1 2017-10-04 Nathan Sidwell <nathan@acm.org>
3 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10 check type for aarch64_simd_valid_immediate.
11 (aarch64_output_simd_mov_immediate): Update prototype.
12 (aarch64_simd_valid_immediate): Update prototype.
13 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
14 support for ORR-immediate.
15 (and<mode>3): modified pattern to add support for BIC-immediate.
16 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
17 now checks for valid immediate for BIC and ORR based on new enum
19 (aarch64_output_simd_mov_immediate): Function now used to output
20 BIC/ORR imm as well based on new enum argument.
21 * config/aarch64/constraints.md (Do): New vector immediate constraint.
23 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
24 (aarch64_reg_or_bic_imm): Likewise.
26 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
29 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
31 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
35 PR rtl-optimization/82396
36 * haifa-sched.c (autopref_multipass_init): Simplify
38 (autopref_rank_data): Simplify sort order.
39 * sched-int.h (autopref_multipass_data_): Remove
40 multi_mem_insn_p, min_offset and max_offset.
42 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
44 * doc/sourcebuild.texi: Document vect_peeling_profitable.
46 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
48 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
51 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
53 * doc/sourcebuild.texi: Document vect_long_mult.
55 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
57 PR tree-optimization/82413
58 * fold-const.c (build_range_check): Use widest_int when comparing
59 the maximum ETYPE value with HIGH.
61 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
63 PR rtl-optimization/82396
64 * haifa-sched.c (autopref_multipass_init): Simplify
66 (autopref_rank_data): Simplify sort order.
67 * sched-int.h (autopref_multipass_data_): Remove
68 multi_mem_insn_p, min_offset and max_offset.
70 2017-10-04 Jakub Jelinek <jakub@redhat.com>
72 PR tree-optimization/82381
73 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
74 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
77 PR tree-optimization/82374
78 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
79 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
80 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
81 current_function_decl to the new decl.
83 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
85 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
86 helper macro for IEEE float128 hardware built-in functions.
87 (SQRTF128_ODD): Add built-in functions with the round-to-odd
89 (TRUNCF128_ODD): Likewise.
90 (ADDF128_ODD): Likewise.
91 (SUBF128_ODD): Likewise.
92 (MULF128_ODD): Likewise.
93 (DIVF128_ODD): Likewise.
94 (FMAF128_ODD): Likewise.
95 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
96 UNSPEC_TRUNC_ROUND_TO_ODD.
97 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
98 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
99 floating point round to odd instructions.
100 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
101 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
102 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
103 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
104 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
105 (trunc<mode>sf2_hw): Change the truncate with round to odd
106 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
107 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
108 to odd hardware instructions.
109 (sub<mode>3_odd): Likewise.
110 (mul<mode>3_odd): Likewise.
111 (div<mode>3_odd): Likewise.
112 (sqrt<mode>2_odd): Likewise.
113 (fma<mode>4_odd): Likewise.
114 (fms<mode>4_odd): Likewise.
115 (nfma<mode>4_odd): Likewise.
116 (nfms<mode>4_odd): Likewise.
117 (trunc<mode>df2_odd): Change the truncate with round to odd
118 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
120 * doc/extend.texi (PowerPC built-in functions): Update documentation
121 for existing IEEE float128-bit built-in functions. Add built-in
122 functions that generate the IEEE 128-bit floating point round to
125 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
127 PR rtl-optimization/77729
128 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
129 to (X&(C1&~C2))|C2 transformations.
131 2017-10-03 Martin Jambor <mjambor@suse.cz>
133 PR tree-optimization/82363
134 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
135 mismatch, mark lacc written regardless of racc.
137 2017-10-03 Jakub Jelinek <jakub@redhat.com>
139 PR tree-optimization/82381
140 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
141 stmt_to_insert nor wheather SSA_NAMEs are default defs.
142 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
143 fallthrough into reassoc_stmt_dominates_stmt_p.
146 * combine.c (combine_instructions): Don't combine in unreachable
149 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
152 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
153 function to not use the have_cpu variable. Do not set cpu_index,
154 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
156 (rs6000_valid_attribute_p): Remove duplicate initializations of
157 old_optimize and func_optimize.
158 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
159 (rs6000_activate_target_options): Make global.
160 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
163 2017-10-02 Jakub Jelinek <jakub@redhat.com>
165 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
166 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
167 Set *poffset to 0 and *psize and *pmax_size to -1 if
168 *poffset + *psize overflows in HOST_WIDE_INT.
170 PR tree-optimization/82387
171 PR tree-optimization/82388
172 PR tree-optimization/82389
173 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
174 instead of live_bytes non-NULL.
176 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
179 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
182 2017-10-02 Richard Biener <rguenther@suse.de>
184 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
185 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
186 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
188 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
190 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
191 requested precision matches the type's.
192 * calls.c (alloc_max_size): Calculate the new candidate size as
193 a widest_int and use wi::to_widest when comparing it with the
194 current candidate size.
195 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
196 zero rather than integer_zero_node.
197 * match.pd: Check for a no-op conversion before using wi::add
198 rather than after. Use tree_to_uhwi when summing small shift
199 counts into an unsigned int.
201 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
202 Alan Hayward <alan.hayward@arm.com>
203 David Sherwood <david.sherwood@arm.com>
206 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
207 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
208 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
211 2017-10-02 Richard Biener <rguenther@suse.de>
213 PR tree-optimization/82355
214 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
215 a mapping for the enclosing loop but avoid generating one for
217 (copy_bb_and_scalar_dependences): Remove premature codegen
218 error on PHIs in blocks duplicated into multiple places.
219 * graphite-scop-detection.c
220 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
221 in the region use it as loop and nest to analyze the DR in.
222 (try_generate_gimple_bb): Likewise.
223 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
224 (add_loop_constraints): For blocks in a loop not in the region
225 create a dimension with a single iteration.
226 * sese.h (gbb_loop_at_index): Remove assert.
228 2017-10-01 Kevin Buettner <kevinb@redhat.com>
230 * omp-expand.c (adjust_context_scope): New function.
231 (expand_parallel_call): Call adjust_context_scope.
233 2017-10-01 Jeff Law <law@redhat.com>
235 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
236 dom_opt_dom_walker class with direct access to private members.
237 Add comments. Call test_for_singularity.
238 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
239 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
240 m_dummy_cond anymore.
241 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
243 (pass_dominator:execute): Build the dummy_cond here and pass it
244 to the dom_opt_dom_walker ctor.
245 (test_for_singularity): New function.
247 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
248 Maya Rashish <coypu@sdf.org>
250 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
251 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
252 (alpha*-*-netbsd*) Use nbsd_tm_file.
253 (arm*-*-netbsdelf*) Likewise.
254 (i[34567]86-*-netbsdelf*) Likewise.
255 (x86_64-*-netbsd*) Likewise.
256 (mips*-*-netbsd*) Likewise.
257 (powerpc-*-netbsd*) Likewise.
258 (sh*-*-netbsd*) Likewise.
259 (sparc-*-netbsdelf*) Likewise.
260 (sparc64-*-netbsd*) Likewise.
261 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
263 (vax-*-netbsdelf*) Likewise.
264 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
265 (UINT_FAST8_TYPE) Likewise.
266 (INT_FAST16_TYPE) Check CHAR_FAST16.
267 (UINT_FAST16_TYPE) Likewise.
269 2017-09-30 Jakub Jelinek <jakub@redhat.com>
272 * config/i386/i386.md
273 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
274 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
275 *divmodsi4_zext_2): New define_insn_and_split.
276 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
277 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
278 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
279 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
280 New define_insn_and_split.
281 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
282 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
283 operands[1] having DImode when mode is SImode.
285 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
286 always SImode for DIV and MOD in REG_EQUAL notes.
288 2017-09-29 Yury Gribov <tetra2005@gmail.com>
291 * match.pd: Fix handling of NaNs in pattern.
293 2017-09-29 Jeff Law <law@redhat.com>
295 * sbitmap.c (bitmap_bit_in_range_p): New function.
296 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
297 * tree-ssa-dse.c (live_bytes_read): New function.
298 (dse_classify_store): Ignore reads of dead bytes.
300 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
301 typos and whitespace errors.
302 * config/i386/predicates.md (address_no_seg_operand): Likewise.
303 * config/s390/s390.c (s390_emit_prologue): Likewise.
305 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
308 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
309 with a symbol for LRA.
311 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
313 PR rtl-optimization/82338
314 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
316 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
318 * genmodes.c (calc_wider_mode): Suppress qsort macro.
319 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
320 (qsort_chk): Declare.
321 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
322 (qsort_chk): New function.
324 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
326 PR tree-optimization/82337
327 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
328 phi definition if the PHI result appears in an abnormal PHI.
329 (find_basis_for_base_expr): Don't record a basis if the LHS of the
330 basis appears in an abnormal PHI.
332 2017-09-29 Richard Biener <rguenther@suse.de>
334 * graphite-isl-ast-to-gimple.c
335 (translate_isl_ast_to_gimple::set_codegen_error): New function.
336 (binary_op_to_tree): Use it.
337 (get_rename_from_scev): Likewise.
338 (copy_loop_phi_nodes): Likewise.
339 (copy_bb_and_scalar_dependences): Likewise.
340 (translate_pending_phi_nodes): Likewise.
342 2017-09-29 Jakub Jelinek <jakub@redhat.com>
345 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
346 for movabsq $(i32 << shift), r64.
348 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
350 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
351 index when encoding %esp as %rsp to avoid 0x67 prefix.
353 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
355 * config/i386/i386.md (*movsf_internal, *movdf_internal):
356 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
358 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
360 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
361 Extensions with more than 16 double VFP registers.
362 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
363 to_clear_mask and all code related to it. Replace the remaining
364 entry by a sbitmap and adapt code accordingly.
366 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
368 * brig-builtins.def: Change pure attributes to const.
370 2017-09-28 Joseph Myers <joseph@codesourcery.com>
372 * config.gcc (default_gnu_indirect_function): Default to yes for
373 sparc*-*-linux* with glibc.
375 2017-09-28 Joseph Myers <joseph@codesourcery.com>
377 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
378 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
379 when creating .init_array and .fini_array sections with priority
382 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
385 * config/aarch64/aarch64.c
386 (aarch64_builtin_support_vector_misalignment): Always return false
387 when misalignment is unknown.
389 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
391 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
392 this function to return false if the definition used by the swap
393 instruction is artificial, or if the memory address from which the
394 constant value is loaded is not represented by a base address held
395 in a register or if the base address register is a frame or stack
396 pointer. Additionally, return false if the base address of the
397 loaded constant is a SYMBOL_REF but is not considered to be a
399 (replace_swapped_load_constant): New function.
400 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
401 loaded constant vector with a load of a swapped constant vector.
403 2017-09-27 Carl Love <cel@us.ibm.com>
405 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
406 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
407 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
410 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
412 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
413 first, always call autopref_rank_data otherwise.
415 2017-09-27 Richard Biener <rguenther@suse.de>
417 * graphite-scop-detection.c (find_scop_parameters): Move
418 loop bound handling ...
419 (gather_bbs::before_dom_children): ... here, avoiding the need
420 to build scop_info->loop_nest.
421 (record_loop_in_sese): Remove.
422 * sese.h (sese_info_t::loop_nest): Remove.
423 * sese.c (new_sese_info): Do not allocate loop_nest.
424 (free_sese_info): Do not free loop_nest.
426 2017-09-27 Jakub Jelinek <jakub@redhat.com>
429 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
430 lhs from calls if the lhs has addressable type.
432 2017-09-27 Richard Biener <rguenther@suse.de>
434 * graphite.h (scop::max_alias_set): New member.
435 * graphite-scop-detection.c: Remove references to non-existing
437 (build_alias_sets): Record the maximum alias set used for drs.
438 (build_scops): Support zero as unlimited for
439 --param graphite-max-arrays-per-scop.
440 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
442 (build_poly_sr_1): ... here. Compute alias set based on the
443 maximum alias set used for drs rather than
444 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
446 2017-09-27 Richard Biener <rguenther@suse.de>
448 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
449 --param loop-block-tile-size=0 to disable tiling.
451 2017-09-27 Richard Biener <rguenther@suse.de>
453 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
454 (graphite-max-nb-scop-params): Document special value zero.
455 * domwalk.h (dom_walker::STOP): New symbolical constant.
456 (dom_walker::dom_walker): Add optional parameter for bb to
458 (dom_walker::~dom_walker): Declare.
459 (dom_walker::before_dom_children): Document STOP return value.
460 (dom_walker::m_user_bb_to_rpo): New member.
461 (dom_walker::m_bb_to_rpo): Likewise.
462 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
463 mapping here if not provided by the user.
464 (dom_walker::~dom_walker): Free bb to RPO mapping if not
465 provided by the user.
466 (dom_walker::STOP): Define.
467 (dom_walker::walk): Do not compute bb to RPO mapping here.
468 Support STOP return value from before_dom_children to stop
470 * graphite-optimize-isl.c (optimize_isl): If the schedule
471 is the same still generate code if -fgraphite-identity
472 or -floop-parallelize-all are given.
473 * graphite-scop-detection.c: Include cfganal.h.
474 (gather_bbs::gather_bbs): Get and pass through bb to RPO
476 (gather_bbs::before_dom_children): Return STOP for BBs
478 (build_scops): Compute bb to RPO mapping and pass it to
479 the domwalk. Treat --param graphite-max-nb-scop-params=0
480 as not limiting the number of params.
481 * graphite.c (graphite_initialize): Remove limit on the
482 number of basic-blocks in a function.
483 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
484 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
487 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
489 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
490 Adjust code to eliminate needing to do the shift right 32-bits
491 operation after XSCVDPSPN.
493 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
495 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
496 ((X / Y) != 0 -> X >= Y): Likewise.
498 2017-09-26 Carl Love <cel@us.ibm.com>
500 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
501 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
502 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
503 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
504 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
505 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
506 definitions and overloading.
507 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
508 statement for P9V_BUILTIN_XST_LEN_R.
509 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
510 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
511 define_expand and define_insn for the instructions and builtins.
512 * doc/extend.texi: Update the built-in documentation file for the new
514 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
515 define_insn for the instructions
517 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
520 * gcc/config/netbsd-protos.h: New file.
521 * gcc/config/netbsd.c: New file.
522 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
523 * gcc/config/t-netbsd: New file.
524 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
525 (tmake_file) Add t-netbsd.
526 (extra_objs) Add netbsd.o.
528 2017-09-26 Janus Weil <janus@gcc.gnu.org>
532 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
534 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
536 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
537 sign extension from a vector register to a GPR by doing a 32-bit
538 direct move and then an EXTSW.
539 (extendsi<mode>2 splitter): Likewise.
540 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
541 right or vector extract after doing XSCVDPSPN. Use
542 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
544 (movdi_from_sf_zero_ext): Likewise.
545 (reload_gpr_from_vsxsf): Likewise.
546 (p8_mfvsrd_4_disf): Delete, no longer used.
547 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
548 then needing to move the SFmode to a GPR to use the XSCVDPSP
549 instruction instead of FRSP and XSCVDPSPN.
550 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
551 it is adjacent to the other XSCVSPDP insns.
552 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
553 SFmode to be in traditional Altivec registers.
554 (vsx_xscvdpspn): Eliminate useless alternative constraint.
555 (vsx_xscvspdpn): Likewise.
556 (vsx_xscvspdpn_scalar): Likewise.
558 2017-09-26 Martin Jambor <mjambor@suse.cz>
560 * tree-sra.c (compare_access_positions): Put integral types first,
561 stabilize sorting of integral types, remove conditions putting
562 non-full-precision integers last.
563 (sort_and_splice_var_accesses): Disable scalarization if a
564 non-integert would be represented by a non-full-precision integer.
566 2017-09-26 Joseph Myers <joseph@codesourcery.com>
568 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
569 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
570 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
571 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
572 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
573 conditionals inside the function instead of around it. Call
574 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
575 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
577 2017-09-26 Richard Biener <rguenther@suse.de>
579 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
581 (scop_detection::build_scop_breadth): ... this. Removed.
582 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
583 (scop_detection::harmful_stmt_in_bb): Likewise.
584 (scop_detection::graphite_can_represent_stmt): Likewise.
585 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
586 (scop_detection::can_represent_loop): Remove recursion, fold in ...
587 (scop_detection::can_represent_loop_1): ... this. Removed.
588 (scop_detection::harmful_loop_in_region): Simplify after inlining
589 the above and remove more quadraticness.
590 (build_scops): Adjust.
591 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
594 2017-09-26 Jakub Jelinek <jakub@redhat.com>
597 * config/i386/i386.c (ix86_print_operand_address_as): Only test
598 REGNO (base) == SP_REG if base is a REG.
601 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
602 if it is different SSA_NAME.
603 (optimize_range_tests_cmp_bitwise): New function.
604 (optimize_range_tests): Call it.
606 2017-09-26 Richard Biener <rguenther@suse.de>
608 PR tree-optimization/82321
609 * graphite.c (canonicalize_loop_closed_ssa): Properly check
610 for the def being inside the loop.
612 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
614 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
616 * config/s390/s390-builtins.def: Fix constraint on op4.
618 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
620 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
621 independent expanders.
622 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
623 ("vec_ordered", "vec_unordered"): New expanders.
625 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
627 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
630 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
632 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
633 vec_unpacks_lo_v16qi.
634 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
636 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
638 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
639 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
640 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
642 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
644 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
646 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
648 ("*vec_slb<mode>"): New insn pattern.
649 ("vec_shr_<mode>"): New expander.
650 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
651 and force the shift count operand to V16QImode.
652 ("vec_srb<mode>"): Set shift count mode to V16QI.
654 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
656 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
657 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
658 ("vec_widen_smult_hi_<mode>"): New expander definitions.
660 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
663 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
665 2017-09-26 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/82320
668 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
671 2017-09-25 Jeff Law <law@redhat.com>
673 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
674 prototype for new argument.
675 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
676 mostly extracted from rs6000_emit_allocate_stack.
677 (rs6000_emit_probe_stack_range_stack_clash): New function.
678 (rs6000_emit_allocate_stack): Call
679 rs6000_emit_probe_stack_range_stack_clash as needed.
680 (rs6000_emit_probe_stack_range): Add additional argument
681 to call to gen_probe_stack_range{si,di}.
682 (output_probe_stack_range): New.
683 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
684 (output_probe_stack_range_stack_clash): New.
685 (rs6000_emit_prologue): Emit notes into dump file as requested.
686 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
687 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
688 Add additional operand and pass it to output_probe_stack_range.
690 2017-09-25 Bin Cheng <bin.cheng@arm.com>
692 PR tree-optimization/82163
693 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
694 (checking_verify_loop_closed_ssa): New parameter.
695 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
696 (check_loop_closed_ssa_stmt): Delete.
697 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
698 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
699 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
702 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
704 * brig-builtins.def: Treat HSAIL barrier builtins as
705 setjmp/longjump style functions.
707 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
709 * target.def (constant_alignment): New hook.
710 * defaults.h (CONSTANT_ALIGNMENT): Delete.
711 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
712 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
713 * doc/tm.texi: Regenerate.
714 * targhooks.h (default_constant_alignment): Declare.
715 (constant_alignment_word_strings): Likewise.
716 * targhooks.c (default_constant_alignment): New function.
717 (constant_alignment_word_strings): Likewise.
718 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
719 instead of CONSTANT_ALIGNMENT.
720 * varasm.c (align_variable, get_variable_align, build_constant_desc)
721 (force_const_mem): Likewise.
722 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
723 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
724 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
725 (TARGET_CONSTANT_ALIGNMENT): Redefine.
726 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
728 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
729 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
730 constant_alignment_word_strings.
731 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
732 (CONSTANT_ALIGNMENT): Likewise.
733 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
734 (arm_constant_alignment): New function.
735 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
736 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
737 constant_alignment_word_strings.
738 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
739 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
740 constant_alignment_word_strings.
741 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
742 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
743 (cris_constant_alignment): New function.
744 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
745 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
746 (epiphany_constant_alignment): New function.
747 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
748 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
749 constant_alignment_word_strings.
750 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
751 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
752 constant_alignment_word_strings.
753 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
754 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
755 constant_alignment_word_strings.
756 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
757 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
758 * config/i386/i386.c (ix86_constant_alignment): Make static.
759 Use the same interface as the target hook.
760 (TARGET_CONSTANT_ALIGNMENT): Redefine.
761 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
762 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
763 constant_alignment_word_strings.
764 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
765 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
766 (TARGET_CONSTANT_ALIGNMENT): Redefine.
767 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
768 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
769 constant_alignment_word_strings.
770 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
771 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
772 constant_alignment_word_strings.
773 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
774 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
775 constant_alignment_word_strings.
776 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
777 * config/microblaze/microblaze.c (microblaze_constant_alignment):
779 (TARGET_CONSTANT_ALIGNMENT): Redefine.
780 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
781 * config/mips/mips.c (mips_constant_alignment): New function.
782 (TARGET_CONSTANT_ALIGNMENT): Redefine.
783 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
784 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
785 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
786 (mmix_constant_alignment): Make static. Use the same interface
788 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
789 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
790 constant_alignment_word_strings.
791 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
792 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
793 constant_alignment_word_strings.
794 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
795 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
796 constant_alignment_word_strings.
797 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
798 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
799 (rs6000_constant_alignment): New function.
800 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
801 * config/riscv/riscv.c (riscv_constant_alignment): New function.
802 (TARGET_CONSTANT_ALIGNMENT): Redefine.
803 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
804 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
805 (rs6000_constant_alignment): New function.
806 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
807 * config/s390/s390.c (s390_constant_alignment): New function.
808 (TARGET_CONSTANT_ALIGNMENT): Redefine.
809 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
810 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
811 constant_alignment_word_strings.
812 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
813 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
814 (sparc_constant_alignment): New function.
815 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
816 * config/spu/spu.c (spu_constant_alignment): New function.
817 (TARGET_CONSTANT_ALIGNMENT): Redefine.
818 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
819 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
820 constant_alignment_word_strings.
821 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
822 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
823 constant_alignment_word_strings.
824 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
825 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
826 constant_alignment_word_strings.
827 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
828 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
829 (visium_constant_alignment): New function.
830 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
831 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
832 (xtensa_constant_alignment): New function.
833 * system.h (CONSTANT_ALIGNMENT): Poison.
835 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
837 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
838 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
839 (rs6000_builtin_valid_without_lhs): New helper function.
840 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
841 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
843 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
845 * target.h (vec_perm_indices): Use unsigned short rather than
847 (auto_vec_perm_indices): Likewise.
848 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
849 Use unsigned int rather than unsigned char.
850 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
852 2017-09-25 Richard Biener <rguenther@suse.de>
854 * cfgloop.h (sort_sibling_loops): Declare.
855 * cfgloop.c (sort_sibling_loops_cmp): New helper.
856 (sort_sibling_loops): New function sorting the sibling loop list
858 * graphite.c (graphite_transform_loops): Sort sibling loops.
860 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
862 * target.def (vec_perm_const_ok): Change sel parameter to
864 * optabs-query.c (can_vec_perm_p): Update accordingly.
865 * doc/tm.texi: Regenerate.
866 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
867 auto_vec_perm_indices and remove separate nelt field.
868 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
869 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
870 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
871 (aarch64_expand_vec_perm_const): Update accordingly.
872 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
874 * config/arm/arm.c (expand_vec_perm_d): Change perm to
875 auto_vec_perm_indices and remove separate nelt field.
876 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
877 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
878 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
880 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
882 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
883 sel to vec_perm_indices.
884 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
885 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
886 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
888 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
891 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
894 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
895 on the FUNCTION_DECL function context if it has a DIE that is a
898 2017-09-25 Richard Biener <rguenther@suse.de>
900 PR tree-optimization/82285
901 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
904 2017-09-25 Tom de Vries <tom@codesourcery.com>
908 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
911 2017-09-25 Richard Biener <rguenther@suse.de>
913 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
914 ISL errors other than isl_error_quota happen. Dump if the
915 schedule is the same.
916 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
917 errors instead of aborting inside ISL.
919 2017-09-25 Iain Sandoe <iain@codesourcery.com>
922 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
923 of libgcc_eh for m64.
924 * config/i386/darwin64.h: Likewise.
926 2017-09-25 Richard Biener <rguenther@suse.de>
929 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
930 attribute for incomplete types nor twice for complete ones.
932 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
935 * config/i386/i386.c (ix86_print_operand_address_as): Encode
936 %esp as %rsp to avoid 0x67 prefix if there is no index or base
939 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
942 * config/i386/i386.opt (mprefer-avx256): Use
943 ix86_target_flags variable.
944 * config/i386/i386.c (ix86_target_string): Move
945 -mprefer-avx256 to flag2_opts.
947 2017-09-22 Jakub Jelinek <jakub@redhat.com>
950 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
951 and x != -1 | y != -1 into (x & y) != -1.
953 2017-09-22 Steve Ellcey <sellcey@cavium.com>
955 * config.gcc: Add new case statement to set
956 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
957 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
958 s390x-*-linux* case statements. Added aarch64 to the list of
959 supported architectures.
961 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
963 PR tree-optimization/82289
964 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
965 STMT_VINFO_RELEVANT_P.
967 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
968 Alan Hayward <alan.hayward@arm.com>
969 David Sherwood <david.sherwood@arm.com>
971 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
972 for VR_RANGE only; don't allow VR_ANTI_RANGE.
973 (extract_range_from_binary_expr_1): Don't call
974 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
976 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
977 Alan Hayward <alan.hayward@arm.com>
978 David Sherwood <david.sherwood@arm.com>
980 * target.def (preferred_vector_alignment): New hook.
981 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
983 * doc/tm.texi: Regenerate.
984 * targhooks.h (default_preferred_vector_alignment): Declare.
985 * targhooks.c (default_preferred_vector_alignment): New function.
986 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
988 (DR_TARGET_ALIGNMENT): New macro.
989 (aligned_access_p): Update commentary.
990 (vect_known_alignment_in_bytes): New function.
991 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
993 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
994 Calculate the misalignment based on the target alignment rather than
996 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
997 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
998 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
999 the target alignment, rather than masking the element misalignment
1000 with the number of elements in a vector. Also use the target
1001 alignment when calculating the maximum number of peels.
1002 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1003 instead of TYPE_ALIGN_UNIT.
1004 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1005 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1006 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1007 (vect_create_data_ref_ptr): Likewise.
1008 (vect_setup_realignment): Realign by ANDing with
1009 -DR_TARGET_MISALIGNMENT.
1010 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1011 the number of peels based on DR_TARGET_ALIGNMENT.
1012 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1013 with the guaranteed alignment boundary when deciding whether
1015 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1016 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1017 (ensure_base_align): Remove stmt_info parameter. Get the
1018 target base alignment from DR_TARGET_ALIGNMENT.
1019 (vectorizable_store): Update call accordingly. Interpret
1020 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1022 (vectorizable_load): Likewise.
1024 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1025 Alan Hayward <alan.hayward@arm.com>
1026 David Sherwood <david.sherwood@arm.com>
1028 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1029 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1030 (vect_enhance_data_refs_alignment): Likewise.
1032 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1034 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1035 error. Only quit immediately if parsing is complete.
1036 (BEGIN): Initialize fatal_err and parse_done.
1037 (begin fpu, end fpu): Check number of arguments.
1038 (begin arch, end arch): Likewise.
1039 (begin cpu, end cpu): Likewise.
1040 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1041 (optalias): Likewise.
1043 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1045 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1046 * config/arm/arm-isa.h: Delete. Move definitions to ...
1047 * arm-cpus.in: ... here. Use new feature and fgroup values.
1048 * config/arm/arm.c (arm_option_override): Use lower case for feature
1050 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1051 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1052 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1054 (print_isa_bits_for): New function.
1055 (gen_isa): New function.
1056 (gen_comm_data): Use print_isa_bits_for.
1057 (define feature): New keyword.
1058 (define fgroup): New keyword.
1059 * config/arm/t-arm (TM_H): Remove.
1060 (GTM_H): Add arm-isa.h.
1061 (arm-isa.h): Add rule to generate file.
1062 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1063 case for feature bit names.
1065 2017-09-22 Richard Biener <rguenther@suse.de>
1067 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1069 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1070 print of no dependency loops ...
1071 * graphite.c (graphite_transform_loops): ... here.
1072 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1074 (same_close_phi_node, remove_duplicate_close_phi,
1075 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1076 (canonicalize_loop_closed_ssa): ... here and simplify.
1077 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1078 (optimize_isl): Use dump_printf_loc to tell when we stopped
1079 optimizing because of an ISL timeout.
1081 2017-09-22 Richard Biener <rguenther@suse.de>
1083 PR tree-optimization/82291
1084 * tree-if-conv.c (predicate_mem_writes): Make sure to
1085 remove writes in blocks predicated with false.
1087 2017-09-22 Richard Biener <rguenther@suse.de>
1089 * sese.c: Include cfganal.h.
1090 (if_region_set_false_region): Remove.
1091 (create_if_region_on_edge): Likewise.
1092 (move_sese_in_condition): Re-implement without destroying
1095 2017-09-22 Richard Biener <rguenther@suse.de>
1097 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1098 Verify both BBs contain loop PHI nodes before dispatching to
1100 (graphite_regenerate_ast_isl): Do not recompute dominators,
1101 do not verify three times. Restructure for clarity.
1102 * graphite-scop-detection.c (same_close_phi_node,
1103 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1104 defined_in_loop_p, canonicalize_loop_closed_ssa,
1105 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1106 checking and SSA rewrite, move to ...
1107 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1108 (graphite_initialize): Do not pass in ctx, do not reset the
1109 SCEV cache, compute only dominators.
1110 (graphite_transform_loops): Allocate ISL ctx after
1111 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1112 Maintain post-dominators only around build_scops.
1113 * sese.c (if_region_set_false_region): Make static. Free
1114 and recompute dominators.
1115 (move_sese_in_condition): Assert we don't get called with
1116 post-dominators computed.
1117 * sese.h (if_region_set_false_region): Remove.
1119 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1121 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1122 mode attribute for TARGET_AVX512VL.
1124 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1126 * config/i386/i386.opt (mprefer-avx256): New option.
1127 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1129 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1130 for TARGET_PREFER_AVX256.
1131 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1133 2017-09-21 Jeff Law <law@redhat.com>
1135 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1136 Fix dump output if the only stack space is for pushed registers.
1138 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1140 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1143 2017-09-21 Martin Sebor <msebor@redhat.com>
1146 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1147 code (in C++) or code that triggers warnings.
1149 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1151 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1153 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1155 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1156 * sched-rgn.c: Ditto.
1157 * sel-sched-ir.c: Ditto.
1159 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1161 * toplev.h (set_random_seed): Adjust return type.
1162 * toplev.c (init_local_tick): Move eager initialization of random_seed
1163 to get_random_seed. Adjust comment.
1164 (init_random_seed): Inline to get_random_seed, delete.
1165 (get_random_seed): Initialize random_seed lazily.
1166 (set_random_seed): Do not return previous value.
1167 (print_switch_value): Do not call get_random_seed.
1169 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1171 * cgraph.c (delete_function_version): New, broken out from...
1172 (cgraph_node::delete_function_version): ...here. Rename to
1173 cgraph_node::delete_function_version_by_decl. Update all uses.
1174 (cgraph_node::remove): Call delete_function_version.
1176 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1179 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1180 VAR_DECLs to which addressable non-volatile parameters are mapped
1181 and for id->retvar after the return value assignment. Clear
1182 id->retval and id->retbnd after inlining.
1184 2017-09-21 Richard Biener <rguenther@suse.de>
1186 PR tree-optimization/82276
1187 PR tree-optimization/82244
1188 * tree-vrp.c (build_assert_expr_for): Set
1189 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1191 (remove_range_assertions): Revert earlier change.
1193 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1196 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1198 2017-09-21 Richard Biener <rguenther@suse.de>
1200 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1201 Restore valid IL after code generation errors.
1202 * graphite.c (graphite_transform_loops): Diagnose code
1203 generation issues as MSG_MISSED_OPTIMIZATION and continue
1204 with processing SCOPs.
1206 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1207 Alan Hayward <alan.hayward@arm.com>
1208 David Sherwood <david.sherwood@arm.com>
1210 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1211 rather than choosing between plus_constant and gen_rtx_<CODE>.
1212 * expr.c (emit_push_insn): Likewise.
1213 (expand_expr_real_2): Likewise.
1215 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1216 Alan Hayward <alan.hayward@arm.com>
1217 David Sherwood <david.sherwood@arm.com>
1219 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1221 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1222 Alan Hayward <alan.hayward@arm.com>
1223 David Sherwood <david.sherwood@arm.com>
1225 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1226 calling tree_to_uhwi.
1228 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1229 Alan Hayward <alan.hayward@arm.com>
1230 David Sherwood <david.sherwood@arm.com>
1232 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1233 INTEGER_CST rather than a negative test for ADDR_EXPR.
1235 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1236 Alan Hayward <alan.hayward@arm.com>
1237 David Sherwood <david.sherwood@arm.com>
1239 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1240 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1242 2017-09-21 Richard Biener <rguenther@suse.de>
1244 PR tree-optimization/71351
1245 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1246 graphite_create_new_loop_guard): Remove, fold remaining parts
1248 (translate_isl_ast_node_for): ... here and simplify.
1250 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1253 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1254 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1255 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1256 alternative always use QI mode, for -Os imov (=R,R) alternative
1257 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1260 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1261 Jeff Law <law@redhat.com>
1263 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1264 (allocate_stack_space): New function, partially extracted from
1266 (s390_emit_prologue): Track offset to most recent stack probe.
1267 Code to allocate space moved into allocate_stack_space.
1268 Dump actions when no stack is allocated.
1269 (s390_prologue_plus_offset): New function.
1270 (s390_emit_stack_probe): Likewise.
1272 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1274 * common.opt (Wa, Wl, Wp, g, gz=): Add
1276 (gno-column-info): Remove.
1277 (gcolumn-info): Drop RejectNegative.
1279 (gno-record-gcc-switches): Remove.
1280 (grecord-gcc-switches): Drop RejectNegative.
1281 (gno-split-dwarf): Remove.
1282 (gsplit-dwarf): Drop RejectNegative.
1283 (gno-strict-dwarf): Remove.
1284 (gstrict-dwarf): Drop RejectNegative.
1285 * config/darwin.opt (gfull, gused): Add RejectNegative.
1286 * dwarf2out.c (gen_producer_string): Drop
1287 gno-record-gcc-switches handler.
1288 * optc-gen.awk: Add g to prefixes with negative forms.
1289 * opts-common.c (remapping_prefix_p): New.
1290 (find_opt): Check it.
1291 (generate_canonical_option): Test g prefix.
1292 (option_map): Add -gno- mapping.
1293 (add_misspelling_candidates): Check remapping_prefix_p.
1295 2017-09-20 Jeff Law <law@redhat.com>
1297 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1298 thinko in stack clash protection support.
1300 * explow.c (compute_stack_clash_protection_loop_data): Use
1301 CONST_INT_P instead of explicit test. Verify object is a
1302 CONST_INT_P before looking at INTVAL.
1303 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1304 instead of explicit test.
1306 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1309 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1310 address instead of to r1 and r11.
1312 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1314 * config.gcc: Support "knm".
1315 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1316 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1318 * config/i386/i386.c (m_KNM): Define.
1319 (processor_target_table): Add "knm".
1321 (ix86_option_override_internal): Add "knm".
1322 (ix86_issue_rate): Add PROCESSOR_KNM.
1323 (ix86_adjust_cost): Ditto.
1324 (ia32_multipass_dfa_lookahead): Ditto.
1325 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1326 (fold_builtin_cpu): Add M_INTEL_KNM.
1327 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1328 (processor_type): Add PROCESSOR_KNM.
1329 * config/i386/x86-tune.def: Add m_KNM.
1330 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1332 2017-09-20 Richard Biener <rguenther@suse.de>
1334 PR tree-optimization/80213
1335 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1336 are allowed in empty BBs as well.
1337 (canonicalize_loop_closed_ssa): Also look for other complex
1339 (scop_detection::get_sese): Include the loop-closed PHI block
1341 (scop_detection::merge_sese): Remove code adding extra blocks.
1342 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1343 (build_scops): Assert the final returned scop is invalid.
1345 2017-09-20 Richard Biener <rguenther@suse.de>
1347 PR tree-optimization/82264
1348 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1350 (vn_phi_lookup): Likewise.
1351 (vn_phi_insert): Likewise.
1353 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1355 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1356 that fits into uhwi or shwi, add DW_AT_const_value regardless
1357 of early_dwarf without going through RTL, using add_AT_unsigned
1360 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1361 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1362 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1363 (DEBUG_MACRO_SECTION): Likewise.
1364 (DEBUG_LTO_MACRO_SECTION): Likewise.
1365 (DEBUG_STR_DWO_SECTION): Likewise.
1366 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1367 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1368 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1369 (DEBUG_LTO_LINE_STR_SECTION): Define.
1370 (init_sections_and_labels): Initialize debug_line_str_section
1371 variable. Initialize debug_loc_section for -gdwarf-5 to
1372 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1374 2017-09-20 Richard Biener <rguenther@suse.de>
1376 * graphite-sese-to-poly.c (extract_affine): Properly handle
1377 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1379 2017-09-20 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/81373
1382 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1383 Force SESE live-out defs to be handled even if they are
1386 2017-09-19 Jeff Law <law@redhat.com>
1388 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1389 nothing for stack adjustments with REG_STACK_CHECK.
1390 * sched-deps.c (parse_add_or_inc): Reject insns with
1391 REG_STACK_CHECK from dependency breaking.
1392 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1393 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1394 * reg-notes.def (STACK_CHECK): New note.
1396 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1397 (ix86_expand_prologue): Dump stack clash info as needed.
1398 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1400 * function.c (dump_stack_clash_frame_info): New function.
1401 * function.h (dump_stack_clash_frame_info): Prototype.
1402 (enum stack_clash_probes): New enum.
1404 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1405 flag_stack_clash_protection.
1406 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1407 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1408 (arm_frame_pointer_required): Likewise.
1409 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1410 (ia64_expand_prologue): Likewise.
1411 * config/mips/mips.c (mips_expand_prologue): Likewise.
1412 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1413 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1414 (sparc_flat_expand_prologue): Likewise.
1415 * config/spu/spu.c (spu_expand_prologue): Likewise.
1417 * explow.c: Include "params.h".
1418 (anti_adjust_stack_and_probe_stack_clash): New function.
1419 (get_stack_check_protect): Likewise.
