization/71437
22579 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22581 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22583 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22584 (dom_opt_dom_walker::thread_across_edge): Remove
22585 handle_dominating_asserts argument. All callers changed.
22586 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22587 changes. Remove calls to lhs_of_dominating_assert. Other
22588 uses of handle_dominating_asserts turn into unconditional code
22589 (simplify_control_stmt_condition_1): Likewise.
22590 (simplify_control_stmt_condition): Likewise.
22591 (thread_through_normal_block, thread_across_edge): Likewise.
22592 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22593 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22594 object if it is not an SSA_NAME.
22595 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22596 before calling into the VRP specific simplifiers.
22597 (identify_jump_threads): Remove handle_dominating_asserts
22600 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22603 * tree-diagnostic.c (default_tree_printer): No longer static.
22604 * tree-diagnostic.h (default_tree_printer): New prototype.
22606 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22608 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22609 Change ins into fmov.
22611 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22613 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22614 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22615 Use h_con constraint for operand 1.
22616 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22617 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22619 2017-03-15 Jeff Law <law@redhat.com>
22621 PR tree-optimization/71437
22622 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22623 (record_temporary_equivalences): Use it.
22625 PR tree-optimization/71437
22626 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22627 tree-ssa-scopedtables.
22628 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22629 (record_conditions, record_cond, vuse_eq): Likewise.
22630 (record_edge_info): Adjust to API tweak of record_conditions.
22631 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22632 (record_temporary_equivalences, optimize_stmt): Likewise.
22633 (eliminate_redundant_computations): Likewise.
22634 (record_equivalences_from_stmt): Likewise.
22635 * tree-ssa-scopedtables.c: Include options.h and params.h.
22636 (vuse_eq): New function, moved from tree-ssa-dom.c
22637 (build_and_record_new_cond): Likewise.
22638 (record_conditions): Likewise. Accept vector of conditions rather
22639 than edge_equivalence structure for first argument.
22640 for the first argument.
22641 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22642 from tree-ssa-dom.c.
22643 (avail_exprs_stack::record_cond): Likewise.
22644 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22645 from tree-ssa-dom.c.
22646 (avail_exprs_stack): Add new member functions lookup_avail_expr
22648 (record_conditions): Declare.
22650 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22653 * lra-constraints.c (process_alt_operands): Increase reject for
22654 reloading an input/output operand.
22656 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22659 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22660 insns to convert from signed/unsigned char/short to IEEE 128-bit
22662 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22664 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22667 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22668 subreg of inner mode for values already in registers.
22670 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22672 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22673 iteration reg is used after the loop.
22675 2017-03-14 Martin Sebor <msebor@redhat.com>
22677 PR tree-optimization/79800
22678 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22679 precision in negative-positive range.
22680 (format_floating): Call non-const overload with adjusted precision.
22682 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22685 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22688 2017-03-14 Martin Sebor <msebor@redhat.com>
22690 PR middle-end/80020
22691 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22692 * builtins.def (aligned_alloc): Use it.
22695 * Makefile.in (GTFILES): Add calls.c.
22696 * calls.c: Include "gt-calls.h".
22698 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22700 PR rtl-optimization/79728
22701 * regs.h (struct target_regs): New field
22702 x_contains_allocatable_regs_of_mode.
22703 (contains_allocatable_regs_of_mode): New macro.
22704 * reginfo.c (init_reg_sets_1): Initialize it, and change
22705 contains_reg_of_mode so it includes global regs as well.
22706 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22707 rather than contains_regs_of_mode.
22709 2017-03-14 Martin Liska <mliska@suse.cz>
22711 * doc/invoke.texi: Document options that can't be combined with
22712 -fcheck-pointer-bounds.
22714 2017-03-14 Martin Liska <mliska@suse.cz>
22716 PR middle-end/79831
22717 * doc/invoke.texi (-Wchkp): Document the option.
22719 2017-03-14 Martin Liska <mliska@suse.cz>
22721 * Makefile.in: Install gcov-dump.
22723 2017-03-14 Martin Liska <mliska@suse.cz>
22725 * multiple_target.c (expand_target_clones): Bail out for
22726 an invalid attribute.
22728 2017-03-14 Richard Biener <rguenther@suse.de>
22730 * alias.c (struct alias_set_entry): Pack properly.
22731 * cfgloop.h (struct loop): Likewise.
22732 * cse.c (struct set): Likewise.
22733 * ipa-utils.c (struct searchc_env): Likewise.
22734 * loop-invariant.c (struct invariant): Likewise.
22735 * lra-remat.c (struct cand): Likewise.
22736 * recog.c (struct change_t): Likewise.
22737 * rtl.h (struct address_info): Likewise.
22738 * symbol-summary.h (function_summary): Likewise.
22739 * tree-loop-distribution.c (struct partition): Likewise.
22740 * tree-object-size.c (struct object_size_info): Likewise.
22741 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22742 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22743 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22744 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22745 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22746 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22747 (struct _stmt_vec_info): Likewise.
22749 2017-03-14 Martin Liska <mliska@suse.cz>
22752 * multiple_target.c (create_dispatcher_calls): Check that
22753 a target can create a function dispatcher.
22755 2017-03-14 Martin Liska <mliska@suse.cz>
22758 * multiple_target.c (expand_target_clones): Drop local.local
22759 flag for default implementation.
22761 2017-03-14 Richard Biener <rguenther@suse.de>
22763 PR tree-optimization/80030
22764 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22766 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22768 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22769 gcc_fallthrough() instead of __attribute__((fallthrough));
22771 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22773 * doc/gcc.texi: Remove "up" link to (DIR).
22774 * doc/gccint.texi: Ditto.
22776 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22778 * doc/install.texi (Specific) <avr>: Remove reference to
22781 2017-03-13 Jeff Law <law@redhat.com>
22783 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22784 attribute rather than comments.
22786 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22787 match_scratch operand is highest.
22789 2017-03-13 Martin Liska <mliska@suse.cz>
22791 PR middle-end/78339
22792 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22793 is a CHKP clone, use original declaration.
22795 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22797 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22798 (arc_conditional_register_usage): Use a different allocation order
22799 when optimizing for size.
22800 * common/config/arc/arc-common.c (arc_option_optimization_table):
22801 Section anchors default on when optimizing for size.
22803 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22805 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22807 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22809 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22810 * config/arc/arc.md (cpu_facility): Add cd variant.
22811 (*movqi_insn): Add code density variant.
22812 (*movhi_insn): Likewise.
22813 (*movqi_insn): Likewise.
22814 (*addsi3_mixed): Likewise.
22815 (subsi3_insn): Likewise.
22817 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22819 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22821 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22823 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22824 expressions with MINUS and UNARY ops.
22826 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22829 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22831 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22832 between vec_select and vector argument.
22833 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22834 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22835 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22836 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22837 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22838 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22840 2017-03-13 Richard Biener <rguenther@suse.de>
22843 * params.def (vect-max-peeling-for-alignment): Fix typo.
22845 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22847 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22848 issue that only occurred with binutils below 2.18.
22850 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22852 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22853 refer to binutils 2.11/2.12 minimum.
22855 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22857 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22858 ftp.kernel.org and simplify binutils requirement.
22860 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22862 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22864 (Optimize Options): Ditto. Also remove redundancy.
22866 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22868 PR translation/79848
22869 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22871 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22872 to G_ to avoid double translation.
22874 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22876 PR translation/79923
22877 * auto-profile.c (get_combined_location): Convert leading
22878 character of diagnostics to lower case and remove trailing period.
22879 (read_profile): Likewise for various diagnostics.
22880 * config/arm/arm.c (arm_option_override): Remove trailing period
22881 from various diagnostics.
22882 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22883 (msp430_expand_delay_cycles): Likewise.
22885 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22888 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22889 full command-line argument, rather than just "str".
22890 (aarch64_validate_march): Likewise.
22891 (aarch64_validate_mtune): Likewise.
22893 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22895 PR rtl-optimization/78911
22896 * lra-assigns.c (must_not_spill_p): New function.
22897 (spill_for): Use it.
22899 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22901 PR tree-optimization/79981
22902 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22903 ATOMIC_COMPARE_EXCHANGE ifn result.
22904 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22905 IFN_ATOMIC_COMPARE_EXCHANGE.
22907 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22910 * opts.c (parse_sanitizer_options): Add missing question mark to
22911 "did you mean" message.
22913 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22915 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22917 (VMULEUH_UNS): Likewise.
22918 (VMULOUB_UNS): Likewise.
22919 (VMULOUH_UNS): Likewise.
22920 * config/rs6000/rs6000.c (builtin_function_type): Remove
22921 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22923 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22926 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22927 x with result of extra_parsing_for_operand_code_0.
22928 (function_reader::extra_parsing_for_operand_code_0): Convert
22929 return type from void to rtx, returning x. When reading
22930 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22931 larger size containing struct block_symbol.
22933 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22935 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22936 -mfloat128-hardware without -m64.
22938 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22941 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22942 entries to the case statement that marks unsigned arguments to
22943 overloaded functions.
22945 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22948 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22950 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22953 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22954 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22956 2017-03-10 Martin Liska <mliska@suse.cz>
22960 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22961 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22964 2017-03-10 Olivier Hainque <hainque@adacore.com>
22966 * tree-switch-conversion (array_value_type): Start by resetting
22967 candidate type to it's main variant.
22969 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22971 PR rtl-optimization/79909
22972 * combine.c (try_combine): Use simplify_replace_rtx on individual
22973 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22974 of the whole CALL_INSN_FUNCTION_USAGE.
22976 PR tree-optimization/79972
22977 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22978 get_range_info on SSA_NAMEs. Formatting fixes.
22980 2017-03-10 Richard Biener <rguenther@suse.de>
22981 Jakub Jelinek <jakub@redhat.com>
22983 PR tree-optimization/77975
22984 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22985 edge to be constant.
22986 (get_val_for): For constant x return it. Formatting fix.
22987 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22988 would use the same bases as the current one.
22990 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22992 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22993 instead of vec_select for V1TImode.
22994 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22996 (VSX_LE_128): Add V1TI to this mode iterator.
22997 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22998 (*vsx_le_perm_store_<mode>): Likewise.
22999 (pre-reload splitter for VSX stores): Likewise.
23000 (post-reload splitter for VSX stores): Likewise.
23001 (*vsx_xxpermdi2_le_<mode>): Likewise.
23002 (*vsx_lxvd2x2_le_<mode>): Likewise.
23003 (*vsx_stxvd2x2_le_<mode>): Likewise.
23005 2017-03-09 Michael Eager <eager@eagercon.com>
23007 Correct failures with --enable-checking=yes,rtl.
23009 * config/microblaze/microblaze.c (microblaze_expand_shift):
23010 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23011 test for const0_rtx.
23012 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23013 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23015 2017-03-09 Richard Biener <rguenther@suse.de>
23017 PR tree-optimization/79977
23018 * graphite-scop-detection.c (scop_detection::merge_sese):
23019 Handle the case of extra exits to blocks dominating the entry.
23021 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23023 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23026 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23028 PR rtl-optimization/79949
23029 * lra-constraints.c (process_alt_operands): Check memory when
23030 trying to predict a cycle. Print about the overall increase.
23032 2017-03-09 Richard Biener <rguenther@suse.de>
23034 PR middle-end/79971
23035 * gimple-expr.c (useless_type_conversion_p): Preserve
23036 TYPE_SATURATING for fixed-point types.
23038 2017-03-09 Richard Biener <rguenther@suse.de>
23041 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23042 alignment of BLKmode params.
23044 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23047 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23048 (VALL_NO_V2Q): Likewise.
23050 * config/aarch64/aarch64-simd.md
23051 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23053 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23054 VALL_NO_V2Q mode iterator.
23055 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23057 2017-03-09 Martin Liska <mliska@suse.cz>
23059 PR tree-optimization/79631
23060 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23061 tree_int_cst_sign_bit just for INTEGER constants.
23063 2017-03-09 Martin Liska <mliska@suse.cz>
23067 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23070 2017-03-09 Marek Polacek <polacek@redhat.com>
23073 * tree.c (inchash::add_expr): Handle TREE_VEC.
23075 2017-03-09 Martin Liska <mliska@suse.cz>
23078 (chkp_narrow_size_and_offset): New function.
23079 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23080 (void chkp_parse_bit_field_ref): New function.
23081 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23082 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23084 2017-03-09 Martin Liska <mliska@suse.cz>
23087 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23088 (chkp_find_bounds_1): Remove gcc_unreachable.
23090 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23093 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23094 BUILT_IN_SYNC*, determine the access type from the size suffix and
23095 always build a MEM_REF with that type. Handle forgotten
23096 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23099 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23100 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23101 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23102 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23103 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23104 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23105 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23106 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23107 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23108 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23109 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23110 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23111 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23112 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23113 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23114 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23115 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23116 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23117 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23118 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23119 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23120 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23121 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23122 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23123 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23124 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23125 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23126 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23127 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23128 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23129 definitions outside of __OPTIMIZE__ guarded section.
23132 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23133 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23134 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23135 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23138 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23140 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23141 ("vfenez<mode>"): Add missing constraints.
23143 2017-03-08 Martin Sebor <msebor@redhat.com>
23146 * config/nds32/nds32.c (nds32_option_override):
23147 Fix misspelled diagnostic.
23149 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23152 * gimplify.c (gimplify_omp_for): Replace index var in outer
23153 taskloop statement with an artificial variable and add
23154 OMP_CLAUSE_PRIVATE clause for it.
23156 2017-03-08 Richard Biener <rguenther@suse.de>
23158 PR tree-optimization/79955
23159 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23160 for accesses that are completely outside of the variable.
23162 2017-03-08 Andrew Haley <aph@redhat.com>
23164 PR tree-optimization/79943
23165 * tree-ssa-loop-split.c (compute_new_first_bound): When
23166 calculating the new upper bound, (END-BEG) should be added, not
23169 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23171 * config/avr/avr.md (setmemhi): Make sure match_dup
23172 operand number comes before match_scratch.
23174 2017-03-08 Richard Biener <rguenther@suse.de>
23176 PR tree-optimization/79920
23177 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23178 with ncopies == 1 to ...
23179 (vect_transform_slp_perm_load): ... here. Properly compute
23180 all element loads by iterating VF times over the group. Do
23181 not handle ncopies (computed in a broken way) in
23182 vect_create_mask_and_perm.
23184 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23187 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23188 is a uniform vector, use uniform_vector_p return value instead of
23189 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23191 2017-03-07 Marek Polacek <polacek@redhat.com>
23193 PR middle-end/79809
23194 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23195 (alloca_call_type): Likewise.
23197 2017-03-07 Martin Liska <mliska@suse.cz>
23199 * gcov.c (process_args): Put comment to correct location.
23201 2017-03-07 Martin Liska <mliska@suse.cz>
23203 PR middle-end/68270
23204 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23205 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23206 (chkp_narrow_bounds_for_field): Likewise.
23207 (chkp_parse_array_and_component_ref): Pass one more argument to
23210 2017-03-07 Richard Biener <rguenther@suse.de>
23212 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23215 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23217 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23218 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23220 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23223 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23224 to end of description.
23225 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23227 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23229 PR rtl-optimization/79901
23230 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23232 (*avx512f_<code><mode>3<mask_name>): ... this.
23233 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23234 iterator instead of VI8_AVX2_AVX512BW.
23236 PR rtl-optimization/79901
23237 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23238 min/max expander, expand it using expand_vec_cond_expr.
23241 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23244 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23247 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23248 to void * for PCH reasons.
23249 * dwarf2out.c (output_loc_operands, output_die): Cast
23250 v.val_vec.array to unsigned char *.
23252 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23255 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23258 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23260 PR rtl-optimization/79571
23261 * lra-constraints.c (process_alt_operands): Calculate static
23262 reject and subtract it from overall when only addresses will be
23265 2017-03-06 Julia Koval <julia.koval@intel.com>
23268 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23269 incoming stack boundary to 128 for 64-bit targets.
23271 2017-03-06 Richard Biener <rguenther@suse.de>
23273 PR tree-optimization/79894
23274 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23275 to NULL after folding it.
23277 2017-03-06 Richard Biener <rguenther@suse.de>
23279 PR tree-optimization/79824
23280 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23281 check disabling peeling for gaps.
23283 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23285 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23286 attributes): Document gettimeofday.
23288 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23290 * config/s390/s390.c (s390_option_override_internal): Set
23291 PARAM_MIN_VECT_LOOP_BOUND
23293 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23295 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23296 * config/s390/s390.md: Likewise.
23298 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23301 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23302 (<avx2_avx512>_perm<mode>): Rename to ...
23303 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23305 (<avx512>_perm<mode>_mask): Rename to ...
23306 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23308 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23309 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23310 instead of VI8F_256_512.
23311 (avx512f_perm<mode>): New define_expand.
23312 (avx512f_perm<mode>_mask): Likewise.
23313 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23314 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23316 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23318 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23319 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23321 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23323 2017-03-06 Martin Liska <mliska@suse.cz>
23326 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23327 when having a SSA NAME w/o VAR_DECL assigned to it.
23329 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23331 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23332 msa_dpsub_<su>_d): Fix MODE for vec_select.
23334 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23336 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23338 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23340 2017-03-06 Richard Biener <rguenther@suse.de>
23342 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23343 * plugin.c (register_plugin_info): Likewise.
23344 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23346 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23348 * config/i386/sse.md (sse_storehps, sse_storelps,
23349 avx_<castmode><avxsizesuffix>_<castmode>,
23350 avx512f_<castmode><avxsizesuffix>_<castmode>,
23351 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23352 in condition that at least one operand is not a MEM.
23354 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23356 PR middle-end/79805
23357 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23358 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23360 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23361 gimple_call_nothrow_p flag based on whether original builtin can throw.
23362 If it can, emit following stmts on the fallthrough edge.
23363 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23364 don't create new bb if inserting just debug stmts on the edge, try to
23365 insert them on the fallthru bb or just reset debug stmts.
23367 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23370 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23371 restore recog_data (including the operand rtxes inside it) around
23372 the call to get_insn_template.
23374 2017-03-03 Martin Sebor <msebor@redhat.com>
23376 PR tree-optimization/79699
23377 * context.c (context::~context): Free MPFR caches to avoid
23378 a memory leak on program exit.
23380 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23382 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23383 Use wide_int::ulow () instead of .elt (0).
23385 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23387 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23388 (*pushxf): Limit oF constraint to 32bit targets and add oC
23389 constraint for 64bit targets.
23390 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23391 (*pushdf): Change rmF constraint to rmC.
23393 2017-03-03 Martin Liska <mliska@suse.cz>
23395 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23396 Remove unused variable.
23398 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23401 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23402 is a memory operand, increase num_memory.
23403 (ix86_expand_args_builtin): Likewise.
23405 2017-03-03 Jan Hubicka <jh@suse.cz>
23408 * ipa-devirt.c (maybe_record_node): Properly handle
23409 __cxa_pure_virtual visibility.
23411 2017-03-03 Martin Liska <mliska@suse.cz>
23413 PR tree-optimization/79803
23414 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23416 (pass_loop_prefetch::execute): Disabled optimization if an
23417 assumption about L1 cache size is not met.
23419 2017-03-03 Martin Liska <mliska@suse.cz>
23421 PR rtl-optimization/79574
23422 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23423 (hash_scan_set): Likewise.
23424 (dump_hash_table): Likewise.
23425 (hoist_code): Likewise.
23427 2017-03-03 Richard Biener <rguenther@suse.de>
23429 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23430 in place of uhigh ().
23431 (fixed_convert_from_real): Likewise.
23433 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23436 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23438 2017-03-03 Richard Biener <rguenther@suse.de>
23440 PR middle-end/79818
23441 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23442 TYPE_OVERFLOW_UNDEFINED check.