1420 (compute_stack_clash_protection_loop_data): Likewise.
1421 (emit_stack_clash_protection_loop_start): Likewise.
1422 (emit_stack_clash_protection_loop_end): Likewise.
1423 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1424 Use anti_adjust_stack_and_probe_stack_clash.
1425 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1426 (emit_stack_clash_protection_loop_start): Likewise.
1427 (emit_stack_clash_protection_loop_end): Likewise.
1428 * rtl.h (get_stack_check_protect): Prototype.
1429 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1430 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1431 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1433 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1435 * doc/tm.texi: Rebuilt.
1436 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1437 get_stack_check_protect.
1438 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1439 * config/arm/arm.c (arm_expand_prologue): Likewise.
1440 (arm_frame_pointer_required): Likewise.
1441 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1442 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1443 * config/mips/mips.c (mips_expand_prologue): Likewise.
1444 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1445 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1446 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1447 (sparc_flat_expand_prologue): Likewise.
1449 * common.opt (-fstack-clash-protection): New option.
1450 * flag-types.h (enum stack_check_type): Note difference between
1451 -fstack-check= and -fstack-clash-protection.
1452 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1453 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1454 * toplev.c (process_options): Issue warnings/errors for cases
1455 not handled with -fstack-clash-protection.
1456 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1457 (-fstack-check): Note additional problem with -fstack-check=generic.
1458 Note that -fstack-check is primarily for Ada and refer users
1459 to -fstack-clash-protection for stack-clash-protection.
1460 Document new params for stack clash protection.
1462 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1464 * config/i386/i386.c (ix86_split_long_move): Do not handle
1465 address used for LEA in a special way.
1467 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1469 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1471 2017-09-19 Martin Sebor <msebor@redhat.com>
1474 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1475 of incompatible types.
1477 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1479 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1480 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1481 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1482 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1484 2017-09-19 Richard Biener <rguenther@suse.de>
1486 PR tree-optimization/82244
1487 * tree-vrp.c (remove_range_assertions): Do not propagate
1488 a constant to abnormals but replace the assert with a copy.
1490 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1492 PR rtl-optimization/57878
1493 PR rtl-optimization/68988
1494 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1495 avoidance test involving non_reload_pseudos. Move frequency test
1496 below the general fragmentation avoidance test.
1498 2017-09-19 Richard Biener <rguenther@suse.de>
1500 PR tree-optimization/69728
1501 * graphite-sese-to-poly.c (schedule_error): New global.
1502 (add_loop_schedule): Handle empty domain by failing the
1504 (build_original_schedule): Handle schedule_error.
1506 2017-09-19 Richard Biener <rguenther@suse.de>
1508 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1509 Do not iterate to sibling loops but only to siblings of inner
1512 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1515 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1516 that the registers are different.
1518 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1520 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1521 to processor_model and "amdfam17h" to arch_names_table.
1522 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1524 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1527 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1529 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1530 Alan Hayward <alan.hayward@arm.com>
1531 David Sherwood <david.sherwood@arm.com>
1533 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1535 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1537 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1538 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1539 vect_schedule_slp_instance.
1540 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1541 Update call to vect_slp_analyze_node_operations. Simplify return
1543 (vect_slp_analyze_bb_1): Update call accordingly.
1544 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1545 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1546 (vect_schedule_slp): Update call accordingly.
1548 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1549 Alan Hayward <alan.hayward@arm.com>
1550 David Sherwood <david.sherwood@arm.com>
1552 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1553 with types that aren't in fact scalar.
1555 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1557 * tree-vect-slp.c (vect_record_max_nunits): New function,
1559 (vect_build_slp_tree_1): ...here.
1560 (vect_build_slp_tree_2): Call it for phis too.
1562 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1564 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1565 to vect_get_vec_def_for_operand when getting the mask operand.
1567 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1568 Alan Hayward <alan.hayward@arm.com>
1569 David Sherwood <david.sherwood@arm.com>
1571 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1574 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1575 Alan Hayward <alan.hayward@arm.com>
1576 David Sherwood <david.sherwood@arm.com>
1578 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1579 calculation for vector booleans.
1581 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1582 Alan Hayward <alan.hayward@arm.com>
1583 David Sherwood <david.sherwood@arm.com>
1585 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1587 (vect_transform_stmt): ...here.
1588 (vect_analyze_stmt): Use it instead of calling
1589 vectorizable_live_operation directly.
1591 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1593 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1594 non-SIMT targets in acc vector loops.
1596 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1598 * configure.ac: Add arc and check if assembler supports gdwarf2.
1599 * configure: Regenerate.
1601 2017-09-18 Richard Biener <rguenther@suse.de>
1603 PR tree-optimization/82220
1604 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1605 epilogue niters from the min_profitable_iters compute.
1607 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1610 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1611 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1612 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1614 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1617 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1618 switching to a new text section.
1620 2017-09-18 Richard Biener <rguenther@suse.de>
1622 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1624 (build_alias_set): Reject aliases with no access function.
1626 2017-09-18 Richard Biener <rguenther@suse.de>
1628 PR tree-optimization/79622
1629 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1631 (build_cross_bb_scalars_use): Likewise.
1633 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1635 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1637 2017-09-18 Alan Modra <amodra@gmail.com>
1640 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1641 stack_pointer_rtx for count 0. Update comments. Break up
1642 large rtl expression.
1644 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1646 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1647 Increase to 20 bytes.
1648 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1649 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1650 or avx version of the stub.
1652 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1655 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1656 compute the minimum stack alignment. Also update preferred stack
1657 boundary for leaf functions.
1659 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1661 PR tree-optimization/82228
1662 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1665 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1667 * common/config/nds32/nds32-common.c
1668 (nds32_option_optimization_table): Refine formatting.
1669 (nds32_option_optimization_table): Use -fsched-pressure and
1670 -fomit-frame-pointer for specific optimization level.
1672 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1674 * config/nds32/nds32.c: Refine formatting and comments.
1675 * config/nds32/nds32.h: Likewise.
1676 * config/nds32/nds32.md: Likewise.
1677 * config/nds32/nds32-cost.c: Likewise.
1678 * config/nds32/nds32-isr.c: Likewise.
1679 * config/nds32/nds32-md-auxiliary.c: Likewise.
1680 * config/nds32/nds32-multiple.md: Likewise.
1681 * config/nds32/nds32-predicates.c: Likewise.
1683 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1684 Jakub Jelinek <jakub@redhat.com>
1686 Add support for -std=c++2a.
1687 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1689 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1691 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1694 * doc/extend.texi (Common Function Attributes): Add
1695 references to ARM, AArch64, and S/390 specific attributes.
1696 (Function Specific Option Pragmas): Add AArch64 and S/390
1697 to list of back ends that support the target pragma.
1699 2017-09-15 Nathan Sidwell <nathan@acm.org>
1701 * doc/standards.texi: Fix C++17 description. Update URLs for
1704 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1706 * common.opt (Wcast-align=strict): New warning option.
1707 * doc/invoke.texi: Document -Wcast-align=strict.
1709 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1711 * cgraph.h (cgraph_thunk_info): Add comments.
1712 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1713 assert for VIRTUAL_* arguments stricter.
1715 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1717 PR tree-optimization/71026
1718 * match.pd: Move RDIV patterns from fold-const.c
1719 * fold-const.c (distribute_real_division): Removed.
1720 (fold_binary_loc): Remove calls to distribute_real_divison.
1722 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1724 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1725 c++1z and gnu++1z as deprecated. Change other references to
1726 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1727 Change -Wc++1z-compat to -Wc++17-compat.
1728 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1729 * dwarf2out.c (highest_c_language): Handle C++17.
1730 (gen_compile_unit_die): Likewise.
1732 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1734 PR rtl-optimization/82192
1735 * combine.c (make_extraction): Don't look through non-paradoxical
1736 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1739 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * target.def (function_arg_offset): New hook.
1744 * targhooks.h (default_function_arg_offset): Declare.
1745 * targhooks.c (default_function_arg_offset): New function.
1746 * function.c (locate_and_pad_parm): Use
1747 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1748 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1749 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1750 * doc/tm.texi: Regenerate.
1751 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1752 * config/spu/spu.c (spu_function_arg_offset): New function.
1753 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1754 * system.h (FUNCTION_ARG_OFFSET): Poison.
1756 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1757 Alan Hayard <alan.hayward@arm.com>
1758 David Sherwood <david.sherwood@arm.com>
1760 * target.def (truly_noop_truncation): New hook.
1761 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1762 than TRULY_NOOP_TRUNCATION.
1763 * hooks.h (hook_bool_uint_uint_true): Declare.
1764 * hooks.c (hook_bool_uint_uint_true): New function.
1765 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1766 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1767 * doc/tm.texi: Regenerate.
1768 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1769 rather than TRULY_NOOP_TRUNCATION in comments.
1770 (simplify_comparison): Likewise.
1771 (record_truncated_value): Likewise.
1772 * expmed.c (extract_bit_field_1): Likewise.
1773 (extract_split_bit_field): Likewise.
1774 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1775 instead of TRULY_NOOP_TRUNCATION.
1776 * function.c (assign_parm_setup_block): Likewise.
1777 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1778 * rtlhooks.c: Include target.h.
1779 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1780 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1781 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1782 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1783 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1784 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1785 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1786 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1787 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1788 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1789 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1790 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1791 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1792 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1793 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1794 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1795 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1796 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1797 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1798 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1799 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1800 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1801 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1802 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1803 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1804 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1805 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1806 rather than TRULY_NOOP_TRUNCATION in comments.
1807 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1808 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1809 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1810 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1811 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1812 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1813 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1814 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1815 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1816 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1817 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1818 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1819 rather than TRULY_NOOP_TRUNCATION in comments.
1820 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1821 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1822 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1823 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1824 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1825 TRULY_NOOP_TRUNCATION condition.
1826 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1827 (TRULY_NOOP_TRUNCATION): Delete.
1828 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1829 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1830 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1831 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1832 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1833 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1834 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1835 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1836 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1837 rather than TRULY_NOOP_TRUNCATION in comments.
1838 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1839 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1840 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1841 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1842 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1843 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1845 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1848 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1849 (*cmp_ior): Likewise.
1850 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1851 (*ior_scc_scc_cmp): Likewise.
1852 (*and_scc_scc): Likewise.
1853 (*and_scc_scc_cmp): Likewise.
1855 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1856 Alan Hayard <alan.hayward@arm.com>
1857 David Sherwood <david.sherwood@arm.com>
1859 * target.def (can_change_mode_class): New hook.
1860 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1861 (hard_regno_nregs): Likewise.
1862 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1863 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1864 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1865 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1866 (LOAD_EXTEND_OP): Update accordingly.
1867 * doc/tm.texi: Regenerate.
1868 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1869 CANNOT_CHANGE_MODE_CLASS.
1870 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1871 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1872 * combine.c (simplify_set): Update accordingly.
1873 * emit-rtl.c (validate_subreg): Likewise.
1874 * recog.c (general_operand): Likewise.
1875 * regcprop.c (mode_change_ok): Likewise.
1876 * reload1.c (choose_reload_regs): Likewise.
1877 (inherit_piecemeal_p): Likewise.
1878 * rtlanal.c (simplify_subreg_regno): Likewise.
1879 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1880 instead of CANNOT_CHANGE_MODE_CLASS.
1881 (reload_cse_simplify_operands): Likewise.
1882 * reload.c (push_reload): Use targetm.can_change_mode_class
1883 instead of CANNOT_CHANGE_MODE_CLASS.
1884 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1885 REG_CANNOT_CHANGE_MODE_P.
1886 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1887 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1888 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1889 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1890 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1891 (arm_can_change_mode_class): New function.
1892 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1893 than CANNOT_CHANGE_MODE_CLASS in comments.
1894 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1895 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1896 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1897 (ix86_can_change_mode_class): ...this new function, inverting the
1898 sense of the return value.
1899 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1900 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1901 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1902 (ia64_can_change_mode_class): New function.
1903 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1904 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1905 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1906 (m32c_can_change_mode_class): ...this new function, inverting the
1907 sense of the return value.
1908 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1909 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1910 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1911 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1912 (mips_can_change_mode_class): ...this new function, inverting the
1913 sense of the return value.
1914 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1915 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1916 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1917 (msp430_can_change_mode_class): New function.
1918 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1919 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1920 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1921 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1922 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1923 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1924 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1925 (pa_cannot_change_mode_class): Replace with...
1926 (pa_can_change_mode_class): ...this new function, inverting the
1927 sense of the return value.
1928 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1929 than CANNOT_CHANGE_MODE_CLASS in comments.
1930 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1931 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1932 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1933 (pdp11_cannot_change_mode_class): Replace with...
1934 (pdp11_can_change_mode_class): ...this new function, inverting the
1935 sense of the return value.
1936 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1937 * config/powerpcspe/powerpcspe-protos.h
1938 (rs6000_cannot_change_mode_class_ptr): Delete.
1939 * config/powerpcspe/powerpcspe.c
1940 (rs6000_cannot_change_mode_class_ptr): Delete.
1941 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1942 (rs6000_option_override_internal): Assign to
1943 targetm.can_change_mode_class instead of
1944 rs6000_cannot_change_mode_class_ptr.
1945 (rs6000_cannot_change_mode_class): Replace with...
1946 (rs6000_can_change_mode_class): ...this new function, inverting the
1947 sense of the return value.
1948 (rs6000_debug_cannot_change_mode_class): Replace with...
1949 (rs6000_debug_can_change_mode_class): ...this new function.
1950 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1951 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1952 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1953 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1954 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1956 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1957 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1958 (rs6000_option_override_internal): Assign to
1959 targetm.can_change_mode_class instead of
1960 rs6000_cannot_change_mode_class_ptr.
1961 (rs6000_cannot_change_mode_class): Replace with...
1962 (rs6000_can_change_mode_class): ...this new function, inverting the
1963 sense of the return value.
1964 (rs6000_debug_cannot_change_mode_class): Replace with...
1965 (rs6000_debug_can_change_mode_class): ...this new function.
1966 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1967 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1968 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1969 (s390_can_change_mode_class): ...this new function, inverting the
1970 sense of the return value.
1971 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1972 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1973 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1974 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1975 (sh_cannot_change_mode_class): Replace with...
1976 (sh_can_change_mode_class): ...this new function, inverting the
1977 sense of the return value.
1978 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1979 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1980 (sparc_can_change_mode_class): New function.
1981 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1982 * config/spu/spu.c (spu_can_change_mode_class): New function.
1983 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1984 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1985 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1986 (visium_can_change_mode_class): New function.
1987 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1989 2017-09-15 Richard Biener <rguenther@suse.de>
1991 PR tree-optimization/82217
1992 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1993 but not undefined case.
1995 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1998 * postreload.c (reload_cse_simplify_operands): Skip
1999 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2001 2017-09-15 Richard Biener <rguenther@suse.de>
2003 PR tree-optimization/68823
2004 * graphite-scop-detection.c (build_alias_set): If we have a
2005 possible dependence check whether we can handle them by just
2006 looking at the DRs DR_ACCESS_FNs.
2007 (build_scops): If build_alias_set fails, fail the SCOP.
2009 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2011 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2012 to support float128 built-in functions that require the ISA 3.0
2014 (BU_FLOAT128_3_HW): Likewise.
2015 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2017 (FMAF128): Likewise.
2019 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2020 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2021 floating point instructions.
2022 (rs6000_invalid_builtin): Likewise.
2023 (rs6000_builtin_mask_names): Likewise.
2024 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2025 (RS6000_BTM_FLOAT128_HW): Likewise.
2026 (RS6000_BTM_COMMON): Likewise.
2027 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2029 * doc/extend.texi (RS/6000 built-in functions): Document the
2030 IEEE 128-bit floating point square root and fused multiply-add
2033 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2035 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2036 reg (r2) isn't in the set of registers defined in the prologue.
2038 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2039 Alan Hayward <alan.hayward@arm.com>
2040 David Sherwood <david.sherwood@arm.com>
2042 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2043 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2044 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2045 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2046 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2048 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2049 max_vectorization_factor.
2050 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2052 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2053 Alan Hayward <alan.hayward@arm.com>
2054 David Sherwood <david.sherwood@arm.com>
2056 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2057 (vect_worthwhile_without_simd_p): Declare.
2058 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2059 (vectorizable_reduction): Use it.
2060 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2061 (vectorizable_operation): Likewise.
2063 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2064 Alan Hayward <alan.hayward@arm.com>
2065 David Sherwood <david.sherwood@arm.com>
2067 * tree-vectorizer.h (vect_get_num_copies): New function.
2068 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2069 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2070 (vectorizable_induction): Likewise.
2071 (vectorizable_live_operation): Likewise.
2072 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2073 (vectorizable_bswap): Likewise.
2074 (vectorizable_call): Likewise.
2075 (vectorizable_conversion): Likewise.
2076 (vectorizable_assignment): Likewise.
2077 (vectorizable_shift): Likewise.
2078 (vectorizable_operation): Likewise.
2079 (vectorizable_store): Likewise.
2080 (vectorizable_load): Likewise.
2081 (vectorizable_condition): Likewise.
2082 (vectorizable_comparison): Likewise.
2083 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2085 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2090 of vect_init_vector.
2092 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2093 Alan Hayward <alan.hayward@arm.com>
2094 David Sherwood <david.sherwood@arm.com>
2096 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2097 an inline wrapper that provides a location.
2098 (gimple_build_vector): Likewise.
2099 * gimple-fold.c (gimple_build_vector_from_val): New function.
2100 (gimple_build_vector): Likewise.
2101 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2102 functions to build the initial value. Always return a gimple value.
2103 (get_initial_defs_for_reduction): Likewise. Only compute
2105 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2106 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2107 (vectorizable_induction): Use gimple_build_vector rather than
2110 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2111 Alan Hayward <alan.hayward@arm.com>
2112 David Sherwood <david.sherwood@arm.com>
2114 * target.h (vec_perm_indices): New typedef.
2115 (auto_vec_perm_indices): Likewise.
2116 * optabs-query.h: Include target.h
2117 (can_vec_perm_p): Take a vec_perm_indices *.
2118 * optabs-query.c (can_vec_perm_p): Likewise.
2119 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2120 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2121 * tree-vect-generic.c (lower_vec_perm): Likewise.
2122 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2123 (vect_grouped_load_supported): Likewise.
2124 (vect_shift_permute_load_chain): Likewise.
2125 (vect_permute_store_chain): Use auto_vec_perm_indices.
2126 (vect_permute_load_chain): Likewise.
2127 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2128 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2129 Update uses of can_vec_perm_p.
2130 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2131 mode with a number of elements. Take a vec_perm_indices *.
2132 (vect_create_epilog_for_reduction): Update accordingly.
2133 Use auto_vec_perm_indices.
2134 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2135 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2136 (vect_transform_slp_perm_load): Likewise.
2137 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2138 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2139 (vect_gen_perm_mask_checked): Likewise.
2140 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2141 (vect_gen_perm_mask_checked): Likewise.
2142 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2143 (vectorizable_store): Likewise.
2144 (vectorizable_load): Likewise.
2145 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2146 (vectorizable_bswap): Likewise.
2148 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2149 Alan Hayward <alan.hayward@arm.com>
2150 David Sherwood <david.sherwood@arm.com>
2152 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2153 * tree.c (build_vector): Likewise.
2154 (build_vector_from_ctor): Update accordingly.
2155 (build_vector_from_val): Likewise.
2156 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2157 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2158 * tree-vect-generic.c (add_rshift): Likewise.
2159 (expand_vector_divmod): Likewise.
2160 (optimize_vector_constructor): Likewise.
2161 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2162 (vect_transform_slp_perm_load): Likewise.
2163 (vect_schedule_slp_instance): Likewise.
2164 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2165 (vectorizable_call): Likewise.
2166 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2167 * expmed.c (make_tree): Likewise.
2168 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2169 a vector passed to build_vector.
2170 (fold_convert_const): Likewise.
2171 (exact_inverse): Likewise.
2172 (fold_ternary_loc): Likewise.
2173 (fold_relational_const): Likewise.
2174 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2175 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2176 (const_unop): Likewise. Store the reduction accumulator in a
2177 variable rather than an array.
2178 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2179 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2180 the new vector, rather than constructing it after the input arrays.
2181 (native_interpret_vector): Use auto_vec<tree> when building
2182 a vector passed to build_vector. Add elements in order.
2183 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2184 auto_vec<tree> when building a vector passed to build_vector.
2185 (vect_create_epilog_for_reduction): Likewise.
2186 (vectorizable_induction): Likewise.
2187 (get_initial_def_for_reduction): Likewise. Fix indentation of
2189 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2191 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2192 passed to build_vector.
2194 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2195 Alan Hayward <alan.hayward@arm.com>
2196 David Sherwood <david.sherwood@arm.com>
2198 * tree-core.h (tree_base::u): Add an "nelts" field.
2199 (tree_vector): Use VECTOR_CST_NELTS as the length.
2200 * tree.c (tree_size): Likewise.
2201 (make_vector): Initialize VECTOR_CST_NELTS.
2202 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2203 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2204 TYPE_VECTOR_SUBPARTS.
2205 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2206 as the number of units, setting it from VECTOR_CST_NELTS.
2207 (const_vector_from_tree): Likewise.
2208 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2209 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2210 (fold_negate_expr_1): Likewise.
2211 (fold_convert_const): Likewise.
2212 (const_binop): Likewise. Differentiate the number of output and
2214 (const_unop): Likewise.
2215 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2216 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2217 in cases that did the opposite.
2219 2017-09-14 Richard Biener <rguenther@suse.de>
2221 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2224 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2226 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2228 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2231 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2232 if and where to split a bb, except for splitting before debug insn
2233 sequences followed by non-label real insn. Delete debug insns
2234 in between basic blocks.
2236 * combine.c (make_compound_operation_int): Formatting fixes.
2238 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2239 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2240 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2241 * config/sol2.h (LINK_EH_SPEC): Likewise.
2242 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2243 * config/s390/linux.h (LINK_SPEC): Likewise.
2244 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2245 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2246 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2247 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2248 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2249 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2250 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2251 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2253 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2255 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2257 (ENDFILE_LINUX_SPEC): Likewise.
2258 (LINK_EH_SPEC): Likewise.
2259 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2260 (LINK_OS_LINUX_SPEC32): Likewise.
2261 (LINK_OS_LINUX_SPEC64): Likewise.
2262 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2263 (LINK_OS_LINUX_SPEC): Likewise.
2265 2017-09-13 Martin Liska <mliska@suse.cz>
2268 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2269 CASE_HIGH is NULL_TREE.
2271 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * target.def (secondary_memory_needed): New hook.
2276 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2277 instead of SECONDARY_MEMORY_NEEDED.
2278 (secondary_memory_needed_mode): Likewise.
2279 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2280 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2281 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2282 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2283 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2284 * doc/tm.texi: Regenerate.
2285 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2286 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2287 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2288 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2289 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2290 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2291 mode argument first and change the reg_class arguments to reg_class_t.
2292 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2293 Make static. Update the call to inline_secondary_memory_needed.
2294 (ix86_register_move_cost): Update the call to
2295 inline_secondary_memory_needed.
2296 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2297 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2299 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2300 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2302 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2303 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2304 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2305 and match hook interface. Add comment from mips.h.
2306 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2307 * config/mmix/mmix.md (truncdfsf2): Refer to
2308 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2310 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2311 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2312 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2313 (pa_secondary_memory_needed): New function.
2314 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2315 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2316 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2317 (pdp11_secondary_memory_needed): Make static and match hook interface.
2318 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2319 * config/powerpcspe/powerpcspe-protos.h
2320 (rs6000_secondary_memory_needed_ptr): Delete.
2321 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2323 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2324 (rs6000_option_override_internal): Assign to
2325 targetm.secondary_memory_needed rather than
2326 rs6000_secondary_memory_needed_ptr.
2327 (rs6000_secondary_memory_needed): Match hook interface.
2328 (rs6000_debug_secondary_memory_needed): Likewise.
2329 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2330 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2331 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2332 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2333 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2334 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2336 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2337 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2338 (rs6000_option_override_internal): Assign to
2339 targetm.secondary_memory_needed rather than
2340 rs6000_secondary_memory_needed_ptr.
2341 (rs6000_secondary_memory_needed): Match hook interface.
2342 (rs6000_debug_secondary_memory_needed): Likewise.
2343 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2344 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2345 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2346 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2347 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2348 (sparc_secondary_memory_needed): New function.
2349 * lra-constraints.c (check_and_process_move): Refer to
2350 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2352 (curr_insn_transform): Likewise.
2353 (process_alt_operands): Use targetm.secondary_memory_needed
2354 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2355 (check_secondary_memory_needed_p): Likewise.
2356 (choose_split_class): Likewise.
2357 * reload.c: Unconditionally include code that was previously
2358 conditional on SECONDARY_MEMORY_NEEDED.
2359 (push_secondary_reload): Use targetm.secondary_memory_needed
2360 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2361 (push_reload): Likewise.
2362 * reload1.c: Unconditionally include code that was previously
2363 conditional on SECONDARY_MEMORY_NEEDED.
2364 (choose_reload_regs): Use targetm.secondary_memory_needed
2365 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2366 (gen_reload): Likewise.
2367 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2369 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2370 Alan Hayward <alan.hayward@arm.com>
2371 David Sherwood <david.sherwood@arm.com>
2373 * target.def (secondary_memory_needed_mode): New hook:
2374 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2375 * targhooks.h (default_secondary_memory_needed_mode): New function.
2376 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2377 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2378 * doc/tm.texi: Regenerate.
2379 * lra-constraints.c (check_and_process_move): Use
2380 targetm.secondary_memory_needed_mode instead of
2381 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2382 (curr_insn_transform): Likewise.
2383 * reload.c (get_secondary_mem): Likewise.
2384 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2385 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2387 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2388 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2389 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2390 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2391 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2393 * config/powerpcspe/powerpcspe-protos.h
2394 (rs6000_secondary_memory_needed_mode): Delete.
2395 * config/powerpcspe/powerpcspe.c
2396 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2397 (rs6000_secondary_memory_needed_mode): Make static.
2398 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2399 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2401 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2403 (rs6000_secondary_memory_needed_mode): Make static.
2404 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2405 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2406 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2407 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2408 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2410 (sparc_secondary_memory_needed_mode): New function.
2411 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2413 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2415 * config/aarch64/constraints.md (Umq): New constraint.
2416 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2418 (mov<mode>): Update condition.
2420 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2422 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2423 when bitposition is the same.
2425 2017-09-13 Richard Biener <rguenther@suse.de>
2427 * dwarf2out.c (output_die_symbol): Remove.
2428 (output_die): Do not output a DIEs symbol.
2430 2017-09-13 Richard Biener <rguenther@suse.de>
2433 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2434 default-def to avoid breaking iterator update with the weird
2435 interaction with cgraph_update_edges_for_call_stmt_node.
2437 2017-09-13 Richard Biener <rguenther@suse.de>
2439 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2440 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2441 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2442 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2444 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2446 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2447 Disable pc relative literal load irrespective of
2448 TARGET_FIX_ERR_A53_84341 for default.
2450 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2452 * config/sparc/sparc.c (output_return): Output the source location of
2453 the insn in the delay slot, if any.
2454 (output_sibcall): Likewise.
2456 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2459 * common.opt (-static-pie): New alias.
2460 (shared): Negate static-pie.
2461 (-no-pie): Update help text.
2463 (static-pie): New option.
2464 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2465 -static-pie support.
2466 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2467 (LINK_EH_SPEC): Likewise.
2468 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2469 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2470 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2471 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2472 (init_gcc_specs): Likewise.
2473 (init_spec): Likewise.
2474 (display_help): Update help message for -pie.
2475 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2478 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2480 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2481 (movdi_aarch64): Likewise.
2482 (movti_aarch64): Likewise.
2484 2017-09-12 Simon Wright <simon@pushface.org>
2487 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2488 calculation of the minor version, always output as 0.
2490 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2493 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2494 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2495 on it early, rather than manual conversion late. For
2496 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2497 instead of performing manual conversion.
2499 2017-09-12 Carl Love <cel@us.ibm.com>
2501 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2502 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2504 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2505 VMULOSW): Add definitions.
2506 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2507 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2508 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2510 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2512 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2514 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2516 (movti_aarch64): Likewise.
2517 (movdf_aarch64): Likewise.
2518 (movtf_aarch64): Likewise.
2519 (load_pairdi): Likewise.
2520 (store_pairdi): Likewise.
2521 (load_pairdf): Likewise.
2522 (store_pairdf): Likewise.
2523 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2524 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2525 (ldr_got_small_<mode>): Likewise.
2526 (ldr_got_small_28k_<mode>): Likewise.
2527 (ldr_got_tiny): Likewise.
2528 * config/aarch64/iterators.md (ldst_sz): New.
2529 (ldpstp_sz): Likewise.
2530 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2532 (thunderx_load): Split load_8 to load_16.
2533 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2535 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2536 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2537 (xgene1_store_pair): Split store_8 to store_16.
2538 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2539 (falkor_st_0_st_sd): Split store_8 to store_16.
2541 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2543 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2544 and store1/2/3/4 to store_4/8/12/16.
2545 * config/aarch64/aarch64.md: Update for rename.
2546 * config/arm/arm.md: Likewise.: Likewise.
2547 * config/arm/arm.c: Likewise.
2548 * config/arm/thumb1.md: Likewise.
2549 * config/arm/thumb2.md: Likewise.
2550 * config/arm/vfp.md: Likewise.
2551 * config/arm/arm-generic.md: Likewise.
2552 * config/arm/arm1020e.md: Likewise.
2553 * config/arm/arm1026ejs.md: Likewise.
2554 * config/arm/arm1136jfs.md: Likewise.
2555 * config/arm/arm926ejs.md: Likewise.
2556 * config/arm/cortex-a15.md: Likewise.
2557 * config/arm/cortex-a17.md: Likewise.
2558 * config/arm/cortex-a5.md: Likewise.
2559 * config/arm/cortex-a53.md: Likewise.
2560 * config/arm/cortex-a57.md: Likewise.
2561 * config/arm/cortex-a7.md: Likewise.
2562 * config/arm/cortex-a8.md: Likewise.
2563 * config/arm/cortex-a9.md: Likewise.
2564 * config/arm/cortex-m4.md: Likewise.
2565 * config/arm/cortex-m7.md: Likewise.
2566 * config/arm/cortex-r4.md: Likewise.
2567 * config/arm/exynos-m1.md: Likewise.
2568 * config/arm/fa526.md: Likewise.
2569 * config/arm/fa606te.md: Likewise.
2570 * config/arm/fa626te.md: Likewise.
2571 * config/arm/fa726te.md: Likewise.
2572 * config/arm/fmp626.md: Likewise.
2573 * config/arm/iwmmxt.md: Likewise.
2574 * config/arm/ldmstm.md: Likewise.
2575 * config/arm/marvell-pj4.md: Likewise.
2576 * config/arm/xgene1.md: Likewise.
2577 * config/aarch64/thunderx.md: Likewise.
2578 * config/aarch64/thunderx2t99.md: Likewise.
2579 * config/aarch64/falkor.md: Likewise.
2581 2017-09-12 Martin Liska <mliska@suse.cz>
2583 * attribs.c (private_lookup_attribute): New function.
2584 * attribs.h (private_lookup_attribute): Declared here.
2585 (lookup_attribute): Called from this place.
2587 2017-09-12 Richard Biener <rguenther@suse.de>
2589 PR tree-optimization/82157
2590 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2591 stmts with side-effects.
2593 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2594 Alan Hayward <alan.hayward@arm.com>
2595 David Sherwood <david.sherwood@arm.com>
2597 * target.def (hard_regno_nregs): New hook.
2598 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2599 * targhooks.h (default_hard_regno_nregs): Declare.
2600 * targhooks.c (default_hard_regno_nregs): New function.
2601 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2602 (TARGET_HARD_REGNO_NREGS): ...this hook.
2603 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2604 (CLASS_MAX_NREGS): Likewise.
2605 * doc/tm.texi: Regenerate.
2606 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2607 instead of HARD_REGNO_NREGS.
2608 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2609 HARD_REGNO_NREGS in the comment.
2610 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2611 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2612 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2613 Return an unsigned int.
2614 (TARGET_HARD_REGNO_NREGS): Redefine.
2615 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2616 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2617 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2618 (arc_hard_regno_nregs): New function.
2619 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2620 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2621 (arm_hard_regno_nregs): New function.
2622 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2623 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2624 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2625 (TARGET_HARD_REGNO_NREGS): Redefine.
2626 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2627 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2628 (HARD_REGNO_NREGS): Delete.
2629 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2630 (cr16_hard_regno_nregs): New function.
2631 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2632 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2633 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2634 (cris_hard_regno_nregs): New function.
2635 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2636 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2637 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2638 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2639 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2640 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2641 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2642 (frv_hard_regno_nregs): Make static. Take and return an
2644 (frv_class_max_nregs): Remove outdated copy of documentation.
2645 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2646 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2647 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2648 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2649 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2650 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2651 (TARGET_HARD_REGNO_NREGS): Redefine.
2652 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2653 (CLASS_MAX_NREGS): Update comment.
2654 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2655 (ia64_hard_regno_nregs): New function.
2656 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2657 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2658 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2659 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2660 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2662 (m32c_hard_regno_nregs): Likewise. Make static.
2663 (TARGET_HARD_REGNO_NREGS): Redefine.
2664 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2665 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2666 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2667 (m68k_hard_regno_nregs): New function.
2668 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2669 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2670 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2671 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2672 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2673 Take and return an unsigned int.
2674 (TARGET_HARD_REGNO_NREGS): Redefine.
2675 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2676 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2677 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2678 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2679 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2680 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2681 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2682 (msp430_hard_regno_nregs): Make static. Take and return an
2684 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2685 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2686 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2687 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2688 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2689 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2690 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2691 (TARGET_HARD_REGNO_NREGS): Redefine.
2692 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2693 (PA_HARD_REGNO_NREGS): ...this.
2694 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2695 (PA_HARD_REGNO_NREGS): ...this.
2696 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2697 (pa_hard_regno_nregs): New function.
2698 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2699 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2700 (pdp11_hard_regno_nregs): New function.
2701 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2702 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2703 (rs6000_hard_regno_nregs_hook): New function.
2704 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2705 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2706 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2707 Take and return an unsigned int. Move earlier in file.
2708 (TARGET_HARD_REGNO_NREGS): Redefine.
2709 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2710 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2711 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2712 (rl78_hard_regno_nregs): Make static. Take and return an
2714 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2715 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2716 (rs6000_hard_regno_nregs_hook): New function.
2717 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2718 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2719 (TARGET_HARD_REGNO_NREGS): Redefine.
2720 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2721 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2722 instead of HARD_REGNO_NREGS.
2723 (s390_hard_regno_nregs): New function.
2724 (s390_hard_regno_mode_ok): Add comment from s390.h.
2725 (TARGET_HARD_REGNO_NREGS): Redefine.
2726 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2727 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2728 (sh_hard_regno_nregs): New function.
2729 (sh_pass_in_reg_p): Use it.
2730 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2731 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2732 (sparc_hard_regno_nregs): New function.
2733 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2734 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2735 (spu_function_arg_advance): Use it, supplying a valid register number.
2736 (TARGET_HARD_REGNO_NREGS): Redefine.
2737 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2738 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2739 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2740 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2741 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2742 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2743 (CLASS_MAX_NREGS): Remove copy of old documentation.
2744 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2745 (visium_hard_regno_nregs): New function.
2746 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2747 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2748 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2749 xtensa_hard_regno_nregs): New function.
2750 * system.h (HARD_REGNO_NREGS): Poison.
2752 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2754 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2755 hard_regno_nregs instead of HARD_REGNO_NREGS.
2756 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2757 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2758 (c6x_expand_epilogue): Likewise.
2759 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2760 (frv_read_iacc_argument): Likewise.
2761 * config/sh/sh.c: Include regs.h.
2762 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2763 (regs_used): Likewise.
2764 (output_stack_adjust): Likewise.
2765 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2766 * expmed.c: Include regs.h.
2767 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2768 * ree.c: Include regs.h.
2769 (combine_reaching_defs): Use hard_regno_nregs instead of
2771 (add_removable_extension): Likewise.
2773 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2775 * regs.h (hard_regno_nregs): Turn into a function.
2776 (end_hard_regno): Update accordingly.
2777 * caller-save.c (setup_save_areas): Likewise.
2778 (save_call_clobbered_regs): Likewise.
2779 (replace_reg_with_saved_mem): Likewise.
2780 (insert_restore): Likewise.
2781 (insert_save): Likewise.
2782 * combine.c (can_change_dest_mode): Likewise.
2783 (move_deaths): Likewise.
2784 (distribute_notes): Likewise.
2785 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2786 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2787 (rs6000_split_multireg_move): Likewise.
2788 (rs6000_register_move_cost): Likewise.
2789 (rs6000_memory_move_cost): Likewise.
2790 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2791 (rs6000_split_multireg_move): Likewise.
2792 (rs6000_register_move_cost): Likewise.
2793 (rs6000_memory_move_cost): Likewise.
2794 * cselib.c (cselib_reset_table): Likewise.
2795 (cselib_lookup_1): Likewise.
2796 * emit-rtl.c (set_mode_and_regno): Likewise.
2797 * function.c (aggregate_value_p): Likewise.
2798 * ira-color.c (setup_profitable_hard_regs): Likewise.
2799 (check_hard_reg_p): Likewise.
2800 (calculate_saved_nregs): Likewise.
2801 (assign_hard_reg): Likewise.
2802 (improve_allocation): Likewise.
2803 (calculate_spill_cost): Likewise.
2804 * ira-emit.c (modify_move_list): Likewise.
2805 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2806 (ira_hard_reg_in_set_p): Likewise.
2807 * ira.c (setup_reg_mode_hard_regset): Likewise.
2808 (clarify_prohibited_class_mode_regs): Likewise.
2809 (check_allocation): Likewise.
2810 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2811 (lra_setup_reg_renumber): Likewise.
2812 (setup_try_hard_regno_pseudos): Likewise.
2813 (spill_for): Likewise.
2814 (assign_hard_regno): Likewise.
2815 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2816 * lra-constraints.c (in_class_p): Likewise.
2817 (lra_constraint_offset): Likewise.
2818 (simplify_operand_subreg): Likewise.
2819 (lra_constraints): Likewise.
2820 (split_reg): Likewise.
2821 (split_if_necessary): Likewise.
2822 (invariant_p): Likewise.