23444 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23446 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23448 (vector_ae_<mode>_p): Likewise.
23449 (vector_nez_<mode>_p): Likewise.
23450 (vector_ne_v2di_p): Likewise.
23451 (vector_ae_v2di_p): Likewise.
23452 (vector_ne_<mode>_p): Likewise.
23453 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23455 (vsx_tsqrt<mode>2_fe): Likewise.
23457 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23460 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23462 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23464 PR rtl-optimization/79780
23465 * cprop.c (one_cprop_pass): When second and further conditional trap
23466 in a single basic block is turned into an unconditional trap, turn it
23467 into a deleted note to avoid RTL verification failures.
23469 2017-03-02 Richard Biener <rguenther@suse.de>
23471 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23473 2017-03-02 Richard Biener <rguenther@suse.de>
23475 PR tree-optimization/79345
23477 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23478 param and abort the walk, returning -1 if it is hit.
23479 (walk_aliased_vdefs): Take a limit param and pass it on.
23480 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23481 defaulting to 0 and return a signed int.
23482 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23483 (check_defs): New helper.
23484 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23485 about uninitialized memory.
23486 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23487 bogus uninitialized warning.
23488 (fixed_convert_from_real): Likewise.
23490 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23492 PR tree-optimization/66768
23493 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23494 iv_use if base object can't be determined.
23496 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23498 PR tree-optimization/79345
23499 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23500 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23501 (get_pattern_stats): Initialize it.
23502 * genemit.c (gen_expand): Verify match_scratch numbers come after
23503 match_operand/match_dup numbers.
23504 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23505 match_scratch numbers.
23506 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23508 * config/s390/s390.md (trunctdsd2): Likewise.
23510 2017-03-02 Richard Biener <rguenther@suse.de>
23512 * wide-int.h (wide_int_storage::operator=): Implement in terms
23515 2017-03-02 Richard Biener <rguenther@suse.de>
23517 PR tree-optimization/79777
23518 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23519 the to insert expression to sth existing.
23521 2017-03-01 Martin Sebor <msebor@redhat.com>
23523 PR middle-end/79692
23524 * gimple-ssa-sprintf.c
23525 (directive::known_width_and_precision): New function.
23526 (format_integer): Use it.
23527 (get_mpfr_format_length): Consider the full range of precision
23528 when computing %g output with the # flag. Set the likely byte
23529 count to 3 rather than 1 when precision is indeterminate.
23530 (format_floating): Correct the lower bound of precision.
23532 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23534 * doc/invoke.texi: Document default code model for 64-bit Linux.
23536 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23539 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23540 udiv rather than div since input pattern is unsigned.
23542 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23544 * config/i386/i386.c (print_reg): Warn for values of
23545 unsupported size in integer register.
23547 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23550 * config/rs6000/predicates.md (current_file_function_operand): Do
23551 not allow self calls to be local if the function is replaceable.
23553 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23556 * config/rs6000/altivec.h (vec_ctz and others): Change the
23557 preprocessor macro that controls conditional compilation from
23558 _ARCH_PWR9 to __POWER9_VECTOR__.
23559 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23560 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23561 control (instead of _ARCH_PWR9 control) so that template
23562 definition uses power9-specific function.
23563 (vec_any_eq): Likewise.
23564 (vec_all_ne): Change macro definition to use a power9-specific
23565 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23566 _ARCH_PWR9 control).
23567 (vec_any_eq) Likewise.
23568 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23569 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23570 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23571 support for xvcmpnedp instruction.
23572 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23573 macro expansion so that Power9 implementation of vec_all_ne does
23574 not use the AltiVec predicate framework.
23575 (VCMPNEH_P): Likewise.
23576 (VCMPNEW_P): Likewise.
23577 (VCMPNED_P): Likewise.
23578 (VCMPNEFP_P): Likewise.
23579 (VCMPNEDP_P): Likewise.
23580 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23581 implementation of vec_any_eq to not use AltiVec predicate
23583 (VCMPAEH_P): Likewise.
23584 (VCMPAEW_P): Likewise.
23585 (VCMPAED_P): Likewise.
23586 (VCMPAEFP_P): Likewise.
23587 (VCMPAEDP_P): Likewise.
23588 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23589 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23590 not use the AltiVec predicate framework.
23591 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23592 of vec_any_eq to not use AltiVec predicate framework.
23593 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23594 support for predefined __POWER9_VECTOR__ macro to indicate that
23595 Power9 instruction selection is enabled.
23596 (altivec_overloaded_builtins): Remove extraneous
23597 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23598 function argument types RS6000_BTI_bool_V16QI and
23599 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23600 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23601 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23602 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23603 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23604 Power9 for implementations of vec_cmpne. Change the signature for
23605 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23606 (representing vec_all_ne) to remove the previously described first
23607 argument of type RS6000_BTI_INTSI, as this was an artifact of
23608 reliance on the AltiVec predicate framework, which is no longer
23609 used in the implementation of these functions. Add
23610 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23611 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23612 since, unlike the AltiVec predicate framework implementation, we
23613 do not share function descriptors between vec_alle and vec_anyeq.
23614 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23615 set of modes that receive special treatment even when
23616 TARGET_P9_VECTOR is true. The special treatment emits code that
23617 does not depend on Power9 instructions.
23618 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23619 define_expand to not rely on AltiVec predicate framework.
23620 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23622 (vector_ne_v2di_p): Change this define_expand to not rely on
23623 AltiVec predicate framework.
23624 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23626 (vector_ne_<mode>_p): Change this define_expand to not rely on
23627 AltiVec predicate framework.
23628 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23630 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23631 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23632 define_insn pattern.
23633 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23634 define_insn pattern because the xvcmpne<VSs>. instruction is not
23636 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23637 instruction is not supported.
23639 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23641 * config/nvptx/nvptx.c: Include intl.h.
23643 2017-03-01 Martin Jambor <mjambor@suse.cz>
23646 * ipa-prop.h (ipa_bits): Removed field known.
23647 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23648 to pointers. Adjusted their comments to warn about their sharing.
23649 (ipcp_transformation_summary): Change bits to a vector of pointers.
23650 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23651 (ipa_get_ipa_bits_for_value): Declare.
23652 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23653 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23654 (ipa_bits_hash_table): Likewise.
23655 (ipa_vr_ggc_hash_traits): Likewise.
23656 (ipa_vr_hash_table): Likewise.
23657 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23658 being pointers and vr_known being removed.
23659 (ipa_set_jf_unknown): Likewise.
23660 (ipa_get_ipa_bits_for_value): New function.
23661 (ipa_set_jfunc_bits): Likewise.
23662 (ipa_get_value_range): New overloaded functions.
23663 (ipa_set_jfunc_vr): Likewise.
23664 (ipa_compute_jump_functions_for_edge): Use the above functions to
23665 construct bits and vr parts of jump functions.
23666 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23667 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23669 (ipcp_grow_transformations_if_necessary): Also allocate
23670 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23672 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23673 them. Fix too long lines.
23674 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23675 vr_known being removed.
23676 (ipa_read_jump_function): Use new setter functions to construct bits
23677 and vr parts of jump functions or set them to NULL.
23678 (write_ipcp_transformation_info): Adjust for bits being pointers.
23679 (read_ipcp_transformation_info): Likewise.
23680 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23682 Include gt-ipa-prop.h.
23683 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23685 (ipcp_store_bits_results): Likewise.
23686 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23687 Do not write to existing jump functions but use a temporary instead.
23689 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23692 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23693 attempt to use its first operand as BIT_FIELD_REF base.
23695 2017-03-01 Richard Biener <rguenther@suse.de>
23697 PR middle-end/79721
23698 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23699 interpolating formula in wrapping arithmetic.
23700 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23702 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23704 PR tree-optimization/79734
23705 * tree-vect-generic.c (expand_vector_condition): Optimize
23706 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23707 Handle VEC_COND_EXPR where comparison has different inner width from
23708 type's inner width.
23710 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23712 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23713 markup, and similar issues. Remove @opindex entries for things
23714 that aren't options. Add missing -mmpy-option entries.
23716 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23718 PR tree-optimization/79737
23719 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23720 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23721 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23722 instead of byte_size. Formatting fix.
23723 (shift_bytes_in_array_right): Formatting fix.
23725 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23728 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23729 condition on optimize for the leaf function test.
23731 2017-02-28 Martin Liska <mliska@suse.cz>
23734 * read-rtl-function.c (function_reader::handle_unknown_directive):
23735 Bail out when one uses -flto.
23737 2017-02-28 Martin Liska <mliska@suse.cz>
23739 * common.opt: Replace space with tabular for options of <number>
23741 * config/i386/i386.opt: Show <number> value for
23742 -mlarge-data-threshold.
23743 * opts.c (print_filtered_help): Do not display number in hexadecimal
23746 2017-02-28 Martin Liska <mliska@suse.cz>
23748 * common.opt: Fix --help=option -Q for options which are of
23751 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23753 * config/i386/i386.c (print_reg): Error out for values
23754 of 8-bit size in invalid integer register.
23756 2017-02-28 Martin Sebor <msebor@redhat.com>
23758 PR tree-optimization/79691
23759 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23761 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23764 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23765 gcc_unreachable with output_operand_lossage.
23767 2017-02-28 Richard Biener <rguenther@suse.de>
23769 PR tree-optimization/79740
23770 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23772 (visit_nary_op): Insert the nary into the hashtable if we
23773 pattern-matched sth.
23774 * tree-ssa-pre.c (eliminate_insert): Robustify.
23776 2017-02-28 Richard Biener <rguenther@suse.de>
23778 PR middle-end/79731
23779 * fold-const.c (decode_field_reference): Reject out-of-bound
23782 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23784 * config/i386/i386.c: Include intl.h.
23785 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23786 instead of just cond ? "..." : "...".
23787 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23788 * coverage.c (read_counts_file): Likewise.
23789 * omp-offload.c: Include intl.h.
23790 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23791 of just cond ? "..." : "...".
23792 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23793 of just cond ? "..." : "...".
23795 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23798 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23800 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23801 'tune for' CPU name.
23802 * config/arm/arm-cpu-data.h: Regenerated.
23804 2017-02-28 Richard Biener <rguenther@suse.de>
23806 PR tree-optimization/79732
23807 * tree-inline.c (expand_call_inline): Do not shadow var.
23809 2017-02-28 Richard Biener <rguenther@suse.de>
23811 PR tree-optimization/79723
23812 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23813 address-space properly.
23815 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23817 * doc/optinfo.texi (Optimization groups): Fix option used for
23819 * doc/invoke.texi (-fopt-info): Document "omp".
23820 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23821 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23822 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23823 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23824 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23826 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23828 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23829 OPTGROUP_OMP with "omp".
23831 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23834 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23835 for arithmetic shift of unsigned V2DI.
23837 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23839 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23840 arc/linux.h headers.
23841 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23842 (LINK_SPEC): Likewise.
23843 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23844 (EXTRA_SPECS): Likewise.
23845 (STARTFILE_SPEC): Likewise.
23846 (ENDFILE_SPEC): Likewise.
23847 (LIB_SPEC): Likewise.
23848 (TARGET_SDATA_DEFAULT): Likewise.
23849 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23850 (MULTILIB_DEFAULTS): Likewise.
23851 (DWARF2_UNWIND_INFO): Likewise.
23852 * config/arc/big.h: New file.
23853 * config/arc/elf.h: Likewise.
23854 * config/arc/linux.h: Likewise.
23855 * config/arc/t-uClibc: Remove.
23857 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23859 PR tree-optimization/77536
23860 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23861 (tree_transform_and_unroll_loop): Use above function to compute the
23862 estimated niter of unrolled loop and use it when scaling profile.
23863 Also use count info rather than frequency if it's non-zero.
23864 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23865 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23866 (vect_transform_loop): Call above function.
23868 2017-02-27 Richard Biener <rguenther@suse.de>
23870 PR tree-optimization/45397
23871 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23872 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23873 (visit_nary_op): Add pattern matching for CSEing sign-changed
23874 or truncated operations with wider ones.
23876 2017-02-27 Richard Biener <rguenther@suse.de>
23878 PR tree-optimization/79690
23879 * tree-vect-stmts.c (vectorizable_store): Use vector type
23880 built from the DR with address-space.
23882 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23884 * doc/invoke.texi (Optimize Options): Refine the description
23885 of asan-use-after-return.
23887 2017-02-25 Alan Modra <amodra@gmail.com>
23889 PR rtl-optimization/79584
23890 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23891 base, not ad->base_term, the reg within base. Remove assertion
23892 that ad->base == ad->base_term. Replace gen_int_mode using
23893 bogus mode with const0_rtx.
23895 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23897 PR middle-end/79396
23898 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23899 FMA_EXPR like tcc_binary or tcc_unary.
23901 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23904 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23906 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23907 (output_loc_operands): Handle DW_OP_call_ref and
23908 DW_OP_GNU_variable_value.
23909 (struct variable_value_struct): New type.
23910 (struct variable_value_hasher): Likewise.
23911 (variable_value_hash): New variable.
23912 (string_types): Remove.
23913 (copy_loc_descr): New function.
23914 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23915 (prepend_loc_descr_to_each): New function.
23916 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23917 instead of add_loc_descr_to_each if the first argument is single
23918 location list and the second has multiple.
23919 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23920 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23921 when looking for variable value which doesn't have other location info.
23922 (loc_list_from_tree): Formatting fix.
23923 (gen_array_type_die): Simplify DW_AT_string_length handling.
23924 (adjust_string_types): Remove.
23925 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23926 string_types. Call resolve_variable_values.
23927 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23928 (resolve_addr_in_expr): Likewise. Add A argument.
23929 (copy_deref_exprloc): Remove deref argument. Adjust for the
23930 original expression being DW_OP_GNU_variable_value with optionally
23931 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23932 optionally after it.
23933 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23934 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23935 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23936 (variable_value_hasher::hash, variable_value_hasher::equal): New
23938 (resolve_variable_value_in_expr, resolve_variable_value,
23939 resolve_variable_values, note_variable_value_in_expr,
23940 note_variable_value): New functions.
23941 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23944 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23947 * opts.h (handle_generated_option): Add GENERATED_P argument.
23948 * opts-common.c (handle_option): Adjust function comment.
23949 (handle_generated_option): Add GENERATED_P argument, pass it to
23951 (control_warning_option): Pass false to handle_generated_option
23953 * opts.c (maybe_default_option): Pass true to handle_generated_option
23955 * optc-gen.awk: Likewise.
23957 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23959 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23960 a REG, look at the REG it is a SUBREG of.
23961 (splitter for cmpeqsi_t): Ditto.
23963 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23965 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23966 the special USEs with the pattern of the insn, not the insn itself.
23968 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23971 * doc/invoke.texi: Document -mload-store-pairs.
23973 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23974 Sandra Loosemore <sandra@codesourcery.com>
23976 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23977 argument isn't a CONST_INT.
23978 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23979 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23980 (nios2_validate_compare): Bypass alternate compare logic if *op2
23981 is not a CONST_INT.
23982 (ldstwm_operation_p): Return false if first_base is not a REG or
23983 if first_offset is not a CONST_INT.
23985 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23987 * config/cris/cris.md: Use correct operand in a define_peephole2.
23989 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23991 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23993 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23995 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23996 this_insn if it is an INSN or JUMP_INSN.
23997 (force_offsettable): Look at base, not at addr.
23998 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23999 on things that aren't necessarily CONST_INTs.
24001 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24003 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24004 -mfpmath=sse is the default also for x86-32 targets with SSE2
24005 instruction set when @option{-ffast-math} is enabled
24007 2017-02-24 Jeff Law <law@redhat.com>
24009 PR rtl-optimizatoin/79286
24010 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24013 2017-02-24 Richard Biener <rguenther@suse.de>
24015 PR tree-optimization/79389
24016 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24019 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24021 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24022 function comment to reflect reality.
24023 (loop_exits_before_overflow): Fix typo in function description.
24025 2017-02-24 Richard Biener <rguenther@suse.de>
24027 PR tree-optimization/79389
24028 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24029 properly that a threading opportunity exists. Detect conditional
24030 copy/constant propagation opportunities.
24032 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24034 * config/visium/visium.md (type): Add trap.
24035 (b): New mode attribute.
24036 (*btst): Rename into...
24037 (*btst<mode>): ...this and adjust.
24038 (*cbranchsi4_btst_insn): Rename into...
24039 (*cbranch<mode>4_btst_insn): ...this and adjust.
24040 (trap): New define_insn.
24042 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24044 PR tree-optimization/79389
24045 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24046 (noce_reversed_cond_code): New function.
24047 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24048 reversed_comparison_code. Formatting fix.
24049 (noce_try_store_flag): Test rev_cond != NULL in addition to
24050 reversed_comparison_code.
24051 (noce_try_store_flag_constants): Likewise.
24052 (noce_try_store_flag_mask): Likewise.
24053 (noce_try_addcc): Use rev_cond if non-NULL instead of
24054 reversed_comparison_code.
24055 (noce_try_cmove_arith): Likewise. Formatting fixes.
24056 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24057 (noce_find_if_block): Initialize rev_cond.
24058 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24059 instead of false as last argument never attempt to reverse it
24062 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24064 PR tree-optimization/79663
24065 * tree-predcom.c (combine_chains): Process refs in reverse order
24066 only for ZERO length chains, and add explaining comment.
24068 2017-02-23 Jeff Law <law@redhat.com>
24070 PR tree-optimization/79578
24071 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24072 in call to operand_equal_p.
24074 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24077 * config/i386/cpuid.h: Fix another undefined behavior.
24079 2017-02-23 Richard Biener <rguenther@suse.de>
24081 PR tree-optimization/79683
24082 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24083 vector types for data-refs.
24085 2017-02-23 Martin Liska <mliska@suse.cz>
24087 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24089 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24091 PR middle-end/79665
24092 * internal-fn.c (get_range_pos_neg): Moved to ...
24093 * tree.c (get_range_pos_neg): ... here. No longer static.
24094 * tree.h (get_range_pos_neg): New prototype.
24095 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24096 are known to be in between 0 and signed maximum inclusive, try to
24097 expand both unsigned and signed divmod and use the cheaper one from
24100 2017-02-22 Jeff Law <law@redhat.com>
24102 PR tree-optimization/79578
24103 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24104 to compare base operands.
24106 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24109 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24110 gpc_reg_operand instead of fpr_reg_operand.
24112 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24114 * config/mips/mips.c (mips_return_in_memory): Force FP
24115 vector types to be returned in memory for o32 ABI.
24117 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24119 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24120 instead of DW_TAG_member for static data member declarations and don't
24121 set no_linkage_name for static inline data members.
24122 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24125 2017-02-22 Martin Liska <mliska@suse.cz>
24127 * doc/invoke.texi: Replace inequality signs with square brackets
24130 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24132 PR tree-optimization/68644
24133 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24135 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24138 * lra-constraints.c (simplify_operand_subreg): Handle
24139 WORD_REGISTER_OPERATIONS targets.
24141 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24144 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24145 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24146 elimination by swapping fld*.
24148 2017-02-22 Richard Biener <rguenther@suse.de>
24150 PR tree-optimization/79673
24151 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24152 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24153 irrelevant address-space qualifiers and avoiding a
24154 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24156 2017-02-22 Richard Biener <rguenther@suse.de>
24158 PR tree-optimization/79666
24159 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24160 to not symbolically negate if that may introduce undefined
24163 2017-02-22 Martin Liska <mliska@suse.cz>
24166 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24167 * data-streamer-out.c (streamer_write_gcov_count_stream):
24169 * value-prof.c (stream_out_histogram_value): Make assert more
24170 precise based on type of counter.
24172 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24175 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24176 Use nonimmediate_operand instead of memory_operand for operand 1.
24177 (float-extend standard_x87sse_constant_load splitter): Ditto.