2823 (inherit_in_ebb): Likewise.
2824 * lra-lives.c (process_bb_lives): Likewise.
2825 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2826 (get_hard_regs): Likewise.
2827 (do_remat): Likewise.
2828 * lra-spills.c (assign_spill_hard_regs): Likewise.
2829 * mode-switching.c (create_pre_exit): Likewise.
2830 * postreload.c (reload_combine_recognize_pattern): Likewise.
2831 * recog.c (peep2_find_free_register): Likewise.
2832 * regcprop.c (kill_value_regno): Likewise.
2833 (set_value_regno): Likewise.
2834 (copy_value): Likewise.
2835 (maybe_mode_change): Likewise.
2836 (find_oldest_value_reg): Likewise.
2837 (copyprop_hardreg_forward_1): Likewise.
2838 * regrename.c (check_new_reg_p): Likewise.
2839 (regrename_do_replace): Likewise.
2840 * reload.c (push_reload): Likewise.
2841 (combine_reloads): Likewise.
2842 (find_dummy_reload): Likewise.
2843 (operands_match_p): Likewise.
2844 (find_reloads): Likewise.
2845 (find_equiv_reg): Likewise.
2846 (reload_adjust_reg_for_mode): Likewise.
2847 * reload1.c (count_pseudo): Likewise.
2848 (count_spilled_pseudo): Likewise.
2849 (find_reg): Likewise.
2850 (clear_reload_reg_in_use): Likewise.
2851 (free_for_value_p): Likewise.
2852 (allocate_reload_reg): Likewise.
2853 (choose_reload_regs): Likewise.
2854 (reload_adjust_reg_for_temp): Likewise.
2855 (emit_reload_insns): Likewise.
2856 (delete_output_reload): Likewise.
2857 * rtlanal.c (subreg_get_info): Likewise.
2858 * sched-deps.c (sched_analyze_reg): Likewise.
2859 * sel-sched.c (init_regs_for_mode): Likewise.
2860 (mark_unavailable_hard_regs): Likewise.
2861 (choose_best_reg_1): Likewise.
2862 (verify_target_availability): Likewise.
2863 * valtrack.c (dead_debug_insert_temp): Likewise.
2864 * var-tracking.c (track_loc_p): Likewise.
2865 (emit_note_insn_var_location): Likewise.
2866 * varasm.c (make_decl_rtl): Likewise.
2867 * reginfo.c (choose_hard_reg_mode): Likewise.
2868 (init_reg_modes_target): Refer directly to
2869 this_target_regs->x_hard_regno_nregs.
2871 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2873 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2874 instead of hard_regno_nregs.
2876 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2878 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2879 end_hard_regno instead of hard_regno_nregs.
2880 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2881 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2882 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2883 * ira-color.c (improve_allocation): Likewise.
2884 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2885 * lra-lives.c (mark_regno_live): Likewise.
2886 (mark_regno_dead): Likewise.
2887 * lra-remat.c (operand_to_remat): Likewise.
2888 * lra.c (collect_non_operand_hard_regs): Likewise.
2889 * postreload.c (reload_combine_note_store): Likewise.
2890 (move2add_valid_value_p): Likewise.
2891 * reload.c (regno_clobbered_p): Likewise.
2893 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2895 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2897 * config/v850/v850.c (v850_reorg): Likewise.
2898 * reload.c (refers_to_regno_for_reload_p): Likewise.
2899 (find_equiv_reg): Likewise.
2900 * reload1.c (reload_reg_reaches_end_p): Likewise.
2902 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2904 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2906 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2907 * config/arm/arm.c (output_move_neon): Likewise.
2908 (arm_attr_length_move_neon): Likewise.
2909 (neon_split_vcombine): Likewise.
2910 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2911 (c6x_mark_reg_written): Likewise.
2912 (c6x_dwarf_register_span): Likewise.
2913 * config/i386/i386.c (ix86_save_reg): Likewise.
2914 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2915 (rws_access_reg): Likewise.
2916 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2917 * mode-switching.c (create_pre_exit): Likewise.
2918 * ree.c (combine_reaching_defs): Likewise.
2919 (add_removable_extension): Likewise.
2920 * regcprop.c (find_oldest_value_reg): Likewise.
2921 (copyprop_hardreg_forward_1): Likewise.
2922 * reload.c (reload_inner_reg_of_subreg): Likewise.
2923 (push_reload): Likewise.
2924 (combine_reloads): Likewise.
2925 (find_dummy_reload): Likewise.
2926 (reload_adjust_reg_for_mode): Likewise.
2927 * reload1.c (find_reload_regs): Likewise.
2928 (forget_old_reloads_1): Likewise.
2929 (reload_reg_free_for_value_p): Likewise.
2930 (reload_adjust_reg_for_temp): Likewise.
2931 (emit_reload_insns): Likewise.
2932 (delete_output_reload): Likewise.
2933 * sel-sched.c (choose_best_reg_1): Likewise.
2934 (choose_best_pseudo_reg): Likewise.
2936 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2941 * target.def (slow_unaligned_access): New hook.
2942 * targhooks.h (default_slow_unaligned_access): Declare.
2943 * targhooks.c (default_slow_unaligned_access): New function.
2944 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2945 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2946 * doc/tm.texi: Regenerate.
2947 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2948 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2949 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2951 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2952 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2954 (rs6000_slow_unaligned_access): New function.
2955 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2956 (expand_block_compare): Likewise.
2957 (expand_strn_compare): Likewise.
2958 (rs6000_rtx_costs): Likewise.
2959 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2960 (riscv_slow_unaligned_access): Likewise.
2961 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2962 (riscv_slow_unaligned_access_p): ...this and make static.
2963 (riscv_option_override): Update accordingly.
2964 (riscv_slow_unaligned_access): New function.
2965 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2966 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2967 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2968 (rs6000_slow_unaligned_access): New function.
2969 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2970 (rs6000_rtx_costs): Likewise.
2971 * config/rs6000/rs6000-string.c (expand_block_compare)
2972 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2973 of SLOW_UNALIGNED_ACCESS.
2974 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2975 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2976 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2977 of SLOW_UNALIGNED_ACCESS.
2978 * expmed.c (simple_mem_bitfield_p): Likewise.
2979 * expr.c (alignment_for_piecewise_move): Likewise.
2980 (emit_group_load_1): Likewise.
2981 (emit_group_store): Likewise.
2982 (copy_blkmode_from_reg): Likewise.
2983 (emit_push_insn): Likewise.
2984 (expand_assignment): Likewise.
2985 (store_field): Likewise.
2986 (expand_expr_real_1): Likewise.
2987 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2988 * lra-constraints.c (simplify_operand_subreg): Likewise.
2989 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2990 * gimple-ssa-store-merging.c: Likewise in block comment at start
2992 * tree-ssa-strlen.c: Include target.h.
2993 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2994 of SLOW_UNALIGNED_ACCESS.
2995 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2999 PR rtl-optimization/82185
3000 * expmed.c (emit_store_flag_int): Only test tem if it has been
3003 2017-09-12 Richard Biener <rguenther@suse.de>
3006 * match.pd ((FTYPE) N CMP CST): Fix typo.
3008 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3010 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3012 (mips_near_type_p): Add 'short_call' attribute as a synonym
3014 * doc/extend.texi (short_call): Document new function attribute.
3016 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3019 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3020 assertion check that in the condition.
3021 (get_atomic_generic_size): Likewise. Before testing if parameter
3022 has pointer type, if it has array type, call for C++
3023 default_conversion to perform array-to-pointer conversion.
3025 2017-09-12 Richard Biener <rguenther@suse.de>
3027 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3028 for operations we cannot scalarize.
3030 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3032 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3033 vectors heap vectors. Clean up comments.
3034 Make visited_bbs a reference.
3035 (profitable_jump_thread_path): Make GC
3036 vectors heap vectors. Clean up comments.
3038 (convert_and_register_jump_thread_path): Make GC vectors heap
3040 (check_subpath_and_update_thread_path): Same. Clean up comments.
3041 Make visited_bbs a reference.
3042 (handle_phi): Abstract common code to to
3043 register_jump_thread_path_if_profitable.
3044 Rename VAR_BB to DEF_BB.
3046 Make GC vectors heap vectors.
3047 Make visited_bbs a reference.
3048 (handle_assignment): Same.
3049 (register_jump_thread_path_if_profitable): New.
3050 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3052 Make GC vectors heap vectors. Clean up comments.
3053 Make visited_bbs a reference.
3054 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3055 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3058 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3061 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3062 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3064 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3066 Revert r251800 and r251799.
3068 2017-09-11 Martin Jambor <mjambor@suse.cz>
3071 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3072 arguments in advance.
3073 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3074 use it to find predecessor edges.
3075 (naive_outof_ssa): Collect vector of predecessors.
3077 2017-09-08 Jason Merrill <jason@redhat.com>
3079 PR c++/70029 - ICE with ref-qualifier and -flto
3080 * langhooks.h (struct lang_hooks_for_types): Add
3081 copy_lang_qualifiers.
3082 * attribs.c (build_type_attribute_qual_variant): Use it.
3083 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3085 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3086 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3088 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3091 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3092 (*mulsi3_sp64): New instruction.
3093 (mulsi3): New expander.
3095 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3097 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3099 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3101 * sancov.c: Include memmodel.h.
3103 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3106 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3109 2017-09-07 Carl Love <cel@us.ibm.com>
3111 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3112 the sldi instruction.
3114 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3116 * sancov.c: Include tm_p.h.
3118 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3121 * output.h (switch_to_other_text_partition): New declaration.
3122 * varasm.c (switch_to_other_text_partition): New function.
3123 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3124 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3125 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3126 to the other text partition before emitting LCL label and switch back
3127 after emitting the word after it.
3129 2017-09-07 Richard Biener <rguenther@suse.de>
3131 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3132 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3133 critical edges here, after loop init.
3134 (pass_data_pre): Remove PROP_no_crit_edges flags.
3135 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3136 for valueization of call args to avoid leaking VN_TOP.
3137 (visit_use): Assert we do not visit default defs.
3138 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3139 Use error_mark_node to more easily detect leaking VN_TOP.
3140 All default-defs are varying, not VN_TOP. Mark them visited.
3141 (run_scc_vn): Make code match comment.
3143 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3145 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3146 OPTION_MASK_FLOAT128_KEYWORD.
3147 (POWERPC_MASKS): Likewise.
3148 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3149 support for the -mfloat128-type option, and make -mfloat128
3150 default on PowerPC Linux systems. Define or undefine
3151 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3152 Define __float128 to be __ieee128 if IEEE 128-bit support is
3153 enabled, or undefine it.
3154 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3155 Delete defining __FLOAT128_TYPE__.
3156 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3157 -mfloat128-type option and make -mfloat128 default on PowerPC
3159 (TARGET_FLOAT128_TYPE): Likewise.
3160 (-mfloat128-type): Likewise.
3161 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3162 Delete the -mfloat128-type option and make -mfloat128 default on
3163 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3164 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3165 128-bit floating point is enabled. Change tests from using
3166 -mfloat128-type to -mfloat128.
3167 (rs6000_mangle_type): Use the correct mangling for the __float128
3168 type even if normal long double is restricted to 64-bits.
3169 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3171 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3172 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3173 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3175 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3176 documentation for -mfloat128.
3178 2017-09-06 Olivier Hainque <hainque@adacore.com>
3180 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3182 2017-09-06 Wish Wu <wishwu007@gmail.com>
3183 Jakub Jelinek <jakub@redhat.com>
3185 * asan.c (initialize_sanitizer_builtins): Add
3186 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3187 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3188 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3189 BT_FN_VOID_UINT64_PTR variables.
3190 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3191 (BT_FN_VOID_UINT16_UINT16): Likewise.
3192 (BT_FN_VOID_UINT32_UINT32): Likewise.
3193 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3194 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3195 (BT_FN_VOID_UINT64_PTR): Likewise.
3196 * common.opt (flag_sanitize_coverage): New variable.
3197 (fsanitize-coverage=trace-pc): Remove.
3198 (fsanitize-coverage=): Add.
3199 * flag-types.h (enum sanitize_coverage_code): New enum.
3200 * fold-const.c (fold_range_test): Disable non-short-circuit
3201 optimization if flag_sanitize_coverage.
3202 (fold_truth_andor): Likewise.
3203 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3204 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3205 (coverage_sanitizer_opts): New array.
3206 (get_closest_sanitizer_option): Add OPTS argument, handle also
3207 OPT_fsanitize_coverage_.
3208 (parse_sanitizer_options): Adjusted to also handle
3209 OPT_fsanitize_coverage_.
3210 (common_handle_option): Add OPT_fsanitize_coverage_.
3211 * sancov.c (instrument_comparison, instrument_switch): New function.
3212 (sancov_pass): Add trace-cmp support.
3213 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3214 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3215 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3216 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3217 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3218 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3219 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3220 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3221 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3222 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3224 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3226 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3227 error. Only quit immediately if parsing is complete.
3228 (BEGIN): Initialize fatal_err and parse_done.
3229 (begin fpu, end fpu): Check number of arguments.
3230 (begin arch, end arch): Likewise.
3231 (begin cpu, end cpu): Likewise.
3232 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3233 (optalias): Likewise.
3235 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3237 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3238 * config/arm/arm-isa.h: Delete. Move definitions to ...
3239 * arm-cpus.in: ... here. Use new feature and fgroup values.
3240 * config/arm/arm.c (arm_option_override): Use lower case for feature
3242 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3243 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3244 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3246 (print_isa_bits_for): New function.
3247 (gen_isa): New function.
3248 (gen_comm_data): Use print_isa_bits_for.
3249 (define feature): New keyword.
3250 (define fgroup): New keyword.
3251 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3252 (arm-isa.h): Add rule to generate file.
3253 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3254 case for feature bit names.
3256 2017-09-06 Richard Biener <rguenther@suse.de>
3258 * tree-ssa-pre.c (NECESSARY): Remove.
3259 (create_expression_by_pieces): Do not touch pass-local flags.
3260 (insert_into_preds_of_block): Likewise.
3261 (do_pre_regular_insertion): Likewise.
3262 (eliminate_insert): Likewise.
3263 (eliminate_dom_walker::before_dom_children): Likewise.
3264 (fini_eliminate): Do not look at inserted_exprs.
3265 (mark_operand_necessary): Remove.
3266 (remove_dead_inserted_code): Replace with simple work-list
3267 algorithm based on inserted_exprs and SSA uses.
3268 (pass_pre::execute): Re-order fini_eliminate and
3269 remove_dead_inserted_code.
3271 2017-09-06 Olivier Hainque <hainque@adacore.com>
3273 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3274 for VxWorks 7. Adjust surrounding comments.
3276 2017-09-06 Richard Biener <rguenther@suse.de>
3278 * gimple-ssa-strength-reduction.c
3279 (find_candidates_dom_walker::before_dom_children): Also allow
3282 2017-09-06 Richard Biener <rguenther@suse.de>
3284 PR tree-optimization/82108
3285 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3286 for gap in the non-permutation SLP case.
3288 2017-09-06 Martin Jambor <mjambor@suse.cz>
3290 PR tree-optimization/82078
3291 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3292 add_access_to_work_queue...
3293 (build_accesses_from_assign): ...here.
3294 (propagate_all_subaccesses): Make sure racc is the group
3295 representative, if there is one.
3297 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3300 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3303 2017-09-06 Richard Biener <rguenther@suse.de>
3305 * gimple-ssa-strength-reduction.c
3306 (find_candidates_dom_walker::before_doom_children): Use a
3307 type and not a mode check.
3309 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3312 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3314 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3316 2017-09-05 Jeff Law <law@redhat.com>
3318 PR tree-optimization/64910
3319 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3320 cases where we have 3 or more operands.
3322 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3325 * omp-low.c (lower_omp_for): Recompute tree invariant if
3326 gimple_omp_for_initial/final is ADDR_EXPR.
3329 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3330 into gimple val before gimplification fo the COND_EXPR.
3332 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3334 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3335 REGION_COPY argument.
3336 (thread_through_all_blocks): Remove unused argument to
3337 duplicate_thread_path.
3339 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3340 Alan Hayward <alan.hayward@arm.com>
3341 David Sherwood <david.sherwood@arm.com>
3343 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3344 Take a scalar_mode rather than a machine_mode.
3345 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3346 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3347 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3348 (aarch64_gen_adjusted_ldpstp): Likewise.
3349 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3351 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3352 Alan Hayward <alan.hayward@arm.com>
3353 David Sherwood <david.sherwood@arm.com>
3355 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3356 Take a scalar_int_mode instead of a machine_mode.
3357 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3358 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3359 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3360 (aarch64_simd_attr_length_rglist): Delete.
3361 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3362 a scalar_int_mode instead of a machine_mode.
3363 (aarch64_add_offset): Likewise.
3364 (aarch64_internal_mov_immediate): Likewise
3365 (aarch64_add_constant_internal): Likewise.
3366 (aarch64_add_constant): Likewise.
3367 (aarch64_movw_imm): Likewise.
3368 (aarch64_rtx_arith_op_extract_p): Likewise.
3369 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3370 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3371 Remove assert that the mode isn't a vector.
3372 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3373 (aarch64_expand_mov_immediate): Update calls after above changes.
3374 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3375 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3376 (aarch64_move_imm): Likewise.
3377 (aarch64_can_const_movi_rtx_p): Likewise.
3378 (aarch64_strip_extend): Likewise.
3379 (aarch64_extr_rtx_p): Likewise.
3380 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3381 a CONST_INT when the mode parameter is VOIDmode.
3382 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3384 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3386 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3387 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3388 (bitwise_type_for_mode): Update accordingly.
3390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3392 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3393 * stor-layout.c (mode_for_size_tree): Likewise.
3394 (mode_for_array): Update accordingly.
3395 (layout_decl): Likewise.
3396 (compute_record_mode): Likewise. Only set the mode once.
3398 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3400 * target.def (get_mask_mode): Change return type to opt_mode.
3402 * doc/tm.texi: Regenerate.
3403 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3404 * targhooks.c (default_get_mask_mode): Likewise.
3405 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3406 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3407 targetm.get_mask_mode.
3408 * tree.c (build_truth_vector_type): Likewise.
3410 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3412 * machmode.h (mode_for_vector): Return an opt_mode.
3413 * stor-layout.c (mode_for_vector): Likewise.
3414 (mode_for_int_vector): Update accordingly.
3415 (layout_type): Likewise.
3416 * config/i386/i386.c (emit_memmov): Likewise.
3417 (ix86_expand_set_or_movmem): Likewise.
3418 (ix86_expand_vector_init): Likewise.
3419 (ix86_get_mask_mode): Likewise.
3420 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3422 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3423 * expmed.c (extract_bit_field_1): Likewise.
3424 * expr.c (expand_expr_real_2): Likewise.
3425 * optabs-query.c (can_vec_perm_p): Likewise.
3426 (can_vec_mask_load_store_p): Likewise.
3427 * optabs.c (expand_vec_perm): Likewise.
3428 * targhooks.c (default_get_mask_mode): Likewise.
3429 * tree-vect-stmts.c (vectorizable_store): Likewise.
3430 (vectorizable_load): Likewise.
3431 (get_vectype_for_scalar_type_and_size): Likewise.
3433 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3435 * machmode.h (mode_for_int_vector): New function.
3436 * stor-layout.c (mode_for_int_vector): Likewise.
3437 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3438 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3439 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3440 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3441 (s390_expand_vcond): Likewise.
3443 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3445 * machmode.h (opt_machine_mode): New type.
3446 (opt_mode<T>): Allow construction from anything that can be
3448 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3449 (mode_for_size): Return an opt_machine_mode.
3450 * stor-layout.c (mode_for_size): Likewise.
3451 (mode_for_size_tree): Update call accordingly.
3452 (bitwise_mode_for_mode): Likewise.
3453 (make_fract_type): Likewise.
3454 (make_accum_type): Likewise.
3455 * caller-save.c (replace_reg_with_saved_mem): Update call
3457 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3458 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3459 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3460 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3461 * expmed.c (extract_bit_field_1): Likewise.
3462 * reload.c (get_secondary_mem): Likewise.
3463 * varasm.c (assemble_integer): Likewise.
3464 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3467 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3469 * machmode.h (decimal_float_mode_for_size): New function.
3470 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3471 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3472 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3473 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3474 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3475 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3477 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3479 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3480 (get_builtin_sync_mode): Likewise.
3481 (expand_ifn_atomic_compare_exchange): Likewise.
3482 (expand_builtin_atomic_clear): Likewise.
3483 (expand_builtin_atomic_test_and_set): Likewise.
3484 (fold_builtin_atomic_always_lock_free): Likewise.
3485 * calls.c (compute_argument_addresses): Likewise.
3486 (emit_library_call_value_1): Likewise.
3487 (store_one_arg): Likewise.
3488 * combine.c (combine_instructions): Likewise.
3489 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3490 * config/arm/arm.c (arm_function_value): Likewise.
3491 (aapcs_allocate_return_reg): Likewise.
3492 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3493 * config/i386/i386.c (construct_container): Likewise.
3494 (ix86_gimplify_va_arg): Likewise.
3495 (ix86_expand_sse_cmp): Likewise.
3496 (emit_memmov): Likewise.
3497 (emit_memset): Likewise.
3498 (expand_small_movmem_or_setmem): Likewise.
3499 (ix86_expand_pextr): Likewise.
3500 (ix86_expand_pinsr): Likewise.
3501 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3502 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3504 * config/mips/mips.c (mips_function_value_1) Likewise.
3505 (mips_block_move_straight): Likewise.
3506 (mips_expand_ins_as_unaligned_store): Likewise.
3507 * config/powerpcspe/powerpcspe.c
3508 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3509 (rs6000_darwin64_record_arg_flush): Likewise.
3510 * config/rs6000/rs6000.c
3511 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3512 (rs6000_darwin64_record_arg_flush): Likewise.
3513 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3514 (sparc_function_value_1): Likewise.
3515 * config/spu/spu.c (adjust_operand): Likewise.
3516 (spu_emit_branch_or_set): Likewise.
3517 (arith_immediate_p): Likewise.
3518 * emit-rtl.c (gen_lowpart_common): Likewise.
3519 * expr.c (expand_expr_real_1): Likewise.
3520 * function.c (assign_parm_setup_block): Likewise.
3521 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3522 * reload1.c (alter_reg): Likewise.
3523 * stor-layout.c (mode_for_vector): Likewise.
3524 (layout_type): Likewise.
3526 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3528 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3529 (spu_convert_move): Likewise.
3530 * lower-subreg.c (resolve_simple_move): Likewise.
3532 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3535 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3536 define_insn to a define_expand.
3537 (altivec_vsum2sws_direct): New define_insn.
3538 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3540 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3542 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3545 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3550 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3551 realign stack if stack alignment needed is less than incoming
3554 2017-09-05 Marek Polacek <polacek@redhat.com>
3557 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3560 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3562 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3564 2017-09-05 Richard Biener <rguenther@suse.de>
3566 PR tree-optimization/82084
3567 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3569 2017-09-05 Richard Biener <rguenther@suse.de>
3571 PR tree-optimization/82102
3572 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3574 2017-09-05 Martin Liska <mliska@suse.cz>
3576 PR tree-optimization/82032
3577 * tree-cfg.c (generate_range_test): New function.
3578 * tree-cfg.h (generate_range_test): Declared here.
3579 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3580 (cleanup_control_expr_graph): Use it.
3581 * tree-switch-conversion.c (try_switch_expansion): Remove
3583 (emit_case_nodes): Use generate_range_test.
3585 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3588 * config/i386/i386.md (*<btsc><mode>_mask): Add
3589 TARGET_USE_BT to insn constraint.
3590 (*btr<mode>_mask): Ditto.
3592 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3594 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3595 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3597 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3600 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3601 TARGET_NEON and TARGET_IWMMXT.
3602 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3603 TARGET_NEON and TARGET_IWMMXT.
3604 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3606 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3608 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3609 (ix86_rewrite_tls_address): Ditto.
3610 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3611 (ix86_rewrite_tls_address_1): Ditto.
3612 (ix86_rewrite_tls_address): Ditto.
3613 * config/i386/predicates.md (tls_address_pattern): New predicate.
3614 * config/i386/i386.md (TLS address splitter): New splitter.
3616 2017-09-04 Richard Biener <rguenther@suse.de>
3618 PR tree-optimization/82084
3619 * fold-const.h (can_native_encode_string_p): Declare.
3620 * fold-const.c (can_native_encode_string_p): Factor out from ...
3621 (native_encode_string): ... here.
3622 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3623 vectorizing stores from constants we later cannot handle.
3625 2017-09-04 Marek Polacek <polacek@redhat.com>
3628 * doc/invoke.texi: Update -Wtautological-compare documentation.
3630 2017-09-04 Jeff Law <law@redhat.com>
3632 PR tree-optimization/64910
3633 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3634 swap the first and last operand if the last is a constant.
3636 2017-09-04 Marek Polacek <polacek@redhat.com>
3639 * convert.c (do_narrow): When sanitizing signed integer overflows,
3640 bail out for signed types.
3641 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3643 2017-09-04 Richard Biener <rguenther@suse.de>
3645 PR tree-optimization/82060
3646 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3647 Move devirtualization after stmt folding and before EH/AB/noreturn
3648 cleanup to get the stmt refs canonicalized. Use a bool instead
3649 of gimple_modified_p since that doesn't work for NOPs. Schedule
3650 NOPs generated by folding for removal.
3652 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3653 Alan Hayward <alan.hayward@arm.com>
3654 David Sherwood <david.sherwood@arm.com>
3656 * coretypes.h (pad_direction): New enum.
3657 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3658 (FUNCTION_ARG_PADDING): Likewise.
3659 * target.def (function_arg_padding): New hook.
3660 * targhooks.h (default_function_arg_padding): Declare.
3661 * targhooks.c (default_function_arg_padding): New function.
3662 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3663 (TARGET_FUNCTION_ARG_PADDING): ...this.
3664 * doc/tm.texi: Regenerate.
3665 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3666 instead of direction.
3667 (compute_argument_addresses): Likewise.
3668 (load_register_parameters): Likewise.
3669 (emit_library_call_value_1): Likewise.
3670 (store_one_arg): Use targetm.calls.function_arg_padding instead
3671 of FUNCTION_ARG_PADDING.
3672 (must_pass_in_stack_var_size_or_pad): Likewise.
3673 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3674 (emit_group_store): Likewise.
3675 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3676 instead of FUNCTION_ARG_PADDING.
3677 (emit_push_insn): Likewise, and propagate enum change throughout
3679 * function.h (direction): Delete.
3680 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3682 * function.c (assign_parm_find_stack_rtl): Likewise.
3683 (assign_parm_setup_block_p): Likewise.
3684 (assign_parm_setup_block): Likewise.
3685 (gimplify_parameters): Likewise.
3686 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3687 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3689 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3690 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3691 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3692 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3693 (aarch64_function_arg_padding): ...this new function.
3694 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3695 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3696 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3697 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3698 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3699 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3700 (arm_pad_arg_upward): Replace with...
3701 (arm_function_arg_padding): ...this new function.
3702 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3704 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3705 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3706 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3707 (ia64_hpux_function_arg_padding): Replace with...
3708 (ia64_function_arg_padding): ...this new function. Use pad_direction
3709 instead of direction. Check for TARGET_HPUX.
3710 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3711 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3712 (iq2000_function_arg_padding): New function.
3713 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3714 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3715 (mips_function_arg_padding): ...this new function.
3716 (mips_pad_reg_upward): Update accordingly.
3717 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3718 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3719 targetm.calls.function_arg_padding.
3720 (FUNCTION_ARG_PADDING): Delete.
3721 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3722 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3723 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3724 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3725 (nios2_block_reg_padding): Return pad_direction instead of direction.
3726 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3727 instead of direction.
3728 (nios2_function_arg_padding): Likewise. Make static.
3729 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3730 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3731 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3732 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3733 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3734 (pa_function_arg_padding): Make static. Return pad_direction instead
3736 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3737 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3738 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3739 instead of direction. Use targetm.calls.function_arg_padding.
3740 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3741 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3742 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3743 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3744 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3746 (function_arg_padding): Rename to...
3747 (rs6000_function_arg_padding): ...this. Make static. Return
3748 pad_direction instead of direction.
3749 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3750 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3751 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3752 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3753 instead of direction. Use targetm.calls.function_arg_padding.
3754 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3755 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3756 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3757 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3758 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3759 (function_arg_padding): Rename to...
3760 (rs6000_function_arg_padding): ...this. Make static. Return
3761 pad_direction instead of direction.
3762 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3763 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3764 * config/s390/s390.c (s390_function_arg_padding): New function.
3765 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3766 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3767 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3768 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3769 (function_arg_padding): Rename to...
3770 (sparc_function_arg_padding): ...this. Make static. Return
3771 pad_direction instead of direction.
3772 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3773 * config/spu/spu.c (spu_function_arg_padding): New function.
3774 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3775 * system.h (FUNCTION_ARG_PADDING): Poison.
3777 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3778 Alan Hayward <alan.hayward@arm.com>
3779 David Sherwood <david.sherwood@arm.com>
3781 * target.def (modes_tieable_p): New hook.
3782 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3783 (TARGET_MODES_TIEABLE_P): ...this.
3784 * doc/tm.texi.in: Regenerate.
3785 * hooks.h (hook_bool_mode_mode_true): Declare.
3786 * hooks.c (hook_bool_mode_mode_true): New function.
3787 * combine.c (subst): Use targetm.modes_tieable_p instead of
3789 * dse.c (find_shift_sequence): Likewise.
3790 * expmed.c (extract_low_bits): Likewise.
3791 * lower-subreg.c: Include target.h.
3792 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3794 * rtlanal.c (rtx_cost): Likewise.
3795 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3796 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3797 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3798 (TARGET_MODES_TIEABLE_P): Redefine.
3799 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3800 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3801 (TARGET_MODES_TIEABLE_P): Redefine.
3802 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3803 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3804 (arc_modes_tieable_p): New function.
3805 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3806 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3807 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3808 (arm_modes_tieable_p): Make static.
3809 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3810 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3811 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3812 (TARGET_MODES_TIEABLE_P): Redefine.
3813 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3814 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3815 (TARGET_MODES_TIEABLE_P): Redefine.
3816 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3817 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3818 (cr16_modes_tieable_p): New function.
3819 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3820 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3821 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3822 (TRULY_NOOP_TRUNCATION): Update comment.
3823 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3824 (TRULY_NOOP_TRUNCATION): Update comment.
3825 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3826 (frv_modes_tieable_p): New function.
3827 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3828 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3829 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3830 (TARGET_MODES_TIEABLE_P): Redefine.
3831 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3832 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3833 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3834 (TARGET_MODES_TIEABLE_P): Redefine.
3835 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3836 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3837 (ia64_modes_tieable_p): New function.
3838 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3839 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3840 (iq2000_modes_tieable_p): New function.
3841 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3842 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3843 (lm32_modes_tieable_p): New function.
3844 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3845 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3846 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3847 (TARGET_MODES_TIEABLE_P): Redefine.
3848 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3849 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3850 (m32r_modes_tieable_p): New function.
3851 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3852 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3853 (m68k_modes_tieable_p): New function.
3854 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3855 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3856 (mcore_modes_tieable_p): New function.
3857 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3858 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3860 (TARGET_MODES_TIEABLE_P): Redefine.
3861 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3862 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3863 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3864 (TARGET_MODES_TIEABLE_P): Redefine.
3865 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3866 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3867 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3868 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3869 (mn10300_modes_tieable_p): ...this and make static.
3870 (TARGET_MODES_TIEABLE_P): Redefine.
3871 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3872 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3873 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3874 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3875 (msp430_modes_tieable_p): Make static.
3876 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3877 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3878 (TARGET_MODES_TIEABLE_P): Redefine.
3879 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3880 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3881 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3882 (TARGET_MODES_TIEABLE_P): Redefine.
3883 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3884 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3885 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3886 (TARGET_MODES_TIEABLE_P): Redefine.
3887 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3888 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3889 (pdp11_modes_tieable_p): New function.
3890 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3891 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3892 (rs6000_modes_tieable_p): New function.
3893 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3894 * config/powerpcspe/powerpcspe.md: Update comment.
3895 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3896 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3897 (TARGET_MODES_TIEABLE_P): Redefine.
3898 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3899 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3900 (rl78_modes_tieable_p): New function.
3901 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3902 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3903 (rs6000_modes_tieable_p): New function.
3904 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3905 * config/rs6000/rs6000.md: Update comment.
3906 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3907 * config/rx/rx.c (rx_modes_tieable_p): New function.
3908 (TARGET_MODES_TIEABLE_P): Redefine.
3909 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3910 * config/s390/s390.c (s390_modes_tieable_p): New function.
3911 (TARGET_MODES_TIEABLE_P): Redefine.
3912 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3913 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3914 (sh_modes_tieable_p): New function.
3915 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3916 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3917 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3918 (sparc_modes_tieable_p): Make static.
3919 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3920 * config/spu/spu.c (spu_modes_tieable_p): New function.
3921 (TARGET_MODES_TIEABLE_P): Redefine.
3922 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3923 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3924 (TARGET_MODES_TIEABLE_P): Redefine.
3925 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3926 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3927 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3928 * config/v850/v850.c (v850_modes_tieable_p): New function.
3929 (TARGET_MODES_TIEABLE_P): Redefine.
3930 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3931 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3932 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3933 (visium_modes_tieable_p): New function.
3934 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3935 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3936 (xtensa_modes_tieable_p): New function.
3937 * system.h (MODES_TIEABLE_P): Poison.
3939 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * target.def (hard_regno_mode_ok): New hook.
3944 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3945 (TARGET_HARD_REGNO_MODE_OK): ...this.
3946 * doc/tm.texi.in: Regenerate.
3947 * hooks.h (hook_bool_uint_mode_true): Declare.
3948 * hooks.c (hook_bool_uint_mode_true): New function.
3949 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3951 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3952 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3953 instead of HARD_REGNO_MODE_OK.
3954 * caller-save.c: Include target.h.
3955 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3957 * combine.c (can_combine_p): Likewise.
3958 (combinable_i3pat): Likewise.
3959 (can_change_dest_mode): Likewise.
3960 * expr.c (init_expr_target): Likewise.
3961 (convert_move): Likewise.
3962 (convert_modes): Likewise.
3963 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3964 (setup_prohibited_mode_move_regs): Likewise.
3965 * ira.h (target_ira): Likewise.
3966 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3967 * lra-constraints.c (process_alt_operands): Likewise.
3968 (split_reg): Likewise.
3969 * recog.c (peep2_find_free_register): Likewise.
3970 * ree.c (combine_reaching_defs): Likewise.
3971 * regcprop.c (maybe_mode_change): Likewise.
3972 * reginfo.c (init_reg_sets_1): Likewise.
3973 (choose_hard_reg_mode): Likewise.
3974 (simplifiable_subregs): Likewise.
3975 * regrename.c (check_new_reg_p): Likewise.
3976 * reload.c (find_valid_class): Likewise.
3977 (find_valid_class_1): Likewise.
3978 (reload_inner_reg_of_subreg): Likewise.
3979 (push_reload): Likewise.
3980 (combine_reloads): Likewise.
3981 (find_dummy_reload): Likewise.
3982 (find_reloads): Likewise.
3983 * reload1.c (find_reg): Likewise.
3984 (set_reload_reg): Likewise.
3985 (allocate_reload_reg): Likewise.
3986 (choose_reload_regs): Likewise.
3987 (reload_adjust_reg_for_temp): Likewise.
3988 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3989 (simplify_subreg_regno): Likewise.
3990 * sel-sched.c (init_regs_for_mode): Likewise.
3991 * varasm.c (make_decl_rtl): Likewise.
3992 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3993 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3995 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3996 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3997 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3998 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3999 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4001 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4002 (arc_mode_class): Delete.
4003 (HARD_REGNO_MODE_OK): Delete.
4004 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4005 (arc_hard_regno_mode_ok): Rename old array to...
4006 (arc_hard_regno_mode_ok_modes): ...this.
4007 (arc_conditional_register_usage): Update accordingly.
4008 (arc_mode_class): Make static.
4009 (arc_hard_regno_mode_ok): New function.
4010 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4011 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4012 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4013 (arm_hard_regno_mode_ok): Make static.
4014 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4016 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4017 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4018 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4021 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4022 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4023 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4024 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4025 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4026 * config/bfin/predicates.md (valid_reg_operand): Use
4027 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4028 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4029 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4031 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4032 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4033 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4034 (cr16_hard_regno_mode_ok): Make static and return a bool.
4035 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4036 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4037 (cris_hard_regno_mode_ok): New function.
4038 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4039 (epiphany_mode_class): Delete.
4040 (HARD_REGNO_MODE_OK): Delete.
4041 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4042 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4043 (hard_regno_mode_ok): Rename to...
4044 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4045 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4046 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4048 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4049 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4050 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4051 (frv_hard_regno_mode_ok): Make static and return a bool.
4052 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4054 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4055 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4056 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4057 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4060 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4061 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4062 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4065 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4066 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4067 (ia64_hard_regno_mode_ok): New function.
4068 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4069 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4070 (iq2000_hard_regno_mode_ok): New function.
4071 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4072 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4073 (lm32_hard_regno_mode_ok): New function.
4074 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4075 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4076 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4077 instead of HARD_REGNO_MODE_OK.
4078 (m32c_hard_regno_ok): Rename to...
4079 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4080 (m32c_cannot_change_mode_class): Update accordingly.
4081 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4082 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4083 (m32r_mode_class): Delete.
4084 (HARD_REGNO_MODE_OK): Delete.
4085 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4086 (m32r_hard_regno_mode_ok): Rename to...
4087 (m32r_hard_regno_modes): ...this.
4088 (m32r_mode_class): Make static.
4089 (m32r_hard_regno_mode_ok): New function.
4090 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4091 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4092 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4093 (m68k_hard_regno_mode_ok): Make static.
4094 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4095 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4096 (mcore_hard_regno_mode_ok): New function.
4097 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4098 (HARD_REGNO_MODE_OK): Delete.
4099 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4101 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4102 (microblaze_hard_regno_mode_ok): New function.
4103 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4104 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4105 (mips_hard_regno_mode_ok): Delete.
4106 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4107 (mips_hard_regno_mode_ok_p): ...this and make static.
4108 (mips_hard_regno_mode_ok_p): Rename to...
4109 (mips_hard_regno_mode_ok_uncached): ...this.
4110 (mips_hard_regno_mode_ok): New function.