24179 2017-02-21 Jeff Law <law@redhat.com>
24181 PR tree-optimization/79621
24182 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24183 blocks with edges to themselves.
24185 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24188 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24189 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24190 Use gimple_call_builtin_p.
24193 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24194 on temporarily removed DEBUG_INSNs.
24196 PR tree-optimization/79649
24197 * tree-loop-distribution.c (classify_partition): Give up on
24198 non-generic address space loads/stores.
24200 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24202 * doc/loop.texi (Loop manipulation): Remove nonexistent
24203 tree_ssa_loop_version from the documentation.
24204 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24206 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24209 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24210 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24211 * config/rs6000/rs6000.c: Include except.h.
24212 (rs6000_expand_split_stack_prologue): Call
24213 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24215 2017-02-21 Martin Jambor <mjambor@suse.cz>
24218 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24219 have been analyzed.
24221 2017-02-21 Martin Jambor <mjambor@suse.cz>
24223 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24224 for backward compatibility only.
24225 * doc/invoke.texi (Option Summary): Remove all references to
24226 -fipa-cp-alignment.
24228 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24232 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24234 * lra-constraints.c (curr_insn_transform): Handle
24235 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24237 2017-02-21 Martin Liska <mliska@suse.cz>
24239 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24241 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24243 PR translation/79638
24244 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24246 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24249 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24250 (arm_function_ok_for_sibcall): Return false for an indirect call by
24251 descriptor if all the argument registers are used.
24252 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24253 alignment of the function.
24255 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24257 PR tree-optimization/61441
24258 * simplify-rtx.c (simplify_const_unary_operation): For
24259 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24260 the sNaN unmodified.
24262 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24264 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24265 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24266 instead of SYSTEM_HEADER_DIR.
24268 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24269 Martin Liška <mliska@suse.cz>
24271 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24272 Fix typos and grammar, use active voice, and clarify.
24274 2017-02-20 Marek Polacek <polacek@redhat.com>
24276 PR middle-end/79537
24277 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24280 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24282 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24285 * config/i386/i386.c (ix86_expand_builtin): Handle
24286 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24287 ix86_builtins_isa[fcode].isa as a requirement of those
24288 flags and any other flag in the bitmask.
24289 (ix86_init_mmx_sse_builtins): Use 0 instead of
24290 ~OPTION_MASK_ISA_64BIT as mask.
24291 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24292 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24293 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24294 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24296 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24299 * lra-constraints.c (split_reg): Check requested split mode
24300 is supported by the register.
24302 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24304 * lra-constraints.c (simplify_operand_subreg): Remove early
24307 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24310 * lra-constraints.c (curr_insn_transform): Tighten condition
24311 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24313 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24316 * lra-constraints.c (curr_insn_transform): Handle
24317 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24319 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24322 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24324 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24326 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24329 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24332 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24334 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24337 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24339 * final.c (last_columnnum, override_columnnum): New variables.
24340 (final_start_function): Set last_columnnum, pass it to begin_prologue
24341 hook and pass 0 to dwarf2out_begin_prologue.
24342 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24343 to source_line debug hook.
24344 (notice_source_line): Compute last_columnnum and for debug_column_info
24345 return true on column changes.
24346 * debug.h (struct gcc_debug_hooks): Add column argument to
24347 source_line and begin_prologue hooks.
24348 (debug_nothing_int_charstar_int_bool): Remove prototype.
24349 (debug_nothing_int_int_charstar,
24350 debug_nothing_int_int_charstar_int_bool): New prototypes.
24351 (dwarf2out_begin_prologue): Add column argument.
24352 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24353 begin_prologue hooks.
24354 (debug_nothing_int_charstar_int_bool): Remove.
24355 (debug_nothing_int_int_charstar,
24356 debug_nothing_int_int_charstar_int_bool): New functions.
24357 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24358 through to dwarf2out_source_line.
24359 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24360 (dwarf2out_source_line): Add column argument, emit it if requested.
24361 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24363 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24364 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24365 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24366 through to dwarf2out_begin_prologue.
24367 (vmsdbgout_source_line): Add column argument, pass it through to
24368 dwarf2out_source_line.
24369 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24370 dbxout_source_line caller.
24371 (dbxout_source_line): Add column argument.
24373 * common.opt (gno-column-info, gcolumn-info): New options.
24374 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24375 (check_die): Also test for multiple DW_AT_decl_column attributes.
24376 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24377 DW_AT_decl_column if requested.
24378 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24380 (gen_variable_die): Likewise.
24381 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24382 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24385 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24386 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24387 (ix86_handle_option): Handle OPT_m3dnowa.
24388 * doc/invoke.texi (-m3dnowa): Document.
24389 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24390 -m3dnowa instead of -m3dnow -march=athlon.
24393 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24394 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24396 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24399 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24400 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24401 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24402 generator for vsx_xxpermdi_<mode>_be.
24403 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24404 force big-endian semantics.
24405 (vsx_xxpermdi_<mode>_be): New define_expand with same
24406 implementation as previous version of vsx_xxpermdi_<mode>.
24408 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24410 PR tree-optimization/79327
24411 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24412 variable, its initialization and use.
24414 2017-02-17 Julia Koval <julia.koval@intel.com>
24416 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24417 (OPTION_MASK_ISA_PKU_UNSET): New.
24418 (ix86_handle_option): Handle -mrdpid.
24419 * config/i386/cpuid.h (bit_RDPID): New.
24420 * config/i386/driver-i386.c (host_detect_local_cpu):
24421 Detect RDPID feature.
24422 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24423 * config/i386/i386-c.c (ix86_target_macros_internal):
24425 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24426 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24427 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24428 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24429 * config/i386/i386.md (define_insn "rdpid"): New.
24430 * config/i386/i386.opt Add -mrdpid.
24431 * config/i386/immintrin.h (_rdpid_u32): New.
24433 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24435 PR rtl-optimization/79541
24436 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24437 instead of transforming it into USE.
24439 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24441 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24442 If HONOR_SNANS (SFmode) force the input to a register.
24443 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24444 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24445 an frsp or similar insn.
24447 2017-02-17 Martin Liska <mliska@suse.cz>
24449 PR rtl-optimization/79577
24450 * params.def (selsched-max-sched-times): Increase minimum to 1.
24452 2017-02-17 Martin Liska <mliska@suse.cz>
24454 PR rtl-optimization/79574
24455 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24457 2017-02-17 Martin Liska <mliska@suse.cz>
24459 PR tree-optimization/79529
24460 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24461 ssa_defined_default_def_p to handle cases which are implicitly
24463 * tree-ssa.c (ssa_defined_default_def_p): New function.
24464 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24465 which are implicitly defined.
24466 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24468 2017-02-17 Richard Biener <rguenther@suse.de>
24470 PR middle-end/79576
24471 * params.def (max-ssa-name-query-depth): Limit to 10.
24473 2017-02-17 Richard Biener <rguenther@suse.de>
24475 PR tree-optimization/79552
24476 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24479 2017-02-17 Richard Biener <rguenther@suse.de>
24482 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24484 2017-02-17 Marek Polacek <polacek@redhat.com>
24486 PR middle-end/79536
24487 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24488 (fold_negate_expr): New wrapper.
24490 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24492 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24493 Correct terminology and de-emphasize pre-standard behavior.
24495 2017-02-16 Alan Modra <amodra@gmail.com>
24497 PR rtl-optimization/79286
24498 * ira.c (def_dominates_uses): New function.
24499 (update_equiv_regs): Don't create an equivalence for insns that
24500 may trap where the register def does not dominate the use.
24502 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24504 PR rtl-optimization/78127
24505 * lra.c (lra): Call lra_eliminate before finish the loop after
24508 2017-02-16 Richard Biener <rguenther@suse.de>
24510 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24512 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24513 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24514 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24515 (isl_val_int_from_wi): New function.
24516 (extract_affine_gmp): Rename to ...
24517 (extract_affine_wi): ... this, take a widest_int.
24518 (extract_affine_int): Just wrap extract_affine_wi.
24519 (add_param_constraints): Use isl_val_int_from_wi.
24520 (add_loop_constraints): Likewise, and extract_affine_wi.
24522 2017-02-15 Jeff Law <law@redhat.com>
24524 PR middle-end/79521
24525 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24526 ira_init_register_move_cost_if_necessary.
24528 2017-02-15 Martin Sebor <msebor@redhat.com>
24530 PR middle-end/32003
24531 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24532 removed in a prior commit.
24534 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24536 PR tree-optimization/79347
24537 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24538 counters during peeling.
24540 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24542 * Makefile.in (site.exp): Remove "set ISLVER".
24544 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24547 * real.c (real_from_integer): Call real_convert even for decimal.
24549 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24552 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24554 2017-02-14 Andrew Pinski <apinski@cavium.com>
24556 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24557 cores and change the partno/implementer to be correct.
24558 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24559 the 'B" as the implementer.
24560 * config/aarch64/aarch64-tune.md: Regenerate.
24562 2017-02-14 Carl Love <cel@us.ibm.com>
24564 * config/rs6000/rs6000.c: Add case statement entry to make the
24565 xvcvuxdsp built-in argument unsigned.
24566 * config/rs6000/vsx.md: Fix the source and return operand types so they
24567 match the instruction definitions from the ISA document. Fix typo
24568 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24571 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24574 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24575 member early_clobber_alts.
24576 * lra-lives.c (reg_early_clobber_p): New.
24577 (process_bb_lives): Use it.
24578 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24579 (debug_operand_data): Initialize early_clobber_alts.
24580 (setup_operand_alternative): Set up early_clobber_alts.
24581 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24582 alternatives to new_insn_reg.
24583 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24585 (lra_update_insn_regno_info): Pass the new arg.
24587 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24589 PR middle-end/79505
24590 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24591 (new_oacc_loop_raw): Don't clear already cleared fields.
24594 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24595 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24596 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24598 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24601 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24603 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24606 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24607 the extra instruction to the right place to store 128-bit constant
24610 2017-02-14 Martin Sebor <msebor@redhat.com>
24612 PR middle-end/79448
24613 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24614 warning for strings of unknown length.
24616 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24618 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24620 2017-02-14 Jeff Law <law@redhat.com>
24623 * ira-costs.c (scan_one_insn): Initialize register move costs
24624 for pseudos seen in USE/CLOBBER insns.
24626 PR tree-optimization/79095
24627 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24628 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24629 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24630 if the operands are known to be not equal, then the resulting range
24632 (intersect_ranges): If the new range is ~[0,0] and the old range is
24633 wide, then prefer ~[0,0].
24634 * tree-vrp.c (overflow_comparison_p_1): New function.
24635 (overflow_comparison_p): New function.
24636 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24637 if NAME is used in an overflow test.
24638 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24639 overflow check that can be expressed as an equality test, then adjust
24640 ops to be that equality test.
24642 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24644 * config/s390/s390-builtin-types.def: Remove flags argument.
24645 * config/s390/s390.c (s390_init_builtins): Likewise.
24647 2017-02-14 Martin Liska <mliska@suse.cz>
24649 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24650 vector. Fix trailing white spaces.
24652 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24654 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24657 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24659 PR rtl-optimization/68664
24660 * config/arm/arm.c (arm_sched_can_speculate_insn):
24661 New function. Declare prototype.
24662 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24664 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24666 PR rtl-optimization/68664
24667 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24669 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24671 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24673 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24674 max skip bytes for function, loop and jump.
24676 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24678 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24679 ABS_EXPR for gimple dump.
24681 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24684 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24686 PR tree-optimization/79408
24687 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24688 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24689 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24690 also if rhs1 is INTEGER_CST.
24692 2017-02-14 Richard Biener <rguenther@suse.de>
24694 PR middle-end/79432
24695 * tree-into-ssa.c (insert_phi_nodes): When the function can
24696 have abnormal edges rewrite SSA names with broken use-def
24697 dominance out of SSA and register them for PHI insertion.
24699 2017-02-13 Martin Sebor <msebor@redhat.com>
24701 PR middle-end/79496
24702 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24703 clearing info.nowrite flag when snprintf size argument is a range.
24705 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24707 * cprop.c (cprop_jump): Add missing space in string literal.
24708 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24709 (get_constraint_for_component_ref): Likewise.
24710 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24711 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24712 * lra-constraints.c (process_alt_operands): Likewise.
24713 * ipa-inline.c (inline_small_functions): Likewise.
24714 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24715 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24716 * trans-mem.c (diagnose_tm_1_op): Likewise.
24717 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24718 (grid_parallel_clauses_gridifiable): Likewise.
24720 * config/nvptx/mkoffload.c (process): Add space in between
24723 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24724 "MOD4_SSE_REGS" and "ALL_REGS".
24726 * spellcheck.c (test_data): Add , in between "foo" and "food".
24728 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24731 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24732 boundary crossing check and subsequent code generation agree.
24734 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24736 * config/aarch64/aarch64.c (has_memory_op): Delete.
24737 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24740 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24742 PR rtl-optimization/79388
24743 PR rtl-optimization/79450
24744 * combine.c (distribute_notes): When removing TEM_INSN for which
24745 corresponding dest has last value recorded, invalidate that last
24748 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24750 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24751 of explicit '@'. Add missing assembly comment marker on branch costs
24754 2017-02-13 Nathan Sidwell <nathan@acm.org>
24756 * gengtype-lex.l (<in_struct>): Add '/'.
24758 2017-02-13 Martin Liska <mliska@suse.cz>
24761 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24763 2017-02-13 Richard Biener <rguenther@suse.de>
24765 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24767 * configure: Re-generate.
24768 * config.in: Likewise.
24769 * graphite-dependences.c: Simplify as if
24770 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24771 * graphite-isl-ast-to-gimple.c: Likewise.
24772 * graphite-optimize-isl.c: Likewise.
24773 * graphite-poly.c: Likewise.
24774 * graphite-sese-to-poly.c: Likewise.
24775 * graphite.h: Likewise.
24776 * toplev.c: Include isl/version.h and use isl_version () for
24777 printing the ISL version.
24778 * doc/install.texi: Update ISL requirement.
24780 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24782 * doc/standards.texi (Standards): Update reference to
24785 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24787 * doc/extend.texi (Named Address Spaces): sourceware.org now
24789 * doc/install.texi (Binaries): Ditto.
24792 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24794 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24795 standard terminology "stringize"/"stringizing" throughout.
24796 * doc/cppinternals.texi: Likewise.
24798 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24800 * doc/extend.texi: Fix some spelling mistakes and typos.
24801 * doc/invoke.texi: Likewise.
24803 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24806 * params.def (inline-min-speedup) Change from 10 to 8.
24808 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24810 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24813 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24816 * ipa-inline-analysis.c (get_minimal_bb): New function.
24817 (record_modified): Use it.
24818 (remap_edge_change_prob): Handle also ancestor functions.
24820 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24822 * doc/contrib.texi (Contributors): Remove broken link into
24823 the Mauve CVS repository.
24825 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24827 PR middle-end/79454
24828 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24829 result computation whenever lhs doesn't have vector mode, not
24830 just when it has BLKmode.
24832 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24834 * doc/makefile.texi (profiledbootstrap): Refer to the
24835 installation instructions only in textual form.
24837 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24840 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24842 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24844 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24845 (Specific): Update mingw-w64 reference.
24847 (Specific): Remove broken link to Renesas RX processor.
24849 2017-02-10 Richard Biener <rguenther@suse.de>
24851 * toplev.c (process_options): Do not mention obsolete graphite
24852 options when printing sorry message about missing graphite support.
24853 Mention -floop-nest-optimize.
24855 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24857 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24858 (vtst_p16): Likewise.
24859 (vtstq_p8): Likewise.
24860 (vtstq_p16): Likewise.
24862 (vtstq_p64): Likewise.
24863 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24864 (vset_lane_p64): New.
24865 (vsetq_lane_p64): New.
24867 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24869 PR tree-optimization/79411
24870 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24871 stmt operands are SSA_NAMEs used in abnormal phis.
24872 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24875 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24878 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24881 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24883 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24885 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24887 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24891 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24892 not arbitrary TREE_CONSTANT.
24895 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24896 "omp declare target link" attribute unless is_global_var.
24897 * omp-offload.c (find_link_var_op): Likewise.
24899 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24900 Chung-Lin Tang <cltang@codesourcery.com>
24902 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24904 (gimplify_adjust_omp_clauses): Don't delete TILE.
24905 (gimplify_omp_for): Deal with TILE.
24906 * internal-fn.c (expand_GOACC_TILE): New function.
24907 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24909 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24910 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24912 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24913 avoid DIV for outermost collapse var.
24914 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24915 Remove out of date comments, fix whitespace.
24916 * omp-general.c (omp_extract_for_data): Deal with tiling.
24917 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24918 adjust OLF_DIM_BASE value.
24919 (struct omp_for_data): Add tiling field.
24920 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24921 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24922 for auto loops. Remove default auto determining, moved to
24923 oacc_loop_fixed_partitions.
24924 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24925 stmts, add e_mask field.
24926 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24927 (oacc_thread_numbers): Use oacc_dim_call.
24928 (oacc_xform_tile): New.
24929 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24930 (finish_oacc_loop): Adjust for ifns vector.
24931 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24932 add case for GOACC_TILE fns.
24933 (oacc_loop_xform_loop): Delete.
24934 (oacc_loop_process): Iterate over call list directly, and add
24935 handling for GOACC_TILE fns.
24936 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24938 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24939 vector partitioning to outer loops. Assign 2 partitions to loops
24940 when available. Add TILE handling.
24941 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24942 (execite_oacc_device_lower): Process GOACC_TILE fns,
24943 ignore unknown specs.
24944 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24945 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24946 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24948 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24950 * configure.ac (ACX_BUGURL): Update.
24951 * configure: Regenerate.
24953 2017-02-09 Richard Biener <rguenther@suse.de>
24955 PR tree-optimization/69823
24956 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24957 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24959 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24961 * config/arc/arc-c.def: Add __NPS400__ definition.
24962 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24963 (TARGET_NPS400): Define.
24965 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24967 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24969 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24970 pointer, arch_info.
24971 (arc_cpu_types): Fill the arch_info field with a pointer into the
24972 arc_arch_types table.
24973 (arc_selected_cpu): Declare.
24974 * config/arc/arc.c (arc_selected_cpu): Make global.
24975 (arc_selected_arch): Delete.
24976 (arc_base_cpu): Delete.
24977 (arc_override_options): Remove references to deleted variables,
24978 update access to arch information.
24979 (ARC_OPT): Update access to arch information.
24980 (ARC_OPTX): Likewise.
24981 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24982 (TARGET_ARC600): Update access to arch information.
24983 (TARGET_ARC601): Likewise.
24984 (TARGET_ARC700): Likewise.
24985 (TARGET_EM): Likewise.
24986 (TARGET_HS): Likewise.
24987 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24990 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24993 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24994 condition/operands for integer GE/LE/GEU/LEU operations.
24996 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24998 PR translation/79397
24999 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25002 2017-02-08 Martin Jambor <mjambor@suse.cz>
25005 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25006 whether allocation happened.
25007 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25008 nothing was allocated.
25010 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25012 PR tree-optimization/79408
25013 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25014 constant, but SSA_NAME with a known integer range, use the minimum
25015 of that range instead of op1 to determine if modulo can be replaced
25016 with its first operand.
25018 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25020 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25022 2017-02-08 Richard Biener <rguenther@suse.de>
25024 PR tree-optimization/71824
25025 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25026 Check all loops contained in the merged region.
25028 2017-02-07 Andrew Pinski <apinski@cavium.com>
25030 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25032 2017-02-07 Andrew Pinski <apinski@cavium.com>
25034 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25035 (thunderxt88): Likewise.
25036 (thunderxt81): Disable LSE and change v8.1 to v8.
25037 (thunderxt83): Likewise.
25039 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25040 Richard Biener <rguenther@suse.de>
25042 PR middle-end/79399
25043 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25044 type from int to size_t.
25045 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25047 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25049 PR rtl-optimization/79386
25050 * cprop.c (bypass_conditional_jumps): Initialize
25051 bypass_last_basic_block already before splitting bbs after
25052 unconditional traps...
25053 (bypass_conditional_jumps): ... rather than here.
25056 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25057 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25058 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25059 fix -masm=intel patterns.
25061 2017-02-07 Richard Biener <rguenther@suse.de>
25063 PR tree-optimization/79256
25064 PR middle-end/79278
25065 * builtins.c (get_object_alignment_2): Use min_align_of_type
25066 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25067 and ADJUST_FIELD_ALIGN.
25069 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25071 * doc/tm.texi: Regenerate.
25072 * stor-layout.c (layout_decl): Adjust.
25073 (update_alignment_for_field): Likewise.
25074 (place_field): Likewise.
25075 (min_align_of_type): Likewise.
25076 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25077 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25078 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25079 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25080 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25081 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25082 * config/i386/i386.c (x86_field_alignment): Likewise.
25083 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25084 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25085 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25086 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25087 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25088 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25092 2017-01-30 Richard Biener <rguenther@suse.de>
25094 PR tree-optimization/79256
25095 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25096 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25099 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25101 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25102 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25103 builtins to SImode and emit a zero-extend, if necessary.
25105 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25107 * docs/invoke.texi (RISC-V Options): Alphabetize.
25109 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25111 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25114 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25116 * config/riscv/riscv.c: New file.
25117 * common/config/riscv/riscv-common.c: Likewise.
25118 * config.gcc: Likewise.
25119 * config/riscv/constraints.md: Likewise.
25120 * config/riscv/elf.h: Likewise.
25121 * config/riscv/generic.md: Likewise.
25122 * config/riscv/linux.h: Likewise.
25123 * config/riscv/multilib-generator: Likewise.
25124 * config/riscv/peephole.md: Likewise.
25125 * config/riscv/pic.md: Likewise.
25126 * config/riscv/predicates.md: Likewise.
25127 * config/riscv/riscv-builtins.c: Likewise.
25128 * config/riscv/riscv-c.c: Likewise.
25129 * config/riscv/riscv-ftypes.def: Likewise.
25130 * config/riscv/riscv-modes.def: Likewise.
25131 * config/riscv/riscv-opts.h: Likewise.
25132 * config/riscv/riscv-protos.h: Likewise.
25133 * config/riscv/riscv.h: Likewise.
25134 * config/riscv/riscv.md: Likewise.
25135 * config/riscv/riscv.opt: Likewise.
25136 * config/riscv/sync.md: Likewise.
25137 * config/riscv/t-elf-multilib: Likewise.
25138 * config/riscv/t-linux: Likewise.
25139 * config/riscv/t-linux-multilib: Likewise.
25140 * config/riscv/t-riscv: Likewise.
25141 * configure.ac: Likewise.
25142 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25143 Waterman as RISC-V maintainers.
25144 * doc/install.texi: Add RISC-V entries.
25145 * doc/invoke.texi: Add RISC-V options section.
25146 * doc/md.texi: Add RISC-V constraints section.
25147 * configure: Regenerated.
25149 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25152 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25153 false values to be constant vectors with all 0 or all 1 bits set.
25154 (vcondu<mode><mode>): Likewise.
25155 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25157 (fpmask_comparison_operator): Update comment.
25158 (vecint_comparison_operator): New predicate.
25159 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25160 vector conditionals when the true and false values are constant
25161 vectors with all 0 bits or all 1 bits set.
25163 2017-02-06 Martin Sebor <msebor@redhat.com>
25165 PR tree-optimization/79376
25166 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25168 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25170 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25171 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25172 to simplify split condition.
25174 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25176 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25177 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25180 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25182 PR rtl-optimization/68664
25183 * target.def (can_speculate_insn): New hook.
25184 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25185 * doc/tm.texi: Regenerate.
25186 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25187 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25188 (rs6000_sched_can_speculate_insn): New function.
25190 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25192 PR tree-optimization/79284
25193 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25194 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25195 vectorizable_mask_load_store, vectorizable_operation,
25196 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25197 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25198 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25199 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25200 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25201 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25202 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25203 is_gimple_assign (stmt). Replace another such test with
25204 is_gimple_assign (stmt).
25206 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25209 * config/avr/avr.c (rtl-iter.h): Include it.
25210 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25211 (avr_legitimate_combined_insn): ...and implementation.
25213 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25215 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25216 * config/s390/s390.c (s390_const_operand_ok)
25217 (s390_canonicalize_comparison, s390_extract_part)
25218 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25219 (s390_contiguous_bitmask_p, s390_rtx_costs)
25220 (legitimize_pic_address): Likewise.
25221 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25222 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25223 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25224 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25225 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25227 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25229 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25230 REGNO($0) == REGNO($1).
25232 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25234 * config/s390/linux.h(SIZE_TYPE): Add comment.
25236 2017-02-06 Julian Brown <julian@codesourcery.com>
25237 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25238 Virendra Pathak <virendra.pathak@broadcom.com>
25240 * config/aarch64/aarch64-cores.def: Change the scheduler
25242 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25243 * config/aarch64/thunderx2t99.md: New file.
25245 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25247 * doc/standards.texi (Go Language): Update link to language
25250 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25252 * tree-eh.c (lower_resx): Sanitize profile.
25253 (cleanup_empty_eh_move_lp): Likewise.
25255 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25258 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25260 * cfgloopmanip.h (loop_version): Update prototype.
25261 * modulo-sched.c (sms_schedule): Update call of loop_version.
25262 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25263 * tree-parloops.c (gen_parallel_loop): Likewise.
25264 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25265 * tree-ssa-loop-split.c (split_loop): Likewise.
25266 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25267 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25269 2017-02-05 Martin Liska <mliska@suse.cz>
25272 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25274 (print_operand_address): Initialize a struct to zero.
25276 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25278 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25279 garbage collector only in textual form.
25281 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25283 * doc/extend.texi (x86 specific memory model extensions for
25284 transactional memory): Simplify a phrase.
25286 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25289 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25290 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25291 (atomic_storedi_1): Likewise.
25293 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25295 PR tree-optimization/79338
25296 * tree-parloops.c (gather_scalar_reductions): Don't call
25297 vect_analyze_loop_form for loop->inner before destroying loop's
25300 2017-02-03 Martin Sebor <msebor@redhat.com>
25302 PR tree-optimization/79327
25303 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25304 when precision has resulted in leading zeros.
25305 (format_integer): Adjust the likely counter to assume an unknown
25306 argument that may be zero is non-zero.
25308 2017-02-03 Jason Merrill <jason@redhat.com>
25311 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25312 avoid copying non-taken branch.
25314 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25316 PR tree-optimization/79340
25317 * tree-vect-loop.c (vectorizable_reduction): Release
25318 vec_defs elements after safe_splicing them into other vectors.
25321 PR tree-optimization/79327
25322 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25323 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25325 (format_integer): Use wide_int_to_tree instead of build_int_cst
25326 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25327 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25328 of shortest and longest sequence.
25330 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25332 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25333 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25335 2017-02-03 Walter Lee <walt@tilera.com>
25338 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25339 after initial stackframe link reg save.
25340 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25342 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25345 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25346 wu for stxssp alternative.
25348 2017-02-03 Martin Sebor <msebor@redhat.com>
25350 PR tree-optimization/79352
25351 * gimple-fold.c (get_range_strlen): Add argument.
25352 (get_range_strlen): Change return type to bool.
25353 (get_maxval_strlen): Pass in a dummy argument.
25354 * gimple-fold.h (get_range_strlen): Change return type to bool.
25355 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25356 * tree.h (array_at_struct_end_p): Add argument.
25357 * tree.c (array_at_struct_end_p): Handle it.
25359 2017-02-03 Martin Liska <mliska@suse.cz>
25362 * multiple_target.c (create_dispatcher_calls): Redirect edge
25363 from a caller of a dispatcher.
25364 (expand_target_clones): Make the clones local.
25365 (ipa_target_clone): Do both target clones and resolvers.
25366 (ipa_dispatcher_calls): Remove the pass.
25367 (pass_dispatcher_calls::gate): Likewise.
25368 (make_pass_dispatcher_calls): Likewise.
25369 * passes.def (pass_target_clone): Put as very first IPA early
25372 2017-02-03 Martin Liska <mliska@suse.cz>
25374 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25375 in case of a function with ifunc attribute.
25377 2017-02-03 Martin Liska <mliska@suse.cz>
25379 * cgraph.c (cgraph_node::dump): Dump function version info.
25380 * symtab.c (symtab_node::dump_base): Add missing new line.
25382 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25384 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25385 (ifcombine_ifandif): Use it.
25387 2017-02-03 Martin Liska <mliska@suse.cz>
25389 * doc/invoke.texi: Document default value for
25390 use-after-scope-direct-emission-threshold.
25392 2017-02-03 Martin Liska <mliska@suse.cz>
25394 PR tree-optimization/79339
25395 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25396 (format_floating): Likewise.
25398 2017-02-03 Martin Liska <mliska@suse.cz>
25401 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25403 (ipa_node_params_t::remove): Likewise.
25404 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25405 initialization from removed ipa_node_params_t::insert.
25406 (ipa_node_params::~ipa_node_params): Move from removed
25407 ipa_node_params_t::release.
25408 * symbol-summary.h (symbol_summary::m_released): New member.
25409 Do not release a summary twice. Do not allow to call finalizer
25410 for types of a summary that live in GGC memory.
25412 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25414 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25417 2017-02-02 Martin Sebor <msebor@redhat.com>
25419 PR middle-end/79275
25420 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25421 (format_string): Tighten up the range of output for non-constant
25422 strings and correct the expected range for wide non-constant strings.
25424 2017-02-02 Martin Sebor <msebor@redhat.com>
25426 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25428 PR middle-end/32003
25429 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25431 (-fdump-tree-@var): Add to index and document how to come up
25432 with pass-specific option and dump file names.
25433 (-fdump-passes): Clarify where to look for output.
25435 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25437 PR middle-end/77445
25438 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25439 statistics of the analyzed path; allow threading for speed when
25440 any of BBs along the path are optimized for speed.
25442 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25444 PR middle-end/78468
25445 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25446 settings of the virtual registers.
25449 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25451 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25452 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25455 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25457 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25458 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25460 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25462 * config/s390/s390.md: Add missing comments with the expanded
25464 * config/s390/vector.md: Likewise.
25465 * config/s390/vx-builtins.md: Likewise.
25467 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25470 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25471 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25472 conditions on a single line.
25474 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25476 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25477 __S390_VX__ to __VX__.
25479 2017-02-01 Andrew Pinski <apinski@cavium.com>
25481 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25482 stmt_info to record_stmt_cost.
25483 (vect_get_known_peeling_cost): Pass stmt_info if known to
25485 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25486 cpu_vector_cost field into
25487 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25488 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25489 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25490 splitting of scalar_stmt_cost and vec_stmt_cost.
25491 (thunderx_vector_cost): Likewise.
25492 (cortexa57_vector_cost): LIkewise.
25493 (exynosm1_vector_cost): Likewise.
25494 (xgene1_vector_cost): Likewise.
25495 (thunderx2t99_vector_cost): Improve after the splitting of the two
25497 (aarch64_builtin_vectorization_cost): Update for the splitting of
25498 scalar_stmt_cost and vec_stmt_cost.
25500 2017-02-01 Torvald Riegel <triegel@redhat.com>
25501 Richard Henderson <rth@redhat.com>
25503 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25504 conditional on existance of a fast atomic load.
25505 * optabs-query.c (can_atomic_load_p): New function.
25506 * optabs-query.h (can_atomic_load_p): Declare it.
25507 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25508 no fast atomic load is available for the particular size of access.
25509 (expand_atomic_compare_and_swap): Likewise.
25510 (expand_atomic_load): Likewise.
25511 (expand_atomic_store): Likewise.
25512 (expand_atomic_fetch_op): Likewise.
25513 * testsuite/lib/target-supports.exp
25514 (check_effective_target_sync_int_128): Remove x86 because it provides
25515 no fast atomic load.
25516 (check_effective_target_sync_int_128_runtime): Likewise.
25518 2017-02-01 Richard Biener <rguenther@suse.de>
25520 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25521 (graphite_transform_loops): Provide opt-info for optimized nests.
25522 * tree-parloop.c (parallelize_loops): Provide opt-info for
25523 parallelized loops.
25525 2017-02-01 Richard Biener <rguenther@suse.de>
25527 PR middle-end/79315
25528 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25529 was not set before.
25531 2017-02-01 Richard Biener <rguenther@suse.de>
25533 PR tree-optimization/71824
25534 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25535 Verify the loops are valid in the merged SESE region.
25536 (scop_detection::can_represent_loop_1): Check analyzing the
25537 evolution of the number of iterations in the region succeeds.
25539 2017-01-31 Ian Lance Taylor <iant@golang.org>
25541 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25542 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25544 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25546 PR preprocessor/79210
25547 * input.c (get_substring_ranges_for_loc): Replace line_width
25548 assertion with error-handling.
25550 2017-01-31 Richard Biener <rguenther@suse.de>
25552 PR tree-optimization/77318
25553 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25554 (create_pw_aff_from_tree): Take loop parameter.
25555 (add_condition_to_pbb): Pass loop of the condition to
25556 create_pw_aff_from_tree.
25558 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25560 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25561 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25563 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25567 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25569 (convert_int_to_float128): Likewise.
25570 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25571 (convert_int_to_float128): Likewise.
25572 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25573 (UNSPEC_IEEE128_CONVERT): Likewise.
25574 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25575 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25576 Use local variables for IBM extended format.
25577 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25578 (fix_trunc<mode>si2_fprs): Likewise.
25579 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25580 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25581 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25582 to know that we can now have integers of all sizes in vector
25584 (fix<uns>_<mode>di2_hw): Likewise.
25585 (float<uns>_<mode>si2_hw): Likewise.
25586 (fix_<mode>si2_hw): Likewise.
25587 (fixuns_<mode>si2_hw): Likewise.
25588 (float<uns>_<mode>di2_hw): Likewise.
25589 (float_<mode>di2_hw): Likewise.
25590 (float_<mode>si2_hw): Likewise.
25591 (floatuns_<mode>di2_hw): Likewise.
25592 (floatuns_<mode>si2_hw): Likewise.
25593 (xscvqp<su>wz_<mode>): Delete, no longer used.
25594 (xscvqp<su>dz_<mode>): Likewise.
25595 (xscv<su>dqp_<mode>): Likewise.
25596 (ieee128_mfvsrd_64bit): Likewise.
25597 (ieee128_mfvsrd_32bit): Likewise.
25598 (ieee128_mfvsrwz): Likewise.
25599 (ieee128_mtvsrw): Likewise.
25600 (ieee128_mtvsrd_64bit): Likewise.
25601 (ieee128_mtvsrd_32bit): Likewise.
25603 2017-01-31 Martin Liska <mliska@suse.cz>
25606 * ipa-prop.c (ipa_free_all_node_params): Call release method
25607 instead of ~sumbol_summary to not to trigger double times
25610 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25612 PR tree-optimization/71691
25613 * bitmap.h (class auto_bitmap): New.
25614 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25615 is_maybe_undefined instead of ssa_undefined_value_p.
25617 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25619 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25620 __S390_ARCH_LEVEL__ to __ARCH__.
25622 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25624 PR tree-optimization/79267
25625 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25626 if should_remove_lhs_p is true.
25628 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25631 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25632 (add_alignment_attribute): New.
25633 (base_type_die): Add alignment attribute.
25634 (subrange_type_die): Likewise.
25635 (modified_type_die): Likewise.
25636 (gen_array_type_die): Likewise.
25637 (gen_descr_array_type_die: Likewise.
25638 (gen_enumeration_type_die): Likewise.
25639 (gen_subprogram_die): Likewise.
25640 (gen_variable_die): Likewise.
25641 (gen_field_die): Likewise.
25642 (gen_ptr_to_mbr_type_die): Likewise.
25643 (gen_struct_or_union_type_die): Likewise.
25644 (gen_subroutine_type_die): Likewise.
25645 (gen_typedef_die): Likewise.
25646 (base_type_cmp): Compare alignment attribute.
25648 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25651 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25652 (setb_unsigned) New pattern for setb with CCUNS.
25653 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25654 subfc./subfe sequence to avoid overflow problems. Generate a
25655 shorter sequence with cmpld/setb for power9.
25656 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25657 for generating subfc. instruction.
25658 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25659 now uses this instruction.
25661 2017-01-30 Ian Lance Taylor <iant@google.com>
25664 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25665 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25667 2017-01-30 Martin Sebor <msebor@redhat.com>
25669 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25670 Move constant to the right of a relational operator.
25671 (get_mpfr_format_length, format_character, format_string): Ditto.
25672 (should_warn_p, maybe_warn): Same.
25674 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25676 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25679 * asan.c (get_translation_unit_decl): Remove function.
25680 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25682 2017-01-30 Martin Liska <mliska@suse.cz>
25684 PR gcov-profile/79259
25685 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25686 -fprofile-generate.
25688 2017-01-30 Martin Liska <mliska@suse.cz>
25691 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25692 Initialize variables with NULL value.
25694 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25697 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25699 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25701 2017-01-30 Richard Biener <rguenther@suse.de>
25703 PR tree-optimization/79276
25704 * tree-vrp.c (process_assert_insertions): Properly adjust common
25705 when removing a duplicate.
25707 * gcc.dg/torture/pr79276.c: New testcase.
25709 2017-01-30 Richard Biener <rguenther@suse.de>
25711 PR tree-optimization/79256
25712 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25713 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25715 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25717 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25720 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25721 ("*r<noxa>sbg_<mode>_sll_bitmask")
25722 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25723 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25724 Use contiguous_bitmask_nowrap_operand.
25726 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25729 * config/rs6000/altivec.h (vec_xl): Revise #define.
25730 (vec_xst): Likewise.
25732 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25734 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25736 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25738 PR rtl-optimization/79194
25739 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25740 traps before call to bypass_conditional_jumps.
25742 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25744 PR tree-optimization/71374
25745 * lra-constraints.c (check_conflict_input_operands): New.
25746 (match_reload): Use it.
25748 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25751 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25752 account to calculate conflict_set.
25754 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25756 PR rtl-optimization/78559
25757 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25758 other_insn in combine.
25760 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25762 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25763 uint16_type_node for BT_UINT16.
25765 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25767 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25768 "RTL Tests" to menu.
25769 (GIMPLE Tests): New node.
25770 (RTL Tests): New node.
25772 2017-01-27 Richard Biener <rguenther@suse.de>
25774 PR tree-optimization/79245
25775 * tree-loop-distribution.c (distribute_loop): Apply cost
25776 modeling also to detected patterns.
25778 2017-01-27 Richard Biener <rguenther@suse.de>
25780 PR tree-optimization/71433
25781 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25782 (compare_assert_loc): New function.
25783 (process_assert_insertions): Sort and optimize assert locations
25784 to remove duplicates and push down identical assertions on
25785 edges to their destination block.
25787 2017-01-27 Richard Biener <rguenther@suse.de>
25789 PR tree-optimization/79244
25790 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25791 out SSA names even if abnormal.
25793 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25795 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25796 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25797 instead of MPFR_RNDN.
25799 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25802 * arm.c (arm_option_override): Don't call build_target_option_node
25803 until after doing all option overrides.
25804 (arm_valid_target_attribute_tree): Likewise.
25806 2017-01-27 Martin Liska <mliska@suse.cz>
25808 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25809 for {cd}tors and C++ {cd}tors.