4111 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4112 of HARD_REGNO_MODE_OK.
4113 (mips_option_override): Update after above name changes.
4114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4115 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4116 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4117 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4118 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4120 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4121 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4122 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4123 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4124 (msp430_hard_regno_mode_ok): Make static and return a bool.
4125 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4126 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4127 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4129 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4130 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4131 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4132 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4133 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4134 (PA_HARD_REGNO_MODE_OK): ...this
4135 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4136 (PA_HARD_REGNO_MODE_OK): ...this.
4137 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4138 (pa_hard_regno_mode_ok): New function.
4139 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4140 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4141 (pdp11_hard_regno_mode_ok): New function.
4142 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4143 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4145 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4147 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4148 (rs6000_hard_regno_mode_ok): Rename to...
4149 (rs6000_hard_regno_mode_ok_uncached): ...this.
4150 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4151 (rs6000_hard_regno_mode_ok): New function.
4152 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4153 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4154 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4155 (riscv_hard_regno_mode_ok): ...this and make static.
4156 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4157 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4158 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4159 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4160 (rl78_hard_regno_mode_ok): Make static and return bool.
4161 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4162 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4164 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4165 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4166 (rs6000_hard_regno_mode_ok): Rename to...
4167 (rs6000_hard_regno_mode_ok_uncached): ...this.
4168 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4169 (rs6000_hard_regno_mode_ok): New function.
4170 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4171 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4172 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4173 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4174 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4175 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4177 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4178 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4179 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4180 (sh_hard_regno_mode_ok): Make static.
4181 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4182 instead of HARD_REGNO_MODE_OK.
4183 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4184 (sparc_mode_class): Delete.
4185 (HARD_REGNO_MODE_OK): Delete.
4186 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4187 (hard_regno_mode_classes): Make static.
4188 (sparc_mode_class): Likewise.
4189 (sparc_hard_regno_mode_ok): New function.
4190 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4191 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4192 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4194 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4195 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4196 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4197 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4198 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4200 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4201 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4202 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4203 (visium_hard_regno_mode_ok): New function.
4204 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4205 instead of HARD_REGNO_MODE_OK.
4206 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4207 (HARD_REGNO_MODE_OK): Delete.
4208 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4209 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4210 (xtensa_option_override): Update accordingly.
4211 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4212 (xtensa_hard_regno_mode_ok): New function.
4213 * system.h (HARD_REGNO_MODE_OK): Poison.
4215 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4216 Alan Hayward <alan.hayward@arm.com>
4217 David Sherwood <david.sherwood@arm.com>
4219 * target.def (hard_regno_call_part_clobbered): New hook.
4220 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4221 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4222 * doc/tm.texi: Regenerate.
4223 * hooks.h (hook_bool_uint_mode_false): Declare.
4224 * hooks.c (hook_bool_uint_mode_false): New function.
4225 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4226 * cselib.c (cselib_process_insn): Use
4227 targetm.hard_regno_call_part_clobbered instead of
4228 HARD_REGNO_CALL_PART_CLOBBERED.
4229 * ira-conflicts.c (ira_build_conflicts): Likewise.
4230 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4231 * lra-constraints.c (need_for_call_save_p): Likewise.
4232 * lra-lives.c: Include target.h.
4233 (check_pseudos_live_through_calls): Use
4234 targetm.hard_regno_call_part_clobbered instead of
4235 HARD_REGNO_CALL_PART_CLOBBERED.
4236 * regcprop.c: Include target.h.
4237 (copyprop_hardreg_forward_1): Use
4238 targetm.hard_regno_call_part_clobbered instead of
4239 HARD_REGNO_CALL_PART_CLOBBERED.
4240 * reginfo.c (choose_hard_reg_mode): Likewise.
4241 * regrename.c (check_new_reg_p): Likewise.
4242 * reload.c (find_equiv_reg): Likewise.
4243 * reload1.c (emit_reload_insns): Likewise.
4244 * sched-deps.c (deps_analyze_insn): Likewise.
4245 * sel-sched.c (init_regs_for_mode): Likewise.
4246 (mark_unavailable_hard_regs): Likewise.
4247 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4248 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4249 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4251 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4252 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4253 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4255 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4257 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4258 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4259 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4261 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4262 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4263 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4265 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4266 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4268 * config/powerpcspe/powerpcspe.c
4269 (rs6000_hard_regno_call_part_clobbered): New function.
4270 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4271 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4272 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4274 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4275 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4276 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4278 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4279 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4280 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4282 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4283 Alan Hayward <alan.hayward@arm.com>
4284 David Sherwood <david.sherwood@arm.com>
4286 * rtl.h (subreg_memory_offset): Declare.
4287 * emit-rtl.c (subreg_memory_offset): New function.
4288 * expmed.c (store_bit_field_1): Use it.
4289 * expr.c (undefined_operand_subword_p): Likewise.
4290 * simplify-rtx.c (simplify_subreg): Likewise.
4292 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4294 PR rtl-optimization/57448
4297 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4298 using atomic_load pattern.
4299 (expand_atomic_store): Likewise.
4301 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4304 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4305 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4306 handling. Use replace_call_with_value with NULL instead of
4307 gsi_replace, unlink_stmt_vdef and release_defs.
4309 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4312 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4314 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4317 * rtl.h (emit_library_call_value_1): Declare.
4318 (emit_library_call): Replace declaration with a series of overloads.
4319 Remove the parameter count argument.
4320 (emit_library_call_value): Likewise.
4321 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4322 with an "rtx_mode_t *".
4323 (emit_library_call_value): Delete.
4324 (emit_library_call): Likewise.
4325 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4326 (asan_emit_allocas_unpoison): Likewise.
4327 * builtins.c (expand_builtin_powi): Likewise.
4328 (expand_asan_emit_allocas_unpoison): Likewise.
4329 * cfgexpand.c (expand_main_function): Likewise.
4330 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4331 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4332 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4333 * config/arm/arm.c (arm_trampoline_init): Likewise.
4334 (arm_call_tls_get_addr): Likewise.
4335 (arm_expand_divmod_libfunc): Likewise.
4336 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4337 (smulsi3_highpart): Likewise.
4338 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4339 (c6x_expand_compare): Likewise.
4340 (c6x_expand_movmem): Likewise.
4341 * config/frv/frv.c (frv_trampoline_init): Likewise.
4342 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4343 (ix86_expand_divmod_libfunc): Likewise.
4344 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4345 (ia64_expand_compare): Likewise.
4346 (ia64_profile_hook): Likewise.
4347 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4348 (nonlocal_goto): Likewise.
4349 (restore_stack_nonlocal): Likewise.
4350 * config/m32r/m32r.c (block_move_call): Likewise.
4351 (m32r_trampoline_init): Likewise.
4352 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4353 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4354 (m68k_call_m68k_read_tp): Likewise.
4355 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4356 (microblaze_expand_divide): Likewise.
4357 * config/mips/mips.h (mips_args): Likewise.
4358 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4359 (MIPS_ICACHE_SYNC): Likewise.
4360 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4361 (nios2_trampoline_init): Likewise.
4362 * config/pa/pa.c (hppa_tls_call): Likewise.
4363 (pa_trampoline_init): Likewise.
4364 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4365 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4366 (expand_strn_compare): Likewise.
4367 (rs6000_generate_compare): Likewise.
4368 (rs6000_expand_float128_convert): Likewise.
4369 (output_profile_hook): Likewise.
4370 (rs6000_trampoline_init): Likewise.
4371 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4372 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4373 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4374 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4375 (rs6000_generate_compare): Likewise.
4376 (rs6000_expand_float128_convert): Likewise.
4377 (output_profile_hook): Likewise.
4378 (rs6000_trampoline_init): Likewise.
4379 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4380 * config/sh/sh.c (sh_trampoline_init): Likewise.
4381 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4382 (sparc_emit_float_lib_cmp): Likewise.
4383 (sparc32_initialize_trampoline): Likewise.
4384 (sparc64_initialize_trampoline): Likewise.
4385 (sparc_profile_hook): Likewise.
4386 * config/spu/spu.c (ea_load_store): Likewise.
4387 * config/spu/spu.md (floatunssidf2): Likewise.
4388 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4389 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4390 * config/visium/visium.c (expand_block_move_4): Likewise.
4391 (expand_block_move_2): Likewise.
4392 (expand_block_move_1): Likewise.
4393 (expand_block_set_4): Likewise.
4394 (expand_block_set_2): Likewise.
4395 (expand_block_set_1): Likewise.
4396 (visium_trampoline_init): Likewise.
4397 (visium_profile_hook): Likewise.
4398 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4399 (xtensa_setup_frame_addresses): Likewise.
4400 (xtensa_trampoline_init): Likewise.
4401 * except.c (sjlj_emit_function_enter): Likewise.
4402 (sjlj_emit_function_exit): Likewise.
4403 * explow.c (allocate_dynamic_stack_space): Likewise.
4404 (probe_stack_range): Likewise.
4405 * expr.c (convert_mode_scalar): Likewise.
4406 * optabs.c (expand_binop): Likewise.
4407 (expand_twoval_binop_libfunc): Likewise.
4408 (expand_unop): Likewise.
4409 (prepare_cmp_insn): Likewise.
4410 (prepare_float_lib_cmp): Likewise.
4411 (expand_float): Likewise.
4412 (expand_fix): Likewise.
4413 (expand_fixed_convert): Likewise.
4414 (maybe_emit_sync_lock_test_and_set): Likewise.
4415 (expand_atomic_compare_and_swap): Likewise.
4416 (expand_mem_thread_fence): Likewise.
4417 (expand_atomic_fetch_op): Likewise.
4419 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4421 * doc/generic.texi (OpenACC): Adjust URL.
4422 * doc/invoke.texi (C Dialect Options): Ditto.
4424 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4426 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4427 predicate for operand 1. Add (m,<S>) constraint.
4428 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4429 Prevent memory operand 1 with register operand 2.
4431 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4433 PR rtl-optimization/82024
4434 * combine.c (try_combine): If the combination result is a PARALLEL,
4435 and we only need to retain the SET in there that would be placed
4436 at I2, check that we can place that at I3 instead, before doing so.
4438 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4441 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4443 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4444 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4447 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4448 Jeff Law <law@redhat.com>
4450 * varasm.c (bss_initializer_p): Do not put constants into .bss
4451 (categorize_decl_for_section): Handle bss_initializer_p returning
4452 false when DECL_INITIAL is NULL.
4454 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4457 * config/s390/s390.c (s390_can_inline_p): New function.
4459 2017-09-01 Jeff Law <law@redhat.com>
4461 PR tree-optimization/82052
4462 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4463 Always initialize the returned slot after a hash table miss
4464 when INSERT is true.
4466 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4468 * config/s390/s390.md (mem_signal_fence): Remove.
4469 * doc/md.texi (mem_signal_fence): Remove.
4470 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4472 * target-insns.def (mem_signal_fence): Remove.
4474 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4477 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4480 * asan.c (create_odr_indicator): Strip name encoding from assembler
4481 name before appending it after __odr_asan_.
4483 2017-09-01 Martin Liska <mliska@suse.cz>
4485 PR tree-optimization/82059
4486 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4487 frequency only when an edge is redirected.
4489 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4491 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4492 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4493 (arc_conditional_register_usage): Remove ARC600 lp_count
4495 (arc_file_start): Emit Tag_ARC_CPU_variation.
4496 (arc_can_use_doloop_p): New conditions to use ZOLs.
4497 (hwloop_fail): New function.
4498 (hwloop_optimize): Likewise.
4499 (hwloop_pattern_reg): Likewise.
4500 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4501 (arc_reorg_loops): New function, calls reorg_loops.
4502 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4503 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4505 (arc_loop_hazard): Remove function, functionality moved into
4507 (arc_hazard): Remove arc_loop_hazard call.
4508 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4509 into hwloop_optimize.
4510 (arc_label_align): Remove ZOL handling.
4511 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4512 * config/arc/arc.md (doloop_begin): Remove pattern.
4513 (doloop_begin_i): Likewise.
4514 (doloop_end_i): Likewise.
4515 (doloop_fallback): Likewise.
4516 (doloop_fallback_m): Likewise.
4517 (doloop_end): Reimplement expand.
4518 (arc_lp): New pattern for LP instruction.
4519 (loop_end): New pattern.
4520 (loop_fail): Likewise.
4521 (decrement_and_branch_until_zero): Likewise.
4522 * config/arc/arc.opt (mlpc-width): New option.
4523 * doc/invoke.texi (mlpc-width): Document option.
4525 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4527 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4528 (arc_ccfsm_advance): Fix checking for delay slots.
4529 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4531 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4533 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4535 (movhi_insn): Update store instruction constraint which are saving
4536 6-bit short immediates.
4537 (movsi_insn): Consider also short scaled load operations.
4538 (zero_extendhisi2_i): Use Usd constraint instead of T.
4539 (extendhisi2_i): Add q constraint.
4540 (arc_clzsi2): Add type and length attributes.
4541 (arc_ctzsi2): Likewise.
4542 * config/arc/constraints.md (Usc): Update constraint, the
4543 assembler can parse two relocations for a single instruction.
4545 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4547 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4548 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4550 2017-08-31 Olivier Hainque <hainque@adacore.com>
4552 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4553 match as powerpc-wrs-vxworks*.
4555 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4557 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4558 register constraint for by-element operand.
4559 (aarch64_mls_elt_merge<mode>): Likewise.
4561 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4563 * config/arc/arc.c (arc_can_follow_jump): Check for short
4566 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4568 * config.gcc: Use g.opt for arc.
4569 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4570 functionality moved to ...
4571 (legitimate_scaled_address_p): New function, ...here.
4572 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4573 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4574 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4576 (arc_override_options): Handle G option.
4577 (arc_output_pic_addr_const): Correct function definition.
4578 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4579 (arc_decl_anon_ns_mem_p): Delete.
4580 (arc_in_small_data_p): Overhaul this function to take into
4581 consideration the value given via G option.
4582 (arc_rewrite_small_data_1): Renamed and corrected old
4583 arc_rewrite_small_data function.
4584 (arc_rewrite_small_data): New function.
4585 (small_data_pattern): Don't use pic_offset_table_rtx.
4586 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4587 * config/arc/simdext.md (movmisalignv2hi): Use
4588 prepare_move_operands function.
4590 (movmisalign*): Likewise.
4591 * doc/invoke.texi (ARC options): Document -G option.
4593 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4595 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4597 * config/arc/arc.c (arc_print_operand): Output scalled address for
4598 sdata whenever is possible.
4599 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4600 load/stores are available.
4601 (compact_sda_memory_operand): Check for the alignment required by
4602 code density instructions.
4603 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4605 * config/arc/constraints.md (Usd): Update constraint.
4606 (Us0): New constraint.
4607 (Usc): Update constraint.
4609 2017-08-31 Richard Biener <rguenther@suse.de>
4612 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4615 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4617 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4618 Resize type_signature.
4620 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4621 Alan Hayward <alan.hayward@arm.com>
4622 David Sherwood <david.sherwood@arm.com>
4624 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4625 subregs whose inner modes can be stored in GPRs.
4626 (aarch64_classify_index): Likewise.
4628 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4629 Alan Hayward <alan.hayward@arm.com>
4630 David Sherwood <david.sherwood@arm.com>
4632 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4633 (V_INT_EQUIV): ...this.
4634 (v_cmp_result): Rename to...
4635 (v_int_equiv): ...this.
4636 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4637 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4638 (copysign<mode>3): Likewise.
4639 (aarch64_simd_bsl<mode>_internal): Likewise.
4640 (aarch64_simd_bsl<mode>): Likewise.
4641 (vec_cmp<mode><mode>): Likewise.
4642 (vcond<mode><mode>): Likewise.
4643 (vcond<v_cmp_mixed><mode>): Likewise.
4644 (vcondu<mode><v_cmp_mixed>): Likewise.
4645 (aarch64_cm<optab><mode>): Likewise.
4646 (aarch64_cmtst<mode>): Likewise.
4647 (aarch64_fac<optab><mode>): Likewise.
4648 (vec_perm_const<mode>): Likewise.
4649 (vcond_mask_<mode><v_cmp_result>): Rename to...
4650 (vcond_mask_<mode><v_int_equiv>): ...this.
4651 (vec_cmp<mode><v_cmp_result>): Rename to...
4652 (vec_cmp<mode><v_int_equiv>): ...this.
4654 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4655 Alan Hayward <alan.hayward@arm.com>
4656 David Sherwood <david.sherwood@arm.com>
4658 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4660 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4661 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4662 (UNSPEC_LD4_DREG): New unspecs.
4663 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4664 (aarch64_ld2<mode>_dreg_be): Replace with...
4665 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4667 (aarch64_ld3<mode>_dreg_le)
4668 (aarch64_ld3<mode>_dreg_be): Replace with...
4669 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4671 (aarch64_ld4<mode>_dreg_le)
4672 (aarch64_ld4<mode>_dreg_be): Replace with...
4673 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4676 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4678 PR tree-optimization/81987
4679 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4680 insert an initializer in a location not dominated by the stride
4683 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4685 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4686 on the entire header of the finally block in the fallthru case.
4688 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4690 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4692 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4694 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4695 rs6000_emit_move_from_cr and call renamed function.
4696 (rs6000_emit_prologue): Call renamed functions.
4697 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4698 movesi_from_cr, remove volatile CRs.
4700 2017-08-30 Jon Beniston <jon@beniston.com>
4701 Richard Biener <rguenther@suse.de>
4703 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4704 of VECTOR_MODE_P check.
4705 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4706 element vector types.
4708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4710 * df.h (df_read_modify_subreg_p): Remove in favor of...
4711 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4712 const_rtx instead of an rtx.
4713 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4714 * df-problems.c (df_word_lr_mark_ref): Likewise.
4715 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4716 (mark_pseudo_reg_dead): Likewise.
4717 (mark_ref_dead): Likewise.
4718 * reginfo.c (init_subregs_of_mode): Likewise.
4719 * sched-deps.c (sched_analyze_1): Likewise.
4720 * df-scan.c (df_def_record_1): Likewise.
4721 (df_uses_record): Likewise.
4722 (df_read_modify_subreg_p): Remove in favor of...
4723 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4724 const_rtx instead of an rtx.
4726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4727 Alan Hayward <alan.hayward@arm.com>
4728 David Sherwood <david.sherwood@arm.com>
4730 * rtl.h (partial_subreg_p): New function.
4731 * caller-save.c (save_call_clobbered_regs): Use it.
4732 * calls.c (expand_call): Likewise.
4733 * combine.c (combinable_i3pat): Likewise.
4734 (simplify_set): Likewise.
4735 (make_extraction): Likewise.
4736 (make_compound_operation_int): Likewise.
4737 (gen_lowpart_or_truncate): Likewise.
4738 (force_to_mode): Likewise.
4739 (make_field_assignment): Likewise.
4740 (reg_truncated_to_mode): Likewise.
4741 (record_truncated_value): Likewise.
4742 (move_deaths): Likewise.
4743 * cse.c (record_jump_cond): Likewise.
4744 (cse_insn): Likewise.
4745 * cselib.c (cselib_lookup_1): Likewise.
4746 * expmed.c (extract_bit_field_using_extv): Likewise.
4747 * function.c (assign_parm_setup_reg): Likewise.
4748 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4749 * ira-build.c (create_insn_allocnos): Likewise.
4750 * lra-coalesce.c (merge_pseudos): Likewise.
4751 * lra-constraints.c (match_reload): Likewise.
4752 (simplify_operand_subreg): Likewise.
4753 (curr_insn_transform): Likewise.
4754 * lra-lives.c (process_bb_lives): Likewise.
4755 * lra.c (new_insn_reg): Likewise.
4756 (lra_substitute_pseudo): Likewise.
4757 * regcprop.c (mode_change_ok): Likewise.
4758 (maybe_mode_change): Likewise.
4759 (copyprop_hardreg_forward_1): Likewise.
4760 * reload.c (push_reload): Likewise.
4761 (find_reloads): Likewise.
4762 (find_reloads_subreg_address): Likewise.
4763 * reload1.c (alter_reg): Likewise.
4764 (eliminate_regs_1): Likewise.
4765 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4767 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4769 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4770 back to if statements, including unpack.
4772 2017-08-30 Martin Liska <mliska@suse.cz>
4775 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4777 (func_checker::compare_asm_inputs_outputs): ... this function.
4778 (func_checker::compare_gimple_asm): Use the function to compare
4779 also ASM constrains.
4780 * ipa-icf-gimple.h: Rename the function.
4782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4783 Alan Hayward <alan.hayward@arm.com>
4784 David Sherwood <david.sherwood@arm.com>
4786 * coretypes.h (complex_mode): New type.
4787 * gdbhooks.py (build_pretty_printer): Handle it.
4788 * machmode.h (complex_mode): New class.
4789 (complex_mode::includes_p): New function.
4790 (is_complex_int_mode): Likewise.
4791 (is_complex_float_mode): Likewise.
4792 * genmodes.c (get_mode_class): Handle complex mode classes.
4793 * function.c (expand_function_end): Use is_complex_int_mode.
4795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4796 Alan Hayward <alan.hayward@arm.com>
4797 David Sherwood <david.sherwood@arm.com>
4799 * coretypes.h (scalar_mode_pod): New typedef.
4800 * gdbhooks.py (build_pretty_printer): Handle it.
4801 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4802 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4803 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4804 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
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 * machmode.h (mode_for_vector): Take a scalar_mode instead
4813 * stor-layout.c (mode_for_vector): Likewise.
4814 * explow.c (promote_mode): Use as_a <scalar_mode>.
4815 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4818 Alan Hayward <alan.hayward@arm.com>
4819 David Sherwood <david.sherwood@arm.com>
4821 * target.def (preferred_simd_mode): Take a scalar_mode
4822 instead of a machine_mode.
4823 * targhooks.h (default_preferred_simd_mode): Likewise.
4824 * targhooks.c (default_preferred_simd_mode): Likewise.
4825 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4826 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4827 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4828 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4829 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4830 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4831 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4832 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4834 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4835 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4836 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4837 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4838 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4839 * doc/tm.texi: Regenerate.
4840 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4844 Alan Hayward <alan.hayward@arm.com>
4845 David Sherwood <david.sherwood@arm.com>
4847 * target.def (scalar_mode_supported_p): Take a scalar_mode
4848 instead of a machine_mode.
4849 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4850 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4851 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4852 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4853 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4854 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4855 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4856 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4857 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4858 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4859 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4860 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4861 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4862 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4864 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4865 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4866 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4867 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4868 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4870 * doc/tm.texi: Regenerate.
4872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4873 Alan Hayward <alan.hayward@arm.com>
4874 David Sherwood <david.sherwood@arm.com>
4876 * coretypes.h (opt_scalar_mode): New typedef.
4877 * gdbhooks.py (build_pretty_printers): Handle it.
4878 * machmode.h (mode_iterator::get_2xwider): Add overload for
4880 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4882 * expr.c (convert_mode_scalar): Likewise.
4883 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4884 * optabs.c (expand_float): Likewise.
4885 (expand_fix): Likewise.
4886 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4889 Alan Hayward <alan.hayward@arm.com>
4890 David Sherwood <david.sherwood@arm.com>
4892 * optabs.c (expand_float): Explicitly check for scalars before
4893 using a branching expansion.
4894 (expand_fix): 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 * expr.c (convert_mode): Split scalar handling out into...
4901 (convert_mode_scalar): ...this new function. Treat the modes
4904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4905 Alan Hayward <alan.hayward@arm.com>
4906 David Sherwood <david.sherwood@arm.com>
4908 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4910 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4913 Alan Hayward <alan.hayward@arm.com>
4914 David Sherwood <david.sherwood@arm.com>
4916 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4917 rather than a machine_mode.
4918 (fixed_from_string): Likewise.
4919 (fixed_convert): Likewise.
4920 (fixed_convert_from_int): Likewise.
4921 (fixed_convert_from_real): Likewise.
4922 (real_convert_from_fixed): Likewise.
4923 * fixed-value.c (fixed_from_double_int): Likewise.
4924 (fixed_from_string): Likewise.
4925 (fixed_convert): Likewise.
4926 (fixed_convert_from_int): Likewise.
4927 (fixed_convert_from_real): Likewise.
4928 (real_convert_from_fixed): Likewise.
4929 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4932 Alan Hayward <alan.hayward@arm.com>
4933 David Sherwood <david.sherwood@arm.com>
4935 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4936 of separate mode class checks. Do not allow vector modes here.
4937 (immed_wide_int_const): Use as_a <scalar_mode>.
4938 * explow.c (trunc_int_for_mode): Likewise.
4939 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4940 (wi::shwi): Likewise.
4941 (wi::min_value): Likewise.
4942 (wi::max_value): Likewise.
4943 * dwarf2out.c (loc_descriptor): Likewise.
4944 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4948 Alan Hayward <alan.hayward@arm.com>
4949 David Sherwood <david.sherwood@arm.com>
4951 * tree.h (SCALAR_TYPE_MODE): New macro.
4952 * expr.c (expand_expr_addr_expr_1): Use it.
4953 (expand_expr_real_2): Likewise.
4954 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4955 (fold_convert_const_fixed_from_int): Likewise.
4956 (fold_convert_const_fixed_from_real): Likewise.
4957 (native_encode_fixed): Likewise
4958 (native_encode_complex): Likewise
4959 (native_encode_vector): Likewise.
4960 (native_interpret_fixed): Likewise.
4961 (native_interpret_real): Likewise.
4962 (native_interpret_complex): Likewise.
4963 (native_interpret_vector): Likewise.
4964 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4965 (simd_clone_adjust_argument_types): Likewise.
4966 (simd_clone_init_simd_arrays): Likewise.
4967 (simd_clone_adjust): Likewise.
4968 * stor-layout.c (layout_type): Likewise.
4969 * tree.c (build_minus_one_cst): Likewise.
4970 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4971 * tree-inline.c (estimate_move_cost): Likewise.
4972 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4973 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4974 (vectorizable_reduction): Likewise.
4975 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4976 (vect_recog_mixed_size_cond_pattern): Likewise.
4977 (check_bool_pattern): Likewise.
4978 (adjust_bool_pattern): Likewise.
4979 (search_type_for_mask_1): Likewise.
4980 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4981 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4982 (vectorizable_load): Likewise.
4983 (vectorizable_store): Likewise.
4984 * ubsan.c (ubsan_encode_value): Likewise.
4985 * varasm.c (output_constant): Likewise.
4987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4988 Alan Hayward <alan.hayward@arm.com>
4989 David Sherwood <david.sherwood@arm.com>
4991 * coretypes.h (scalar_mode): New class.
4992 * machmode.h (scalar_mode): Likewise.
4993 (scalar_mode::includes_p): New function.
4994 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4995 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4996 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4997 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4998 * expmed.c (store_bit_field_1): Likewise.
4999 (extract_bit_field_1): Likewise.
5000 * expr.c (write_complex_part): Likewise.
5001 (read_complex_part): Likewise.
5002 (emit_move_complex_push): Likewise.
5003 (expand_expr_real_2): Likewise.
5004 * function.c (assign_parm_setup_reg): Likewise.
5005 (assign_parms_unsplit_complex): Likewise.
5006 * optabs.c (expand_binop): Likewise.
5007 * rtlanal.c (subreg_get_info): Likewise.
5008 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5009 * varasm.c (output_constant_pool_2): Likewise.
5011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5012 Alan Hayward <alan.hayward@arm.com>
5013 David Sherwood <david.sherwood@arm.com>
5015 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5017 (expmed_mult_highpart_optab): Likewise.
5018 (expmed_mult_highpart): Likewise.
5020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5021 Alan Hayward <alan.hayward@arm.com>
5022 David Sherwood <david.sherwood@arm.com>
5024 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5025 instead of a machine_mode.
5026 (builtin_memset_read_str): Likewise.
5027 * builtins.c (c_readstr): Likewise.
5028 (builtin_memcpy_read_str): Likewise.
5029 (builtin_strncpy_read_str): Likewise.
5030 (builtin_memset_read_str): Likewise.
5031 (builtin_memset_gen_str): Likewise.
5032 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5033 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5034 instead of a machine_mode.
5035 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5037 (make_extraction): Likewise.
5038 (try_widen_shift_mode): Take and return scalar_int_modes instead
5040 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5041 a scalar_int_mode instead of a machine_mode.
5042 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5043 (avr_addr_space_pointer_mode): Likewise.
5044 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5045 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5046 (msp430_unwind_word_mode): Likewise.
5047 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5048 (spu_addr_space_pointer_mode): Likewise.
5049 (spu_addr_space_address_mode): Likewise.
5050 (spu_libgcc_cmp_return_mode): Likewise.
5051 (spu_libgcc_shift_count_mode): Likewise.
5052 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5053 (rl78_addr_space_pointer_mode): Likewise.
5054 (fl78_unwind_word_mode): Likewise.
5055 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5057 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5058 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5059 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5060 (mips_valid_pointer_mode): Likewise.
5061 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5062 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5063 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5065 (ft32_addr_space_address_mode): Likewise.
5066 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5067 scalar_int_mode instead of a machine_mode.
5068 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5070 (m32c_addr_space_address_mode): Likewise.
5071 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5072 (rs6000_eh_return_filter_mode): Likewise.
5073 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5074 (rs6000_eh_return_filter_mode): Likewise.
5075 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5076 (s390_libgcc_shift_count_mode): Likewise.
5077 (s390_unwind_word_mode): Likewise.
5078 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5080 * target.def (mode_rep_extended): Likewise.
5081 (valid_pointer_mode): Likewise.
5082 (addr_space.valid_pointer_mode): Likewise.
5083 (eh_return_filter_mode): Return a scalar_int_mode rather than
5085 (libgcc_cmp_return_mode): Likewise.
5086 (libgcc_shift_count_mode): Likewise.
5087 (unwind_word_mode): Likewise.
5088 (addr_space.pointer_mode): Likewise.
5089 (addr_space.address_mode): Likewise.
5090 * doc/tm.texi: Regenerate.
5091 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5093 (do_jump): Use scalar_int_mode for local variables.
5094 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5095 rather than a machine_mode.
5096 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5097 (scompare_loc_descriptor_wide): Likewise.
5098 (scompare_loc_descriptor_narrow): Likewise.
5099 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5101 * except.c (sjlj_emit_dispatch_table): Likewise.
5102 (expand_builtin_eh_copy_values): Likewise.
5103 * explow.c (convert_memory_address_addr_space_1): Likewise.
5104 Take a scalar_int_mode rather than a machine_mode.
5105 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5106 than a machine_mode.
5107 (memory_address_addr_space): Use scalar_int_mode for local variables.
5108 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5109 rather than a machine_mode.
5110 * expmed.c (mask_rtx): Likewise.
5111 (init_expmed_one_conv): Likewise.
5112 (expand_mult_highpart_adjust): Likewise.
5113 (extract_high_half): Likewise.
5114 (expmed_mult_highpart_optab): Likewise.
5115 (expmed_mult_highpart): Likewise.
5116 (expand_smod_pow2): Likewise.
5117 (expand_sdiv_pow2): Likewise.
5118 (emit_store_flag_int): Likewise.
5119 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5121 (extract_low_bits): Likewise.
5122 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5124 * expr.c (pieces_addr::adjust): Likewise.
5125 (can_store_by_pieces): Likewise.
5126 (store_by_pieces): Likewise.
5127 (clear_by_pieces_1): Likewise.
5128 (expand_expr_addr_expr_1): Likewise.
5129 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5130 (expand_expr_real_1): Likewise.
5131 (try_casesi): Likewise.
5132 * final.c (shorten_branches): Likewise.
5133 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5134 type of "mode" to machine_mode.
5135 * internal-fn.c (expand_arith_overflow_result_store): Take a
5136 scalar_int_mode rather than a machine_mode.
5137 (expand_mul_overflow): Use scalar_int_mode for local variables.
5138 * loop-doloop.c (doloop_modify): Likewise.
5139 (doloop_optimize): Likewise.
5140 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5141 than a machine_mode.
5142 (expand_doubleword_shift_condmove): Likewise.
5143 (expand_doubleword_shift): Likewise.
5144 (expand_doubleword_clz): Likewise.
5145 (expand_doubleword_popcount): Likewise.
5146 (expand_doubleword_parity): Likewise.
5147 (expand_absneg_bit): Use scalar_int_mode for local variables.
5148 (prepare_float_lib_cmp): Likewise.
5149 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5150 rather than a machine_mode.
5151 (convert_memory_address_addr_space): Likewise.
5152 (get_mode_bounds): Likewise.
5153 (get_address_mode): Return a scalar_int_mode rather than a
5155 * rtlanal.c (get_address_mode): Likewise.
5156 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5157 than a machine_mode.
5158 * targhooks.c (default_mode_rep_extended): Likewise.
5159 (default_valid_pointer_mode): Likewise.
5160 (default_addr_space_valid_pointer_mode): Likewise.
5161 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5162 than a machine_mode.
5163 (default_libgcc_cmp_return_mode): Likewise.
5164 (default_libgcc_shift_count_mode): Likewise.
5165 (default_unwind_word_mode): Likewise.
5166 (default_addr_space_pointer_mode): Likewise.
5167 (default_addr_space_address_mode): Likewise.
5168 * targhooks.h (default_eh_return_filter_mode): Likewise.
5169 (default_libgcc_cmp_return_mode): Likewise.
5170 (default_libgcc_shift_count_mode): Likewise.
5171 (default_unwind_word_mode): Likewise.
5172 (default_addr_space_pointer_mode): Likewise.
5173 (default_addr_space_address_mode): Likewise.
5174 (default_mode_rep_extended): Take a scalar_int_mode rather than
5176 (default_valid_pointer_mode): Likewise.
5177 (default_addr_space_valid_pointer_mode): Likewise.
5178 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5180 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5181 rather than a machine_mode.
5182 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5183 for local variables.
5184 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5185 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5186 than a machine_mode.
5188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5189 Alan Hayward <alan.hayward@arm.com>
5190 David Sherwood <david.sherwood@arm.com>
5192 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5193 the mode argument to scalar_int_mode.
5194 (do_jump_by_parts_zero_rtx): Likewise.
5195 (do_jump_by_parts_equality_rtx): Likewise.
5196 (do_jump_by_parts_greater): Take a mode argument.
5197 (do_jump_by_parts_equality): Likewise.
5198 (do_jump_1): Update calls accordingly.
5200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5201 Alan Hayward <alan.hayward@arm.com>
5202 David Sherwood <david.sherwood@arm.com>
5204 * is-a.h (safe_dyn_cast): New function.
5205 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5206 (jump_table_for_label): Likewise.
5207 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5208 instead of an rtx_insn *.
5209 (shorten_branches): Use dyn_cast instead of LABEL_P and
5210 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5211 rtx_jump_table_data::get_data_mode.
5212 (final_scan_insn): Likewise.
5214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5215 Alan Hayward <alan.hayward@arm.com>
5216 David Sherwood <david.sherwood@arm.com>
5218 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5219 trying to combine a full-register integer set with a subreg
5222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5223 Alan Hayward <alan.hayward@arm.com>
5224 David Sherwood <david.sherwood@arm.com>
5226 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5227 that is always either address_mode or pointer_mode.
5229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5230 Alan Hayward <alan.hayward@arm.com>
5231 David Sherwood <david.sherwood@arm.com>
5233 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5234 when the two are known to be equal.
5236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5237 Alan Hayward <alan.hayward@arm.com>
5238 David Sherwood <david.sherwood@arm.com>
5240 * simplify-rtx.c (simplify_const_unary_operation): Use
5241 is_a <scalar_int_mode> instead of checking for a nonzero
5242 precision. Forcibly convert op_mode to a scalar_int_mode
5243 in that case. More clearly differentiate the operand and
5244 result modes and use the former when deciding what the value
5245 of a count-bits operation should be. Use is_int_mode instead
5246 of checking for a MODE_INT. Remove redundant check for whether
5247 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
5254 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5255 (widen_bswap): Likewise.
5256 (expand_parity): Likewise.
5257 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5258 (expand_ffs): Likewise.
5259 (epand_unop): Check for scalar integer modes before calling the
5262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5263 Alan Hayward <alan.hayward@arm.com>
5264 David Sherwood <david.sherwood@arm.com>
5266 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5268 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5275 value_mode arguments. Use scalar_int_mode internally.
5276 (store_bit_field_1): Rename the new integer mode from imode
5277 to op0_mode and use it instead of GET_MODE (op0). Update calls
5278 to store_split_bit_field, store_bit_field_using_insv and
5279 store_fixed_bit_field.
5280 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5281 Use scalar_int_mode internally. Use a bit count rather than a mode
5282 when calculating the largest bit size for get_best_mode.
5283 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5284 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5285 Remove assertion that OP0 has a scalar integer mode.
5286 (store_split_bit_field): Add op0_mode and value_mode arguments.
5287 Update calls to extract_fixed_bit_field.
5288 (extract_bit_field_using_extv): Add an op0_mode argument.
5289 Use scalar_int_mode internally.
5290 (extract_bit_field_1): Rename the new integer mode from imode to
5291 op0_mode and use it instead of GET_MODE (op0). Update calls to
5292 extract_split_bit_field, extract_bit_field_using_extv and
5293 extract_fixed_bit_field.
5294 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5295 to extract_split_bit_field and extract_fixed_bit_field_1.
5296 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5297 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5299 (extract_split_bit_field): Add an op0_mode argument. Update call
5300 to extract_fixed_bit_field.
5302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5303 Alan Hayward <alan.hayward@arm.com>
5304 David Sherwood <david.sherwood@arm.com>
5306 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5307 * explow.c (hard_function_value): Likewise.
5308 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5309 convert_to_mode call outside the loop.
5310 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5311 for the mode iterator. Require the mode specified by max_pieces
5313 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5315 (copy_blkmode_to_reg): Likewise.
5316 (set_storage_via_setmem): Likewise.
5317 * optabs.c (prepare_cmp_insn): Likewise.
5318 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5319 * stor-layout.c (finish_bitfield_representative): Likewise.
5321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5322 Alan Hayward <alan.hayward@arm.com>
5323 David Sherwood <david.sherwood@arm.com>
5325 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5326 * expr.c (convert_move): Use them.
5327 (convert_modes): Likewise.
5328 (store_expr_with_bounds): Likewise.
5330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5331 Alan Hayward <alan.hayward@arm.com>
5332 David Sherwood <david.sherwood@arm.com>
5334 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5335 parameter for the mode of "x". Remove the "known_x", "known_mode"
5336 and "known_ret" arguments. Change the type of the mode argument
5338 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5339 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5340 (reg_num_sign_bit_copies_for_combine): Likewise.
5341 * rtlanal.c (nonzero_bits1): Likewise.
5342 (num_sign_bit_copies1): Likewise.
5343 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5344 (reg_num_sign_bit_copies_general): Likewise.
5345 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5346 (reg_nonzero_bits_general): Likewise.
5348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5349 Alan Hayward <alan.hayward@arm.com>
5350 David Sherwood <david.sherwood@arm.com>
5352 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5353 than in subroutines. Return 1 for non-integer modes.
5354 (cached_num_sign_bit_copies): Change the type of the mode parameter
5356 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5357 classes. Handle CONST_INT_P first and then check whether X also
5358 has a scalar integer mode. Check the same thing for inner registers
5359 of a SUBREG and for values that are being extended or truncated.
5361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5362 Alan Hayward <alan.hayward@arm.com>
5363 David Sherwood <david.sherwood@arm.com>
5365 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5366 in subroutines. Return the mode mask for non-integer modes.
5367 (cached_nonzero_bits): Change the type of the mode parameter
5369 (nonzero_bits1): Likewise. Remove early exit for other mode
5370 classes. Handle CONST_INT_P first and then check whether X
5371 also has a scalar integer mode.
5373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5374 Alan Hayward <alan.hayward@arm.com>
5375 David Sherwood <david.sherwood@arm.com>
5377 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5378 Return a scalar_int_mode and assert that the size is greater than
5380 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5381 (op_by_pieces_d::op_by_pieces_d): Likewise.
5382 (op_by_pieces_d::run): Likewise.
5383 (can_store_by_pieces): Likewise.
5385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5386 Alan Hayward <alan.hayward@arm.com>
5387 David Sherwood <david.sherwood@arm.com>
5389 * combine.c (extract_left_shift): Add a mode argument and update
5391 (make_compound_operation_int): Change the type of the mode parameter
5392 to scalar_int_mode and update the call to extract_left_shift.
5394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5395 Alan Hayward <alan.hayward@arm.com>
5396 David Sherwood <david.sherwood@arm.com>
5398 * combine.c (simplify_and_const_int): Change the type of the mode
5399 parameter to scalar_int_mode.
5400 (simplify_and_const_int_1): Likewise. Update recursive call.
5402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5403 Alan Hayward <alan.hayward@arm.com>
5404 David Sherwood <david.sherwood@arm.com>
5406 * combine.c (simplify_compare_const): Check that the mode is a
5407 scalar_int_mode (rather than VOIDmode) before testing its
5409 (simplify_comparison): Move COMPARISON_P handling out of the
5410 loop and restrict the latter part of the loop to scalar_int_modes.
5411 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5412 and when considering SUBREG_REGs. Use is_int_mode instead of
5413 checking GET_MODE_CLASS against MODE_INT.
5415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5416 Alan Hayward <alan.hayward@arm.com>
5417 David Sherwood <david.sherwood@arm.com>
5419 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5420 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5421 shift_unit_mode and for modes involved in scalar shifts.
5423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5424 Alan Hayward <alan.hayward@arm.com>
5425 David Sherwood <david.sherwood@arm.com>
5427 * combine.c (force_int_to_mode): New function, split out from...
5428 (force_to_mode): ...here. Keep xmode up-to-date and use it
5429 instead of GET_MODE (x).
5431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5432 Alan Hayward <alan.hayward@arm.com>
5433 David Sherwood <david.sherwood@arm.com>
5435 * optabs-query.h (extraction_insn::struct_mode): Change type to
5436 opt_scalar_int_mode and update comment.
5437 (extraction_insn::field_mode): Change type to scalar_int_mode.
5438 (extraction_insn::pos_mode): Likewise.
5439 * combine.c (make_extraction): Update accordingly.
5440 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5441 (get_optab_extraction_insn): Likewise.
5442 * recog.c (simplify_while_replacing): Likewise.
5443 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5444 parameter to opt_scalar_int_mode.
5446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5447 Alan Hayward <alan.hayward@arm.com>
5448 David Sherwood <david.sherwood@arm.com>
5450 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5451 to a scalar_int_mode instead of a machine_mode.
5452 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5453 (get_best_mode): Return a boolean and use a pointer argument to store
5454 the selected mode. Replace the limit mode parameter with a bit limit.
5455 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5456 for the values returned by bit_field_mode_iterator::next_mode.
5457 (store_bit_field): Update call to get_best_mode.
5458 (store_fixed_bit_field): Likewise.
5459 (extract_fixed_bit_field): Likewise.
5460 * expr.c (optimize_bitfield_assignment_op): Likewise.
5461 * fold-const.c (optimize_bit_field_compare): Likewise.
5462 (fold_truth_andor_1): Likewise.
5463 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5464 Update for new type of m_mode.
5465 (get_best_mode): As above.
5467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5468 Alan Hayward <alan.hayward@arm.com>
5469 David Sherwood <david.sherwood@arm.com>
5471 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5472 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5473 (store_bit_field): Check is_a <scalar_int_mode> before calling
5474 strict_volatile_bitfield_p.
5475 (extract_bit_field): Likewise.
5477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5478 Alan Hayward <alan.hayward@arm.com>
5479 David Sherwood <david.sherwood@arm.com>
5481 * target.def (cstore_mode): Return a scalar_int_mode.
5482 * doc/tm.texi: Regenerate.
5483 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5484 * targhooks.h (default_cstore_mode): Likewise.
5485 * targhooks.c (default_cstore_mode): Likewise, using a forced
5487 * expmed.c (emit_cstore): Expect the target of the cstore to be
5490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5491 Alan Hayward <alan.hayward@arm.com>
5492 David Sherwood <david.sherwood@arm.com>
5494 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5496 (niter_desc): Likewise mode.
5497 (iv_analyze): Add a mode parameter.
5499 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5500 and change its type to scalar_int_mode.
5501 * loop-iv.c: Update commentary at head of file.
5502 (iv_constant): Pass the mode paraeter before the rtx it describes
5503 and change its type to scalar_int_mode. Remove VOIDmode handling.
5504 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5505 (iv_extend): Likewise.
5506 (shorten_into_mode): Likewise.
5507 (iv_add): Use scalar_int_mode.
5508 (iv_mult): Likewise.
5509 (iv_shift): Likewise.
5510 (canonicalize_iv_subregs): Likewise.
5511 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5512 it describes and change its mode to scalar_int_mode. Also change
5513 the type of the returned inner_mode to scalar_int_mode.
5514 (get_biv_step): Likewise, turning outer_mode from a pointer
5515 into a direct parameter. Update call to get_biv_step_1.
5516 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5517 iv_constant and get_biv_step.
5518 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5519 and change its type to scalar_int_mode. Don't initialise iv->mode
5520 to VOIDmode and remove later checks for its still being VOIDmode.
5521 Update calls to iv_analyze_op and iv_analyze_expr. Check
5522 is_a <scalar_int_mode> when changing the mode under consideration.
5523 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5524 Update call to iv_analyze_expr.
5525 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5526 inner register is not also a scalar_int_mode. Update call to
5528 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5529 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5530 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5531 separate mode class checks. Update calls to iv_analyze. Remove
5532 fix-up of VOIDmodes after iv_analyze_biv.
5533 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5534 don't have a scalar_int_mode. Update call to biv_p.
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 * cfgexpand.c (convert_debug_memory_address): Use
5541 as_a <scalar_int_mode>.
5542 * combine.c (expand_compound_operation): Likewise.
5543 (make_extraction): Likewise.
5544 (change_zero_ext): Likewise.
5545 (simplify_comparison): Likewise.
5546 * cse.c (cse_insn): Likewise.
5547 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5548 (mem_loc_descriptor): Likewise.
5549 (loc_descriptor): Likewise.
5550 * expmed.c (init_expmed_one_mode): Likewise.
5551 (synth_mult): Likewise.
5552 (emit_store_flag_1): Likewise.
5553 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5554 of a comparison with size.
5555 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5556 (reduce_to_bit_field_precision): Likewise.
5557 * function.c (expand_function_end): Likewise.
5558 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5559 * loop-doloop.c (doloop_modify): Likewise.
5560 * optabs.c (expand_binop): Likewise.
5561 (expand_unop): Likewise.
5562 (expand_copysign_absneg): Likewise.
5563 (prepare_cmp_insn): Likewise.
5564 (maybe_legitimize_operand): Likewise.
5565 * recog.c (const_scalar_int_operand): Likewise.
5566 * rtlanal.c (get_address_mode): Likewise.
5567 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5568 (simplify_cond_clz_ctz): Likewise.
5569 * tree-nested.c (get_nl_goto_field): Likewise.
5570 * tree.c (build_vector_type_for_mode): Likewise.
5571 * var-tracking.c (use_narrower_mode): Likewise.
5573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5574 Alan Hayward <alan.hayward@arm.com>
5575 David Sherwood <david.sherwood@arm.com>
5577 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5578 * builtins.c (expand_builtin_signbit): Use it.
5579 * cfgexpand.c (expand_debug_expr): Likewise.
5580 * dojump.c (do_jump): Likewise.
5581 (do_compare_and_jump): Likewise.
5582 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5583 * expmed.c (make_tree): Likewise.
5584 * expr.c (expand_expr_real_2): Likewise.
5585 (expand_expr_real_1): Likewise.
5586 (try_casesi): Likewise.
5587 * fold-const-call.c (fold_const_call_ss): Likewise.
5588 * fold-const.c (unextend): Likewise.
5589 (extract_muldiv_1): Likewise.
5590 (fold_single_bit_test): Likewise.
5591 (native_encode_int): Likewise.
5592 (native_encode_string): Likewise.
5593 (native_interpret_int): Likewise.
5594 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5595 * internal-fn.c (expand_addsub_overflow): Likewise.
5596 (expand_neg_overflow): Likewise.
5597 (expand_mul_overflow): Likewise.
5598 (expand_arith_overflow): Likewise.
5599 * match.pd: Likewise.
5600 * stor-layout.c (layout_type): Likewise.
5601 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5602 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5603 * tree-ssanames.c (get_range_info): Likewise.
5604 * tree-switch-conversion.c (array_value_type) Likewise.
5605 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5606 (vect_recog_divmod_pattern): Likewise.
5607 (vect_recog_mixed_size_cond_pattern): Likewise.
5608 * tree-vrp.c (extract_range_basic): Likewise.
5609 (simplify_float_conversion_using_ranges): Likewise.
5610 * tree.c (int_fits_type_p): Likewise.
5611 * ubsan.c (instrument_bool_enum_load): Likewise.
5612 * varasm.c (mergeable_string_section): Likewise.
5613 (narrowing_initializer_constant_valid_p): Likewise.
5614 (output_constant): Likewise.
5616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5617 Alan Hayward <alan.hayward@arm.com>
5618 David Sherwood <david.sherwood@arm.com>
5620 * machmode.h (NARROWEST_INT_MODE): New macro.
5621 * expr.c (alignment_for_piecewise_move): Use it instead of
5622 GET_CLASS_NARROWEST_MODE (MODE_INT).
5623 (push_block): Likewise.
5624 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5626 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5629 Alan Hayward <alan.hayward@arm.com>
5630 David Sherwood <david.sherwood@arm.com>
5632 * postreload.c (move2add_valid_value_p): Change the type of the
5633 mode parameter to scalar_int_mode.
5634 (move2add_use_add2_insn): Add a mode parameter and use it instead
5636 (move2add_use_add3_insn): Likewise.
5637 (reload_cse_move2add): Update accordingly.
5639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5640 Alan Hayward <alan.hayward@arm.com>
5641 David Sherwood <david.sherwood@arm.com>
5643 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5645 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5646 * optabs.c (expand_unop): Likewise.
5648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5649 Alan Hayward <alan.hayward@arm.com>
5650 David Sherwood <david.sherwood@arm.com>
5652 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5653 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5654 (popcount_loc_descriptor): Likewise.
5655 (bswap_loc_descriptor): Likewise.
5656 (rotate_loc_descriptor): Likewise.
5657 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5658 calling the functions above.
5660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5661 Alan Hayward <alan.hayward@arm.com>
5662 David Sherwood <david.sherwood@arm.com>
5664 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5666 (try_combine): Likewise.
5667 (simplify_if_then_else): Likewise.
5668 * cse.c (cse_insn): Likewise.
5669 * dwarf2out.c (mem_loc_descriptor): Likewise.
5670 * emit-rtl.c (gen_lowpart_common): Likewise.
5671 * simplify-rtx.c (simplify_truncation): Likewise.
5672 (simplify_binary_operation_1): Likewise.
5673 (simplify_const_relational_operation): Likewise.
5674 (simplify_ternary_operation): Likewise.
5675 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5678 Alan Hayward <alan.hayward@arm.com>
5679 David Sherwood <david.sherwood@arm.com>
5681 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5682 * reload.c (push_reload): Likewise.
5683 (find_reloads): Likewise.
5685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5686 Alan Hayward <alan.hayward@arm.com>
5687 David Sherwood <david.sherwood@arm.com>
5689 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5690 (make_compound_operation_int): Likewise.
5691 (change_zero_ext): Likewise.
5692 * expr.c (convert_move): Likewise.
5693 (convert_modes): Likewise.
5694 * fwprop.c (forward_propagate_subreg): Likewise.
5695 * loop-iv.c (get_biv_step_1): Likewise.
5696 * optabs.c (widen_operand): Likewise.
5697 * postreload.c (move2add_valid_value_p): Likewise.
5698 * recog.c (simplify_while_replacing): Likewise.
5699 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5700 (simplify_binary_operation_1): Likewise. Remove redundant
5701 mode equality check.
5703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5704 Alan Hayward <alan.hayward@arm.com>
5705 David Sherwood <david.sherwood@arm.com>
5707 * combine.c (combine_simplify_rtx): Add checks for
5708 is_a <scalar_int_mode>.
5709 (simplify_if_then_else): Likewise.
5710 (make_field_assignment): Likewise.
5711 (simplify_comparison): Likewise.
5712 * ifcvt.c (noce_try_bitop): Likewise.
5713 * loop-invariant.c (canonicalize_address_mult): Likewise.
5714 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5717 Alan Hayward <alan.hayward@arm.com>
5718 David Sherwood <david.sherwood@arm.com>
5720 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5721 is_a <scalar_int_mode> instead of != BLKmode.
5723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5724 Alan Hayward <alan.hayward@arm.com>
5725 David Sherwood <david.sherwood@arm.com>
5727 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5728 instead of != VOIDmode.
5729 * combine.c (if_then_else_cond): Likewise.
5730 (change_zero_ext): Likewise.
5731 * dwarf2out.c (mem_loc_descriptor): Likewise.
5732 (loc_descriptor): Likewise.
5733 * rtlanal.c (canonicalize_condition): Likewise.
5734 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5737 Alan Hayward <alan.hayward@arm.com>
5738 David Sherwood <david.sherwood@arm.com>
5740 * simplify-rtx.c (simplify_binary_operation_1): Use
5741 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5744 Alan Hayward <alan.hayward@arm.com>
5745 David Sherwood <david.sherwood@arm.com>
5747 * wide-int.h (int_traits<unsigned char>) New class.
5748 (int_traits<unsigned short>) Likewise.
5749 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5750 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5752 * combine.c (set_nonzero_bits_and_sign_copies): Use
5753 is_a <scalar_int_mode>.
5754 (find_split_point): Likewise.
5755 (combine_simplify_rtx): Likewise.
5756 (simplify_logical): Likewise.
5757 (expand_compound_operation): Likewise.
5758 (expand_field_assignment): Likewise.
5759 (make_compound_operation): Likewise.
5760 (extended_count): Likewise.
5761 (change_zero_ext): Likewise.
5762 (simplify_comparison): Likewise.
5763 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5764 (ucompare_loc_descriptor): Likewise.
5765 (minmax_loc_descriptor): Likewise.
5766 (mem_loc_descriptor): Likewise.
5767 (loc_descriptor): Likewise.
5768 * expmed.c (init_expmed_one_mode): Likewise.
5769 * lra-constraints.c (lra_constraint_offset): Likewise.
5770 * optabs.c (prepare_libcall_arg): Likewise.
5771 * postreload.c (move2add_note_store): Likewise.
5772 * reload.c (operands_match_p): Likewise.
5773 * rtl.h (load_extend_op): Likewise.
5774 * rtlhooks.c (gen_lowpart_general): Likewise.
5775 * simplify-rtx.c (simplify_truncation): Likewise.
5776 (simplify_unary_operation_1): Likewise.
5777 (simplify_binary_operation_1): Likewise.
5778 (simplify_const_binary_operation): Likewise.
5779 (simplify_const_relational_operation): Likewise.
5780 (simplify_subreg): Likewise.
5781 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5782 * var-tracking.c (adjust_mems): Likewise.
5783 (prepare_call_arguments): Likewise.
5785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5786 Alan Hayward <alan.hayward@arm.com>
5787 David Sherwood <david.sherwood@arm.com>
5789 * machmode.h (is_int_mode): New fuction.
5790 * combine.c (find_split_point): Use it.
5791 (combine_simplify_rtx): Likewise.
5792 (simplify_if_then_else): Likewise.
5793 (simplify_set): Likewise.
5794 (simplify_shift_const_1): Likewise.
5795 (simplify_comparison): Likewise.
5796 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5797 * cse.c (notreg_cost): Likewise.
5798 (cse_insn): Likewise.
5799 * cselib.c (cselib_lookup_1): Likewise.
5800 * dojump.c (do_jump_1): Likewise.
5801 (do_compare_rtx_and_jump): Likewise.
5802 * dse.c (get_call_args): Likewise.
5803 * dwarf2out.c (rtl_for_decl_init): Likewise.
5804 (native_encode_initializer): Likewise.
5805 * expmed.c (emit_store_flag_1): Likewise.
5806 (emit_store_flag): Likewise.
5807 * expr.c (convert_modes): Likewise.
5808 (store_field): Likewise.
5809 (expand_expr_real_1): Likewise.
5810 * fold-const.c (fold_read_from_constant_string): Likewise.
5811 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5812 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5813 * optabs.c (expand_binop): Likewise.
5814 (expand_unop): Likewise.
5815 (expand_abs_nojump): Likewise.
5816 (expand_one_cmpl_abs_nojump): Likewise.
5817 * simplify-rtx.c (mode_signbit_p): Likewise.
5818 (val_signbit_p): Likewise.
5819 (val_signbit_known_set_p): Likewise.
5820 (val_signbit_known_clear_p): Likewise.
5821 (simplify_relational_operation_1): Likewise.
5822 * tree.c (vector_type_mode): Likewise.
5824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5825 Alan Hayward <alan.hayward@arm.com>
5826 David Sherwood <david.sherwood@arm.com>
5828 * machmode.h (smallest_mode_for_size): Fix formatting.
5829 (smallest_int_mode_for_size): New function.
5830 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5831 instead of smallest_mode_for_size.
5832 * combine.c (make_extraction): Likewise.
5833 * config/arc/arc.c (arc_expand_movmem): Likewise.
5834 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5835 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5836 * config/s390/s390.c (s390_expand_insv): Likewise.
5837 * config/sparc/sparc.c (assign_int_registers): Likewise.
5838 * config/spu/spu.c (spu_function_value): Likewise.
5839 (spu_function_arg): Likewise.
5840 * coverage.c (get_gcov_type): Likewise.
5841 (get_gcov_unsigned_t): Likewise.
5842 * dse.c (find_shift_sequence): Likewise.
5843 * expmed.c (store_bit_field_1): Likewise.
5844 * expr.c (convert_move): Likewise.
5845 (store_field): Likewise.
5846 * internal-fn.c (expand_arith_overflow): Likewise.
5847 * optabs-query.c (get_best_extraction_insn): Likewise.
5848 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5849 * stor-layout.c (layout_type): Likewise.
5850 (initialize_sizetypes): Likewise.
5851 * targhooks.c (default_get_mask_mode): Likewise.
5852 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5855 Alan Hayward <alan.hayward@arm.com>
5856 David Sherwood <david.sherwood@arm.com>
5858 * machmode.h (opt_mode::else_blk): New function.
5859 (int_mode_for_mode): Declare.
5860 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5861 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5863 * cfgexpand.c (expand_debug_expr): Likewise.
5864 * combine.c (gen_lowpart_or_truncate): Likewise.
5865 (gen_lowpart_for_combine): Likewise.
5866 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5867 * config/avr/avr.c (avr_to_int_mode): Likewise.
5868 (avr_out_plus_1): Likewise.
5869 (avr_out_plus): Likewise.
5870 (avr_out_round): Likewise.
5871 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5872 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5873 (s390_expand_vcond): Likewise.
5874 * config/spu/spu.c (spu_split_immediate): Likewise.
5875 (spu_expand_mov): Likewise.
5876 * dse.c (get_stored_val): Likewise.
5877 * expmed.c (store_bit_field_1): Likewise.
5878 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5880 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5881 (extract_low_bits): Likewise.
5882 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5883 handling rather than repeating the check.
5884 (emit_group_store): Likewise.
5885 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5886 * optabs.c (expand_absneg_bit): Likewise.
5887 (expand_copysign_absneg): Likewise.
5888 (expand_copysign_bit): Likewise.
5889 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5890 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5891 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5892 * var-tracking.c (prepare_call_arguments): Likewise.
5893 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5894 int_mode_for_mode instead of mode_for_size.
5895 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
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 * machmode.h (int_mode_for_size): New function.
5902 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5903 instead of mode_for_size.
5904 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5906 * combine.c (expand_field_assignment): Use int_mode_for_size
5907 instead of mode_for_size.
5908 (make_extraction): Likewise.
5909 (simplify_shift_const_1): Likewise.
5910 (simplify_comparison): Likewise.
5911 * dojump.c (do_jump): Likewise.
5912 * dwarf2out.c (mem_loc_descriptor): Likewise.
5913 * emit-rtl.c (init_derived_machine_modes): Likewise.
5914 * expmed.c (flip_storage_order): Likewise.
5915 (convert_extracted_bit_field): Likewise.
5916 * expr.c (copy_blkmode_from_reg): Likewise.
5917 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5918 * internal-fn.c (expand_mul_overflow): Likewise.
5919 * lower-subreg.c (simple_move): Likewise.
5920 * optabs-libfuncs.c (init_optabs): Likewise.
5921 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5922 * tree.c (vector_type_mode): Likewise.
5923 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5924 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5925 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5926 * tree-vect-stmts.c (vectorizable_load): Likewise.
5927 (vectorizable_store): Likewise.
5929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5930 Alan Hayward <alan.hayward@arm.com>
5931 David Sherwood <david.sherwood@arm.com>
5933 * coretypes.h (pod_mode): New type.
5934 (scalar_int_mode_pod): New typedef.
5935 * machmode.h (pod_mode): New class.
5936 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5937 * genmodes.c (emit_mode_int_n): Update accordingly.
5938 * lower-subreg.h (target_lower_subreg): Change type to
5939 scalar_int_mode_pod.
5940 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5941 scalar_int_mode_pod.
5943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5944 Alan Hayward <alan.hayward@arm.com>
5945 David Sherwood <david.sherwood@arm.com>
5947 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5948 machine_mode to scalar_int_mode.
5949 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5950 (rs6000_option_override_internal): Remove cast to int.
5951 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5952 machine_mode to scalar_int_mode.
5953 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5954 (rs6000_option_override_internal): Remove cast to int.
5955 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5956 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5958 * config/s390/s390.c (s390_expand_builtin): Likewise.
5959 * coretypes.h (scalar_int_mode): New type.
5960 (opt_scalar_int_mode): New typedef.
5961 * machmode.h (scalar_int_mode): New class.
5962 (scalar_int_mode::includes_p): New function.
5963 (byte_mode): Change type to scalar_int_mode.
5964 (word_mode): Likewise.
5965 (ptr_mode): Likewise.
5966 * emit-rtl.c (byte_mode): Likewise.
5967 (word_mode): Likewise.
5968 (ptr_mode): Likewise.
5969 (init_derived_machine_modes): Update accordingly.
5970 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5971 and MODE_PARTIAL_INT.
5972 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5973 opt_scalar_int_mode.
5975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5976 Alan Hayward <alan.hayward@arm.com>
5977 David Sherwood <david.sherwood@arm.com>
5979 * target.def (libgcc_floating_mode_supported_p): Take a
5981 * doc/tm.texi: Regenerate.
5982 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5984 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5985 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5989 Alan Hayward <alan.hayward@arm.com>
5990 David Sherwood <david.sherwood@arm.com>
5992 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5993 * doc/tm.texi: Regenerate.
5994 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5995 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5996 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5997 * targhooks.h (default_floatn_mode): Likewise.
5998 * targhooks.c (default_floatn_mode): Likewise.
5999 * tree.c (build_common_tree_nodes): Update accordingly.
6001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6002 Alan Hayward <alan.hayward@arm.com>
6003 David Sherwood <david.sherwood@arm.com>
6005 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6006 (mode_iterator::iterate_p): Likewise.
6007 (mode_iterator::get_wider): Likewise.
6008 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6011 Alan Hayward <alan.hayward@arm.com>
6012 David Sherwood <david.sherwood@arm.com>
6014 * coretypes.h (opt_scalar_float_mode): New typedef.
6015 * machmode.h (float_mode_for_size): New function.
6016 * emit-rtl.c (double_mode): Delete.
6017 (init_emit_once): Use float_mode_for_size.
6018 * stor-layout.c (layout_type): Likewise.
6019 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6022 Alan Hayward <alan.hayward@arm.com>
6023 David Sherwood <david.sherwood@arm.com>
6025 * output.h (assemble_real): Take a scalar_float_mode.
6026 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6027 * config/arm/arm.md (consttable_4): Likewise.
6028 (consttable_8): Likewise.
6029 (consttable_16): Likewise.
6030 * config/mips/mips.md (consttable_float): Likewise.
6031 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6032 * varasm.c (assemble_real): Take a scalar_float_mode.
6033 (output_constant_pool_2): Update accordingly.
6034 (output_constant): Likewise.
6036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6037 Alan Hayward <alan.hayward@arm.com>
6038 David Sherwood <david.sherwood@arm.com>
6040 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6041 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6042 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6043 (native_encode_real): Likewise.
6044 (native_interpret_real): Likewise.
6045 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6046 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6049 Alan Hayward <alan.hayward@arm.com>
6050 David Sherwood <david.sherwood@arm.com>
6052 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6053 <scalar_float_mode>. Simplify.
6054 (gen_extend_conv_libfunc): Likewise.
6056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6057 Alan Hayward <alan.hayward@arm.com>
6058 David Sherwood <david.sherwood@arm.com>
6060 * coretypes.h (scalar_float_mode): New type.
6061 * machmode.h (mode_traits::from_int): Use machine_mode if
6062 USE_ENUM_MODES is defined.
6063 (is_a): New function.
6065 (dyn_cast): Likewise.
6066 (scalar_float_mode): New class.
6067 (scalar_float_mode::includes_p): New function.
6068 (is_float_mode): Likewise.
6069 * gdbhooks.py (MachineModePrinter): New class.
6070 (build_pretty_printer): Use it for scalar_float_mode.
6071 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6072 (format_helper::format_helper): Turn into a template.
6073 * genmodes.c (get_mode_class): New function.
6074 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6075 or machine_mode if none.
6076 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6077 as_a <scalar_float_mode>.
6078 * dwarf2out.c (mem_loc_descriptor): Likewise.
6079 (insert_float): Likewise.
6080 (add_const_value_attribute): Likewise.
6081 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6082 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6083 (expand_unop): Update accordingly.
6084 (expand_abs_nojump): Likewise.
6085 (expand_copysign_absneg): Take a scalar_float_mode.
6086 (expand_copysign_bit): Likewise.
6087 (expand_copysign): Update accordingly.
6089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6090 Alan Hayward <alan.hayward@arm.com>
6091 David Sherwood <david.sherwood@arm.com>
6093 * coretypes.h (opt_mode): New class.
6094 * machmode.h (opt_mode): Likewise.
6095 (opt_mode::else_void): New function.
6096 (opt_mode::require): Likewise.
6097 (opt_mode::exists): Likewise.
6098 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6099 (GET_MODE_2XWIDER_MODE): Likewise.
6100 (mode_iterator::get_wider): Update accordingly.
6101 (mode_iterator::get_2xwider): Likewise.
6102 (mode_iterator::get_known_wider): Likewise, turning into a template.
6103 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6104 forcing a wider mode to exist.
6105 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6106 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6107 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6108 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6109 * lower-subreg.c (init_lower_subreg): Likewise.
6110 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6111 on the final iteration.
6112 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6113 a wider mode exists before asking for a move pattern.
6114 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6115 forcing a wider mode to exist.
6116 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6117 returning false if no such mode exists.
6118 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6119 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6120 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6121 Avoid checking for a MODE_INT if we already know the mode is not a
6123 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6124 forcing a wider mode to exist.
6125 (expmed_mult_highpart_optab): Likewise.
6126 (expmed_mult_highpart): Likewise.
6127 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6129 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6130 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6131 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6132 * internal-fn.c (expand_mul_overflow): Update use of
6133 GET_MODE_2XWIDER_MODE.
6134 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6135 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6136 GET_MODE_WIDER_MODE.
6137 (convert_plusminus_to_widen): Likewise.
6138 * tree-switch-conversion.c (array_value_type): Likewise.
6139 * var-tracking.c (emit_note_insn_var_location): Likewise.
6140 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6141 Return false inside rather than outside the loop if no wider mode
6143 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6144 and GET_MODE_2XWIDER_MODE
6145 (can_compare_p): Use else_void.
6146 * gdbhooks.py (OptMachineModePrinter): New class.
6147 (build_pretty_printer): Use it for opt_mode.
6149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6150 Alan Hayward <alan.hayward@arm.com>
6151 David Sherwood <david.sherwood@arm.com>
6153 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6154 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6157 Alan Hayward <alan.hayward@arm.com>
6158 David Sherwood <david.sherwood@arm.com>
6160 * machmode.h (mode_traits): New structure.
6161 (get_narrowest_mode): New function.
6162 (mode_iterator::start): Likewise.
6163 (mode_iterator::iterate_p): Likewise.
6164 (mode_iterator::get_wider): Likewise.
6165 (mode_iterator::get_known_wider): Likewise.
6166 (mode_iterator::get_2xwider): Likewise.
6167 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6168 (FOR_EACH_MODE): Likewise.
6169 (FOR_EACH_MODE_FROM): Likewise.
6170 (FOR_EACH_MODE_UNTIL): Likewise.
6171 (FOR_EACH_WIDER_MODE): Likewise.
6172 (FOR_EACH_2XWIDER_MODE): Likewise.
6173 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6174 * combine.c (simplify_comparison): Likewise
6175 * config/i386/i386.c (type_natural_mode): Likewise.
6176 * cse.c (cse_insn): Likewise.
6177 * dse.c (find_shift_sequence): Likewise.
6178 * emit-rtl.c (init_derived_machine_modes): Likewise.
6179 (init_emit_once): Likewise.
6180 * explow.c (hard_function_value): Likewise.
6181 * expmed.c (extract_fixed_bit_field_1): Likewise.
6182 (extract_bit_field_1): Likewise.
6183 (expand_divmod): Likewise.
6184 (emit_store_flag_1): Likewise.
6185 * expr.c (init_expr_target): Likewise.
6186 (convert_move): Likewise.
6187 (alignment_for_piecewise_move): Likewise.
6188 (widest_int_mode_for_size): Likewise.
6189 (emit_block_move_via_movmem): Likewise.
6190 (copy_blkmode_to_reg): Likewise.
6191 (set_storage_via_setmem): Likewise.
6192 (compress_float_constant): Likewise.
6193 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6194 * optabs-query.c (get_best_extraction_insn): Likewise.
6195 * optabs.c (expand_binop): Likewise.
6196 (expand_twoval_unop): Likewise.
6197 (expand_twoval_binop): Likewise.
6198 (widen_leading): Likewise.
6199 (widen_bswap): Likewise.
6200 (expand_parity): Likewise.
6201 (expand_unop): Likewise.
6202 (prepare_cmp_insn): Likewise.
6203 (prepare_float_lib_cmp): Likewise.
6204 (expand_float): Likewise.
6205 (expand_fix): Likewise.
6206 (expand_sfix_optab): Likewise.
6207 * postreload.c (move2add_use_add2_insn): Likewise.
6208 * reg-stack.c (reg_to_stack): Likewise.
6209 * reginfo.c (choose_hard_reg_mode): Likewise.
6210 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6211 * stor-layout.c (mode_for_size): Likewise.
6212 (smallest_mode_for_size): Likewise.
6213 (mode_for_vector): Likewise.
6214 (finish_bitfield_representative): Likewise.
6215 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6216 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6217 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6218 * var-tracking.c (prepare_call_arguments): Likewise.
6220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6221 Alan Hayward <alan.hayward@arm.com>
6222 David Sherwood <david.sherwood@arm.com>
6224 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6225 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6226 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6227 * machmode.h (mode_size): Move earlier in file.
6228 (mode_precision): Likewise.
6229 (mode_inner): Likewise.
6230 (mode_nunits): Likewise.
6231 (mode_unit_size): Likewise.
6232 (unit_unit_precision): Likewise.
6233 (mode_wider): Likewise.
6234 (mode_2xwider): Likewise.
6235 (machine_mode): New class.
6236 (mode_to_bytes): New function.
6237 (mode_to_bits): Likewise.
6238 (mode_to_precision): Likewise.
6239 (mode_to_inner): Likewise.
6240 (mode_to_unit_size): Likewise.
6241 (mode_to_unit_precision): Likewise.
6242 (mode_to_nunits): Likewise.
6243 (GET_MODE_SIZE): Use mode_to_bytes.
6244 (GET_MODE_BITSIZE): Use mode_to_bits.
6245 (GET_MODE_PRECISION): Use mode_to_precision.
6246 (GET_MODE_INNER): Use mode_to_inner.
6247 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6248 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6249 (GET_MODE_NUNITS): Use mode_to_nunits.
6250 * system.h (ALWAYS_INLINE): New macro.
6251 * config/powerpcspe/powerpcspe-c.c
6252 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6253 int for arg1_mode and arg2_mode.
6255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6256 Alan Hayward <alan.hayward@arm.com>
6257 David Sherwood <david.sherwood@arm.com>
6259 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6260 Prefix mode names with E_ in case statements.
6261 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6262 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6263 (aarch64_split_simd_move): Likewise.
6264 (aarch64_gen_storewb_pair): Likewise.
6265 (aarch64_gen_loadwb_pair): Likewise.
6266 (aarch64_gen_store_pair): Likewise.
6267 (aarch64_gen_load_pair): Likewise.
6268 (aarch64_get_condition_code_1): Likewise.
6269 (aarch64_constant_pool_reload_icode): Likewise.
6270 (get_rsqrte_type): Likewise.
6271 (get_rsqrts_type): Likewise.
6272 (get_recpe_type): Likewise.
6273 (get_recps_type): Likewise.
6274 (aarch64_gimplify_va_arg_expr): Likewise.
6275 (aarch64_simd_container_mode): Likewise.
6276 (aarch64_emit_load_exclusive): Likewise.
6277 (aarch64_emit_store_exclusive): Likewise.
6278 (aarch64_expand_compare_and_swap): Likewise.
6279 (aarch64_gen_atomic_cas): Likewise.
6280 (aarch64_emit_bic): Likewise.
6281 (aarch64_emit_atomic_swap): Likewise.
6282 (aarch64_emit_atomic_load_op): Likewise.
6283 (aarch64_evpc_trn): Likewise.
6284 (aarch64_evpc_uzp): Likewise.
6285 (aarch64_evpc_zip): Likewise.
6286 (aarch64_evpc_ext): Likewise.
6287 (aarch64_evpc_rev): Likewise.
6288 (aarch64_evpc_dup): Likewise.
6289 (aarch64_gen_ccmp_first): Likewise.
6290 (aarch64_gen_ccmp_next): Likewise.
6291 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6292 (alpha_emit_xfloating_libcall): Likewise.
6293 (emit_insxl): Likewise.
6294 (alpha_arg_type): Likewise.
6295 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6296 (arc_preferred_simd_mode): Likewise.
6297 (arc_secondary_reload): Likewise.
6298 (get_arc_condition_code): Likewise.
6299 (arc_print_operand): Likewise.
6300 (arc_legitimate_constant_p): Likewise.
6301 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6302 * config/arc/arc.md (casesi_load): Likewise.
6303 (casesi_compact_jump): Likewise.
6304 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6305 (cc_use_register): Likewise.
6306 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6307 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6308 (arm_init_iwmmxt_builtins): Likewise.
6309 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6310 (neon_expand_vector_init): Likewise.
6311 (arm_attr_length_move_neon): Likewise.
6312 (maybe_get_arm_condition_code): Likewise.
6313 (arm_emit_vector_const): Likewise.
6314 (arm_preferred_simd_mode): Likewise.
6315 (arm_output_iwmmxt_tinsr): Likewise.
6316 (thumb1_output_casesi): Likewise.
6317 (thumb2_output_casesi): Likewise.
6318 (arm_emit_load_exclusive): Likewise.
6319 (arm_emit_store_exclusive): Likewise.
6320 (arm_expand_compare_and_swap): Likewise.
6321 (arm_evpc_neon_vuzp): Likewise.
6322 (arm_evpc_neon_vzip): Likewise.
6323 (arm_evpc_neon_vrev): Likewise.
6324 (arm_evpc_neon_vtrn): Likewise.
6325 (arm_evpc_neon_vext): Likewise.
6326 (arm_validize_comparison): Likewise.
6327 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6328 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6329 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6330 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6331 (c6x_preferred_simd_mode): Likewise.
6332 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6333 (epiphany_rtx_costs): Likewise.
6334 * config/epiphany/predicates.md (proper_comparison_operator):
6336 * config/frv/frv.c (condexec_memory_operand): Likewise.
6337 (frv_emit_move): Likewise.
6338 (output_move_single): Likewise.
6339 (output_condmove_single): Likewise.
6340 (frv_hard_regno_mode_ok): Likewise.
6341 (frv_matching_accg_mode): Likewise.
6342 * config/h8300/h8300.c (split_adds_subs): Likewise.
6343 (h8300_rtx_costs): Likewise.
6344 (h8300_print_operand): Likewise.
6345 (compute_mov_length): Likewise.
6346 (output_logical_op): Likewise.
6347 (compute_logical_op_length): Likewise.
6348 (compute_logical_op_cc): Likewise.