25811 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25813 * config/s390/s390.md ("*setmem_long_and")
25814 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25816 2017-01-26 Martin Sebor <msebor@redhat.com>
25818 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25821 2017-01-26 Martin Sebor <msebor@redhat.com>
25823 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25824 HAVE_DFmode before using XFmode or DFmode.
25825 (parse_directive): Avoid using the z length modifier to avoid
25826 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25828 PR middle-end/78703
25829 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25830 to accept adjustment as an array.
25831 (get_int_range): New function.
25832 (struct directive): Make width and prec arrays.
25833 (directive::set_width, directive::set_precision): Call get_int_range.
25834 (format_integer, format_floating): Handle width and precision ranges.
25835 (format_string, parse_directive): Same.
25837 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25840 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25841 comdat_type_p set, just clone them, but keep the children in the
25845 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25846 which have direct callers with -fvar-tracking-assignments enabled
25848 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25849 inside of type units.
25851 2017-01-26 Martin Sebor <msebor@redhat.com>
25853 PR middle-end/78703
25854 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25856 (struct format_result): Replace number_chars, number_chars_min,
25857 and number_chars_max with a single member of struct result_range.
25859 (format_result::operator+=): Adjust.
25860 (struct fmtresult): Remove bounded. Handle likely and unlikely
25862 (fmtresult::adjust_for_width_or_precision): New function.
25863 (fmtresult:type_max_digits): New function.
25864 (bytes_remaining): Handle likely and unlikely counters.
25865 (min_bytes_remaining): Remove.
25866 (format_percent): Simplify.
25867 (format_integer, format_floating): Set likely and unlikely counters.
25868 (get_string_length, format_character, format_string): Same.
25869 (format_plain, should_warn_p): New function.
25870 (maybe_warn): Call should_warn_p. Update diagnostic messages
25871 and handle those for all directives, including plain strings.
25872 (format_directive): Handle likely and unlikely counters.
25873 Remove unnecessary quoting from diagnostics. Add an informational
25875 (add_bytes): Remove.
25876 (pass_sprintf_length::compute_format_length): Simplify.
25877 (try_substitute_return_value): Handle likely and unlikely counters.
25879 2017-01-26 Carl Love <cel@us.ibm.com>
25881 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25882 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25884 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25887 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25888 endianess for subregs into account.
25889 * lra-constraints.c (lra_constraints): Do risky transformations
25890 always on the first iteration.
25891 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25892 last_call_used_reg_set.
25893 (process_bb_lives): Define and use last_call_used_reg_set.
25894 * lra.c (lra): Always continue after lra_constraints on the first
25897 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25899 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25901 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25903 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25905 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25906 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25907 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25908 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25909 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25910 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25911 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25912 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25913 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25915 2017-01-26 Marek Polacek <polacek@redhat.com>
25918 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25919 for the third operand.
25921 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25923 PR middle-end/79236
25924 * omp-low.c (struct omp_context): Add simt_stmt field.
25925 (scan_omp_for): Return omp_context *.
25926 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25927 context to the _simt_ SIMD stmt.
25928 (lower_omp_for): For combined SIMD with sibling _simt_
25929 SIMD, make sure to use the same decls in _looptemp_
25930 clauses as in the sibling.
25932 2017-01-26 David Sherwood <david.sherwood@arm.com>
25934 PR middle-end/79212
25935 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25938 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25941 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25942 emit fld b; fld a; if possible.
25944 * brig-builtins.def: Update copyright years.
25945 * config/arm/arm_acle_builtins.def: Update copyright years.
25947 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25950 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25951 constraint instead of o for the stxsd instruction.
25953 2017-01-25 Carl Love <cel@us.ibm.com>
25955 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25956 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25958 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25960 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25962 2017-01-25 Richard Biener <rguenther@suse.de>
25964 PR tree-optimization/69264
25965 * target.def (vector_alignment_reachable): Improve documentation.
25966 * doc/tm.texi: Regenerate.
25967 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25969 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25970 earlier changes with respect to TYPE_USER_ALIGN.
25971 (vector_alignment_reachable_p): Likewise. Improve dumping.
25973 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25976 * config/arm/arm.md (xordi3): Force constant operand into a register
25979 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25981 * doc/invoke.texi (-fstore-merging): Correct default optimization
25982 levels at which it is enabled.
25983 (-O): Move -fstore-merging from list to...
25986 2017-01-25 Richard Biener <rguenther@suse.de>
25989 * omp-expand.c: Include debug.h.
25990 (expand_omp_taskreg): Make sure to generate early debug before
25991 outlining anything from a function.
25992 (expand_omp_target): Likewise.
25993 (grid_expand_target_grid_body): Likewise.
25995 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25998 * asan.c (get_translation_unit_decl): New function.
25999 (asan_add_global): Extract modules file name from globals
26000 TRANSLATION_UNIT_DECL name.
26002 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26005 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26006 for long calls with APCS frame and VFP.
26008 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26010 * cfg.c (original_copy_tables_initialized_p): New function.
26011 * cfg.h (original_copy_tables_initialized_p): New decl.
26012 * cfgrtl.c (relink_block_chain): Guard the call to
26013 free_original_copy_tables with a call to
26014 original_copy_tables_initialized_p.
26015 * cgraph.h (symtab_node::native_rtl_p): New decl.
26016 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26017 (symtab_node::needed_p): Don't assert for early assembly output
26018 for __RTL functions.
26019 (cgraph_node::finalize_function): Set "force_output" for __RTL
26021 (cgraph_node::analyze): Bail out early for __RTL functions.
26022 (analyze_functions): Update assertion to support __RTL functions.
26023 (cgraph_node::expand): Bail out early for __RTL functions.
26024 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26026 * function.h (struct function): Update comment for field
26028 * gimple-expr.c: Include "tree-pass.h".
26029 (gimple_has_body_p): Return false for __RTL functions.
26030 * Makefile.in (OBJS): Add run-rtl-passes.o.
26031 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26033 (gcc::pass_manager::get_clean_slate): New accessor.
26034 * passes.c: Include "insn-addr.h".
26035 (should_skip_pass_p): Add logging. Update logic for running
26036 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26037 property-provider override so it is only done for gimple passes.
26039 (skip_pass): New function.
26040 (execute_one_pass): Call skip_pass when skipping passes.
26041 * read-md.c (md_reader::read_char): Support filtering
26042 the input to a subset of line numbers.
26043 (md_reader::md_reader): Initialize fields
26044 m_first_line and m_last_line.
26045 (md_reader::read_file_fragment): New function.
26046 * read-md.h (md_reader::read_file_fragment): New decl.
26047 (md_reader::m_first_line): New field.
26048 (md_reader::m_last_line): New field.
26049 * read-rtl-function.c (function_reader::create_function): Only
26050 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26051 curr_properties. Set DECL_INITIAL to a dummy block.
26052 (read_rtl_function_body_from_file_range): New function.
26053 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26055 * run-rtl-passes.c: New file.
26056 * run-rtl-passes.h: New file.
26058 2017-01-24 Jeff Law <law@redhat.com>
26060 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26063 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26065 PR tree-optimization/79159
26066 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26067 (record_nonwrapping_iv): Improve boundary using above function if no
26068 value range information.
26070 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26071 Martin Jambor <mjambor@suse.cz>
26073 * brig-builtins.def: New file.
26074 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26075 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26076 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26077 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26078 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26079 * builtin-types.def (BT_INT8): New.
26080 (BT_INT16): Likewise.
26081 (BT_UINT8): Likewise.
26082 (BT_UINT16): Likewise.
26083 (BT_FN_ULONG): Likewise.
26084 (BT_FN_UINT_INT): Likewise.
26085 (BT_FN_UINT_ULONG): Likewise.
26086 (BT_FN_UINT_LONG): Likewise.
26087 (BT_FN_UINT_PTR): Likewise.
26088 (BT_FN_ULONG_PTR): Likewise.
26089 (BT_FN_INT8_FLOAT): Likewise.
26090 (BT_FN_INT16_FLOAT): Likewise.
26091 (BT_FN_UINT32_FLOAT): Likewise.
26092 (BT_FN_UINT16_FLOAT): Likewise.
26093 (BT_FN_UINT8_FLOAT): Likewise.
26094 (BT_FN_UINT64_FLOAT): Likewise.
26095 (BT_FN_UINT16_UINT32): Likewise.
26096 (BT_FN_UINT32_UINT16): Likewise.
26097 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26098 (BT_FN_INT_PTR_INT): Likewise.
26099 (BT_FN_UINT_PTR_UINT): Likewise.
26100 (BT_FN_LONG_PTR_LONG): Likewise.
26101 (BT_FN_ULONG_PTR_ULONG): Likewise.
26102 (BT_FN_VOID_UINT64_UINT64): Likewise.
26103 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26104 (BT_FN_INT8_INT8_INT8): Likewise.
26105 (BT_FN_INT16_INT16_INT16): Likewise.
26106 (BT_FN_INT_INT_INT): Likewise.
26107 (BT_FN_UINT_FLOAT_UINT): Likewise.
26108 (BT_FN_FLOAT_UINT_UINT): Likewise.
26109 (BT_FN_ULONG_UINT_UINT): Likewise.
26110 (BT_FN_ULONG_UINT_PTR): Likewise.
26111 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26112 (BT_FN_UINT_UINT_UINT): Likewise.
26113 (BT_FN_VOID_UINT_PTR): Likewise.
26114 (BT_FN_UINT_UINT_PTR: Likewise.
26115 (BT_FN_UINT32_UINT64_PTR): Likewise.
26116 (BT_FN_INT_INT_UINT_UINT): Likewise.
26117 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26118 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26119 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26120 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26121 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26122 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26123 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26124 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26125 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26126 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26127 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26128 * doc/frontends.texi: List BRIG FE.
26129 * doc/install.texi (Testing): Add BRIG tesring requirements.
26130 * doc/invoke.texi (Overall Options): Mention BRIG.
26131 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26133 2017-01-24 Richard Biener <rguenther@suse.de>
26135 PR translation/79208
26136 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26138 2017-01-24 Martin Jambor <mjambor@suse.cz>
26141 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26142 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26143 and known_contexts.
26145 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26147 PR middle-end/79123
26148 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26149 casts from signed to unsigned really don't have a range.
26151 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26153 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26154 GMP_RNDx for compatiblity.
26156 2017-01-24 Martin Liska <mliska@suse.cz>
26159 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26160 that would prevent us to call alloca with -1 as argument.
26162 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26164 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26165 Avoid -Wformat-security warning.
26167 2017-01-23 Andrew Pinski <apinski@cavium.com>
26169 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26172 2017-01-23 Martin Sebor <msebor@redhat.com>
26174 PR middle-end/78703
26175 * gimple-ssa-sprintf.c (warn_level): New global.
26176 (format_integer): Use it here and throughout the rest of the file.
26177 Use the same switch to compute sign as base.
26178 (maybe_warn): New function.
26179 (format_directive): Factor out warnings into maybe_warn.
26180 Add debugging output. Use warn_level.
26181 (add_bytes): Use warn_level.
26182 (pass_sprintf_length::compute_format_length): Add debugging output.
26183 (try_substitute_return_value): Same.
26184 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26186 PR middle-end/78703
26187 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26188 (struct fmtresult, format_integer, format_floating): Adjust.
26189 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26190 (get_string_length, format_string,format_directive): Same.
26191 (pass_sprintf_length::compute_format_length): Same.
26192 (try_substitute_return_value): Simplify slightly.
26194 PR middle-end/78703
26195 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26196 (fmtresult::operator+=): Outlined.
26197 (struct fmtresult): Add ctors.
26198 (struct conversion_spec): Rename...
26199 (struct directive): ...to this. Add and remove data members.
26200 (directive::set_width, directive::set_precision): New functions.
26201 (format_percent): Use fmtresult ctor.
26202 (get_width_and_precision): Remove.
26203 (format_integer): Make naming changes. Avoid computing width and
26205 (format_floating): Same. Adjust indentation.
26206 (format_character, format_none): New functions.
26207 (format_string): Moved character handling to format_character.
26208 (format_directive): Remove arguments, change return type.
26209 (parse_directive): New function.
26210 (pass_sprintf_length::compute_format_length): Move directive
26211 parsing to parse_directive.
26213 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26215 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26216 (assign_assembler_name_if_needed): ... this.
26217 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26218 (assign_assembler_name_if_needed): ... this.
26219 (free_lang_data_in_cgraph): Adjust callers.
26220 * cgraphunit.c (cgraph_node::analyze): Likewise.
26221 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26224 2017-01-23 Richard Biener <rguenther@suse.de>
26226 PR tree-optimization/79088
26227 PR tree-optimization/79188
26228 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26229 resetting loop bounds after last path deletion. Reset loop
26230 bounds of the target loop, make code match the comments.
26231 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26232 Make sure loops need no fixups.
26234 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26236 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26237 exponent support with double type for first argument.
26238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26239 type returned by __builtin_vec_extract_sig,
26240 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26241 functions from "vector int" to "vector unsigned int" or from
26242 "vector long long int" to "vector unsigned long long int".
26243 Changed type returned by __builtin_vec_extract_exp,
26244 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26245 functions from "vector int" to "vector unsigned int" or from
26246 "vector long long int" to "vector unsigned long long int".
26247 Changed return type of __builtin_vec_test_data_class,
26248 __builtin_vec_test_data_class_sp, and
26249 __builtin_vec_test_data_class_dp from "vector int" to
26250 "vector bool int" or from "vector long long int" to "vector bool
26251 long long int" and changed second argument type from "unsigned
26252 int" to "int". Added new overloaded function forms "vector float
26253 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26254 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26255 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26256 double, vector unsigned long long int)" and "vector double
26257 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26258 long int)". Changed return type of
26259 __builtin_scalar_test_data_class and
26260 __builtin_scalar_test_data_class_sp and
26261 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26262 int" and changed second argument from "unsigned int" to "int".
26263 Changed type returned by __builtin_scalar_test_neg,
26264 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26265 from "int" to "bool int". Added new overloaded function form
26266 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26267 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26268 exponent double-precision with floating point first argument.
26269 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26270 documentation of scalar_test_data_class, scalar_test_neg,
26271 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26272 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26273 vec_test_data_class built-in functions to reflect refinements in
26274 their type signatures.
26276 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26278 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26280 (aarch64_elf_asm_destructor): Likewise.
26282 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26284 PR rtl-optimization/78634
26285 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26286 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26287 * ifcvt.c (noce_try_cmove): Add missing cost check.
26289 PR rtl-optimization/71724
26290 * combine.c (if_then_else_cond): Look for situations where it is
26291 beneficial to undo the work of one of the recursive calls.
26293 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26295 PR tree-optimization/70754
26296 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26297 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26298 combined stmt before it if not NULL.
26299 (combine_chains): Process refs reversely and compute dominance point
26302 2017-01-23 Martin Liska <mliska@suse.cz>
26304 PR tree-optimization/79196
26305 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26306 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26308 (strlen_optimize_stmt): Call the renamed function.
26310 2017-01-23 Michael Matz <matz@suse.de>
26312 PR tree-optimization/78384
26313 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26315 2017-01-23 Richard Biener <rguenther@suse.de>
26317 PR tree-optimization/79186
26318 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26319 both incoming edges before moving an assert.
26321 2017-01-23 Martin Jambor <mjambor@suse.cz>
26323 * ipa-prop.c (load_from_param_1): Removed.
26324 (load_from_unmodified_param): Bits from load_from_param_1 put back
26326 (load_from_param): Removed.
26327 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26328 with stmt. Reverted back to use of load_from_unmodified_param.
26330 2017-01-23 Martin Jambor <mjambor@suse.cz>
26333 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26334 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26335 field a pointer to garbage collected vector, mark lattices and
26336 ipcp_orig_node with GTY((skip)).
26337 (ipa_get_param_count): Adjust to descriptors being a pointer.
26338 (ipa_get_param): Likewise.
26339 (ipa_get_type): Likewise.
26340 (ipa_get_param_move_cost): Likewise.
26341 (ipa_set_param_used): Likewise.
26342 (ipa_get_controlled_uses): Likewise.
26343 (ipa_set_controlled_uses): Likewise.
26344 (ipa_is_param_used): Likewise.
26345 (ipa_node_params_t): Move into garbage collector. New methods insert
26347 (ipa_node_params_sum): Annotate wth GTY(()).
26348 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26350 (ipa_load_from_parm_agg): Adjust declaration.
26351 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26352 * ipa-profile.c (ipa_profile): Likewise.
26353 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26354 (ipa_populate_param_decls): Make descriptors parameter garbage
26356 (ipa_dump_param): Adjust to descriptors being a pointer.
26357 (ipa_alloc_node_params): Likewise.
26358 (ipa_initialize_node_params): Likewise.
26359 (load_from_param_1): Make descriptors parameter garbage collected.
26360 (load_from_unmodified_param): Likewise.
26361 (load_from_param): Likewise.
26362 (ipa_load_from_parm_agg): Likewise.
26363 (ipa_node_params::~ipa_node_params): Removed.
26364 (ipa_free_all_node_params): Remove call to delete operator.
26365 (ipa_node_params_t::insert): New.
26366 (ipa_node_params_t::remove): Likewise.
26367 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26368 copy known_csts and known_contexts vectors.
26369 (ipa_read_node_info): Adjust to descriptors being a pointer.
26370 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26372 (ipcp_transform_function): Make descriptors variable garbage
26375 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26377 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26378 * config/i386/avx512dqintrin.h: Ditto.
26379 * config/i386/avx512fintrin.h: Ditto.
26380 * config/i386/i386.c: Handle new builtins.
26381 * config/i386/i386-builtin.def: Add new builtins.
26382 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26383 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26385 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26386 Martin Liska <mliska@suse.cz>
26388 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26389 * asan.c (asan_expand_poison_ifn): Support stores and use
26390 appropriate ASAN report function.
26391 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26392 * internal-fn.def (ASAN_POISON_USE): Declare.
26393 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26394 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26395 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26396 ASAN_POISON calls w/o LHS.
26397 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26398 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26399 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26400 * gimplify.c (asan_poison_variables): Add attribute
26401 use_after_scope_memory to variables that really needs to live
26403 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26404 having the attribute.
26406 2017-01-23 Martin Liska <mliska@suse.cz>
26408 * asan.c (create_asan_shadow_var): New function.
26409 (asan_expand_poison_ifn): Likewise.
26410 * asan.h (asan_expand_poison_ifn): New declaration.
26411 * internal-fn.c (expand_ASAN_POISON): Likewise.
26412 * internal-fn.def (ASAN_POISON): New builtin.
26413 * sanopt.c (pass_sanopt::execute): Expand
26414 asan_expand_poison_ifn.
26415 * tree-inline.c (copy_decl_for_dup_finish): Make function
26417 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26418 * tree-ssa.c (is_asan_mark_p): New function.
26419 (execute_update_addresses_taken): Rewrite local variables
26420 (identified just by use-after-scope as addressable) into SSA.
26422 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26424 * doc/install.texi (Specific): opensource.apple.com uses https
26425 now. Remove trailing slash.
26427 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26429 * README.Portability: Remove note on an Irix compatibility issue.
26431 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26433 * gcov.c (INCLUDE_ALGORITHM): Define.
26434 (INCLUDE_VECTOR): Define.
26435 No longer include <vector> and <algorithm> directly.
26437 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26439 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26441 * doc/invoke.texi (Code Gen Options): Ditto.
26443 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26446 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26448 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26450 rtl-optimization/79125
26451 * cprop.c (local_cprop_pass): Handle cases where we make an
26452 unconditional trap.
26454 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26458 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26459 read from, for big endian.
26461 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26463 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26464 register pauth builtins for LP64 only.
26466 2017-01-20 Marek Polacek <polacek@redhat.com>
26469 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26472 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26474 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26476 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26477 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26478 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26480 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26483 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26484 in big-endian mode when they are not a single duplicated value.
26486 2017-01-20 Richard Biener <rguenther@suse.de>
26488 * BASE-VER: Bump to 7.0.1.
26490 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26492 * omp-low.c (omplow_simd_context): New struct. Use it...
26493 (lower_rec_simd_input_clauses): ...here and...
26494 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26495 references to idx, lane, max_vf, is_simt.
26497 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26499 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26502 2017-01-20 Martin Jambor <mjambor@suse.cz>
26504 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26505 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26507 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26508 (GTFILES): Rename hsa.c to hsa-common.c.
26509 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26510 * hsa-dump.c: Likewise.
26511 * hsa-gen.c: Likewise.
26512 * hsa-regalloc.c: Likewise.
26513 * ipa-hsa.c: Likewise.
26514 * omp-expand.c: Likewise.
26515 * omp-low.c: Likewise.
26516 * toplev.c: Likewise.
26518 2017-01-20 Marek Polacek <polacek@redhat.com>
26521 * doc/invoke.texi: Document -Wduplicated-branches.
26522 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26523 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26524 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26525 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26526 return 0 only when not OEP_LEXICOGRAPHIC.
26527 (fold_build_cleanup_point_expr): Use the expression
26528 location when building CLEANUP_POINT_EXPR.
26529 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26530 * tree.c (add_expr): Handle error_mark_node.
26532 2017-01-20 Martin Liska <mliska@suse.cz>
26535 * tree-profile.c (init_ic_make_global_vars): Do not call
26537 (gimple_init_gcov_profiler): Likewise.
26539 2017-01-20 Martin Liska <mliska@suse.cz>
26542 * cgraph.h (maybe_create_reference): Remove argument and
26544 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26546 * ipa-cp.c (create_specialized_node): Likewise.
26547 * symtab.c (symtab_node::maybe_create_reference): Handle
26548 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26550 2017-01-20 Martin Liska <mliska@suse.cz>
26552 * read-rtl-function.c (function_reader::create_function): Use
26553 build_decl instread of build_decl_stat.
26555 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26557 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26558 * config/i386/avx512dqintrin.h: Ditto.
26559 * config/i386/avx512fintrin.h: Ditto.
26560 * config/i386/i386-builtin-types.def: Add new types.
26561 * config/i386/i386.c: Handle new types.
26562 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26563 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26564 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26565 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26566 (__builtin_ia32_kshiftridi): New.
26567 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26569 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26573 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26574 define to rs6000_init_stack_protect_guard.
26575 (rs6000_init_stack_protect_guard): New function.
26577 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26578 Yunqiang Su <yunqiang.su@imgtec.com>
26580 * config.gcc (supported_defaults): Add madd4.
26581 (with_madd4): Add validation.
26582 (all_defaults): Add madd4.
26583 * config/mips/mips.opt (mmadd4): New option.
26584 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26586 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26588 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26589 (ISA_HAS_FUSED_MADD4): Likewise.
26590 * doc/invoke.texi (-mmadd4): Document the new option.
26591 * doc/install.texi (--with-madd4): Document the new option.
26593 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26595 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26596 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26597 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26598 (aarch64_init_pauth_hint_builtins): New.
26599 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26600 (aarch64_expand_builtin): Expand new builtins.
26602 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26604 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26605 * combine-stack-adj.c (no_unhandled_cfa): Handle
26606 REG_CFA_TOGGLE_RA_MANGLE.
26607 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26608 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26609 info for return address signing.
26610 (aarch64_expand_epilogue): Likewise.
26612 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26614 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26615 * config/aarch64/aarch64-protos.h
26616 (aarch64_return_address_signing_enabled): New declaration.
26617 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26619 (aarch64_expand_prologue): Sign return address before it's pushed onto
26621 (aarch64_expand_epilogue): Authenticate return address fetched from
26623 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26624 (aarch64_attributes): New function attributes for "sign-return-address".
26625 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26626 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26627 ("*do_return"): Generate combined instructions according to key index.
26628 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26629 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26631 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26632 * config/aarch64/aarch64.opt (msign-return-address=): New.
26633 * doc/extend.texi (AArch64 Function Attributes): Documents
26634 "sign-return-address=".
26635 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26637 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26639 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26640 overall option summary.
26642 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26644 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26645 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26646 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26647 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26649 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26651 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26652 -mpower9-minmax by default for -mcpu=power9.
26653 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26654 128-bit floating point.
26656 2017-01-20 Alan Modra <amodra@gmail.com>
26658 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26659 optimizing for size.
26661 2017-01-20 Alan Modra <amodra@gmail.com>
26664 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26665 for strcmp and strncmp from corresponding builtin decl.
26667 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26669 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26670 instead of i386/rtems-64.h.
26671 * config/i386/rtems-64.h: Remove.
26673 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26677 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26679 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26681 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26683 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26684 Change int to HOST_WIDE_INT.
26685 * config/aarch64/aarch64-protos.h
26686 (aarch64_simd_gen_const_vector_dup): Likewise.
26687 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26689 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26691 * langhooks-def.h (lhd_type_for_size): New decl.
26692 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26693 * langhooks.c (lhd_type_for_size): New function, taken from
26696 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26698 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26699 define_bypass for CR latency.
26700 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26701 (power9-alu2): Add define_bypass for CR latency.
26703 (power9-mul): Update insn latency.
26704 (power9-mul-compare): Update insn latency, bypass latency and remove
26707 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26709 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26711 * config/aarch64/aarch64.md
26712 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26713 aarch64_nopcrelative_literal_loads.
26714 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26716 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26718 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26719 TARGET_LOONGSON_3A.
26720 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26722 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26725 * config.gcc (supported_defaults): Add lxc1-sxc1.
26726 (with_lxc1_sxc1): Add validation.
26727 (all_defaults): Add lxc1-sxc1.
26728 * config/mips/mips.opt (mlxc1-sxc1): New option.
26729 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26731 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26732 __mips_no_lxc1_sxc1.
26733 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26734 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26735 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26737 2017-01-19 Richard Biener <rguenther@suse.de>
26739 PR tree-optimization/72488
26740 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26741 sure to restore SSA info.
26742 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26744 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26746 PR rtl-optimization/79121
26747 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26748 of the inner type when shifting an extended value.
26750 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26753 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26754 interposable aliases.
26756 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26759 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26760 Use the evmergelohi instruction.
26761 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26762 (mov_sitf_e500_subreg8_2_be): Likewise.
26763 (mov_sitf_e500_subreg12_2_le): Likewise.
26764 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26765 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26766 (mov_sitf_e500_subreg8_2_le): Likewise.
26767 (mov_sitf_e500_subreg12_2_be): Likewise.
26769 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26771 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26772 attribute from vecsimple to vecperm.
26773 (altivec_vbpermq2): Likewise.
26775 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26778 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26780 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26781 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26782 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26783 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26784 case where N arg is SIZE_MAX.
26785 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26786 (cmpstrsi): Add pattern.
26788 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26790 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26791 __builtin_vec_revb builtins.
26792 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26793 built-in functions to support generation of the ISA 3.0 XXBR<x>
26794 vector byte reverse instructions.
26795 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26796 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26797 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26798 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26799 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26800 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26801 (P9V_BUILTIN_VEC_REVB): Likewise.
26802 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26803 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26804 (p9_xxbrq_v16qi): Likewise.
26805 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26806 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26807 (p9_xxbrh_v8hi): Likewise.
26808 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26809 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26810 vec_revb built-in functions.
26812 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26814 PR rtl-optimization/78952
26815 * config/i386/i386.md (any_extract): New code iterator.
26816 (*insvqi_2): Use any_extract for source operand.
26817 (*insvqi_3): Use any_shiftrt for source operand.
26819 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26821 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26823 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26825 2017-01-18 Matthias Klose <doko@ubuntu.com>
26827 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26829 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26831 * config/rs6000/altivec.h (vec_bperm): Change #define.
26832 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26833 (altivec_vbpermq2): New define_insn.
26834 (altivec_vbpermd): Likewise.
26835 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26836 function interface.
26837 (VBPERMD): Likewise.
26838 (VBPERM): New polymorphic function interface.
26839 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26840 Add entries for P9V_BUILTIN_VEC_VBPERM.
26841 * doc/extend.texi: Add interfaces for vec_bperm.
26843 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26845 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26846 first letter of error messages.
26847 (s390_resolve_overloaded_builtin): Likewise.
26848 * config/s390/s390.c (s390_expand_builtin): Likewise.
26849 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26850 (s390_valid_target_attribute_inner_p): Likewise.
26851 * config/s390/s390.md ("tabort"): Likewise.
26853 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26855 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26856 (ISA_AVOID_DIV_HILO): New macro.
26857 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26858 (ISA_HAS_DDIV): Likewise.
26860 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26862 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26864 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26866 * doc/invoke.texi (fabi-version): Spelling fix.
26868 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26871 * doc/invoke.texi (fabi-version): Mention mangling fix for
26874 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26877 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26879 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26882 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26885 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26887 (stack_protect_test): Ditto.
26888 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26889 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26891 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26892 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26893 -mstack-protector-guard-offset=.
26894 (RS/6000 and PowerPC Options): Ditto.
26896 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26898 * config/i386/i386.h (MASK_CLASS_P): New define.
26899 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26900 there are no registers from different register sets also when
26901 mask registers are used. Update function comment.
26902 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26903 to (*k/*r) and (*k/*km) alternatives.
26905 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26907 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26908 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26909 (EH_RETURN_HANDLER_RTX): New define.
26910 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26911 Force frame pointer in EH return functions.
26912 (aarch64_expand_epilogue): Add barrier for eh_return.
26913 (aarch64_final_eh_return_addr): Remove.
26914 (aarch64_eh_return_handler_rtx): New function.
26915 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26917 (aarch64_eh_return_handler_rtx): New prototype.
26919 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26921 * config/rs6000/altivec.h (vec_rlmi): New #define.
26922 (vec_vrlnm): Likewise.
26923 (vec_rlnm): Likewise.
26924 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26925 (UNSPEC_VRLNM): Likewise.
26926 (VIlong): New mode iterator.
26927 (altivec_vrl<VI_char>mi): New define_insn.
26928 (altivec_vrl<VI_char>nm): Likewise.
26929 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26931 (VRLDNM): Likewise.
26932 (RLNM): New polymorphic function entry.
26933 (VRLWMI): New monomorphic function entry.
26934 (VRLDMI): Likewise.
26935 (RLMI): New polymorphic function entry.
26936 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26937 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26938 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26941 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26944 * dwarf2out.c (field_byte_offset): Restore the
26945 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26946 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26947 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26950 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26953 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26955 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26958 * dwarf2out.c (add_data_member_location_attribute): For constant
26959 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26960 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26961 DW_AT_byte_size attributes.
26963 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26965 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26966 after forcing to constant memory when the code model is medium.
26968 2017-01-17 Julia Koval <julia.koval@intel.com>
26971 * config/i386/avx512fintrin.h
26972 (_mm512_i32gather_ps): Change __addr type to void const*.
26973 (_mm512_mask_i32gather_ps): Ditto.
26974 (_mm512_i32gather_pd): Ditto.
26975 (_mm512_mask_i32gather_pd): Ditto.
26976 (_mm512_i64gather_ps): Ditto.
26977 (_mm512_mask_i64gather_ps): Ditto.
26978 (_mm512_i64gather_pd): Ditto.
26979 (_mm512_mask_i64gather_pd): Ditto.
26980 (_mm512_i32gather_epi32): Ditto.
26981 (_mm512_mask_i32gather_epi32): Ditto.
26982 (_mm512_i32gather_epi64): Ditto.
26983 (_mm512_mask_i32gather_epi64): Ditto.
26984 (_mm512_i64gather_epi32): Ditto.
26985 (_mm512_mask_i64gather_epi32): Ditto.
26986 (_mm512_i64gather_epi64): Ditto.
26987 (_mm512_mask_i64gather_epi64): Ditto.
26988 (_mm512_i32scatter_ps): Change __addr type to void*.
26989 (_mm512_mask_i32scatter_ps): Ditto.
26990 (_mm512_i32scatter_pd): Ditto.
26991 (_mm512_mask_i32scatter_pd): Ditto.
26992 (_mm512_i64scatter_ps): Ditto.
26993 (_mm512_mask_i64scatter_ps): Ditto.
26994 (_mm512_i64scatter_pd): Ditto.
26995 (_mm512_mask_i64scatter_pd): Ditto.
26996 (_mm512_i32scatter_epi32): Ditto.
26997 (_mm512_mask_i32scatter_epi32): Ditto.
26998 (_mm512_i32scatter_epi64): Ditto.
26999 (_mm512_mask_i32scatter_epi64): Ditto.
27000 (_mm512_i64scatter_epi32): Ditto.
27001 (_mm512_mask_i64scatter_epi32): Ditto.
27002 (_mm512_i64scatter_epi64): Ditto.
27003 (_mm512_mask_i64scatter_epi64): Ditto.
27004 * config/i386/avx512pfintrin.h
27005 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27006 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27007 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27008 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27009 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27010 (_mm512_prefetch_i32scatter_ps): Ditto.
27011 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27012 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27013 (_mm512_prefetch_i64scatter_pd): Ditto.
27014 (_mm512_prefetch_i64scatter_ps): Ditto.
27015 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27016 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27017 * config/i386/avx512vlintrin.h
27018 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27019 (_mm_mmask_i32gather_ps): Ditto.
27020 (_mm256_mmask_i32gather_pd): Ditto.
27021 (_mm_mmask_i32gather_pd): Ditto.
27022 (_mm256_mmask_i64gather_ps): Ditto.
27023 (_mm_mmask_i64gather_ps): Ditto.
27024 (_mm256_mmask_i64gather_pd): Ditto.
27025 (_mm_mmask_i64gather_pd): Ditto.
27026 (_mm256_mmask_i32gather_epi32): Ditto.
27027 (_mm_mmask_i32gather_epi32): Ditto.
27028 (_mm256_mmask_i32gather_epi64): Ditto.
27029 (_mm_mmask_i32gather_epi64): Ditto.
27030 (_mm256_mmask_i64gather_epi32): Ditto.
27031 (_mm_mmask_i64gather_epi32): Ditto.
27032 (_mm256_mmask_i64gather_epi64): Ditto.
27033 (_mm_mmask_i64gather_epi64): Ditto.
27034 (_mm256_i32scatter_ps): Change __addr type to void*.
27035 (_mm256_mask_i32scatter_ps): Ditto.
27036 (_mm_i32scatter_ps): Ditto.
27037 (_mm_mask_i32scatter_ps): Ditto.
27038 (_mm256_i32scatter_pd): Ditto.
27039 (_mm256_mask_i32scatter_pd): Ditto.
27040 (_mm_i32scatter_pd): Ditto.
27041 (_mm_mask_i32scatter_pd): Ditto.
27042 (_mm256_i64scatter_ps): Ditto.
27043 (_mm256_mask_i64scatter_ps): Ditto.
27044 (_mm_i64scatter_ps): Ditto.
27045 (_mm_mask_i64scatter_ps): Ditto.
27046 (_mm256_i64scatter_pd): Ditto.
27047 (_mm256_mask_i64scatter_pd): Ditto.
27048 (_mm_i64scatter_pd): Ditto.
27049 (_mm_mask_i64scatter_pd): Ditto.
27050 (_mm256_i32scatter_epi32): Ditto.
27051 (_mm256_mask_i32scatter_epi32): Ditto.
27052 (_mm_i32scatter_epi32): Ditto.
27053 (_mm_mask_i32scatter_epi32): Ditto.
27054 (_mm256_i32scatter_epi64): Ditto.
27055 (_mm256_mask_i32scatter_epi64): Ditto.
27056 (_mm_i32scatter_epi64): Ditto.
27057 (_mm_mask_i32scatter_epi64): Ditto.
27058 (_mm256_i64scatter_epi32): Ditto.
27059 (_mm256_mask_i64scatter_epi32): Ditto.
27060 (_mm_i64scatter_epi32): Ditto.
27061 (_mm_mask_i64scatter_epi32): Ditto.
27062 (_mm256_i64scatter_epi64): Ditto.
27063 (_mm256_mask_i64scatter_epi64): Ditto.
27064 (_mm_i64scatter_epi64): Ditto.
27065 (_mm_mask_i64scatter_epi64): Ditto.
27066 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27067 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27068 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27069 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27070 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27071 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27072 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27073 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27074 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27075 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27076 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27077 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27078 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27079 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27080 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27081 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27082 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27083 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27084 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27085 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27086 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27087 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27088 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27089 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27090 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27091 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27092 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27093 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27094 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27095 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27096 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27097 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27098 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27099 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27100 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27101 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27102 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27103 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27104 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27105 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27106 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27107 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27108 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27109 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27110 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27111 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27112 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27113 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27114 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27115 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27116 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27117 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27118 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27119 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27120 definitions accordingly.
27122 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27123 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27126 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27129 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27132 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27133 hard regs for inner regno.
27135 2017-01-17 Martin Liska <mliska@suse.cz>
27138 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27139 assumption and add comment.
27141 2017-01-17 Nathan Sidwell <nathan@acm.org>
27143 * ipa-visibility.c (localize_node): New function, broken out of ...
27144 (function_and_variable_visibility): ... here. Call it.
27146 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27148 PR middle-end/77445
27149 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27150 correctly set frequency of oudgoing edge.
27151 (duplicate_thread_path): Fix profile updating.
27153 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27156 * configure.ac: Add GCC_BASE_VER.
27157 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27158 version from BASE-VER file.
27159 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27160 (gcc.o): Depend on $(BASEVER).
27161 * common.opt (dumpfullversion): New option.
27162 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27163 * doc/invoke.texi: Document -dumpfullversion.
27164 * doc/install.texi: Document --with-gcc-major-version-only.
27165 * configure: Regenerated.
27167 2017-01-17 Richard Biener <rguenther@suse.de>
27169 PR tree-optimization/71433
27170 * tree-vrp.c (register_new_assert_for): Merge same asserts
27171 on all incoming edges.
27172 (process_assert_insertions_for): Handle insertions at the
27175 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27177 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27178 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27180 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27183 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27186 2017-01-17 Alan Modra <amodra@gmail.com>
27189 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27190 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27191 symbolic stack limit when pic.
27193 2017-01-16 Martin Sebor <msebor@redhat.com>
27195 PR tree-optimization/78608
27196 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27198 2017-01-16 Jeff Law <law@redhat.com>
27201 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27202 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27203 for several include directories that may be relative to sysroot.
27204 * config/i386/x-mingw32 (gplus_includedir): Define.
27205 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27206 (native_system_includedir): Likewise.
27207 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27208 override if TARGET_SYSTEM_ROOT is defined.
27209 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27211 PR tree-optimization/79090
27212 PR tree-optimization/33562
27213 PR tree-optimization/61912
27214 PR tree-optimization/77485
27215 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27216 and computed trims into the dump file.
27218 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27220 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27222 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27225 * gimplify.c (gimplify_init_constructor): If want_value and
27226 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27230 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27231 sequence. Formatting fixes.
27232 (doloop_optimize): Formatting fixes.
27235 * gcc.c (debug_level_greater_than_spec_func): New function.
27236 (static_spec_functions): Add debug-level-gt spec function.
27237 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27239 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27240 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27241 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27242 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27243 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27244 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27246 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27248 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27249 QImode fixups to general and mask registers only.
27251 2017-01-16 Carl Love <cel@us.ibm.com>
27253 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27254 for built-in functions
27255 vector signed char vec_nabs (vector signed char)
27256 vector signed short vec_nabs (vector signed short)
27257 vector signed int vec_nabs (vector signed int)
27258 vector signed long long vec_nabs (vector signed long long)
27259 vector float vec_nabs (vector float)
27260 vector double vec_nabs (vector double)
27261 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27263 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27264 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27265 * doc/extend.texi: Update the documentation file for the new built-in
27268 2017-01-16 Martin Sebor <msebor@redhat.com>
27270 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27273 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27275 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27276 UNSPEC_VSX__XXSPLTD to require special splat handling.