6349 (h8300_shift_needs_scratch_p): Likewise.
6350 (output_a_shift): Likewise.
6351 (compute_a_shift_length): Likewise.
6352 (compute_a_shift_cc): Likewise.
6353 (expand_a_rotate): Likewise.
6354 (output_a_rotate): Likewise.
6355 * config/i386/i386.c (classify_argument): Likewise.
6356 (function_arg_advance_32): Likewise.
6357 (function_arg_32): Likewise.
6358 (function_arg_64): Likewise.
6359 (function_value_64): Likewise.
6360 (ix86_gimplify_va_arg): Likewise.
6361 (ix86_legitimate_constant_p): Likewise.
6362 (put_condition_code): Likewise.
6363 (split_double_mode): Likewise.
6364 (ix86_avx256_split_vector_move_misalign): Likewise.
6365 (ix86_expand_vector_logical_operator): Likewise.
6366 (ix86_split_idivmod): Likewise.
6367 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6368 (ix86_build_const_vector): Likewise.
6369 (ix86_build_signbit_mask): Likewise.
6370 (ix86_match_ccmode): Likewise.
6371 (ix86_cc_modes_compatible): Likewise.
6372 (ix86_expand_branch): Likewise.
6373 (ix86_expand_sse_cmp): Likewise.
6374 (ix86_expand_sse_movcc): Likewise.
6375 (ix86_expand_int_sse_cmp): Likewise.
6376 (ix86_expand_vec_perm_vpermi2): Likewise.
6377 (ix86_expand_vec_perm): Likewise.
6378 (ix86_expand_sse_unpack): Likewise.
6379 (ix86_expand_int_addcc): Likewise.
6380 (ix86_split_to_parts): Likewise.
6381 (ix86_vectorize_builtin_gather): Likewise.
6382 (ix86_vectorize_builtin_scatter): Likewise.
6383 (avx_vpermilp_parallel): Likewise.
6384 (inline_memory_move_cost): Likewise.
6385 (ix86_tieable_integer_mode_p): Likewise.
6386 (x86_maybe_negate_const_int): Likewise.
6387 (ix86_expand_vector_init_duplicate): Likewise.
6388 (ix86_expand_vector_init_one_nonzero): Likewise.
6389 (ix86_expand_vector_init_one_var): Likewise.
6390 (ix86_expand_vector_init_concat): Likewise.
6391 (ix86_expand_vector_init_interleave): Likewise.
6392 (ix86_expand_vector_init_general): Likewise.
6393 (ix86_expand_vector_set): Likewise.
6394 (ix86_expand_vector_extract): Likewise.
6395 (emit_reduc_half): Likewise.
6396 (ix86_emit_i387_round): Likewise.
6397 (ix86_mangle_type): Likewise.
6398 (ix86_expand_round_sse4): Likewise.
6399 (expand_vec_perm_blend): Likewise.
6400 (canonicalize_vector_int_perm): Likewise.
6401 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6402 (expand_vec_perm_1): Likewise.
6403 (expand_vec_perm_interleave3): Likewise.
6404 (expand_vec_perm_even_odd_pack): Likewise.
6405 (expand_vec_perm_even_odd_1): Likewise.
6406 (expand_vec_perm_broadcast_1): Likewise.
6407 (ix86_vectorize_vec_perm_const_ok): Likewise.
6408 (ix86_expand_vecop_qihi): Likewise.
6409 (ix86_expand_mul_widen_hilo): Likewise.
6410 (ix86_expand_sse2_abs): Likewise.
6411 (ix86_expand_pextr): Likewise.
6412 (ix86_expand_pinsr): Likewise.
6413 (ix86_preferred_simd_mode): Likewise.
6414 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6415 * config/i386/sse.md (*andnot<mode>3): Likewise.
6416 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6417 (*<code><mode>3): Likewise.
6418 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6419 (ia64_expand_atomic_op): Likewise.
6420 (ia64_arg_type): Likewise.
6421 (ia64_mode_to_int): Likewise.
6422 (ia64_scalar_mode_supported_p): Likewise.
6423 (ia64_vector_mode_supported_p): Likewise.
6424 (expand_vec_perm_broadcast): Likewise.
6425 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6426 (iq2000_function_arg_advance): Likewise.
6427 (iq2000_function_arg): Likewise.
6428 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6429 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6430 (m68k_libcall_value): Likewise.
6431 (m68k_function_value): Likewise.
6432 (sched_attr_op_type): Likewise.
6433 * config/mcore/mcore.c (mcore_output_move): Likewise.
6434 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6436 (microblaze_function_arg): Likewise.
6437 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6438 (mips_print_operand): Likewise.
6439 (mips_mode_ok_for_mov_fmt_p): Likewise.
6440 (mips_vector_mode_supported_p): Likewise.
6441 (mips_preferred_simd_mode): Likewise.
6442 (mips_expand_vpc_loongson_even_odd): Likewise.
6443 (mips_expand_vec_unpack): Likewise.
6444 (mips_expand_vi_broadcast): Likewise.
6445 (mips_expand_vector_init): Likewise.
6446 (mips_expand_vec_reduc): Likewise.
6447 (mips_expand_msa_cmp): Likewise.
6448 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6449 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6450 (cc_flags_for_mode): Likewise.
6451 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6452 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6453 (nds32_output_casesi_pc_relative): Likewise.
6454 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6455 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6456 (nvptx_gen_unpack): Likewise.
6457 (nvptx_gen_pack): Likewise.
6458 (nvptx_gen_shuffle): Likewise.
6459 (nvptx_gen_wcast): Likewise.
6460 (nvptx_preferred_simd_mode): Likewise.
6461 * config/pa/pa.c (pa_secondary_reload): Likewise.
6462 * config/pa/predicates.md (base14_operand): Likewise.
6463 * config/powerpcspe/powerpcspe-c.c
6464 (altivec_resolve_overloaded_builtin): Likewise.
6465 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6467 (rs6000_preferred_simd_mode): Likewise.
6468 (output_vec_const_move): Likewise.
6469 (rs6000_expand_vector_extract): Likewise.
6470 (rs6000_split_vec_extract_var): Likewise.
6471 (reg_offset_addressing_ok_p): Likewise.
6472 (rs6000_legitimate_offset_address_p): Likewise.
6473 (rs6000_legitimize_address): Likewise.
6474 (rs6000_emit_set_const): Likewise.
6475 (rs6000_const_vec): Likewise.
6476 (rs6000_emit_move): Likewise.
6477 (spe_build_register_parallel): Likewise.
6478 (rs6000_darwin64_record_arg_recurse): Likewise.
6479 (swap_selector_for_mode): Likewise.
6480 (spe_init_builtins): Likewise.
6481 (paired_init_builtins): Likewise.
6482 (altivec_init_builtins): Likewise.
6483 (do_load_for_compare): Likewise.
6484 (rs6000_generate_compare): Likewise.
6485 (rs6000_expand_float128_convert): Likewise.
6486 (emit_load_locked): Likewise.
6487 (emit_store_conditional): Likewise.
6488 (rs6000_output_function_epilogue): Likewise.
6489 (rs6000_handle_altivec_attribute): Likewise.
6490 (rs6000_function_value): Likewise.
6491 (emit_fusion_gpr_load): Likewise.
6492 (emit_fusion_p9_load): Likewise.
6493 (emit_fusion_p9_store): Likewise.
6494 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6495 (fusion_gpr_mem_load): Likewise.
6496 (fusion_addis_mem_combo_load): Likewise.
6497 (fusion_addis_mem_combo_store): Likewise.
6498 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6499 (fusion_gpr_mem_load): Likewise.
6500 (fusion_addis_mem_combo_load): Likewise.
6501 (fusion_addis_mem_combo_store): Likewise.
6502 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6504 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6505 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6506 (rs6000_preferred_simd_mode): Likewise.
6507 (output_vec_const_move): Likewise.
6508 (rs6000_expand_vector_extract): Likewise.
6509 (rs6000_split_vec_extract_var): Likewise.
6510 (reg_offset_addressing_ok_p): Likewise.
6511 (rs6000_legitimate_offset_address_p): Likewise.
6512 (rs6000_legitimize_address): Likewise.
6513 (rs6000_emit_set_const): Likewise.
6514 (rs6000_const_vec): Likewise.
6515 (rs6000_emit_move): Likewise.
6516 (rs6000_darwin64_record_arg_recurse): Likewise.
6517 (swap_selector_for_mode): Likewise.
6518 (paired_init_builtins): Likewise.
6519 (altivec_init_builtins): Likewise.
6520 (rs6000_expand_float128_convert): Likewise.
6521 (emit_load_locked): Likewise.
6522 (emit_store_conditional): Likewise.
6523 (rs6000_output_function_epilogue): Likewise.
6524 (rs6000_handle_altivec_attribute): Likewise.
6525 (rs6000_function_value): Likewise.
6526 (emit_fusion_gpr_load): Likewise.
6527 (emit_fusion_p9_load): Likewise.
6528 (emit_fusion_p9_store): Likewise.
6529 * config/rx/rx.c (rx_gen_move_template): Likewise.
6530 (flags_from_mode): Likewise.
6531 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6532 (s390_slb_comparison): Likewise.
6533 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6534 (s390_vector_mode_supported_p): Likewise.
6535 (s390_cc_modes_compatible): Likewise.
6536 (s390_match_ccmode_set): Likewise.
6537 (s390_canonicalize_comparison): Likewise.
6538 (s390_emit_compare_and_swap): Likewise.
6539 (s390_branch_condition_mask): Likewise.
6540 (s390_rtx_costs): Likewise.
6541 (s390_secondary_reload): Likewise.
6542 (__SECONDARY_RELOAD_CASE): Likewise.
6543 (s390_expand_cs): Likewise.
6544 (s390_preferred_simd_mode): Likewise.
6545 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6546 * config/sh/sh.c (sh_print_operand): Likewise.
6547 (dump_table): Likewise.
6548 (sh_secondary_reload): Likewise.
6549 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6550 * config/sh/sh.md (casesi_worker_1): Likewise.
6551 (casesi_worker_2): Likewise.
6552 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6553 (fcc_comparison_operator): Likewise.
6554 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6555 (emit_soft_tfmode_cvt): Likewise.
6556 (sparc_preferred_simd_mode): Likewise.
6557 (output_cbranch): Likewise.
6558 (sparc_print_operand): Likewise.
6559 (sparc_expand_vec_perm_bmask): Likewise.
6560 (vector_init_bshuffle): Likewise.
6561 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6562 (spu_vector_mode_supported_p): Likewise.
6563 (spu_expand_insv): Likewise.
6564 (spu_emit_branch_or_set): Likewise.
6565 (spu_handle_vector_attribute): Likewise.
6566 (spu_builtin_splats): Likewise.
6567 (spu_builtin_extract): Likewise.
6568 (spu_builtin_promote): Likewise.
6569 (spu_expand_sign_extend): Likewise.
6570 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6571 (tilegx_simd_int): Likewise.
6572 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6573 (tilepro_simd_int): Likewise.
6574 * config/v850/v850.c (const_double_split): Likewise.
6575 (v850_print_operand): Likewise.
6576 (ep_memory_offset): Likewise.
6577 * config/vax/vax.c (vax_rtx_costs): Likewise.
6578 (vax_output_int_move): Likewise.
6579 (vax_output_int_add): Likewise.
6580 (vax_output_int_subtract): Likewise.
6581 * config/visium/predicates.md (visium_branch_operator): Likewise.
6582 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6583 (visium_print_operand_address): Likewise.
6584 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6585 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6586 (xtensa_expand_conditional_branch): Likewise.
6587 (xtensa_copy_incoming_a7): Likewise.
6588 (xtensa_output_literal): Likewise.
6589 * dfp.c (decimal_real_maxval): Likewise.
6590 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6593 Alan Hayward <alan.hayward@arm.com>
6594 David Sherwood <david.sherwood@arm.com>
6596 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6597 (mode_nunits_inline): Likewise.
6598 (mode_inner_inline): Likewise.
6599 (mode_unit_size_inline): Likewise.
6600 (mode_unit_precision_inline): Likewise.
6601 (emit_insn_modes_h): Likewise. Also emit a #define of the
6603 (emit_mode_wider): Add an E_ prefix to mode names.
6604 (emit_mode_complex): Likewise.
6605 (emit_mode_inner): Likewise.
6606 (emit_mode_adjustments): Likewise.
6607 (emit_mode_int_n): Likewise.
6608 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6609 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6610 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6611 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6612 (CRC32_BUILTIN, ENTRY): Likewise.
6613 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6614 (aarch64_pop_regs): Likewise.
6615 (aarch64_process_components): Likewise.
6616 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6617 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6618 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6619 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6620 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6621 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6623 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6624 (pcmpistr): Likewise.
6625 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6626 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6627 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6629 * config/rl78/rl78.c (mduc_regs): Likewise.
6630 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6631 (htm_expand_builtin): Likewise.
6632 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6633 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6634 * config/xtensa/xtensa.c (print_operand): Likewise.
6635 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6636 (NUM_MODE_VECTOR_INT): Likewise.
6637 * genoutput.c (null_operand): Likewise.
6638 (output_operand_data): Likewise.
6639 * genrecog.c (print_parameter_value): Likewise.
6640 * lra.c (debug_operand_data): Likewise.
6642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6644 * dwarf2out.c (scompare_loc_descriptor_wide)
6645 (scompare_loc_descriptor_narrow): New functions, split out from...
6646 (scompare_loc_descriptor): ...here.
6647 * expmed.c (emit_store_flag_int): New function, split out from...
6648 (emit_store_flag): ...here.
6650 2017-08-30 Richard Biener <rguenther@suse.de>
6652 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6653 (dwarf2out_early_finish): Move setting of AT_pubnames from
6654 early debug output to early finish.
6656 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6658 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6659 and -mdata-region to the assembler.
6661 2017-08-30 Richard Biener <rguenther@suse.de>
6663 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6665 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6666 (dwarf2out_early_global_decl): Only generate a DIE for the
6667 abstract origin if it doesn't already exist or is a declaration DIE.
6668 (resolve_addr): Do not add the linkage name twice when
6669 generating a stub DIE for the DW_TAG_GNU_call_site target.
6671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6673 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6674 Use machine_mode rather than int for arg1_mode.
6676 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6679 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6680 that the second argument of the built-in functions to unpack
6681 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6682 switch statement instead a lot of if statements.
6683 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6684 Allow 64-bit values to be in Altivec registers as well as
6685 traditional floating point registers.
6686 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6688 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6690 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6691 MAX_REGS_PER_ADDRESS == 1.
6693 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6695 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6696 * config/i386/i386.c (ix86_option_override_internal): Simplify
6697 setting of opts->x_flag_entry.
6699 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6700 Jakub Jelinek <jakub@redhat.com>
6701 Richard Biener <rguenther@suse.de>
6703 PR tree-optimization/81503
6704 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6705 folded constant fits in the target type; reorder tests for clarity.
6707 2017-08-29 Martin Liska <mliska@suse.cz>
6709 * passes.def: Include pass_lower_switch.
6710 * stmt.c (dump_case_nodes): Remove and move to
6711 tree-switch-conversion.
6712 (case_values_threshold): Likewise.
6713 (expand_switch_as_decision_tree_p): Likewise.
6714 (emit_case_decision_tree): Likewise.
6715 (expand_case): Likewise.
6716 (balance_case_nodes): Likewise.
6717 (node_has_low_bound): Likewise.
6718 (node_has_high_bound): Likewise.
6719 (node_is_bounded): Likewise.
6720 (emit_case_nodes): Likewise.
6721 (struct simple_case_node): New struct.
6722 (add_case_node): Remove.
6723 (emit_case_dispatch_table): Use vector instead of case_list.
6724 (reset_out_edges_aux): Remove.
6725 (compute_cases_per_edge): Likewise.
6726 (expand_case): Build list of simple_case_node.
6727 (expand_sjlj_dispatch_table): Use it.
6728 * tree-switch-conversion.c (struct case_node): Moved from
6729 stmt.c and adjusted.
6730 (emit_case_nodes): Likewise.
6731 (node_has_low_bound): Likewise.
6732 (node_has_high_bound): Likewise.
6733 (node_is_bounded): Likewise.
6734 (case_values_threshold): Likewise.
6735 (reset_out_edges_aux): Likewise.
6736 (compute_cases_per_edge): Likewise.
6737 (add_case_node): Likewise.
6738 (dump_case_nodes): Likewise.
6739 (balance_case_nodes): Likewise.
6740 (expand_switch_as_decision_tree_p): Likewise.
6741 (emit_jump): Likewise.
6742 (emit_case_decision_tree): Likewise.
6743 (try_switch_expansion): Likewise.
6744 (do_jump_if_equal): Likewise.
6745 (emit_cmp_and_jump_insns): Likewise.
6746 (fix_phi_operands_for_edge): New function.
6747 (record_phi_operand_mapping): Likewise.
6748 (class pass_lower_switch): New pass.
6749 (pass_lower_switch::execute): New function.
6750 (make_pass_lower_switch): Likewise.
6751 (conditional_probability):
6752 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6753 * tree-pass.h: Add make_pass_lower_switch.
6755 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6758 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6761 2017-08-29 Richard Biener <rguenther@suse.de>
6763 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6764 we do not add a DW_AT_inline attribute twice.
6765 (gen_subprogram_die): Remove code setting DW_AT_inline on
6766 DECL_ABSTRACT_P nodes.
6768 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6770 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6771 calls to internal functions.
6772 (gimplify_modify_expr): Likewise.
6773 * tree-call-cdce.c (use_internal_fn): Likewise.
6774 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6775 (convert_to_divmod): Set the nothrow flag.
6776 * tree-if-conv.c (predicate_mem_writes): Likewise.
6777 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6778 (vectorizable_call): Likewise.
6779 (vectorizable_store): Likewise.
6780 (vectorizable_load): Likewise.
6781 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6782 (vect_recog_mask_conversion_pattern): Likewise.
6784 2017-08-29 Martin Liska <mliska@suse.cz>
6787 * gcc.c (driver_handle_option): Add new argument.
6788 * opts-common.c (handle_option): Pass
6789 target_option_override_hook.
6790 * opts-global.c (lang_handle_option): Add new option.
6791 (set_default_handlers): Add new argument.
6792 (decode_options): Likewise.
6793 * opts.c (target_handle_option): Likewise.
6794 (common_handle_option): Call target_option_override_hook.
6795 * opts.h (struct cl_option_handler_func): Add hook for
6796 target option override.
6797 (struct cl_option_handlers): Likewise.
6798 (set_default_handlers): Add new argument.
6799 (decode_options): Likewise.
6800 (common_handle_option): Likewise.
6801 (target_handle_option): Likewise.
6802 * toplev.c (toplev::main): Pass targetm.target_option.override
6805 2017-08-29 Richard Biener <rguenther@suse.de>
6806 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6808 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6809 life to the active subtree.
6811 2017-08-28 Jeff Law <law@redhat.com>
6813 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6814 derive_equivalences.
6815 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6817 (edge_info::derive_equivalences): New private member function
6819 * tree-ssa-dom.c (class edge_info): Changed from a struct
6820 to a class. Add ctor/dtor, methods and data members.
6821 (edge_info::edge_info): Renamed from allocate_edge_info.
6822 Initialize additional members.
6823 (edge_info::~edge_info): New.
6824 (free_dom_edge_info): Delete the edge info.
6825 (record_edge_info): Use new class & associated member functions.
6826 Tighten forms for testing for edge equivalences.
6827 (record_temporary_equivalences): Iterate over the simple
6828 equivalences rather than assuming there's only one per edge.
6829 (cprop_into_successor_phis): Iterate over the simple
6830 equivalences rather than assuming there's only one per edge.
6831 (optimize_stmt): Use operand_equal_p rather than pointer
6832 equality for mini-DSE code.
6834 2017-08-28 Nathan Sidwell <nathan@acm.org>
6836 * gcc.c (execute): Fold SIGPIPE handling into switch
6837 statement. Adjust internal error message.
6839 2017-08-28 Richard Biener <rguenther@suse.de>
6842 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6843 Do nothing for removed DIEs.
6845 2017-08-28 Richard Biener <rguenther@suse.de>
6847 PR tree-optimization/81977
6848 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6851 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6854 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6855 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6856 using targetm.gen_mem_thread_fence.
6858 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6861 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6862 predicate to register_operand. Reorder operands.
6863 (*btr<mode>): Ditto.
6864 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6865 (*btr<mode>_mask): Ditto.
6867 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6869 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6870 * config/rs6000/xmmintrin.h: New file.
6871 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6873 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6876 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6877 parameter and_insn and return it.
6878 (recombine_lvx_pattern): Insert a copy to ensure availability of
6879 the base register of the copied masking operation at the point of
6880 the instruction replacement.
6881 (recombine_stvx_pattern): Likewise.
6883 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6885 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6886 undocumented switches.
6887 (-mpower9-dform-vector): Likewise.
6888 (-mpower9-dform): Likewise.
6889 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6890 comments to delete references to -mpower9-dform* switches.
6891 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6892 Delete reference to -mpower9-dform* switches, test for
6893 -mpower9-vector instead.
6894 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6895 (OTHER_P9_VECTOR_MASKS): Likewise.
6896 (POWERPC_MASKS): Likewise.
6897 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6898 tests against -mpower9-dform* to -mpower9-vector. Delete code
6899 that checked for -mpower9-dform* consistancy with other options.
6900 Add test for -mpower9-misc to enable other power9 switches.
6901 (rs6000_init_hard_regno_mode_ok): Likewise.
6902 (rs6000_option_override_internal): Likewise.
6903 (rs6000_emit_prologue): Likewise.
6904 (rs6000_emit_epilogue): Likewise.
6905 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6906 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6907 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6909 (emit_fusion_p9_store): Likewise.
6910 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6911 resetting these macros if the assembler does not support ISA 3.0
6913 (TARGET_P9_DFORM_VECTOR): Likewise.
6914 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6915 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6917 2017-08-25 Alan Modra <amodra@gmail.com>
6919 PR rtl-optimization/81747
6920 * cse.c (cse_extended_basic_block): Don't attempt to record
6921 equivalences for degenerate conditional branches that branch
6922 to their fall-through.
6924 2017-08-24 Martin Sebor <msebor@redhat.com>
6927 * gimple-fold.c (size_must_be_zero_p): New function.
6928 (gimple_fold_builtin_memory_op): Call it.
6930 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6932 * config/rs6000/mm_malloc.h: New file.
6934 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6936 PR tree-optimization/81913
6937 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6938 analysis when either IVs in condition can wrap.
6940 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6942 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6943 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6945 2017-08-24 Richard Biener <rguenther@suse.de>
6948 * targhooks.c (default_target_can_inline_p): Properly
6949 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6950 is present and always compare.
6951 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6952 infer -mfpmath=sse from TARGET_SSE_P.
6953 (ix86_can_inline_p): Properly use target_option_default_node when
6954 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6956 2017-08-24 Richard Biener <rguenther@suse.de>
6959 * dwarf2out.c (output_die): Handle flag_generate_offload like
6961 (output_comp_unit): Likewise.
6962 (gen_array_type_die): Likewise.
6963 (dwarf2out_early_finish): Likewise.
6964 (note_variable_value_in_expr): Likewise.
6965 (dwarf2out_finish): Likewise. Adjust assert.
6966 * cgraphunit.c (symbol_table::compile): Move setting of
6967 flag_generate_offload earlier ...
6968 (symbol_table::finalize_compilation_unit): ... here, before
6969 early debug finalization.
6971 2017-08-24 Richard Biener <rguenther@suse.de>
6973 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6974 and ipa-fnsummary.h.
6975 (ix86_can_inline_p): When ix86_fpmath flags do not match
6976 check whether the callee uses FP math at all.
6978 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6981 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6982 instead of TYPE_PRECISION.
6984 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6985 Alan Hayward <alan.hayward@arm.com>
6986 David Sherwood <david.sherwood@arm.com>
6988 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6990 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6992 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6994 (ix86_frame::stack_realign_allocate): New field.
6995 (struct machine_frame_state): Modify comments.
6996 (machine_frame_state::sp_realigned_fp_end): New field.
6997 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6999 (sp_valid_at): Add assertion to assure no attempt to access invalid
7000 offset of a realigned stack.
7001 (fp_valid_at): Likewise.
7002 (choose_baseaddr): Modify comments.
7003 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7004 ix86_expand_prologue.
7005 (ix86_expand_prologue): Modify stack realignment and allocation.
7006 (ix86_expand_epilogue): Modify comments.
7007 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7008 avx2_runtime, avx512f, and avx512f_runtime.
7010 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7012 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7013 (mstackrealign): Do not init to -1.
7014 * config/i386/i386.c (ix86_option_override_internal):
7015 Check opts_set, not opts when setting default value of
7016 opts->x_ix86_force_align_arg_pointer.
7018 2017-08-23 Richard Biener <rguenther@suse.de>
7020 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7021 lang_hooks.decl_printable_name.
7022 * print-rtl-function.c (print_rtx_function): Likewise.
7023 * tree-pretty-print.c (dump_function_header): Likewise.
7025 2017-08-23 Richard Biener <rguenther@suse.de>
7028 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7029 -g0 at compile-time.
7031 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7034 * doc/sourcebuild.texi (Other hardware attributes):
7037 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7040 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7041 Fix single-use check.
7043 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7045 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7047 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7049 * doc/install.texi: Modify to add more details on running selected
7052 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7054 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7055 is combined with -mabi=ms.
7056 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7059 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7061 PR tree-optimization/81488
7062 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7063 and cached_basis fields.
7064 (MAX_SPREAD): New constant.
7065 (alloc_cand_and_find_basis): Initialize new fields.
7066 (clear_visited): New function.
7067 (create_phi_basis_1): Rename from create_phi_basis, set visited
7068 and cached_basis fields.
7069 (create_phi_basis): New wrapper function.
7070 (phi_add_costs_1): Rename from phi_add_costs, add spread
7071 parameter, set visited field, short-circuit when limits reached.
7072 (phi_add_costs): New wrapper function.
7073 (record_phi_increments_1): Rename from record_phi_increments, set
7075 (record_phi_increments): New wrapper function.
7076 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7077 (phi_incr_cost): New wrapper function.
7078 (all_phi_incrs_profitable_1): Rename from
7079 all_phi_incrs_profitable, set visited field.
7080 (all_phi_incrs_profitable): New wrapper function.
7082 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7083 Alan Hayward <alan.hayward@arm.com>
7084 David Sherwood <david.sherwood@arm.com>
7086 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7087 that takes the outer and inner modes.
7088 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7089 comparison as the canonical test for a paradoxical subreg.
7090 * combine.c (simplify_set): Use paradoxical_subreg_p.
7091 (make_extraction): Likewise.
7092 (force_to_mode): Likewise.
7093 (rtx_equal_for_field_assignment_p): Likewise.
7094 (gen_lowpart_for_combine): Likewise.
7095 (simplify_comparison): Likewise.
7096 * cse.c (equiv_constant): Likewise.
7097 * expmed.c (store_bit_field_1): Likewise.
7098 * final.c (alter_subreg): Likewise.
7099 * fwprop.c (propagate_rtx): Likewise.
7100 (forward_propagate_subreg): Likewise.
7101 * ira-conflicts.c (ira_build_conflicts): Likewise.
7102 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7103 * lra-constraints.c (curr_insn_transform): Likewise.
7104 (split_reg): Likewise.
7105 * lra-eliminations.c (move_plus_up): Likewise.
7106 (lra_eliminate_regs_1): Likewise.
7107 * recog.c (general_operand): Likewise.
7108 * ree.c (combine_reaching_defs): Likewise.
7109 * reload.c (push_reload): Likewise.
7110 (find_reloads): Likewise.
7111 * reload1.c (elimination_effects): Likewise.
7112 (compute_reload_subreg_offset): Likewise.
7113 (choose_reload_regs): Likewise.
7114 * rtlanal.c (subreg_lsb_1): Likewise.
7115 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7116 (simplify_subreg): Likewise.
7117 * var-tracking.c (track_loc_p): Likewise.
7118 * emit-rtl.c (byte_lowpart_offset): Likewise.
7119 (paradoxical_subreg_p): Delete out-of-line definition.
7121 2017-08-22 Jeff Law <law@redhat.com>
7123 PR tree-optimization/81741
7124 PR tree-optimization/71947
7125 * tree-ssa-dom.c: Include tree-inline.h.
7126 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7127 equivalences if one is more expensive to compute than the other.
7128 * tree-ssa-scopedtables.h (class const_or_copies): Make
7129 record_const_or_copy_raw method private.
7130 (class avail_exprs_stack): New method simplify_binary_operation.
7131 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7132 avail_exprs_stack::simplify_binary_operation as needed.
7133 (avail_exprs_stack::simplify_binary_operation): New function.
7135 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7137 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7138 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7139 (DOT_SYMBOLS): Likewise.
7140 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7141 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7142 (RS6000_ABI_NAME): Likewise.
7143 (TARGET_CMODEL): Likewise.
7144 (TOC_SECTION_ASM_OP): Likewise.
7145 (SET_CMODEL): New macro.
7146 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7148 2017-08-22 Richard Biener <rguenther@suse.de>
7150 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7151 to allow for free-lang-data replacements similar to verify_type_variant.
7153 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7154 Alan Hayward <alan.hayward@arm.com>
7155 David Sherwood <david.sherwood@arm.com>
7157 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7158 VOIDmode for the LABEL_REF.
7160 2017-08-22 Richard Biener <rguenther@suse.de>
7162 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7164 2017-08-22 Richard Biener <rguenther@suse.de>
7166 * common.opt (feliminate-dwarf2-dups): Ignore.
7167 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7168 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7169 same_die_p_wrap, compute_section_prefix,
7170 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7171 (comdat_symbol_id, comdat_symbol_number): Likewise.
7172 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7174 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7175 (output_die): Mark unreachable path unreachable.
7176 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7177 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7178 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7179 (dwarf2out_early_finish): Likewise.
7181 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7183 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7185 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7188 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7189 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7190 (avr_attribute_table) <io, io_low, address>: Initialize
7191 .decl_required with true.
7193 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7195 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7196 undocumented debugging options.
7197 (-mvsx-scalar-double): Likewise.
7198 (-mallow-df-permute): Likewise.
7199 (-mvectorize-builtins): Likewise.
7200 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7201 (rs6000_builtin_vectorized_function): Likewise.
7202 (rs6000_builtin_md_vectorized_function): Likewise.
7203 (rs6000_opt_vars): Likewise.
7205 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7208 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7209 (*btrq_imm): Rename from *btrq.
7210 (*btcq_imm): Rename from *btcq.
7211 (btsc): New code attribute.
7212 (*<btsc><mode>): New insn pattern.
7213 (*btr<mode>): Ditto.
7214 (*<btsc><mode>_mask): New insn_and_split pattern.
7215 (*btr<mode>_mask): Ditto.
7217 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7218 Alan Hayward <alan.hayward@arm.com>
7219 David Sherwood <david.sherwood@arm.com>
7221 * function.c (pad_below): Simplify padding calculation.
7223 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7224 Alan Hayward <alan.hayward@arm.com>
7225 David Sherwood <david.sherwood@arm.com>
7227 * target.def (function_prologue): Remove frame size argument.
7228 (function_epilogue): Likewise.
7229 * doc/tm.texi: Regenerate.
7230 * final.c (final_start_function): Update call to function_prologue.
7231 (final_end_function): Update call to function_epilogue.
7232 (default_function_pro_epilogue): Remove frame size argument.
7233 * output.h (default_function_pro_epilogue): Likewise.
7234 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7235 (arm_output_function_prologue): Likewise.
7236 * config/frv/frv.c (frv_function_prologue): Likewise.
7237 (frv_function_epilogue): Likewise.
7238 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7239 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7240 (ia64_output_function_epilogue): Likewise.
7241 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7242 (m32r_output_function_epilogue): Likewise.
7243 * config/microblaze/microblaze.c (microblaze_function_prologue)
7244 (microblaze_function_epilogue): Likewise.
7245 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7246 (mips_output_function_epilogue): Likewise.
7247 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7248 (mmix_target_asm_function_epilogue): Likewise.
7249 * config/msp430/msp430.c (msp430_start_function): Likewise.
7250 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7251 (nds32_asm_function_epilogue): Likewise.
7252 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7253 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7254 (pa_output_function_epilogue): Likewise.
7255 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7256 (rs6000_output_function_epilogue): Likewise.
7257 * config/rl78/rl78.c (rl78_start_function): Likewise.
7258 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7259 (rs6000_output_function_epilogue): Likewise.
7260 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7261 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7262 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7263 (sparc_asm_function_epilogue): Likewise.
7265 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7267 * tree.h (type_has_mode_precision_p): New function.
7268 * convert.c (convert_to_integer_1): Use it.
7269 * expr.c (expand_expr_real_2): Likewise.
7270 (expand_expr_real_1): Likewise.
7271 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7272 * match.pd: Likewise.
7273 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7274 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7275 * tree-tailcall.c (process_assignment): Likewise.
7276 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7277 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7278 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7279 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7280 (vectorizable_assignment): Likewise.
7281 (vectorizable_shift): Likewise.
7282 (vectorizable_operation): Likewise.
7283 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7285 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7287 * match.pd: Add pow (C, x) simplification.
7289 2017-08-21 Richard Biener <rguenther@suse.de>
7291 PR tree-optimization/81900
7292 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7293 for blocks with abnormal predecessors.
7294 (compute_antic): Do not set visited flag prematurely.
7296 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7299 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7301 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7303 * stor-layout.h (vector_type_mode): Move to...
7304 * tree.h (vector_type_mode): ...here.
7305 * stor-layout.c (vector_type_mode): Move to...
7306 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7308 2017-08-21 Richard Biener <rguenther@suse.de>
7310 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7311 register_external_die hooks.
7312 (debug_false_tree_charstarstar_uhwistar): Declare.
7313 (debug_nothing_tree_charstar_uhwi): Likewise.
7314 * debug.c (do_nothing_debug_hooks): Adjust.
7315 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7316 (debug_nothing_tree_charstar_uhwi): Likewise.
7317 * dbxout.c (dbx_debug_hooks): Adjust.
7318 (xcoff_debug_hooks): Likewise.
7319 * sdbout.c (sdb_debug_hooks): Likewise.
7320 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7321 * dwarf2out.c (macinfo_label_base): New global.
7322 (dwarf2out_register_external_die): New function for the
7323 register_external_die hook.
7324 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7325 (dwarf2_debug_hooks): Use them.
7326 (dwarf2_lineno_debug_hooks): Adjust.
7327 (struct die_struct): Add with_offset flag.
7328 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7329 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7330 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7331 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7332 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7333 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7334 defining section names for the early LTO debug variants.
7335 (reset_indirect_string): New helper.
7336 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7337 (print_dw_val): Add support for offsetted symbol references.
7338 (get_ultimate_context): Split out from is_cxx.
7339 (is_cxx): Use get_ultimate_context.
7340 (is_fortran): Add decl overload.
7341 (compute_comp_unit_symbol): Split out worker from
7342 compute_section_prefix.
7343 (compute_section_prefix): Call compute_comp_unit_symbol and
7344 set comdat_type_p here.
7345 (output_die): Skip DIE symbol output for the LTO added one.
7346 Handle DIE symbol references with offset.
7347 (output_comp_unit): Guard section name mangling properly.
7348 For LTO debug sections emit a symbol at the section beginning
7349 which we use to refer to its DIEs.
7350 (add_abstract_origin_attribute): For DIEs registered via
7351 dwarf2out_register_external_die directly refer to the early
7352 DIE rather than indirectly through the shadow one we created.
7353 Remove obsolete call to dwarf2out_abstract_function for
7354 non-function/block origins.
7355 (gen_array_type_die): When generating early LTO debug do
7356 not emit DW_AT_string_length.
7357 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7358 late when in LTO. As suggested place a gcc_unreachable for
7359 the DECL_ABSTRACT_P case.
7360 (gen_subprogram_die): Avoid another specification DIE
7361 for early built declarations/definitions for the late LTO case.
7362 (gen_variable_die): Add type references for late duplicated VLA dies
7364 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7365 we have the abstract instance already.
7366 (process_scope_var): Adjust decl DIE contexts in LTO which
7367 first puts them in limbo.
7368 (gen_decl_die): Do not generate type DIEs late apart from
7369 types for VLAs or for decls we do not yet have a DIE. Do not
7370 call dwarf2out_abstract_function late.
7371 (dwarf2out_early_global_decl): Make sure to create DIEs
7372 for abstract instances of a decl first.
7373 (dwarf2out_late_global_decl): Adjust comment.
7374 (output_macinfo_op): With multiple macro sections use
7375 macinfo_label_base to distinguish labels.
7376 (output_macinfo): Likewise. Update macinfo_label_base.
7377 Pass in the line info label.
7378 (note_variable_value_in_expr): When generating LTO resolve
7379 all variable values here by generating DIEs as needed.
7380 (init_sections_and_labels): Add early LTO debug flag parameter
7381 and generate different sections and names if set. Add generation
7382 counter for the labels so we can have multiple of them.
7383 (reset_dies): Helper to allow DIEs to be output multiple times.
7384 (dwarf2out_finish): When outputting DIEs to the fat part of an
7385 LTO object first reset DIEs.
7386 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7387 (modified_type_die): Check for decl_ultimate_origin being self
7389 (gen_type_die_with_usage): Likewise.
7390 (gen_typedef_die): Allow decl_ultimate_origin being self.
7391 (set_decl_abstract_flags): Remove.
7392 (set_block_abstract_flags): Likewise.
7393 (dwarf2out_abstract_function): Treat the early generated DIEs
7394 as the abstract copy and only add DW_AT_inline and
7395 DW_AT_artificial here and call set_decl_origin_self.
7396 If the DIE has an abstract origin don't do anything.
7397 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7398 if we have none yet (Go fails to build one, PR78628).
7399 (variably_modified_type_p): Prevent endless recursion for Ada
7400 cyclic pointer types.
7401 * lto-streamer-in.c: Include debug.h.
7402 (dref_queue): New global.
7403 (lto_read_tree_1): Stream in DIE references.
7404 (lto_input_tree): Register DIE references.
7405 (input_function): Stream DECL_DEBUG_ARGS.
7406 * lto-streamer-out.c: Include debug.h.
7407 (lto_write_tree_1): Output DIE references.
7408 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7409 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7410 (output_function): Stream DECL_DEBUG_ARGS.
7411 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7412 Stream DECL_ABSTRACT_ORIGIN.