27278 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27281 * system.h: Poison strndup.
27283 2017-01-16 Alan Modra <amodra@gmail.com>
27286 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27289 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27291 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27293 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27295 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27296 call recog here. Assert that INSN_CODE (insn) is non-negative.
27298 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27301 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27303 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27304 in the currently scheduled RTL fragment.
27306 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27308 PR rtl-optimization/78751
27309 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27312 2017-01-14 Jeff Law <law@redhat.com>
27314 PR tree-optimization/79090
27315 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27316 variable length stores.
27317 (compute_trims): Delete dead assignment to *trim_tail.
27318 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27321 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27323 PR rtl-optimization/78626
27324 PR rtl-optimization/78727
27325 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27326 of a block, and split such blocks after everything else is finished.
27328 2017-01-14 Alan Modra <amodra@gmail.com>
27331 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27332 target legitimate_combined_insn.
27333 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27334 (rs6000_legitimate_combined_insn): New function.
27335 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27337 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27338 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27339 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27341 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27343 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27345 2017-01-13 Jeff Law <law@redhat.com>
27347 PR tree-optimization/33562
27348 PR tree-optimization/61912
27349 PR tree-optimization/77485
27350 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27352 (delete_dead_assignment): Likewise.
27353 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27354 statement to delete_dead_call and delete_dead_assignment.
27356 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27359 * substring-locations.c (format_warning_va): Strengthen case 1 so
27360 that both endpoints of the substring must be within the format
27361 range for just the substring to be printed.
27363 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27365 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27366 * config/i386/i386.c (ix86_target_string): Add missing options
27367 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27368 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27369 flags_other and ix86_target_other to flags2_other. Display unknown
27371 (ix86_valid_target_attribute_inner_p): Add missing options and
27372 reorder options by implied ISAs, as in ix86_target_string.
27374 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27376 * hash-table.h (hash_table::too_empty_p): New function.
27377 (hash_table::expand): Use it.
27378 (hash_table::traverse): Likewise.
27379 (hash_table::empty_slot): Use sizeof (value_type) instead of
27380 sizeof (PTR) to convert bytes to elements. Shrink the table
27381 if the current size is excessive for the current number of
27384 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27386 * ira-costs.c (record_reg_classes): Break from the inner loop
27387 early once alt_fail is known to be true. Update outer loop
27388 handling accordingly.
27390 2017-01-13 Jeff Law <law@redhat.com>
27392 * tree-ssa-dse.c (decrement_count): New function.
27393 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27394 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27395 when we know the partially dead statement is a mem* function.
27397 PR tree-optimization/61912
27398 PR tree-optimization/77485
27399 * tree-ssa-dse.c: Include expr.h.
27400 (maybe_trim_constructor_store): New function.
27401 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27403 PR tree-optimization/33562
27404 PR tree-optimization/61912
27405 PR tree-optimization/77485
27406 * doc/invoke.texi: Document new dse-max-object-size param.
27407 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27408 * tree-ssa-dse.c: Include params.h.
27409 (dse_store_status): New enum.
27410 (initialize_ao_ref_for_dse): New, partially extracted from
27412 (valid_ao_ref_for_dse, normalize_ref): New.
27413 (setup_live_bytes_from_ref, compute_trims): Likewise.
27414 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27415 (maybe_trim_partially_dead_store): Likewise.
27416 (maybe_trim_complex_store): Likewise.
27417 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27418 Track what bytes live from the original store. Return tri-state
27419 for dead, partially dead or live.
27420 (dse_dom_walker): Add constructor, destructor and new private members.
27421 (delete_dead_call, delete_dead_assignment): New extracted from
27423 (dse_optimize_stmt): Make a member of dse_dom_walker.
27424 Use initialize_ao_ref_for_dse.
27426 PR tree-optimization/33562
27427 PR tree-optimization/61912
27428 PR tree-optimization/77485
27429 * sbitmap.h (bitmap_count_bits): Prototype.
27430 (bitmap_clear_range, bitmap_set_range): Likewise.
27431 * sbitmap.c (bitmap_clear_range): New function.
27432 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27434 2017-01-13 Martin Liska <mliska@suse.cz>
27437 * function.c (set_cfun): Add new argument force.
27438 * function.h (set_cfun): Likewise.
27439 * ipa-inline-transform.c (inline_call): Use the function when
27440 strict alising from is dropped for function we inline to.
27442 2017-01-13 Richard Biener <rguenther@suse.de>
27444 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27445 for dumping GIMPLE INTEGER_CSTs.
27447 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27449 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27450 to 201112L since C++17.
27452 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27455 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27456 if -fsanitize=kernel-address is present.
27458 2017-01-13 Richard Biener <rguenther@suse.de>
27460 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27461 as _Literal ( type ) number in case usual suffixes do not
27462 preserve all information.
27464 2017-01-13 Richard Biener <rguenther@suse.de>
27466 PR tree-optimization/77283
27467 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27468 and ssa-iterators.h.
27469 (is_feasible_trace): Implement a cost model based on joiner
27472 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27475 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27476 char or short to __float128/_Float128 directly.
27478 2017-01-12 Martin Sebor <msebor@redhat.com>
27480 to -Wformat-overflow.
27481 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27482 (min_bytes_remaining): Same.
27483 (get_string_length): Same.
27484 (format_string): Same.
27485 (format_directive): Same.
27487 (pass_sprintf_length::handle_gimple_call): Same.
27489 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27491 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27492 info.nowrite calls with no lhs that can't throw. Return bool
27493 whether gsi_remove has been called or not.
27494 (pass_sprintf_length::handle_gimple_call): Return bool whether
27495 try_substitute_return_value called gsi_remove. Formatting fix.
27496 (pass_sprintf_length::execute): Don't use gsi_remove if
27497 handle_gimple_call returned true.
27500 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27501 be removed due to side-effects, don't remove following barrier nor
27502 turn the successor edge into fallthru edge.
27504 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27507 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27508 element-reversing loads and stores as not swappable.
27510 2017-01-12 Nathan Sidwell <nathan@acm.org>
27511 Nicolai Stange <nicstange@gmail.com>
27513 * combine.c (try_combine): Don't ignore result of overlap checking
27514 loop. Combine overlap & asm check into single loop.
27516 2017-01-12 Richard Biener <rguenther@suse.de>
27518 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27519 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27521 2017-01-12 Richard Biener <rguenther@suse.de>
27523 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27524 and TS_TARGET_OPTION directly derive from TS_BASE.
27525 * tree-core.h (tree_optimization_option): Derive from tree_base.
27526 (tree_target_option): Likewise.
27528 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27530 * config/i386/i386.c (memory_address_length): Increase len
27531 only when rip_relative_addr_p returns false.
27533 2017-01-11 Julia Koval <julia.koval@intel.com>
27535 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27536 (OPTION_MASK_ISA_SGX_SET): New.
27537 (ix86_handle_option): Handle OPT_msgx.
27538 * config.gcc: Added sgxintrin.h.
27539 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27540 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27541 * config/i386/i386.c (ix86_target_string): Add -msgx.
27543 (ix86_option_override_internal): Handle new options.
27544 (ix86_valid_target_attribute_inner_p): Add sgx.
27545 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27546 * config/i386/i386.opt: Add msgx.
27547 * config/i386/sgxintrin.h: New file.
27548 * config/i386/x86intrin.h: Add sgxintrin.h.
27550 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27553 * fold-const.c (maybe_nonzero_address): Return 1 for function
27555 (tree_single_nonzero_warnv_p): Don't handle function local objects
27559 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27562 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27565 * opts.c: Include "spellcheck.h"
27566 (struct string_fragment): New struct.
27567 (struct edit_distance_traits<const string_fragment &>): New
27569 (get_closest_sanitizer_option): New function.
27570 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27572 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27574 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27576 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27577 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27578 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27579 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27580 for initial die_offset if dwarf_split_debug_info.
27581 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27582 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27584 (output_skeleton_debug_sections): Formatting fix. Use
27585 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27586 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27588 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27590 * config/arm/cortex-a53.md: Add bypasses for
27591 cortex_a53_r2f_cvt.
27592 (cortex_a53_r2f): Only use for transfers.
27593 (cortex_a53_f2r): Likewise.
27594 (cortex_a53_r2f_cvt): Add reservation for conversions.
27595 (cortex_a53_f2r_cvt): Likewise.
27597 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27599 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27600 to all inlined functions, change static to extern.
27602 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27605 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27607 (arm_assemble_integer): Likewise.
27609 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27611 * config.gcc: Use new awk script to check CPU, FPU and architecture
27612 parameters for --with-... options.
27613 * config/arm/parsecpu.awk: New file
27614 * config/arm/arm-cpus.in: New file.
27615 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27617 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27619 * config/arm/t-arm: Update dependency rules.
27620 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27621 of processing .def files.
27622 * config/arm/genopt.sh: Deleted.
27623 * config/arm/gentune.sh: Deleted.
27624 * config/arm/arm-cores.def: Deleted.
27625 * config/arm/arm-arches.def: Deleted.
27626 * config/arm/arm-fpus.def: Deleted.
27627 * config/arm/arm-tune.md: Regenerated.
27628 * config/arm/arm-tables.opt: Regenerated.
27629 * config/arm/arm-cpu.h: New generated file.
27630 * config/arm/arm-cpu-data.h: New generated file.
27631 * config/arm/arm-cpu-cdata.h: New generated file.
27633 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27636 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27638 (input_varpool_node): Unpack dynamically_initialized bit.
27640 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27642 PR rtl-optimization/79032
27643 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27644 the alignment of the adjusted memory reference against that of MODE,
27645 instead of the alignment of the original memory reference.
27647 2017-01-11 Martin Jambor <mjambor@suse.cz>
27649 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27651 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27652 decorated functions.
27654 2017-01-11 Richard Biener <rguenther@suse.de>
27656 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27657 set range/nonnull info for PHI results. Do not set it on
27658 stmts marked for removal.
27660 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27662 * expr.c (store_field): In the bitfield case, fetch the return value
27663 from the registers before applying a single big-endian adjustment.
27664 Always do a final load for a BLKmode value not larger than a word.
27666 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27669 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27670 that we correctly handle column numbers greater than
27671 LINE_MAP_MAX_COLUMN_NUMBER.
27673 2017-01-10 Martin Sebor <msebor@redhat.com>
27675 PR middle-end/78245
27676 * gimple-ssa-sprintf.c (get_destination_size): Call
27677 {init,fini}object_sizes.
27678 * tree-object-size.c (addr_object_size): Adjust.
27679 (pass_through_call): Adjust.
27680 (pass_object_sizes::execute): Adjust.
27681 * tree-object-size.h (fini_object_sizes): Declare.
27683 2017-01-10 Martin Sebor <msebor@redhat.com>
27685 PR tree-optimization/78775
27686 * builtins.c (get_size_range): Move...
27687 * calls.c: ...to here.
27688 (alloc_max_size): Accept zero argument.
27689 (operand_signed_p): Remove.
27690 (maybe_warn_alloc_args_overflow): Call get_size_range.
27691 * calls.h (get_size_range): Declare.
27693 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27695 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27696 from TI's devices.csv file as of September 2016.
27697 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27699 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27701 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27702 * doc/invoke.texi: Likewise.
27703 * doc/md.texi: Likewise.
27704 * doc/objc.texi: Likewise.
27706 2017-01-10 Joshua Conner <joshconner@google.com>
27708 * config/arm/fuchsia-elf.h: New file.
27709 * config/fuchsia.h: New file.
27710 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27711 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27713 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27715 2016-01-10 Richard Biener <rguenther@suse.de>
27717 PR tree-optimization/79034
27718 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27719 Propagate out degenerate PHIs in the joiner.
27721 2017-01-10 Martin Liska <mliska@suse.cz>
27723 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27724 (sort_congruence_classes_by_decl_uid): Likewise.
27725 (sort_congruence_class_groups_by_decl_uid): Likewise.
27726 (sem_item_optimizer::merge_classes): Sort class, groups in these
27727 classes and members in the groups by DECL_UID of declarations.
27728 This would make merge operations stable.
27730 2017-01-10 Martin Liska <mliska@suse.cz>
27732 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27733 usage of m_classes_vec.
27734 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27735 (sem_item_optimizer::get_group_by_hash): Likewise.
27736 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27737 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27738 (sem_item_optimizer::verify_classes): Likewise.
27739 (sem_item_optimizer::process_cong_reduction): Likewise.
27740 (sem_item_optimizer::dump_cong_classes): Likewise.
27741 (sem_item_optimizer::merge_classes): Likewise.
27742 * ipa-icf.h (congruence_class_hash): Rename from
27743 congruence_class_group_hash. Remove declaration of m_classes_vec.
27745 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27748 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27749 * config.gcc: Add avx512vpopcntdqintrin.h.
27750 * config/i386/avx512vpopcntdqintrin.h: New.
27751 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27752 * config/i386/i386-builtin-types.def: Add new types.
27753 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27754 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27755 __builtin_ia32_vpopcountq_v8di_mask): New.
27756 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27757 __AVX512VPOPCNTDQ__.
27758 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27759 (PTA_AVX512VPOPCNTDQ): Define.
27760 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27761 TARGET_AVX512VPOPCNTDQ_P): Define.
27762 * config/i386/i386.opt: Add mavx512vpopcntdq.
27763 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27764 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27766 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27768 PR middle-end/77484
27769 * predict.def (PRED_CALL): Set to 67.
27771 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27773 * expr.c (store_field): In the bitfield case, if the value comes from
27774 a function call and is of an aggregate type returned in registers, do
27775 not modify the field mode; extract the value in all cases if the mode
27776 is BLKmode and the size is not larger than a word.
27778 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27781 * config/i386/cpuid.h: Fix undefined behavior.
27783 2017-01-04 Jeff Law <law@redhat.com>
27785 PR tree-optimization/79007
27786 PR tree-optimization/67955
27787 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27788 conservative for pt.null when flag_non_call_exceptions is on.
27790 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27792 PR translation/79019
27793 PR translation/79020
27794 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27795 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27796 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27798 * config/avr/avr.opt (maccumulate-args): Likewise.
27799 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27800 * common.opt (freport-bug): Likewise.
27801 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27802 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27803 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27804 translatable string.
27805 * config/i386/i386.c (function_value_32): Likewise.
27806 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27807 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27809 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27810 * common/config/msp430/msp430-common.c (msp430_handle_option):
27812 * symtab.c (symtab_node::verify_base): Likewise.
27813 * opts.c (set_debug_level): Likewise.
27814 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27815 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27816 missing whitespace to translatable strings.
27817 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27818 * config/sh/superh.h: Likewise.
27819 * config/i386/xopintrin.h: Likewise.
27820 * config/i386/znver1.md: Likewise.
27821 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27822 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27823 * double-int.h (struct double_int): Likewise.
27824 * double-int.c (div_and_round_double): Likewise.
27825 * wide-int.cc: Likewise.
27826 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27827 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27828 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27829 (crossjumps_occurred): ... this.
27830 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27833 PR tree-optimization/78899
27834 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27835 returning bool return struct loop *, NULL for failure and the new
27837 (versionable_outer_loop_p): Don't version outer loop if it has
27838 dont_vectorized bit set.
27839 (tree_if_conversion): When versioning outer loop, ensure
27840 tree_if_conversion is performed also on the inner loop of the
27841 non-vectorizable outer loop copy.
27842 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27843 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27844 prevent vectorization of it.
27845 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27846 the outer loop vectorization of the non-scalar version is attempted
27847 before vectorization of the inner loop in scalar version. If
27848 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27849 vectorization of its inner loop.
27850 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27851 has 2 inner loops, rename also on edges from bb whose single pred
27852 is outer_loop->header. Fix typo in function comment.
27854 2017-01-09 Martin Sebor <msebor@redhat.com>
27857 * asan.c (asan_emit_stack_protection): Increase local buffer size
27858 to avoid snprintf truncation warning.
27860 2017-01-09 Andrew Pinski <apinski@cavium.com>
27862 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27863 to reference thunderx2t99 for the tuning structure
27864 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27866 (thunderx2t99_extra_costs): This.
27867 * config/aarch64/aarch64-tune.md: Regenerate.
27868 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27869 (vulcan_addrcost_table): This.
27870 (vulcan_regmove_cost): Rename to ...
27871 (thunderx2t99_regmove_cost): This.
27872 (vulcan_vector_cost): Rename to ...
27873 (thunderx2t99_vector_cost): this.
27874 (vulcan_branch_cost): Rename to ...
27875 (thunderx2t99_branch_cost): This.
27876 (vulcan_tunings): Rename to ...
27877 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27878 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27880 2017-01-09 Martin Jambor <mjambor@suse.cz>
27884 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27885 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27886 (propagate_vr_accross_jump_function): Use the above function for all
27887 value range computations for pass-through jump functions and type
27888 converasion from explicit value range values.
27889 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27890 parameters from TYPE_ARG_TYPES.
27891 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27892 (ipa_write_node_info): Stream type of the actual argument.
27893 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27895 2017-01-09 Martin Liska <mliska@suse.cz>
27898 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27899 (lookup_compiler): Do not show error message with have_E.
27901 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27903 PR tree-optimization/78938
27904 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27905 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27906 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27907 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27910 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27912 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27915 2017-01-09 Richard Biener <rguenther@suse.de>
27917 PR tree-optimization/78997
27918 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27919 name condition properly.
27921 2017-01-09 Richard Biener <rguenther@suse.de>
27924 * dwarf2out.c (is_cxx): New overload with context.
27925 (is_naming_typedef_decl): Use it.
27927 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27929 * invoke.texi (Option Summary): Correct spacing in option lists
27930 and add line breaks to fix over-long lines.
27932 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27934 PR middle-end/17660
27936 * extend.texi (Common Variable Attributes): Add xref to GCC
27937 Internals manual to explain mode attribute keywords.
27939 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27942 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27943 and Preprocessor Options.
27944 (Options for Linking): Document -pthread here....
27945 (RS/6000 and PowerPC Options): ...not here.
27946 (Solaris 2 Options): ...or here.
27947 * doc/cppopts.texi: Document -pthread.
27949 2017-01-08 Martin Sebor <msebor@redhat.com>
27951 PR middle-end/77708
27952 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27953 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27954 New member functions.
27955 (format_directive): Used them.
27957 (pass_sprintf_length::handle_gimple_call): Same.
27958 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27959 to avoid truncation for any argument.
27960 (extract_affine_mul): Same.
27961 * tree.c (get_file_function_name): Same.
27963 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27965 PR middle-end/77484
27966 * predict.def (PRED_INDIR_CALL): Set to 86.
27968 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27970 PR preprocessor/54124
27971 * doc/cppopts.texi: Reformat -d subtable to list the full name
27972 of the options. Add cross-reference to the docs for the general
27973 compiler -d options.
27974 * doc/invoke.texi (Developer Options): Add cross-reference to the
27975 preprocessor-specific -d option documentation.
27977 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27979 PR preprocessor/13498
27980 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27981 redudant material, and reflect new command-line options.
27982 (System Headers): Likewise.
27984 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27986 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27987 -isystem, and -idirafter. Copy-edit.
27988 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27989 default for -ftrack-macro-expansion. Delete obsolete and
27990 badly-formatted implementation details about -fdebug-cpp output.
27991 * doc/cppwarnopts.texi: Copy-edit.
27993 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27996 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27997 that the transition from a max line width >= 1<<10 to narrower
27998 lines works correctly.
28000 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28002 * doc/options.texi (PerFunction): New.