7413 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7414 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7415 DECL_CONTEXT for file-scope decls.
7416 * lto-streamer.h (struct dref_entry): Declare.
7417 (dref_queue): Likewise.
7418 * cfgexpand.c (pass_expand::execute): Do not call the
7419 outlining_inline_function hook here.
7420 * lto-wrapper.c (debug_obj): New global.
7421 (tool_cleanup): Unlink it if required.
7422 (debug_objcopy): New function.
7423 (run_gcc): Handle early debug sections in the IL files by
7424 extracting them to separate files, partially linkin them and
7425 feeding the result back as result to the linker.
7426 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7427 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7428 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7429 sections into a separate segment.
7430 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7432 (darwin_asm_dwarf_section): Likewise.
7433 (darwin_asm_output_dwarf_offset): Likewise.
7434 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7436 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7437 Alan Hayward <alan.hayward@arm.com>
7438 David Sherwood <david.sherwood@arm.com>
7440 * read-md.h (md_reader::record_potential_iterator_use): Replace
7441 pointer argument with an rtx and an index.
7442 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7443 (apply_mode_iterator): Likewise.
7444 (apply_code_iterator): Likewise.
7445 (apply_int_iterator): Likewise.
7446 (apply_subst_iterator): Likewise.
7447 (record_iterator_use): Likewise.
7448 (record_attribute_use): Likewise.
7449 (md_reader::record_potential_iterator_use): Likewise. Update calls
7450 to record_iterator_use and apply_iterator.
7451 (iterator_use): Replace ptr with x and index.
7452 (attribute_use): Likewise.
7453 (apply_attribute_uses): Update calls to apply_iterator.
7454 (apply_iterators): Likewise. Update initialization of iterator_use.
7455 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7456 and record_potential_iterator_use.
7457 (rtx_reader::read_rtx_operand): Likewise.
7459 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7460 Alan Hayward <alan.hayward@arm.com>
7461 David Sherwood <david.sherwood@arm.com>
7463 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7466 2017-08-21 Richard Biener <rguenther@suse.de>
7469 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7470 at struct end conservatively when comparing common bases.
7472 2017-08-21 Richard Biener <rguenther@suse.de>
7474 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7475 (mem_ref_in_stmt): Remove.
7476 (determine_max_movement): Use ref index to get at the reference.
7477 (invariantness_dom_walker::before_dom_children): Deal with
7478 lim data already initialized.
7479 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7481 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7483 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7484 (TARGET_ISA_ROUND): Ditto.
7485 (TARGET_ROUND): Ditto.
7486 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7487 * config/i386/i386.md: Ditto.
7488 * config/i386/sse.md: Ditto.
7489 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7490 with OPTION_MASK_ISA_SSE4_1.
7492 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7495 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7496 __builtin_ia32_lzcnt_u16.
7498 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7501 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7502 (rs6000_set_current_function): Rewrite function to use it.
7504 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7507 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7508 and TYPE_WARN_IF_NOT_ALIGN.
7509 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7510 (handle_warn_if_not_align): New.
7511 (place_union_field): Call handle_warn_if_not_align.
7512 (place_field): Call handle_warn_if_not_align.
7513 Copy TYPE_WARN_IF_NOT_ALIGN.
7514 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7515 (layout_type): Likewise.
7516 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7518 (tree_decl_common): Add warn_if_not_align.
7519 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7520 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7521 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7522 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7523 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7524 * doc/extend.texi: Document warn_if_not_aligned attribute.
7525 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7527 2017-08-17 Martin Liska <mliska@suse.cz>
7530 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7531 (get_data_dependence): Use it as pointer type.
7532 (distribute_loop): Likewise.
7534 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7536 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7537 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7538 (p8_vmrgow_<mode>_direct): New define_insn.
7539 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7540 handle endianness for vmrgew and vmrgow permute patterns.
7542 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7544 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7545 * config/rs6000/rs6000-cpus.def: Remove comment.
7546 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7547 (POWERPC_MASKS): Likewise.
7548 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7549 use of TARGET_VSX_TIMODE.
7550 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7551 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7552 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7553 (rs6000_option_override_internal): Remove dead code.
7554 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7555 (rs6000_legitimize_reload_address): Likewise.
7556 (rs6000_legitimate_address_p): Likewise.
7557 (rs6000_opt_masks): Delete "vsx-timode".
7558 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7559 from function comment.
7560 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7561 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7562 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7564 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7565 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7566 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7569 (define_peephole2): Likewise.
7571 2017-08-17 Martin Sebor <msebor@redhat.com>
7574 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7575 past the end of an array.
7576 (test_pp_format): Add test cases.
7578 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7580 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7581 missing ECF_NOTHROW flags.
7583 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7586 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7587 operands residing in integer registers.
7588 (*vsx_le_perm_load_<mode>): Likewise.
7589 (*vsx_le_perm_store_<mode>): Likewise.
7590 (define_peephole2): Add peepholes to optimize the above.
7592 2017-08-17 Marek Polacek <polacek@redhat.com>
7595 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7596 to mimic what shorten_compare did. Change the return type to bool.
7597 (fold_cond_expr_with_comparison): Update call to
7598 operand_equal_for_comparison_p.
7599 (fold_ternary_loc): Likewise.
7601 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7603 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7605 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7607 2017-08-17 Richard Biener <rguenther@suse.de>
7609 * tree-ssa-structalias.c (solve_graph): When propagating
7610 to successors update the graphs succ edges and avoid duplicate work.
7612 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7615 * config/i386/i386.c (ix86_option_override_internal): Save target
7616 specific options after ix86_stack_protector_guard_reg was changed.
7618 2017-08-17 Richard Biener <rguenther@suse.de>
7620 PR tree-optimization/81827
7621 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7622 (new_var_info): Initialize it conservatively.
7623 (get_call_vi): Mark register vars.
7624 (new_scalar_tmp_constraint_exp): Likewise.
7625 (handle_rhs_call): Likewise.
7626 (handle_const_call): Likewise.
7627 (create_function_info_for): Likewise.
7628 (solve_constraints): Sort varinfos to separate register from
7629 non-register vars to pack points-to solution bitmaps during
7632 2017-08-17 Marek Polacek <polacek@redhat.com>
7634 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7636 2017-08-17 Richard Biener <rguenther@suse.de>
7638 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7639 to true when overflow is undefined and we saturated the result.
7641 2017-08-17 Alan Modra <amodra@gmail.com>
7644 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7645 Don't use store multiple if only one reg needs saving.
7646 (interesting_frame_related_regno): New function.
7647 (rs6000_frame_related): Don't emit frame info for regs that
7649 (rs6000_emit_epilogue): Likewise.
7651 2017-08-16 Nathan Sidwell <nathan@acm.org>
7653 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7654 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7655 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7656 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7657 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7658 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7659 (verify_type): Adjust for TYPE_BINFO move.
7660 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7661 process TYPE_BINFO directly.
7662 (hash_tree): Likewise.
7663 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7665 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7668 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7670 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7672 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7675 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7676 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7677 Add ix86_binary_operator_ok to insn constraint.
7678 (*iordi_1_bts): Ditto.
7679 (*xordi_1_btc): Ditto.
7680 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7681 Update corresponding peephole2 pattern.
7685 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7687 PR tree-optimization/81832
7688 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7689 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7691 2017-08-16 Marek Polacek <polacek@redhat.com>
7694 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7695 regarding size_zero_node.
7697 2017-08-16 Martin Liska <mliska@suse.cz>
7700 * config.gcc: Respect previously set extra_objs in case
7703 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7705 PR tree-optimization/81835
7706 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7707 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7708 not depend on the phi.
7710 2017-08-16 Alan Modra <amodra@gmail.com>
7712 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7715 2017-08-16 Alan Modra <amodra@gmail.com>
7717 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7718 (save_reg_p): ..into this. Update all callers.
7719 (first_reg_to_save): Simplify.
7721 2017-08-16 Alan Modra <amodra@gmail.com>
7723 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7726 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7730 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7731 constant count if that count is less than 32.
7733 2017-08-15 Nathan Sidwell <nathan@acm.org>
7735 * gcc.c (execute): Emit friendlier message if inferior is killed
7736 by an external cause.
7738 2017-08-15 Richard Biener <rguenther@suse.de>
7740 PR tree-optimization/81790
7741 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7742 CONSTRUCTORs from simplifying and VN.
7744 2017-08-14 Martin Sebor <msebor@redhat.com>
7746 * builtin-attrs.def: Add comments.
7748 2017-08-14 Martin Sebor <msebor@redhat.com>
7751 * doc/extend.texi (attribute nonstring): Document new attribute.
7753 2017-08-14 Martin Sebor <msebor@redhat.com>
7756 * tree-diagnostic.c (default_tree_printer): Handle %G.
7757 * gimple-pretty-print.h (percent_G_format): Declare new function.
7758 * gimple-pretty-print.c (percent_G_format): Define.
7759 * tree-pretty-print.c (percent_K_format): Add argument.
7761 2017-08-14 Martin Sebor <msebor@redhat.com>
7763 PR translation/79998
7764 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7765 Remove a stray space.
7767 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7770 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7771 (*iordi_1_bts): Ditto.
7772 (*xordi_1_btc): Ditto.
7774 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7777 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7778 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7780 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7781 quoted strings, and make more translator-friendly.
7782 (darwin_rs6000_override_options): Likewise.
7783 (rs6000_option_override_internal): Likewise.
7784 (rs6000_return_in_memory): Fix overlong line.
7785 (init_cmulative_args): Use quoted strings, and make more
7786 translator-friendly.
7787 (rs6000_pass_by_reference): Fix overlong line.
7788 (def_builtin): Use quoted strings.
7789 (altivec_expand_predicate_builtin): Use quoted strings, and make
7790 more translator-friendly.
7791 (htm_expand_builtin): Use quoted strings.
7792 (cpu_expand_builtin): Use quoted strings, and make more
7793 translator-friendly.
7794 (altivec_expand_builtin): Likewise.
7795 (paired_expand_predicate_builtin): Likewise.
7796 (rs6000_invalid_builtin): Likewise.
7797 (builtin_function_type): Use quoted strings.
7798 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7799 more translator-friendly.
7800 (rs6000_trampoline_init): Likewise.
7801 (rs6000_handle_altivec_attribute): Likewise.
7802 (rs6000_inner_target_options): Use quoted strings.
7803 (rs6000_disable_incompatible_switches): Likewise.
7804 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7805 strings, and make more translator-friendly.
7806 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7808 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7810 PR tree-optimization/81799
7811 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7812 cond_expr to simple gimple operand.
7814 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7817 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7819 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7823 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7824 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7825 TARGET_GASISR_PROLOGUES.
7826 * config/avr/avr.c (avr_option_override): Same.
7827 (avr_pass_pre_proep::execute): Same.
7829 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7832 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7833 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7835 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7837 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7838 define_insn_and_split. Split to a memory load from 0 in
7839 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7840 using PTR mode iterator.
7841 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7842 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7843 (*add_tp_<mode>): Redefine as define_insn_and_split.
7844 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7845 address space. Merge with *add_tp_x32 using PTR mode iterator.
7846 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7847 Split to an add with a memory load from 0 in
7848 DEFAULT_TLS_SEG_REG address space.
7850 2017-08-12 Andrew Pinski <apinski@cavium.com>
7852 * config/aarch64/aarch64-option-extensions.def (rdma):
7853 Fix feature string to what Linux prints out in /proc/cpuinfo.
7855 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7858 * dwarf2out.c (modified_type_die): For C typedef types that have
7859 an ultimate origin, process the ultimate origin instead of the
7861 (gen_typedef_die): Assert that input DECLs have no ultimate
7863 (gen_type_die_with_usage): For typedef variants that have an
7864 ultimate origin, just call gen_decl_die on the original DECL.
7865 (process_scope_var): Avoid creating DIEs for local typedefs and
7866 concrete static variables.
7868 2017-08-12 Alan Modra <amodra@gmail.com>
7872 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7873 match gnu-user.h startfile.
7874 (ENDFILE_LINUX_SPEC): Similarly.
7876 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7879 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7881 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7883 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7884 * config/aarch64/aarch64.md (mov<mode>): Change.
7885 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7886 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7887 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7889 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7891 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7892 for storage order barriers.
7894 2017-08-11 Martin Liska <mliska@suse.cz>
7897 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7898 variables of void type.
7900 2017-08-11 Martin Liska <mliska@suse.cz>
7902 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7903 TARGET_SUPPORTS_ALIASES.
7904 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7905 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7906 (optimize_weakref): Likewise.
7907 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7908 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7909 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7911 2017-08-11 Martin Liska <mliska@suse.cz>
7914 * config/i386/i386.c (make_resolver_func): Do complete
7915 refactoring of the function.
7917 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7920 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7921 * config/i386/i386.c (ix86_stack_protect_guard): Use
7922 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7923 * doc/invoke.texi (x86 Options): Document
7924 -mstack-protector-guard-symbol= option.
7926 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7928 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7929 * config/i386/i386.c (ix86_split_stack_guard): New function.
7930 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7931 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7932 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7933 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7934 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7935 (split_stack_space_check): Call ix86_split_stack_guard.
7937 2017-08-10 Martin Sebor <msebor@redhat.com>
7939 * print-tree.c (print_node): Print location using the established
7941 Replace spaces with colons.
7942 (debug_raw, debug): Ditto.
7944 2017-08-10 Martin Sebor <msebor@redhat.com>
7947 * pretty-print.c (pp_format): Correct the handling of %s precision.
7949 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7952 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7954 (ix86_finalize_stack_frame_flags): This. Also clear
7955 frame_pointer_needed if -fno-omit-frame-pointer is used without
7957 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7958 with ix86_finalize_stack_frame_flags.
7959 (ix86_expand_epilogue): Likewise.
7960 (ix86_expand_split_stack_prologue): Likewise.
7961 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7963 2017-08-10 Martin Liska <mliska@suse.cz>
7966 * c-attribs.c (handle_target_attribute):
7967 Report warning for an empty string argument of target attribute.
7969 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7972 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7974 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7975 or DECL_NONLOCAL labels.
7976 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7977 or DECL_NONLOCAL labels here.
7979 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7981 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7982 to indicate when early gimple folding has been disabled.
7983 (rs6000_gimple_fold_builtin): Add debug content.
7984 (rs6000_invalid_builtin): Fix whitespace.
7985 (rs6000_expand_builtin): Fix whitespace.
7986 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7988 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7991 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7992 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7994 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7996 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7997 (qdf24xx): Likewise.
7998 * config/aarch64/aarch64.md: Include falkor.md.
7999 * config/aarch64/falkor.md: New.
8001 2017-08-09 Marek Polacek <polacek@redhat.com>
8004 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8005 * diagnostic.c (emit_diagnostic): Add a comment.
8006 (emit_diagnostic_valist): New function.
8008 2017-08-09 Marek Polacek <polacek@redhat.com>
8011 * input.c (make_location): New overload.
8012 * input.h (make_location): Declare.
8014 2017-08-08 Alan Modra <amodra@gmail.com>
8015 H.J. Lu <hongjiu.lu@intel.com>
8018 * gcc.c (NO_PIE_SPEC): Delete.
8019 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8021 (LINK_PIE_SPEC): ..to here.
8022 (LINK_COMMAND_SPEC): Support -no-pie.
8023 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8024 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8026 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8027 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8028 (ENDFILE_CRTEND_SPEC): Similarly.
8030 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8033 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8034 (mstack-protector-guard-offset=): Ditto.
8035 * config/i386/i386.c (ix86_option_override): Handle
8036 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8038 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8039 ix86_stack_protect_guard_offset variables.
8040 (TARGET_STACK_PROTECT_GUARD): Always define.
8041 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8042 and -mstack-protector-guard-offset= options.
8044 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8046 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8047 boundary case for the last candidate.
8049 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8051 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8052 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8054 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8057 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8058 * config/aarch64/aarch64-builtins.c
8059 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8060 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8061 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8063 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8064 Andrew Pinski <pinskia@gmail.com>
8067 * internal-fn.def (XORSIGN): New.
8068 * optabs.def (xorsign_optab): New.
8069 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8070 (convert_expand_mult_copysign): New.
8071 (pass_optimize_widening_mul::execute): Call
8072 convert_expand_mult_copysign.
8074 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8076 PR tree-optimization/81354
8077 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8078 Insert on edges rather than explicitly creating landing pads.
8079 (analyze_candidates_and_replace): Commit edge inserts.
8081 2017-08-08 Richard Biener <rguenther@suse.de>
8084 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8085 (expand_simple_operations): Also look through ADDR_EXPRs with
8086 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8088 2017-08-08 Richard Biener <rguenther@suse.de>
8090 PR tree-optimization/81723
8091 * tree-vect-slp.c (struct bst_traits): New hash traits.
8092 (bst_fail): New global.
8093 (vect_build_slp_tree_2): New worker, split out from ...
8094 (vect_build_slp_tree): ... this now wrapping it with using
8095 bst_fail set to cache SLP tree build fails. Properly handle
8097 (vect_analyze_slp_instance): Allocate and free bst_fail.
8099 2017-08-08 Martin Liska <mliska@suse.cz>
8102 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8103 LABEL_DECLs that can be from a different function.
8105 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8107 PR tree-optimization/81744
8108 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8109 loop's number of iterations.
8111 2017-08-08 Martin Liska <mliska@suse.cz>
8113 * asan.c: Include header files.
8114 * attribs.c (build_decl_attribute_variant): New function moved
8116 (build_type_attribute_qual_variant): Likewise.
8117 (cmp_attrib_identifiers): Likewise.
8118 (simple_cst_list_equal): Likewise.
8119 (omp_declare_simd_clauses_equal): Likewise.
8120 (attribute_value_equal): Likewise.
8121 (comp_type_attributes): Likewise.
8122 (build_type_attribute_variant): Likewise.
8123 (lookup_ident_attribute): Likewise.
8124 (remove_attribute): Likewise.
8125 (merge_attributes): Likewise.
8126 (merge_type_attributes): Likewise.
8127 (merge_decl_attributes): Likewise.
8128 (merge_dllimport_decl_attributes): Likewise.
8129 (handle_dll_attribute): Likewise.
8130 (attribute_list_equal): Likewise.
8131 (attribute_list_contained): Likewise.
8132 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8133 (lookup_attribute_by_prefix): Likewise.
8134 * bb-reorder.c: Include header files.
8135 * builtins.c: Likewise.
8136 * calls.c: Likewise.
8137 * cfgexpand.c: Likewise.
8138 * cgraph.c: Likewise.
8139 * cgraphunit.c: Likewise.
8140 * convert.c: Likewise.
8141 * dwarf2out.c: Likewise.
8142 * final.c: Likewise.
8143 * fold-const.c: Likewise.
8144 * function.c: Likewise.
8145 * gimple-expr.c: Likewise.
8146 * gimple-fold.c: Likewise.
8147 * gimple-pretty-print.c: Likewise.
8148 * gimple.c: Likewise.
8149 * gimplify.c: Likewise.
8150 * hsa-common.c: Likewise.
8151 * hsa-gen.c: Likewise.
8152 * internal-fn.c: Likewise.
8153 * ipa-chkp.c: Likewise.
8154 * ipa-cp.c: Likewise.
8155 * ipa-devirt.c: Likewise.
8156 * ipa-fnsummary.c: Likewise.
8157 * ipa-inline.c: Likewise.
8158 * ipa-visibility.c: Likewise.
8160 * lto-cgraph.c: Likewise.
8161 * omp-expand.c: Likewise.
8162 * omp-general.c: Likewise.
8163 * omp-low.c: Likewise.
8164 * omp-offload.c: Likewise.
8165 * omp-simd-clone.c: Likewise.
8166 * opts-global.c: Likewise.
8167 * passes.c: Likewise.
8168 * predict.c: Likewise.
8169 * sancov.c: Likewise.
8170 * sanopt.c: Likewise.
8171 * symtab.c: Likewise.
8172 * toplev.c: Likewise.
8173 * trans-mem.c: Likewise.
8174 * tree-chkp.c: Likewise.
8175 * tree-eh.c: Likewise.
8176 * tree-into-ssa.c: Likewise.
8177 * tree-object-size.c: Likewise.
8178 * tree-parloops.c: Likewise.
8179 * tree-profile.c: Likewise.
8180 * tree-ssa-ccp.c: Likewise.
8181 * tree-ssa-live.c: Likewise.
8182 * tree-ssa-loop.c: Likewise.
8183 * tree-ssa-sccvn.c: Likewise.
8184 * tree-ssa-structalias.c: Likewise.
8185 * tree-ssa.c: Likewise.
8186 * tree-streamer-in.c: Likewise.
8187 * tree-vectorizer.c: Likewise.
8188 * tree-vrp.c: Likewise.
8190 * ubsan.c: Likewise.
8191 * varasm.c: Likewise.
8192 * varpool.c: Likewise.
8193 * tree.c: Remove functions moved to attribs.[ch].
8195 * config/aarch64/aarch64.c: Add attrs.h header file.
8196 * config/alpha/alpha.c: Likewise.
8197 * config/arc/arc.c: Likewise.
8198 * config/arm/arm.c: Likewise.
8199 * config/avr/avr.c: Likewise.
8200 * config/bfin/bfin.c: Likewise.
8201 * config/c6x/c6x.c: Likewise.
8202 * config/cr16/cr16.c: Likewise.
8203 * config/cris/cris.c: Likewise.
8204 * config/darwin.c: Likewise.
8205 * config/epiphany/epiphany.c: Likewise.
8206 * config/fr30/fr30.c: Likewise.
8207 * config/frv/frv.c: Likewise.
8208 * config/ft32/ft32.c: Likewise.
8209 * config/h8300/h8300.c: Likewise.
8210 * config/i386/winnt.c: Likewise.
8211 * config/ia64/ia64.c: Likewise.
8212 * config/iq2000/iq2000.c: Likewise.
8213 * config/lm32/lm32.c: Likewise.
8214 * config/m32c/m32c.c: Likewise.
8215 * config/m32r/m32r.c: Likewise.
8216 * config/m68k/m68k.c: Likewise.
8217 * config/mcore/mcore.c: Likewise.
8218 * config/microblaze/microblaze.c: Likewise.
8219 * config/mips/mips.c: Likewise.
8220 * config/mmix/mmix.c: Likewise.
8221 * config/mn10300/mn10300.c: Likewise.
8222 * config/moxie/moxie.c: Likewise.
8223 * config/msp430/msp430.c: Likewise.
8224 * config/nds32/nds32-isr.c: Likewise.
8225 * config/nds32/nds32.c: Likewise.
8226 * config/nios2/nios2.c: Likewise.
8227 * config/nvptx/nvptx.c: Likewise.
8228 * config/pa/pa.c: Likewise.
8229 * config/pdp11/pdp11.c: Likewise.
8230 * config/powerpcspe/powerpcspe.c: Likewise.
8231 * config/riscv/riscv.c: Likewise.
8232 * config/rl78/rl78.c: Likewise.
8233 * config/rx/rx.c: Likewise.
8234 * config/s390/s390.c: Likewise.
8235 * config/sh/sh.c: Likewise.
8236 * config/sol2.c: Likewise.
8237 * config/sparc/sparc.c: Likewise.
8238 * config/spu/spu.c: Likewise.
8239 * config/stormy16/stormy16.c: Likewise.
8240 * config/tilegx/tilegx.c: Likewise.
8241 * config/tilepro/tilepro.c: Likewise.
8242 * config/v850/v850.c: Likewise.
8243 * config/vax/vax.c: Likewise.
8244 * config/visium/visium.c: Likewise.
8245 * config/xtensa/xtensa.c: Likewise.
8247 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8250 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8251 constraints since the -mupper-regs-* switches have been
8253 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8254 into a vector from a double word element that was extracted from
8255 another vector, and eliminate extra XXPERMDI instructions.
8256 (vsx_concat_<mode>_2): Likewise.
8257 (vsx_concat_<mode>_3): Likewise.
8258 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8259 concat to allow optimizing inserts from previous extracts.
8261 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8263 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8264 memory reference to a SSP offset in TLS address space.
8265 (ix86_print_operand) <case '@'>: Remove.
8266 (ix86_print_operand_punct_valid_p): Remove '@' code.
8267 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8269 (stack_tls_protect_set_<mode>): Remove.
8270 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8271 (stack_tls_protect_test_<mode>): Remove.
8272 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8274 2017-08-07 Olivier Hainque <hainque@adacore.com>
8277 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8279 2017-08-07 Douglas Rupp <rupp@adacore.com>
8281 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8282 variable was referenced as multidir in command.
8284 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8287 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8290 2017-08-07 Martin Liska <mliska@suse.cz>
8292 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8293 * config/rl78/rl78.c: Add include of attribs.h.
8294 * config/sh/sh.c: Likewise.
8295 * config/v850/v850.c: Likewise.
8297 2017-08-07 Tom de Vries <tom@codesourcery.com>
8300 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8302 2017-08-07 Martin Liska <mliska@suse.cz>
8304 * config/mips/mips.c: Include attribs.h.
8306 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8309 * doc/invoke.texi: Document change in behvaior for -Ofast for
8312 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8314 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8316 (aarch64_pop_regs): Likewise.
8317 (aarch64_gen_load_pair): Likewise.
8318 (aarch64_save_callee_saves): Likewise.
8319 (aarch64_restore_callee_saves): Likewise.
8321 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8323 * config/i386/i386.c: Revert the last change.
8325 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8328 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8330 (ix86_finalize_stack_frame_flags): This. Also clear
8331 frame_pointer_needed if -fno-omit-frame-pointer is used without
8333 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8334 with ix86_finalize_stack_frame_flags.
8335 (ix86_expand_epilogue): Likewise.
8336 (ix86_expand_split_stack_prologue): Likewise.
8338 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8341 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8342 to P_AES for Westmere.
8344 2017-08-07 Jonathan Yong <10walls@gmail.com>
8346 * config/i386/mingw.opt (fset-stack-executable): Removed.
8347 * config/i386/cygming.opt (fset-stack-executable): Moved
8349 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8351 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8353 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8355 2017-08-07 Marek Polacek <polacek@redhat.com>
8358 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8360 2017-08-07 Martin Liska <mliska@suse.cz>
8362 * attribs.h (canonicalize_attr_name): New function.
8363 (cmp_attribs): Move from c-format.c and adjusted.
8364 (is_attribute_p): Moved from tree.h.
8365 * tree-inline.c: Add new includes.
8366 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8367 (private_is_attribute_p): Remove.
8368 (private_lookup_attribute): Likewise.
8369 (private_lookup_attribute_by_prefix): Simplify.
8370 (remove_attribute): Use is_attribute_p.
8371 * tree.h: Remove removed declarations.
8373 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8376 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8377 instead of computing it in the function. Formatting fix.
8378 (expand_case): Don't rely on default_edge being the first edge,
8379 clear it if removing it, pass default_edge to
8380 emit_case_dispatch_table.
8381 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8384 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8386 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8387 insn in the function, emit NOP after the insn.
8389 2017-08-06 Tom de Vries <tom@codesourcery.com>
8391 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8394 2017-08-06 Tom de Vries <tom@codesourcery.com>
8396 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8399 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8401 PR tree-optimization/57371
8402 * match.pd: New pattern.
8404 2017-08-04 Marek Polacek <polacek@redhat.com>
8407 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8408 perform the computation in offset_int.
8410 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8412 PR tree-optimization/81136
8413 * tree-vectorizer.h: Include tree-hash-traits.h.
8414 (vec_base_alignments): New typedef.
8415 (vec_info): Add a base_alignments field.
8416 (vect_record_base_alignments): Declare.
8417 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8419 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8420 (create_data_ref): Add an is_conditional_in_stmt argument.
8421 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8422 the is_conditional_in_stmt field.
8423 (data_ref_loc): Add an is_conditional_in_stmt field.
8424 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8425 (find_data_references_in_stmt): Update call to create_data_ref.
8426 (graphite_find_data_references_in_stmt): Likewise.
8427 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8428 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8429 (vect_record_base_alignment): New function.
8430 (vect_record_base_alignments): Likewise.
8431 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8432 for nested statements even if we fail to compute a misalignment.
8433 Use pooled base alignments for unconditional references.
8434 (vect_find_same_alignment_drs): Compare base addresses instead
8436 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8437 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8439 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8441 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8442 Add an explicit name for the enum. Use auto_vec for slp_instances
8444 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8446 (_bb_vec_info): Add a constructor and destructor.
8447 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8448 (destroy_loop_vec_info): Delete.
8449 (vect_destroy_datarefs): Likewise.
8450 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8451 (vec_info::vec_info): New function.
8452 (vec_info::~vec_info): Likewise.
8453 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8454 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8455 destroy_loop_vec_info.
8456 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8457 (_loop_vec_info::_loop_vec_info): ...this.
8458 (destroy_loop_vec_info): Replace with...
8459 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8460 the stmt_vec_infos. Leave handling of vec_info information to its
8461 destructor. Remove explicit vector releases.
8462 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8463 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8464 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8465 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8466 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8467 (destroy_bb_vec_info): Replace with...
8468 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8469 information to its destructor.
8470 (vect_slp_analyze_bb_1): Use new and delete instead of
8471 new_bb_vec_info and destroy_bb_vec_info.
8472 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8475 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8477 * tree-data-ref.h (subscript): Add access_fn field.
8478 (data_dependence_relation): Add could_be_independent_p.
8479 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8480 (same_access_functions): Move to tree-data-ref.c.
8481 * tree-data-ref.c (ref_contains_union_access_p): New function.
8482 (access_fn_component_p): Likewise.
8483 (access_fn_components_comparable_p): Likewise.
8484 (dr_analyze_indices): Add a reference to access_fn_component_p.
8485 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8487 (constant_access_functions): Likewise.
8488 (add_other_self_distances): Likewise.
8489 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8490 (initialize_data_dependence_relation): Use XCNEW and remove
8491 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8492 of access functions that have the same type. Allow the
8493 subsequence to end with different bases in some circumstances.
8494 Record the chosen access functions in SUB_ACCESS_FN.
8495 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8496 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8497 (subscript_dependence_tester_1): Likewise dra and drb.
8498 (build_classic_dist_vector): Update calls accordingly.
8499 (subscript_dependence_tester): Likewise.
8500 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8501 DDR_COULD_BE_INDEPENDENT_P.
8502 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8503 comp_alias_ddrs instead of may_alias_ddrs.
8504 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8506 (vect_analyze_data_ref_dependence): Use it if
8507 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8508 distance vectors if that fails.
8509 (dependence_distance_ge_vf): New function.
8510 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8511 LOOP_VINFO_MAY_ALIAS_DDRS.
8513 2017-08-04 Richard Biener <rguenther@suse.de>
8516 * fold-const.c (fold_binary_loc): Properly restrict
8517 minus_var0 && minus_var1 case when associating undefined overflow
8520 2017-08-04 Tom de Vries <tom@codesourcery.com>
8522 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8524 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8527 Don't start diagnostic messages with a capital letter.
8528 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8530 (rs6000_invalid_builtin): Likewise.
8531 (rs6000_trampoline_init): Likewise.
8533 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8536 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8537 after setting changeable df flags.
8539 2017-08-03 Richard Biener <rguenther@suse.de>
8541 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8542 up if the use is in USE - X.
8544 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8546 * toplev.c (dumpfile.h): New include.
8547 (internal_error_reentered): New static function. Use it...
8548 (internal_error_function): ...here to handle reentered internal_error.
8550 2017-08-03 Richard Biener <rguenther@suse.de>
8553 * fold-const.c (split_tree): Add minus_var and minus_con
8554 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8555 here but always use minus_*.
8556 (associate_trees): Assert we never associate with MINUS_EXPR
8557 and NULL first operand. Do not recurse for PLUS_EXPR operands
8558 when associating as MINUS_EXPR either.
8559 (fold_binary_loc): Track minus_var and minus_con.
8561 2017-08-03 Tom de Vries <tom@codesourcery.com>
8564 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8565 ACCEL_COMPILER, apply finish_options on
8566 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8568 2017-08-03 Tom de Vries <tom@codesourcery.com>
8571 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8572 function_entry_patch_area_size > 0.
8574 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8577 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8578 instead of 10??LU, perform unit multiplication in wide_int,
8579 don't change alloc_object_size_limit if the limit is larger
8582 PR tree-optimization/81655
8583 PR tree-optimization/81588
8584 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8585 the case when ranges[i].low and high are 1 for unsigned type with
8589 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8590 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8592 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8594 * tree-vrp.h: Add include guard.
8596 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8599 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8600 (ud2): New insn pattern.
8601 * config/i386/i386.c (ix86_expand_epilogue):
8602 For naked functions, generate ud2 instead of trap insn.
8604 2017-08-02 Marek Polacek <polacek@redhat.com>
8607 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8609 2017-08-02 Tom de Vries <tom@codesourcery.com>
8610 Cesar Philippidis <cesar@codesourcery.com>
8612 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8613 Add missing edge probabilities.
8615 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8617 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8620 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8623 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8624 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8625 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8626 prologue_seq sequences - if any.
8628 2017-08-02 Richard Biener <rguenther@suse.de>
8630 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8631 via vectors if supported, integer extracts via punning if supported
8632 or otherwise vector extracts.
8634 2017-08-02 Richard Biener <rguenther@suse.de>
8636 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8638 (bitmap_insert_into_set): ... this.
8640 2017-08-02 Richard Biener <rguenther@suse.de>
8642 PR tree-optimization/81633
8644 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8646 PR tree-optimization/71752
8647 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8649 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8651 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8652 (machine_function::call_ms2sysv_pad_out): Remove field.
8653 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8654 (ix86_compute_frame_layout): Likewise.
8656 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8659 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8660 attribute with interrupt attribute.
8662 2017-08-01 Andrew Pinski <apinski@cavium.com>
8664 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8665 BIT_INSERT_EXPR's operand 1
8666 to see if the types precision matches.
8668 2017-08-01 Martin Liska <mliska@suse.cz>
8671 * builtins.c (expand_builtin_memcpy_args): Remove.
8672 (expand_builtin_memcpy): Call newly added function
8673 expand_builtin_memory_copy_args.
8674 (expand_builtin_memcpy_with_bounds): Likewise.
8675 (expand_builtin_mempcpy): Remove last argument.
8676 (expand_builtin_mempcpy_with_bounds): Likewise.
8677 (expand_builtin_memory_copy_args): New function created from
8678 expand_builtin_mempcpy_args with small modifications.
8679 (expand_builtin_mempcpy_args): Remove.
8680 (expand_builtin_stpcpy): Remove unused argument.
8681 (expand_builtin): Likewise.
8682 (expand_builtin_with_bounds): Likewise.
8684 2017-08-01 Martin Liska <mliska@suse.cz>
8687 Make mempcpy more optimal (PR middle-end/70140).
8689 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8692 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8693 __builtin_vec_cmpne verify both arguments are compatible vectors
8694 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8695 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8696 move computation of aligned to after checking the argument types.
8700 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8701 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8704 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8705 Doug Rupp <rupp@adacore.com>
8706 Olivier Hainque <hainque@adacore.com>
8708 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8709 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8711 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8712 for TARGET_OS_CPP_BUILTIN.
8713 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8714 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8716 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8717 passing required abi options to the assembler for EABI configurations.
8718 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8719 of .text.hot and .text.unlikely sections for kernel modules when
8720 using ARM style exceptions.
8721 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8722 options. Add EXTRA_CC1_SPEC.
8723 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8725 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8727 (ARM_TARGET2_DWARF_FORMAT): Define.
8728 * config/arm/t-vxworks: Adjust multilib control to removal of the
8729 Diab command line options.
8731 2017-08-01 Martin Liska <mliska@suse.cz>
8733 PR gcov-profile/81561
8734 * gcov.c (unblock): Make unblocking safe as we need to preserve
8735 index correspondence of blocks and block_lists.
8737 2017-08-01 Richard Biener <rguenther@suse.de>
8739 PR tree-optimization/81181
8740 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8741 (compute_antic): ... end of iteration here.
8743 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8745 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8746 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8747 (ftree-slp-vectorize): Likewise.
8748 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8749 can no longer be set independent of flag_tree_loop_vectorize.
8750 * omp-general.c (emp_max_vf): Likewise.
8751 * opts.c (enable_fdo_optimizations): Remove references to
8752 flag_tree_vectorize, these are now implicit.
8753 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8754 and leave it for the options machinery.
8756 2017-08-01 Martin Liska <mliska@suse.cz>
8759 * builtins.c (expand_builtin_memcpy_args): Remove.
8760 (expand_builtin_memcpy): Call newly added function
8761 expand_builtin_memory_copy_args.
8762 (expand_builtin_memcpy_with_bounds): Likewise.
8763 (expand_builtin_mempcpy): Remove last argument.
8764 (expand_builtin_mempcpy_with_bounds): Likewise.
8765 (expand_builtin_memory_copy_args): New function created from
8766 expand_builtin_mempcpy_args with small modifications.
8767 (expand_builtin_mempcpy_args): Remove.
8768 (expand_builtin_stpcpy): Remove unused argument.
8769 (expand_builtin): Likewise.
8770 (expand_builtin_with_bounds): Likewise.
8772 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8775 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8776 print "ds:" only for immediates in generic address space.
8778 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8781 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8782 (ix86_function_ok_for_sibcall): Return false for naked functions.
8784 2017-08-01 Richard Biener <rguenther@suse.de>
8786 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8787 (compute_antic): Seed worklist with exit block predecessors.
8788 * cfganal.c (dfs_find_deadend): For a cycle return the source
8789 of the edge closing it.
8791 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8793 * config/aarch64/aarch64.c
8794 (aarch64_can_const_movi_rtx_p): Move 0 check.
8796 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8798 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8799 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8801 * match.pd: Ditto in address comparison pattern.
8803 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8805 PR tree-optimization/81627
8806 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8807 closed ssa form for store-store chain.
8809 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8811 PR tree-optimization/81620
8812 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8813 for store-store chain.
8815 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8817 PR tree-optimization/81588
8818 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8819 ranges[i].in_p, invert comparison code ccode. For >/>=,
8820 swap rhs1 and rhs2 and comparison code unconditionally,
8821 for </<= don't do that. Don't swap rhs1/rhs2 again if
8822 ranges[i].in_p, instead invert comparison code ccode if
8823 opcode or oe->rank is BIT_IOR_EXPR.
8826 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8827 a direct optab to conversion optab.
8828 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8829 with GET_MODE_INNER as last argument instead of optab_handler.
8830 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8831 vector extraction if possible and optab is available.