28003 * opt-functions.awk (switch_flags): Map both Optimization and
28004 PerFunction to CL_OPTIMIZATION.
28005 * opth-gen.awk: Test for PerFunction flag along with
28007 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28008 it only when the latter is present. Skip those that don't in
28009 the hash function generator.
28010 * common.opt (fvar-tracking): Mark as PerFunction instead of
28012 (fvar-tracking-assignments): Likewise.
28013 (fvar-tracking-assignments-toggle): Likewise.
28014 (fvar-tracking-uninit): Likewise.
28016 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28018 PR translation/79018
28019 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28022 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28025 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28026 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28027 TARGET_LONG_JUMP_TABLE_OFFSETS.
28028 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28029 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28030 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28031 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28032 * config/m68k/m68k.md (tablejump expander): Likewise.
28033 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28034 TARGET_LONG_JUMP_TABLE_OFFSETS.
28035 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28036 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28038 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28039 David Holsgrove <david.holsgrove@xilinx.com>
28041 * common/config/microblaze/microblaze-common.c
28042 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28043 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28045 * config/microblaze/microblaze.c (microblaze_must_save_register)
28046 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28048 (microblaze_eh_return): New function.
28049 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28050 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28051 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28052 * config/microblaze/microblaze.md (eh_return): New pattern.
28054 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28056 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28057 GCC_DIAGNOSTIC_STRINGIFY): Define.
28059 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28061 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28063 * config/arm/arm.md (<mcrr>): New.
28065 * config/arm/arm.c (arm_arch5te): New.
28066 (arm_option_override): Set arm_arch5te.
28067 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28069 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28070 (arm_mcrr_qualifiers): ... this. New.
28071 (MRRC_QUALIFIERS): Define to...
28072 (arm_mrrc_qualifiers): ... this. New.
28073 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28075 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28076 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28077 (MRRCI, mrrc, MRRC): New.
28078 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28079 VUNSPEC_MRRC2): New.
28081 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28083 * config/arm/arm.md (<mcr>): New.
28085 * config/arm/arm.c (arm_coproc_builtin_available): Add
28086 support for mcr, mrc, mcr2 and mrc2.
28087 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28088 (arm_mcr_qualifiers): ... this. New.
28089 (MRC_QUALIFIERS): Define to ...
28090 (arm_mrc_qualifiers): ... this. New.
28091 (MCR_QUALIFIERS): Define to ...
28092 (arm_mcr_qualifiers): ... this. New.
28093 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28095 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28096 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28097 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28098 VUNSPEC_MRC2): New.
28100 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28102 * config/arm/arm.md (*ldc): New.
28106 * config/arm/arm.c (arm_coproc_builtin_available): Add
28107 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28108 (arm_coproc_ldc_stc_legitimate_address): New.
28109 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28110 'qualifier_const_pointer'.
28111 (LDC_QUALIFIERS): Define to...
28112 (arm_ldc_qualifiers): ... this. New.
28113 (STC_QUALIFIERS): Define to...
28114 (arm_stc_qualifiers): ... this. New.
28115 * config/arm/arm-protos.h
28116 (arm_coproc_ldc_stc_legitimate_address): New.
28117 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28118 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28119 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28120 stc2, stcl, stc2l): New.
28121 * config/arm/constraints.md (Uz): New.
28122 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28123 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28124 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28125 VUNSPEC_STC2L): New.
28127 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28129 * config/arm/arm.md (<cdp>): New.
28130 * config/arm/arm.c (neon_const_bounds): Rename this ...
28131 (arm_const_bounds): ... this.
28132 (arm_coproc_builtin_available): New.
28133 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28134 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28135 (CDP_QUALIFIERS): Define to...
28136 (arm_cdp_qualifiers): ... this. New.
28138 (arm_expand_builtin_args): Add case for 6 arguments.
28139 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28140 (arm_const_bounds): ... this.
28141 (arm_coproc_builtin_available): New.
28142 * config/arm/arm_acle.h (__arm_cdp): New.
28144 * config/arm/arm_acle_builtins.def (cdp): New.
28146 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28147 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28148 'arm_const_bounds'.
28149 * config/arm/types.md (coproc): New.
28150 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28151 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28152 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28153 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28155 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28157 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28158 (UBINOP_QUALIFIERS): New.
28160 (acle_builtin_data): New. Change comment.
28161 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28162 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28163 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28164 arm_acle_builtins.def.
28165 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28166 (arm_init_acle_builtins): New.
28167 (CRC32_BUILTIN): Remove.
28168 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28169 crc32cb, crc32ch and crc32cw.
28170 (arm_init_crc32_builtins): Remove.
28171 (arm_init_builtins): Use arm_init_acle_builtins rather
28172 than arm_init_crc32_builtins.
28173 (arm_expand_acle_builtin): New.
28174 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28175 * config/arm/arm_acle_builtins.def: New.
28177 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28179 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28180 (arm_builtin_datum): ... this.
28181 (arm_init_neon_builtin): Rename to ...
28182 (arm_init_builtin): ... this. Add a new parameters PREFIX
28183 and USE_SIG_IN_NAME.
28184 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28185 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28186 'arm_builtin_datum'.
28187 (arm_init_vfp_builtins): Likewise.
28188 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28189 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28190 (arm_expand_neon_args): Rename to ...
28191 (arm_expand_builtin_args): ... this. Rename builtin_arg
28192 enum values and differentiate between ARG_BUILTIN_MEMORY
28193 and ARG_BUILTIN_NEON_MEMORY.
28194 (arm_expand_neon_builtin_1): Rename to ...
28195 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28196 values, arm_expand_builtin_args and add bool parameter NEON.
28197 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28198 (arm_expand_vfp_builtin): Likewise.
28199 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28203 PR middle-end/77484
28204 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28205 * predict.c (tree_estimate_probability_bb): Reverse direction of
28206 polymorphic call predictor.
28208 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28210 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28211 (determine_pass_name_match): ...this new function and...
28212 (should_skip_pass_p): ...this new function.
28214 2017-01-06 Nathan Sidwell <nathan@acm.org>
28216 * ipa-visibility.c (function_and_variable_visibility): Reformat
28217 comments and long lines. Remove extrneous if.
28218 * symtab.c (symtab_node::make_decl_local): Fix code format.
28219 (symtab_node::set_section_for_node): Fix comment typo.
28221 2017-01-06 Martin Liska <mliska@suse.cz>
28224 * lra-constraints.c: Rename invariant to lra_invariant.
28225 * predict.c (set_even_probabilities): Initialize e to NULL.
28227 2017-01-05 Martin Sebor <msebor@redhat.com>
28229 PR tree-optimization/78910
28230 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28231 (format_integer): Correct off-by-one error in the handling
28232 of precision with negative numbers in signed conversions..
28234 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28236 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28238 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28240 PR tree-optimization/71016
28241 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28242 factor_out_conditional_conversion. Formatting fix.
28243 (factor_out_conditional_conversion): Add cond_stmt argument.
28244 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28245 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28248 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28250 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28251 read-rtl-function.o, and selftest-rtl.o.
28252 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28253 (selftest::aarch64_test_loading_full_dump): New function.
28254 (selftest::aarch64_run_selftests): New function.
28255 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28256 selftest::aarch64_run_selftests.
28257 * config/i386/i386.c
28258 (selftest::ix86_test_loading_dump_fragment_1): New function.
28259 (selftest::ix86_test_loading_call_insn): New function.
28260 (selftest::ix86_test_loading_full_dump): New function.
28261 (selftest::ix86_test_loading_unspec): New function.
28262 (selftest::ix86_run_selftests): Call the new functions.
28263 * emit-rtl.c (maybe_set_max_label_num): New function.
28264 * emit-rtl.h (maybe_set_max_label_num): New decl.
28265 * function.c (instantiate_decls): Guard call to
28266 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28267 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28269 * gensupport.c (gen_reader::gen_reader): Pass "false"
28270 for new "compact" param of rtx_reader.
28271 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28272 rather than an empty string for NULL strings.
28273 * read-md.c: Potentially include config.h rather than bconfig.h.
28274 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28275 (have_error): New global, copied from errors.c.
28276 (md_reader::read_name): Rename to...
28277 (md_reader::read_name_1): ...this, adding "out_loc" param,
28278 and converting "missing name or number" to returning false, rather
28280 (md_reader::read_name): Reimplement in terms of read_name_1.
28281 (md_reader::read_name_or_nil): New function.
28282 (md_reader::read_string): Handle "(nil)" by returning NULL.
28283 (md_reader::md_reader): Add new param "compact".
28284 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28285 (md_reader::read_file): New method.
28286 * read-md.h (md_reader::md_reader): Add new param "compact".
28287 (md_reader::read_file): New method.
28288 (md_reader::is_compact): New accessor.
28289 (md_reader::read_name): Convert return type from void to file_location.
28290 (md_reader::read_name_or_nil): New decl.
28291 (md_reader::read_name_1): New decl.
28292 (md_reader::m_compact): New field.
28293 (noop_reader::noop_reader): Pass "false" for new "compact" param
28295 (rtx_reader::rtx_reader): Add new "compact" param.
28296 (rtx_reader::read_rtx_operand): Make virtual and convert return
28297 type from void to rtx.
28298 (rtx_reader::read_until): New decl.
28299 (rtx_reader::handle_any_trailing_information): New virtual function.
28300 (rtx_reader::postprocess): New virtual function.
28301 (rtx_reader::finalize_string): New virtual function.
28302 (rtx_reader::m_in_call_function_usage): New field.
28303 (rtx_reader::m_reuse_rtx_by_id): New field.
28304 * read-rtl-function.c: New file.
28305 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28306 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28307 (selftest::verify_three_block_rtl_cfg): New decl.
28308 * read-rtl-function.h: New file.
28309 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28310 For host, include function.h, memmodel.h, and emit-rtl.h.
28311 (one_time_initialization): New function.
28312 (struct compact_insn_name): New struct.
28313 (compact_insn_names): New array.
28314 (find_code): Handle insn codes in compact dumps.
28315 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28316 (bind_subst_iter_and_attr): Likewise.
28317 (add_condition_to_string): Likewise.
28318 (add_condition_to_rtx): Likewise.
28319 (apply_attribute_uses): Likewise.
28320 (add_current_iterators): Likewise.
28321 (apply_iterators): Likewise.
28322 (initialize_iterators): Guard usage of apply_subst_iterator with
28323 #ifdef GENERATOR_FILE.
28324 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28325 (md_reader::read_mapping): Likewise.
28326 (add_define_attr_for_define_subst): Likewise.
28327 (add_define_subst_attr): Likewise.
28328 (read_subst_mapping): Likewise.
28329 (check_code_iterator): Likewise.
28330 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28332 (one_time_initialization): New function.
28333 (rtx_reader::read_until): New method.
28334 (read_flags): New function.
28335 (parse_reg_note_name): New function.
28336 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28337 Handle reuse_rtx ids.
28338 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28339 Add parsing support for RTL dumps, mirroring the special-cases in
28340 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28341 values, and calling handle_any_trailing_information.
28342 (rtx_reader::read_rtx_operand): Convert return type from void
28343 to rtx, returning return_rtx. Handle case 'e'. Call
28344 finalize_string on XSTR and XTMPL fields.
28345 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28346 "(nil)" values were omitted. Call the postprocess vfunc on the
28348 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28349 class ctor. Initialize m_in_call_function_usage. Call
28350 one_time_initialization.
28351 * rtl-tests.c (selftest::test_uncond_jump): Call
28352 set_new_first_and_last_insn.
28353 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28354 * selftest-rtl.c: New file.
28355 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28356 (selftest::get_insn_by_uid): New decl.
28357 * selftest-run-tests.c (selftest::run_tests): Call
28358 read_rtl_function_c_tests.
28359 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28360 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28363 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28365 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28366 operands in a special way. Assert that pos+len <= mode precision.
28368 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28370 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28371 3 argument Alias with unlimited for the negative form.
28372 (fno-vect-cost-model): Removed.
28374 2017-01-05 Martin Liska <mliska@suse.cz>
28376 * hsa-gen.c (gen_hsa_divmod): New function.
28377 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28379 2017-01-05 Martin Liska <mliska@suse.cz>
28382 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28385 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28387 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28388 small constant length operands.
28390 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28392 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28393 between loop iterations.
28395 2017-01-05 Martin Liska <mliska@suse.cz>
28398 * gimplify.c (gimplify_decl_expr): Compare to
28399 asan_poisoned_variables instread of checking flags.
28400 (gimplify_target_expr): Likewise.
28401 (gimplify_expr): Likewise.
28402 (gimplify_function_tree): Conditionally initialize
28403 asan_poisoned_variables.
28405 2017-01-04 Jeff Law <law@redhat.com>
28407 PR tree-optimizatin/78812
28408 * rtl.h (contains_mem_rtx_p): Prototype.
28409 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28410 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28411 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28412 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28413 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28415 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28417 * input.c (assert_char_at_range): Default-initialize actual_range.
28419 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28421 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28422 to match the caller.
28424 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28426 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28427 insns after final jump in test to emit dummy move.
28429 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28431 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28432 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28434 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28436 * multiple_target.c (create_dispatcher_calls): Init e_next.
28437 * tree-ssa-loop-split.c (split_loop): Init border.
28438 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28441 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28446 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28447 (altivec_register_operand): Do not return true if the operand
28448 contains a SUBREG mixing SImode and SFmode.
28449 (vsx_register_operand): Likewise.
28450 (vsx_reg_sfsubreg_ok): New predicate.
28451 (vfloat_operand): Do not return true if the operand contains a
28452 SUBREG mixing SImode and SFmode.
28453 (vint_operand): Likewise.
28454 (vlogical_operand): Likewise.
28455 (gpc_reg_operand): Likewise.
28456 (int_reg_operand): Likewise.
28457 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28458 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28459 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28461 (rs6000_emit_move_si_sf_subreg): New helper function.
28462 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28463 fixup SUBREGs involving SImode and SFmode.
28464 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28465 numbers for the new peephole2 optimization.
28466 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28467 the GLIBC math library that do AND/IOR/XOR operations on single
28468 precision floating point.
28469 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28470 target macros to say whether we need to avoid SUBREGs mixing
28472 (TARGET_ALLOW_SF_SUBREG): Likewise.
28473 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28474 (UNSPEC_SI_FROM_SF): Likewise.
28475 (iorxor): Change spacing.
28476 (and_ior_xor): New iterator for AND, IOR, and XOR.
28477 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28478 (movdi_from_sf_zero_ext): Likewise.
28479 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28480 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28481 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28482 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28483 (fms<mode>4): Likewise.
28484 (fnma<mode>4): Likewise.
28485 (fnms<mode>4): Likewise.
28486 (nfma<mode>4): Likewise.
28487 (nfms<mode>4): Likewise.
28489 2017-01-04 Marek Polacek <polacek@redhat.com>
28492 * doc/invoke.texi: Document -Wpointer-compare.
28494 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28496 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28499 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28500 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28503 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28506 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28507 documentation of the powerpc_popcntb_ok attribute.
28508 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28509 code to issue warning messages if a requested CPU configuration is
28510 not supported by the binary (assembler and loader) toolchain.
28511 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28512 made to define a built-in function that has been disabled.
28513 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28514 made to define a built-in function that has been disabled.
28515 (altivec_init_builtins): Add comment explaining why definition
28516 of the DST built-in functions is not preceded by an assertion
28517 check. Add assertions to prevent ICE if attempts are made to
28518 define an altivec predicate or an abs* built-in function that has
28520 (htm_init_builtins): Add comment explaining why definition of the
28521 htm built-in functions is not preceded by an assertion check.
28523 2017-01-04 Jeff Law <law@redhat.com>
28525 PR tree-optimizatin/67955
28526 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28527 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28528 the points-to solution does not include pt_null. Use DECL_PT_UID
28531 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28533 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28534 Use gen_int_mode instead of gen_lopwart for const_int operands.
28536 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28538 PR tree-optimization/71563
28539 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28540 out of range value - has low bits known to be zero.
28542 2017-01-04 Alan Modra <amodra@gmail.com>
28544 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28545 * configure: Regenerate.
28546 * config.in: Regenerate.
28548 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28551 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28552 a substring of the message, but strcmp with the whole message. Ifdef
28553 ENABLE_NLS, translate the message first using dgettext.
28555 2017-01-03 Jeff Law <law@redhat.com>
28557 PR tree-optimizatin/78856
28558 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28559 (mark_threaded_blocks): Remove code to truncate thread paths that
28560 cross multiple loop headers. Instead invalidate the cached loop
28561 iteration information and handle case of a thread path walking
28562 into an irreducible region.
28564 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28567 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28568 assertions. Add support for doing the signbit if the IEEE 128-bit
28569 floating point value is in a GPR.
28570 * config/rs6000/rs6000.md (Fsignbit): Delete.
28571 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28572 Update the length attribute if the value is in a GPR.
28573 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28574 the sign or zero extension instruction, since the value is always 0/1.
28575 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28578 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28579 extracting SImode to a GPR register so that we can generate a
28580 store, limit the vector to be in a traditional Altivec register
28581 for the vextuwrx instruction.
28583 2017-01-03 Ian Lance Taylor <iant@google.com>
28585 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28587 2017-01-03 Martin Sebor <msebor@redhat.com>
28589 PR tree-optimization/78696
28590 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28591 precision. Use MPFR for %f for greater fidelity. Correct handling
28593 (pass_sprintf_length::compute_format_length): Set width and precision
28594 specified by asrerisk to void_node for vararg functions.
28595 (try_substitute_return_value): Adjust dump output.
28597 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28599 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28601 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28603 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28604 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28605 -mlra/-mno-lra was passed to the compiler.
28607 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28609 PR rtl-optimization/65618
28610 * emit-rtl.c (try_split): Move initialization of "before" and
28611 "after" to just before the call to emit_insn_after_setloc.
28613 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28615 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28617 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28619 * dwarf2out.c (gen_enumeration_type_die): When
28620 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28622 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28624 PR tree-optimization/78965
28625 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28626 Change first argument from const call_info & to call_info &. For %n
28627 set info.nowrite to false.
28629 PR middle-end/78901
28630 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28631 possibly throwing calls.
28633 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28634 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28635 and fns handling, rather than in a separate case SSA_NAME.
28637 2017-01-02 Jeff Law <law@redhat.com>
28639 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28640 fixes for first_period and second_period variables.
28642 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28645 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28646 (*insvqi_1): New insn pattern.
28647 (*insvqi_1_mem_rex64): Ditto.
28648 (*insvqi_2): Ditto.
28649 (*insvqi_3): Rename from *insvqi.
28651 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28653 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28655 * doc/cfg.texi (Edges): Remove reference to Java.
28656 (Maintaining the CFG): Ditto.
28658 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28660 PR middle-end/77674
28661 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28662 transparent aliases.
28664 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28666 PR middle-end/77484
28667 * predict.def (PRED_CALL): Update hitrate.
28668 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28669 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28670 into direct/indirect/polymorphic variants.
28672 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28674 Update copyright years.
28676 * gcc.c (process_command): Update copyright notice dates.
28677 * gcov-dump.c (print_version): Ditto.
28678 * gcov.c (print_version): Ditto.
28679 * gcov-tool.c (print_version): Ditto.
28680 * gengtype.c (create_file): Ditto.
28681 * doc/cpp.texi: Bump @copying's copyright year.
28682 * doc/cppinternals.texi: Ditto.
28683 * doc/gcc.texi: Ditto.
28684 * doc/gccint.texi: Ditto.
28685 * doc/gcov.texi: Ditto.
28686 * doc/install.texi: Ditto.
28687 * doc/invoke.texi: Ditto.
28689 Copyright (C) 2017 Free Software Foundation, Inc.
28691 Copying and distribution of this file, with or without modification,
28692 are permitted in any medium without royalty provided the copyright
28693 notice and this notice are preserved.