8832 * expr.c (store_constructor): Use convert_optab_handler instead
8833 of optab_handler. Use vector initialization from smaller
8834 vectors if possible and optab is available.
8835 * tree-vect-stmts.c (vectorizable_load): Likewise.
8836 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8838 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8839 of vec_init from half-sized vectors with the same element mode.
8840 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8841 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8842 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8843 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8844 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8845 after mode in gen_vec_extract* calls.
8846 (vec_extract<mode>): Renamed to ...
8847 (vec_extract<mode><ssescalarmodelower>): ... this.
8848 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8849 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8850 element mode after mode in gen_vec_init* calls.
8851 (VEC_INIT_HALF_MODE): New mode iterator.
8852 (vec_init<mode>): Renamed to ...
8853 (vec_init<mode><ssescalarmodelower>): ... this.
8854 (vec_init<mode><ssehalfvecmodelower>): New expander.
8855 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8856 (vec_extractv2sfsf): ... this.
8857 (vec_initv2sf): Renamed to ...
8858 (vec_initv2sfsf): ... this.
8859 (vec_extractv2si): Renamed to ...
8860 (vec_extractv2sisi): ... this.
8861 (vec_initv2si): Renamed to ...
8862 (vec_initv2sisi): ... this.
8863 (vec_extractv4hi): Renamed to ...
8864 (vec_extractv4hihi): ... this.
8865 (vec_initv4hi): Renamed to ...
8866 (vec_initv4hihi): ... this.
8867 (vec_extractv8qi): Renamed to ...
8868 (vec_extractv8qiqi): ... this.
8869 (vec_initv8qi): Renamed to ...
8870 (vec_initv8qiqi): ... this.
8871 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8872 (vec_init<mode>): Renamed to ...
8873 (vec_init<mode><VEC_base_l>): ... this.
8874 (vec_extract<mode>): Renamed to ...
8875 (vec_extract<mode><VEC_base_l>): ... this.
8876 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8877 (vec_initv2sfsf): ... this.
8878 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8879 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8880 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8881 element mode after mode in gen_vec_init* calls.
8882 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8883 (vec_init<mode><Vel>): ... this.
8884 (vec_extract<mode>): Renamed to ...
8885 (vec_extract<mode><Vel>): ... this.
8886 * config/aarch64/iterators.md (Vel): New mode attribute.
8887 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8888 Add element mode after mode in gen_vec_extract* calls.
8889 * config/s390/vector.md (non_vec_l): New mode attribute.
8890 (vec_extract<mode>): Renamed to ...
8891 (vec_extract<mode><non_vec_l>): ... this.
8892 (vec_init<mode>): Renamed to ...
8893 (vec_init<mode><non_vec_l>): ... this.
8894 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8895 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8897 * config/arm/iterators.md (V_elem_l): New mode attribute.
8898 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8899 (vec_extract<mode><V_elem_l>): ... this.
8900 (vec_extractv2di): Renamed to ...
8901 (vec_extractv2didi): ... this.
8902 (vec_init<mode>): Renamed to ...
8903 (vec_init<mode><V_elem_l>): ... this.
8904 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8905 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8906 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8907 Add element mode after gen_vec_extract* calls.
8908 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8909 (vec_init<mode><unitmode>): ... this.
8910 (vec_extract<mode>): Renamed to ...
8911 (vec_extract<mode><unitmode>): ... this.
8912 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8913 (vec_init<mode><unitmode>): ... this.
8914 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8915 (vec_initv2sfsf): ... this.
8916 (vec_extractv2sf): Renamed to ...
8917 (vec_extractv2sfsf): ... this.
8918 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8919 Add element mode after gen_vec_extract* calls.
8920 * config/mips/mips.md (unitmode): New mode iterator.
8921 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8922 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8923 * config/spu/spu.md (inner_l): New mode attribute.
8924 (vec_init<mode>): Renamed to ...
8925 (vec_init<mode><inner_l>): ... this.
8926 (vec_extract<mode>): Renamed to ...
8927 (vec_extract<mode><inner_l>): ... this.
8928 * config/sparc/sparc.md (veltmode): New mode iterator.
8929 (vec_init<VMALL:mode>): Renamed to ...
8930 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8931 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8932 (vec_initv2sisi): ... this.
8933 (vec_initv2sf): Renamed to ...
8934 (vec_initv2sfsf): ... this.
8935 (vec_extractv2sf): Renamed to ...
8936 (vec_extractv2sfsf): ... this.
8937 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8938 (vec_init<mode>): Renamed to ...
8939 (vec_init<mode><VEC_base_l>): ... this.
8940 (vec_extract<mode>): Renamed to ...
8941 (vec_extract<mode><VEC_base_l>): ... this.
8942 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8943 (vec_initv2sfsf): ... this.
8944 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8945 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8946 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8947 gen_vec_init* calls.
8949 2017-08-01 Richard Biener <rguenther@suse.de>
8951 PR tree-optimization/81297
8952 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8953 TREE_OVERFLOW from INTEGER_CSTs.
8955 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8957 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8959 2017-07-31 Carl Love <cel@us.ibm.com>
8961 * config/rs6000/rs6000-c: Add support for built-in functions
8962 vector signed char vec_xl_be (signed long long, signed char *);
8963 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8964 vector signed int vec_xl_be (signed long long, signed int *);
8965 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8966 vector signed long long vec_xl_be (signed long long, signed long long *);
8967 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8968 vector signed short vec_xl_be (signed long long, signed short *);
8969 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8970 vector double vec_xl_be (signed long long, double *);
8971 vector float vec_xl_be (signed long long, float *);
8972 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8973 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8974 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8976 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8977 (altivec_expand_builtin): Add switch statement to call
8978 altivec_expand_xl_be for each builtin.
8979 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8980 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8981 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8982 __builtin_vsx_le_be_v16qi.
8983 * doc/extend.texi: Update the built-in documentation file for the
8984 new built-in functions.
8986 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8989 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8991 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8993 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8995 * config.gcc: Add z14.
8996 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8997 CPU model numbers for z13s and z14.
8998 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9000 * config/s390/s390-opts.h (enum processor_type): Rename
9001 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9002 * config/s390/s390.c (processor_table): Add field for CPU name to
9003 be passed to Binutils.
9004 (s390_asm_output_machine_for_arch): Use the new field in
9005 processor_table for Binutils.
9006 (s390_expand_builtin): Replace arch12 with z14.
9007 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9008 (s390_get_sched_attrmask): Likewise.
9009 (s390_get_unit_mask): Likewise.
9010 * config/s390/s390.opt: Add z14 to processor_type enum.
9012 2017-07-31 Martin Jambor <mjambor@suse.cz>
9015 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9016 regardless of optimization level.
9018 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9019 Martin Liska <mliska@suse.cz>
9021 * predict.def: Remove old comment and adjust probability.
9022 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9025 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9028 * config/i386/i386.c (ix86_function_naked): New function.
9029 (ix86_can_use_return_insn_p): Return false for naked functions.
9030 (ix86_expand_prologue): Skip prologue for naked functions.
9031 (ix86_expand_epilogue): Skip epilogue for naked functions
9032 and emit trap instruction.
9033 (ix86_warn_func_return): New function.
9034 (ix86_attribute_table): Add "naked" attribute specification.
9035 (TARGET_WARN_FUNC_RETURN): Define.
9036 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9038 2017-07-31 Martin Liska <mliska@suse.cz>
9040 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9041 (dump_gimple_bb_header): Always dump BB info.
9042 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9044 2017-07-31 Martin Liska <mliska@suse.cz>
9047 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9048 also with current_function_decl non-null equality.
9050 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9053 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9054 change type to the element type, instead add eltype variable and
9055 use it where we are interested in the element type.
9057 PR tree-optimization/81603
9058 * ipa-polymorphic-call.c
9059 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9060 offset arithmetic in offset_int, bail out if the resulting bit offset
9061 doesn't fit into shwi.
9063 2017-07-31 Martin Liska <mliska@suse.cz>
9065 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9066 (gimplify_save_expr): Fix comment.
9068 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9071 * config/i386/i386.c (ix86_function_arg): Update arguments for
9073 (ix86_compute_frame_layout): Set the initial stack offset to
9074 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9075 INCOMING_FRAME_SP_OFFSET.
9076 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9077 stack before exception handler returns.
9078 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9079 the 'ERROR_CODE' for exception handler.
9081 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9083 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9084 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9085 (ASM_OUTPUT_REG_POP): Ditto.
9086 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9087 instead of asm_fprintf to output pure string.
9089 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9091 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9092 to imported_module_or_decl hook.
9093 (debug_nothing_tree_tree_tree_bool): Remove.
9094 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9095 * debug.c (do_nothing_debug_hooks): Use
9096 debug_nothing_tree_tree_tree_bool_bool instead of
9097 debug_nothing_tree_tree_tree_bool.
9098 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9099 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9100 * sdbout.c (sdb_debug_hooks): Likewise.
9101 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9102 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9104 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9105 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9106 attribute, don't add anything.
9108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9110 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9111 (fold_build2_stat_loc): Likewise.
9112 (fold_build3_stat_loc): Likewise.
9113 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9114 (fold_build1_loc): Remove macro.
9115 (fold_build2_loc): Likewise.
9116 (fold_build3_loc): Likewise.
9118 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9120 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9121 (gimple_build_debug_bind_source_stat): Likewise.
9122 * gimple.h (gimple_build_debug_bind): Remove macro.
9123 (gimple_build_debug_bind_source): Likewise.
9125 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9127 * bitmap.c (bitmap_alloc): Adjust.
9128 (bitmap_gc_alloc): Likewise.
9129 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9131 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9133 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9134 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9135 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9136 (bitmap_gc_alloc_stat): Likewise.
9137 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9141 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9142 * rtl.h (shallow_copy_rtx): Remove macro.
9144 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9146 * emit-rtl.c (gen_raw_REG): Adjust.
9147 * gengenrtl.c (gendef): Likewise.
9148 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9149 * rtl.h (rtx_alloc): Remove macro.
9151 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9153 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9154 (build_tree_list_stat): Likewise.
9155 * tree.h (build_tree_list): Remove macro.
9156 (build_tree_list_vec): Likewise.
9158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9160 * tree.c (make_vector_stat): Remove _stat from name.
9161 (build_vector_stat): Likewise.
9162 * tree.h (make_vector_stat): Remove macro.
9163 (build_vector_stat): Likewise.
9165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9167 * tree.h (build_var_debug_value): Remove prototype.
9169 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9171 * tree.c (tree_cons_stat): Remove _stat from name.
9172 * tree.h (tree_cons): Remove macro.
9174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9176 * tree.c (build_vl_exp_stat): Remove _stat from name.
9177 * tree.h (build_vl_exp): Remove macro.
9179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9181 * tree.c (build_decl_stat): Remove _stat from name.
9182 * tree.h (build_decl): Remove macro.
9184 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9186 * gimple.c (gimple_build_with_ops_stat): Adjust.
9187 (gimple_alloc_stat): Remove _stat from name.
9188 * gimple.h (gimple_alloc): Remove macro.
9190 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9192 * tree.c (make_tree_vec_stat): Remove _stat from name.
9193 (grow_tree_vec_stat): Likewise.
9194 * tree.h (make_tree_vec_stat): Adjust prototype.
9195 (grow_tree_vec_stat): Likewise.
9196 (make_tree_vec): Remove macro.
9197 (grow_tree_vec): Likewise.
9199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9201 * fold-const.c (fold_build1_stat_loc): Adjust.
9202 (fold_build2_stat_loc): Likewise.
9203 (fold_build3_stat_loc): Likewise.
9204 * tree.c (build0_stat): Remove _stat from name.
9205 (build1_stat): Likewise.
9206 (build2_stat): Likewise.
9207 (build3_stat): Likewise.
9208 (build4_stat): Likewise.
9209 (build5_stat): Likewise.
9210 * tree.h (build1_loc): Remove macro, and rename _stat function
9212 (build2_loc): Likewise.
9213 (build3_loc): Likewise.
9214 (build4_loc): Likewise.
9215 (build5_loc): Likewise.
9217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9219 * tree.c (make_int_cst_stat): Remove _stat from name.
9220 * tree.h (make_int_cst_stat): Adjust prototype.
9221 (make_int_cst): Remove macro.
9223 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9225 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9226 * tree.h (make_tree_binfo_stat): Adjust prototype.
9227 (make_tree_binfo): Remove.
9229 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9231 * tree.c (copy_node_stat): Rename to copy_node.
9232 (build_distinct_type_copy): Adjust.
9233 * tree.h (copy_node_stat): Adjust prototype.
9234 (copy_node): Remove macro.
9236 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9238 * tree.c (make_node_stat): rename to make_node.
9239 (build_tree_list_stat): Adjust.
9240 (build0_stat): Likewise.
9241 (build2_stat): Likewise.
9242 (build3_stat): Likewise.
9243 (build4_stat): Likewise.
9244 (build5_stat): Likewise.
9245 (build_decl_stat): Likewise.
9246 * tree.h (make_node_stat): Adjust prototype.
9247 (make_node): remove macro.
9249 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9251 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9252 (PPC_FEATURE2_SCV): Likewise.
9253 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9255 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9257 * config/aarch64/aarch64.c
9258 (aarch64_internal_mov_immediate): Add new special pattern.
9259 * config/aarch64/aarch64.md (*movdi_aarch64):
9260 Add reg/32bit const mov case.
9262 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9263 Richard Sandiford <richard.sandiford@linaro.org>
9265 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9266 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9267 Add integer and movi cases.
9268 (movi-split-hf-df-sf split, fp16): New.
9269 (enabled): Added TARGET_FP_F16INST.
9270 * config/aarch64/iterators.md (GPF_HF): New.
9271 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9273 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9275 * config/aarch64/aarch64.c
9276 (aarch64_simd_container_mode): Add prototype.
9277 (aarch64_expand_mov_immediate): Add HI support.
9278 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9279 (aarch64_can_const_movi_rtx_p): New.
9280 (aarch64_preferred_reload_class):
9281 Remove restrictions of using FP registers for certain SIMD operations.
9282 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9283 (aarch64_valid_floating_const): Add integer move validation.
9284 (aarch64_simd_imm_scalar_p): Remove.
9285 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9286 (aarch64_legitimate_constant_p): Expand list of supported cases.
9287 * config/aarch64/aarch64-protos.h
9288 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9289 (aarch64_reinterpret_float_as_int): New.
9290 (aarch64_simd_imm_scalar_p): Remove.
9291 * config/aarch64/constraints.md (Uvi): New.
9292 (Dd): Split into Ds and new Dd.
9293 * config/aarch64/aarch64.md (*movsi_aarch64):
9295 (*movdi_aarch64): Add SIMD mov case.
9297 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9299 * tree-predcom.c: (struct chain): Handle store-store chain in which
9300 stores for elimination only store loop invariant values.
9301 (execute_pred_commoning_chain): Ditto.
9302 (prepare_initializers_chain_store_elim): Ditto.
9303 (prepare_finalizers): Ditto.
9304 (is_inv_store_elimination_chain): New function.
9305 (initialize_root_vars_store_elim_1): New function.
9307 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9309 * tree-predcom.c: Revise general description of the pass.
9310 (enum chain_type): New enum type for store elimination.
9311 (struct chain): New field supporting store elimination.
9312 (struct component): Ditto.
9313 (dump_chain): Dump store-stores chain.
9314 (release_chain): Release resources.
9315 (split_data_refs_to_components): Compute and create component
9316 contains only stores for elimination.
9317 (get_chain_last_ref_at): New function.
9318 (make_invariant_chain): Initialization.
9319 (make_rooted_chain): Specify chain type in parameter and record it.
9320 (add_looparound_copies): Skip for store-stores chain.
9321 (determine_roots_comp): Compute type of chain and pass it to
9323 (initialize_root_vars_store_elim_2): New function.
9324 (finalize_eliminated_stores): New function.
9325 (remove_stmt): Handle store for elimination.
9326 (execute_pred_commoning_chain): Execute predictive commoning on
9328 (determine_unroll_factor): Skip unroll for store-stores chain.
9329 (prepare_initializers_chain_store_elim): New function.
9330 (prepare_initializers_chain): Hanlde store-store chain.
9331 (prepare_finalizers_chain, prepare_finalizers): New function.
9332 (tree_predictive_commoning_loop): Return integer value indicating
9333 if loop is unrolled or lcssa form is corrupted.
9334 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9336 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9338 * tree-predcom.c (initialize_root): Delete.
9339 (execute_pred_commoning_chain): Initialize root vars and replace
9340 reference of non-combined chain directly, rather than call above
9343 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9345 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9346 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9348 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9350 * tree-predcom.c (struct chain): New field init_seq.
9351 (release_chain): Release init_seq.
9352 (prepare_initializers_chain): Record intialization stmts in above
9354 (insert_init_seqs): New function.
9355 (tree_predictive_commoning_loop): Call insert_init_seqs.
9357 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9359 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9361 2017-07-28 Richard Biener <rguenther@suse.de>
9363 * match.pd: Remove superfluous :c.
9364 * genmatch.c (simplify::id): Add member.
9365 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9367 (current_id): New global.
9368 (dt_node::parent): Move from ...
9369 (dt_operand::parent): ... here. Add for_id member.
9370 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9371 (decision_tree::find_node): Relax order requirement when
9372 merging DT_TRUE nodes to ones inbetween the current simplify
9373 and the one we try to merge with. Add diagnostic whenever
9374 we need to enforce pattern order by not merging.
9375 (decision_tree::insert): Set current_id.
9376 (decision_tree::print_node): Dump parent node and for_id.
9377 (parser::last_id): Add member.
9378 (parser::push_simplify): Assign unique id.
9379 (parser::parser): Initialize last_id.
9381 2017-07-28 Martin Liska <mliska@suse.cz>
9384 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9385 gimple_build_debug_bind.
9387 2017-07-28 Richard Biener <rguenther@suse.de>
9389 PR tree-optimization/81502
9390 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9392 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9394 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9395 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9396 for BIT_FIELD_REF args.
9397 * fold-const.c (make_bit_field_ref): Likewise.
9398 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9400 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9403 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9404 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9405 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9406 Or it into SANITIZER_UNDEFINED.
9407 * ubsan.c: Include gimple-fold.h and varasm.h.
9408 (ubsan_expand_ptr_ifn): New function.
9409 (instrument_pointer_overflow): New function.
9410 (maybe_instrument_pointer_overflow): New function.
9411 (instrument_object_size): Formatting fix.
9412 (pass_ubsan::execute): Call instrument_pointer_overflow
9413 and maybe_instrument_pointer_overflow.
9414 * internal-fn.c (expand_UBSAN_PTR): New function.
9415 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9416 * sanitizer.def (__ubsan_handle_pointer_overflow,
9417 __ubsan_handle_pointer_overflow_abort): New builtins.
9418 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9419 * internal-fn.def (UBSAN_PTR): New internal function.
9420 * opts.c (sanitizer_opts): Add pointer-overflow.
9421 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9422 * fold-const.c (build_range_check): Compute pointer range check in
9423 integral type if pointer arithmetics would be needed. Formatting
9426 2017-07-28 Martin Liska <mliska@suse.cz>
9429 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9430 parameters that are of a variable-length.
9432 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9434 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9436 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9437 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9438 (CRT_CALL_STATIC_FUNCTION): Likewise.
9439 (ASM_DEFAULT_SPEC): New define.
9440 (ASM_SPEC32): Likewise.
9441 (ASM_SPEC64): Likewise.
9442 (ASM_SPEC_COMMON): Likewise.
9443 (ASM_SPEC): Likewise.
9444 (INVALID_64BIT): Likewise.
9445 (LINK_OS_DEFAULT_SPEC): Likewise.
9446 (LINK_OS_SPEC32): Likewise.
9447 (LINK_OS_SPEC64): Likewise.
9448 (POWERPC_LINUX): Likewise.
9449 (PTRDIFF_TYPE): Likewise.
9450 (RESTORE_FP_PREFIX): Likewise.
9451 (RESTORE_FP_SUFFIX): Likewise.
9452 (SAVE_FP_PREFIX): Likewise.
9453 (SAVE_FP_SUFFIX): Likewise.
9454 (SIZE_TYPE): Likewise.
9455 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9456 (TARGET_64BIT): Likewise.
9457 (TARGET_64BIT): Likewise.
9458 (TARGET_AIX): Likewise.
9459 (WCHAR_TYPE_SIZE): Likewise.
9460 (WCHAR_TYPE): Undefine.
9461 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9462 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9463 (CPP_OS_RTEMS_SPEC): Delete.
9464 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9465 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9467 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9469 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9471 PR tree-optimization/81578
9472 * tree-parloops.c (build_new_reduction): Bail out if
9473 reduction_code isn't one of the standard OpenMP reductions.
9474 Move the details printing after that decision.
9476 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9478 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9479 related to reload_in_progress.
9480 (splat_input_operand): Likewise.
9481 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9483 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9485 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9486 (TARGET_INSTANTIATE_DECLS): Likewise.
9487 (legitimate_indexed_address_p): Delete reload_in_progress code.
9488 (rs6000_debug_legitimate_address_p): Likewise.
9489 (rs6000_eliminate_indexed_memrefs): Likewise.
9490 (rs6000_emit_le_vsx_store): Likewise.
9491 (rs6000_emit_move_si_sf_subreg): Likewise.
9492 (rs6000_emit_move): Likewise.
9493 (register_to_reg_type): Likewise.
9494 (rs6000_pre_atomic_barrier): Likewise.
9495 (rs6000_machopic_legitimize_pic_address): Likewise.
9496 (rs6000_allocate_stack_temp): Likewise.
9497 (rs6000_address_for_fpconvert): Likewise.
9498 (rs6000_address_for_altivec): Likewise.
9499 (rs6000_secondary_memory_needed_rtx): Delete function.
9500 (rs6000_check_sdmode): Likewise.
9501 (rs6000_alloc_sdmode_stack_slot): Likewise.
9502 (rs6000_instantiate_decls): Likewise.
9503 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9504 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9505 Delete reload_in_progress.
9506 (*vec_reload_and_plus_<mptrsize>): Likewise.
9507 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9508 (vsx_div_v2di): Likewise.
9509 (vsx_udiv_v2di): Likewise.
9511 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9513 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9514 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9515 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9516 (rs6000_debug_reg_global): Delete print of LRA status.
9517 (rs6000_option_override_internal): Delete dead LRA related code.
9518 (rs6000_lra_p): Delete function.
9519 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9521 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9523 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9524 * config/riscv/rtems.h: New file.
9526 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9527 Sudakshina Das <sudi.das@arm.com>
9529 * config/aarch64/aarch64.md
9530 (define_split for and<mode>3nr_compare): Move
9531 non aarch64_logical_operand to a register.
9532 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9533 register immediate operand to a register.
9534 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9536 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9539 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9541 2017-07-27 Richard Biener <rguenther@suse.de>
9543 PR tree-optimization/81573
9544 PR tree-optimization/81494
9545 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9546 multi defuse cycle case.
9548 2017-07-27 Richard Biener <rguenther@suse.de>
9550 PR tree-optimization/81571
9551 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9554 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9556 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9557 earlier and only if MASK_FPU is set. Adjust formatting.
9559 2017-07-27 Martin Liska <mliska@suse.cz>
9561 * opt-functions.awk: Add validation of value of Init.
9562 * optc-gen.awk: Pass new argument.
9564 2017-07-27 Martin Liska <mliska@suse.cz>
9566 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9567 Fix wrong condition.
9569 2017-07-27 Martin Liska <mliska@suse.cz>
9571 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9572 BBs and edges seen by autoFDO.
9574 2017-07-27 Richard Biener <rguenther@suse.de>
9576 PR tree-optimization/81502
9577 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9578 with incompatible but same sized type.
9579 (execute_update_addresses_taken): Likewise.
9581 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9583 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9584 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9586 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9589 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9590 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9591 Change s_operand to memory_operand.
9593 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9595 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9596 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9597 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9598 Emit instructions rather than returning an expression. Handle TFmode
9599 and KFmode by casting to TImode.
9600 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9601 (rs6000_emit_le_vsx_store): Likewise.
9602 * config/rs6000/vsx.md (VSX_TI): New iterator.
9603 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9604 (*vsx_le_undo_permute_<mode>): Likewise.
9605 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9606 emit the split sequence.
9607 (*vsx_le_perm_store_<mode>): Likewise.
9609 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9611 PR tree-optimization/81555
9612 PR tree-optimization/81556
9613 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9614 if true, force CHANGED for the recursive invocation.
9615 (reassociate_bb): Remember original length of ops array, pass
9616 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9618 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9619 attributes for noipa attribute. For naked attribute use
9620 lookup_attribute first before lookup_attribute_spec.
9621 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9623 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9624 for functions with noipa attribute.
9625 (cgraph_externally_visible_p): Return true for functions with noipa
9627 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9628 for functions with noipa attribute.
9629 * doc/extend.texi: Document noipa function attribute.
9630 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9631 also for functions with noipa attribute.
9632 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9634 2017-07-26 Andrew Pinski <apinski@cavium.com>
9636 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9637 vec_unalign_load_cost and vec_unalign_store_cost.
9639 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9641 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9642 -mvsx-small-integer option.
9643 (ISA_3_0_MASKS_IEEE): Likewise.
9644 (OTHER_VSX_VECTOR_MASKS): Likewise.
9645 (POWERPC_MASKS): Likewise.
9646 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9647 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9648 code, only testing for DImode being allowed in non-VSX floating
9650 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9651 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9653 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9654 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9655 TARGET_P8_VECTOR test.
9656 (rs6000_secondary_reload_simple_move): Likewise.
9657 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9658 since TARGET_P9_VECTOR was already tested.
9659 (rs6000_opt_masks): Remove -mvsx-small-integer.
9660 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9661 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9663 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9664 test for TARGET_VEXTRACTUB was used, and that uses
9666 (p9 extract splitter): Likewise.
9667 (vsx_extract_<mode>_di_p9): Likewise.
9668 (vsx_extract_<mode>_store_p9): Likewise.
9669 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9670 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9671 the elimination of TARGET_VSX_SMALL_INTEGER.
9672 (vsx_extract_<mode>_p8): Likewise.
9673 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9674 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9675 (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.
9680 * config/rs6000/rs6000.md (sign extend splitter): Change
9681 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9682 (floatsi<mode>2_lfiwax_mem): Likewise.
9683 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9684 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9685 since a test for TARGET_P9_VECTOR was used.
9686 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9687 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9688 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9689 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9690 TARGET_P8_VECTOR test.
9691 (fix_trunc<mode>si2_stfiwx): Likewise.
9692 (fix_trunc<mode>si2_internal): Likewise.
9693 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9694 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9696 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9697 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9698 TARGET_P8_VECTOR test.
9699 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9700 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9701 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9703 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9704 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9705 since a test for TARGET_P9_VECTOR was used.
9706 (splitter for loading small constants): Likewise.
9708 2017-07-26 Andrew Pinski <apinski@cavium.com>
9710 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9713 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9716 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9717 (fp_valid_at): Likewise.
9719 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9721 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9722 (qdf24xx_addrcost_table): Likewise.
9723 (cortexa57_tunings): Update to use generic_branch_cost.
9724 (cortexa72_tunings): Likewise.
9725 (cortexa73_tunings): Likewise.
9726 (qdf24xx_tunings): Likewise.
9728 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9730 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9731 (thunderx2t99_branch_cost): Likewise.
9732 (cortexa35_tunings): Update to use generic_branch_cost.
9733 (cortexa53_tunings): Likewise.
9734 (cortexa57_tunings): Likewise.
9735 (cortexa72_tunings): Likewise.
9736 (cortexa73_tunings): Likewise.
9737 (thunderx2t99_tunings): Likewise.
9739 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9741 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9742 (sparc_option_override): Honour MASK_FSMULD.
9743 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9744 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9745 * config/sparc/sparc.opt (mfsmuld): New option.
9746 * doc/invoke.texi (mfsmuld): Document option.
9748 2017-07-26 Marek Polacek <polacek@redhat.com>
9751 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9753 2017-07-26 Richard Biener <rguenther@suse.de>
9755 * gimple-match-head.c (do_valueize): Return OP if valueize
9757 (get_def): New helper to get at the def stmt of a SSA name
9759 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9760 do_valueize to get at the def stmt.
9761 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9763 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9766 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9768 2017-07-26 Martin Liska <mliska@suse.cz>
9771 * function.c (expand_function_start): Make expansion of
9772 nonlocal_goto_save_area after parm_birth_insn.
9774 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9776 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9777 from all CPU target flags enable members.
9779 2017-07-26 Richard Biener <rguenther@suse.de>
9781 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9782 (decision_tree::gen): Make 'type' const.
9783 (write_predicate): Likewise.
9785 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9787 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9788 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9789 (rs6000_option_override_internal): Likewise.
9790 (rs6000_expand_vector_set): Likewise.
9791 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9792 (TARGET_UPPER_REGS_SF): Likewise.
9793 (TARGET_UPPER_REGS_DI): Likewise.
9794 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9795 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9796 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9797 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9798 (Splitters for DI constants in Altivec registers): Likewise.
9799 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9800 (vsx_set_v4sf_p9): Likewise.
9801 (vsx_set_v4sf_p9_zero): Likewise.
9802 (vsx_insert_extract_v4sf_p9): Likewise.
9803 (vsx_insert_extract_v4sf_p9_2): Likewise.
9805 2017-07-25 Carl Love <cel@us.ibm.com>
9807 * doc/extend.texi: Update the built-in documentation file for the
9808 existing built-in functions
9809 vector signed char vec_cnttz (vector signed char);
9810 vector unsigned char vec_cnttz (vector unsigned char);
9811 vector signed short vec_cnttz (vector signed short);
9812 vector unsigned short vec_cnttz (vector unsigned short);
9813 vector signed int vec_cnttz (vector signed int);
9814 vector unsigned int vec_cnttz (vector unsigned int);
9815 vector signed long long vec_cnttz (vector signed long long);
9816 vector unsigned long long vec_cnttz (vector unsigned long long);
9818 2017-07-25 Andrew Pinski <apinski@cavium.com>
9820 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9821 accesses where the use is for the first operand of a BIT_INSERT.
9823 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9826 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9827 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9829 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9831 * config/i386/gstabs.h: Delete.
9832 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9834 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9836 * config/i386/i386.c (ix86_decompose_address): Do not check for
9837 register RTX when looking at index_reg or base_reg.
9838 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9840 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9842 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9843 to update EH info here.
9845 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9847 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9849 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9851 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9853 2017-07-25 Torsten Duwe <duwe@suse.de>
9855 * common.opt: Introduce -fpatchable-function-entry
9856 command line option, and its variables function_entry_patch_area_size
9857 and function_entry_patch_area_start.
9858 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9859 including a two-value parser.
9860 * target.def (print_patchable_function_entry): New target hook.
9861 * targhooks.h (default_print_patchable_function_entry): New function.
9862 * targhooks.c (default_print_patchable_function_entry): Likewise.
9863 * toplev.c (process_options): Switch off IPA-RA if
9864 patchable function entries are being generated.
9865 * varasm.c (assemble_start_function): Look at the
9866 patchable-function-entry command line switch and current
9867 function attributes and maybe generate NOP instructions by
9868 calling the print_patchable_function_entry hook.
9869 * doc/extend.texi: Document patchable_function_entry attribute.
9870 * doc/invoke.texi: Document -fpatchable_function_entry
9871 command line option.
9872 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9874 * doc/tm.texi: Re-generate.
9876 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9879 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9880 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9882 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9884 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9886 2017-07-25 Richard Biener <rguenther@suse.de>
9888 PR tree-optimization/81455
9889 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9890 not walk in cycles when looking for guards.
9892 2017-07-25 Richard Biener <rguenther@suse.de>
9894 PR tree-optimization/81529
9895 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9896 when optimizing backedge uses.
9898 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9900 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9902 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9903 to dl_section_ref. On AIX, append an expression to subtract
9904 the size of the section length to dl_section_ref.
9906 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9908 * configure.ac: If any of the config.* scripts fail, exit 1.
9909 * configure: Regenerate.
9911 2017-07-25 Richard Biener <rguenther@suse.de>
9914 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9915 of immediate uses, be more verbose on errors.
9917 2017-07-25 Richard Biener <rguenther@suse.de>
9919 PR tree-optimization/81510
9920 * tree-vect-loop.c (vect_is_simple_reduction): When the
9921 reduction stmt is not inside the loop bail out.
9923 2017-07-25 Richard Biener <rguenther@suse.de>
9925 PR tree-optimization/81303
9926 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9927 profitability check against LOOP_VINFO_NITERSM1.
9929 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9931 * domwalk.c (cmp_bb_postorder): Simplify.
9932 (sort_bbs_postorder): New function. Use it...
9933 (dom_walker::walk): ...here to optimize common cases.
9935 2017-07-25 Martin Liska <mliska@suse.cz>
9938 * ipa-visibility.c (function_and_variable_visibility): Make the
9939 redirection just on target that supports aliasing.
9940 Fix GNU coding style.
9942 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9945 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9947 * config/arm/rtems.h: Likewise.
9948 * config/bfin/rtems.h: Likewise.
9949 * config/i386/rtemself.h: Likewise.
9950 * config/lm32/rtems.h: Likewise.
9951 * config/m32c/rtems.h: Likewise.
9952 * config/m68k/rtemself.h: Likewise.
9953 * config/microblaze/rtems.h: Likewise.
9954 * config/mips/rtems.h: Likewise.
9955 * config/moxie/rtems.h: Likewise.
9956 * config/nios2/rtems.h: Likewise.
9957 * config/powerpcspe/rtems.h: Likewise.
9958 * config/rs6000/rtems.h: Likewise.
9959 * config/rtems.h: Likewise.
9960 * config/sh/rtems.h: Likewise.
9961 * config/sh/rtemself.h: Likewise.
9962 * config/sparc/rtemself.h: Likewise.
9964 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9967 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9968 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9969 * tree-ssa-structalias.c (alias_get_name): Same.
9971 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9974 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9975 instructions if no du chain is found.
9977 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9979 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9981 2017-07-25 Richard Biener <rguenther@suse.de>
9984 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9987 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9989 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9991 (ISA_2_7_MASKS_SERVER): Likewise.
9992 (ISA_3_0_MASKS_IEEE): Likewise.
9993 (OTHER_P8_VECTOR_MASKS): Likewise.
9994 (OTHER_VSX_VECTOR_MASKS): Likewise.
9995 (POWERPC_MASKS): Likewise.
9996 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9997 duplicate list of options.
9998 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9999 explicit -mupper-regs options.
10000 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10001 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10002 alias for -mupper-regs-df.
10003 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10004 (rs6000_init_hard_regno_mode_ok): Likewise.
10005 (rs6000_option_override_internal): Likewise.
10006 (rs6000_opt_masks): Likewise.
10007 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10008 options in terms of whether -mvsx or -mpower8-vector was used.
10009 (TARGET_UPPER_REGS_DI): Likewise.
10010 (TARGET_UPPER_REGS_SF): Likewise.
10011 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10012 -mupper-regs-* options.
10014 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10016 * passes.c (emergency_dump_function): Print some empty lines and a
10017 header before the RTL dump.
10019 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10021 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10023 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10026 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10027 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10029 2017-07-24 Carl Love <cel@us.ibm.com>
10031 * config/rs6000/rs6000-c.c: Add support for built-in functions
10032 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10033 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10034 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10035 vec_extract_fp_from_shortl): Add defines for the two builtins.
10036 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10037 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10039 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10040 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10041 * doc/extend.texi: Update the built-in documentation file for the
10042 new built-in function.
10044 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10047 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10049 * doc/generic.texi: Likewise.
10050 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10051 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10053 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10055 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10056 (aarch64_mls_elt_merge<mode>): Likewise.
10058 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10060 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10061 having __cxa_atexit.
10063 2017-07-23 Michael Collison <michael.collison@arm.com>
10065 * config/arm/arm.c (arm_option_override): Deprecate
10066 use of -mstructure-size-boundary.
10067 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10068 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10070 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10073 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10074 Reduce cost estimate for direct moves.
10076 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10079 * config/i386/i386.c (ix86_option_override_internal): Disable
10080 BMI, BMI2 and TBM instructions for -m16.
10082 2017-07-21 Carl Love <cel@us.ibm.com>
10084 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10085 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10086 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10087 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10088 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10089 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10090 VMULOSW): New enum "unspec" values.
10091 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10092 altivec_vmulosw): New patterns.
10093 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10094 VMULOSW): Add definitions.
10096 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10098 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10099 (qdf24xx): Likewise.
10100 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10101 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10102 (AARCH64_FL_V8_1): Renumber.
10103 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10104 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10105 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10106 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10107 rdma to feature modifiers list.
10109 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10111 PR middle-end/56727
10112 * ipa-visibility (function_and_variable_visibility): Convert
10113 recursive PLT call to direct call if appropriate.
10115 2017-07-21 Andrew Pinski <apinski@cavium.com>
10117 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10118 operand 1 to see if the types precision matches.
10119 * fold-const.c (operand_equal_p): Likewise.
10121 2017-07-21 Richard Biener <rguenther@suse.de>
10123 PR tree-optimization/81303
10124 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10125 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10126 (vect_peeling_hash_get_lowest_cost): Adjust.
10127 (vect_enhance_data_refs_alignment): Likewise. Use
10128 vect_get_peeling_costs_all_drs to compute the penalty for no
10129 peeling to match up costs.
10131 2017-07-21 Richard Biener <rguenther@suse.de>
10133 PR tree-optimization/81500
10134 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10135 we didn't identify a reduction path.
10137 2017-07-21 Tom de Vries <tom@codesourcery.com>
10138 Cesar Philippidis <cesar@codesourcery.com>
10140 PR gcov-profile/81442
10141 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10144 2017-07-21 Tom de Vries <tom@codesourcery.com>
10147 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10149 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10150 nvptx_override_options_after_change.
10152 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10154 * dwarf2out.c (output_file_names): Avoid double testing for
10155 dwarf_version >= 5.
10157 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10159 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10161 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10163 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10165 (try_crossjump_to_edge): Do not punt on partitioned functions.
10167 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10169 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10170 Put all BBs reachable only via paths crossing cold region to cold
10172 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10174 2016-07-21 Richard Biener <rguenther@suse.de>
10176 PR tree-optimization/81303
10177 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10178 into account prologue and epilogue iterations when raising
10179 min_profitable_iters to sth at least covering one vector iteration.
10181 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10183 * config/arm/arm.c (arm_test_cpu_arch_dat):
10186 2017-07-20 Nathan Sidwell <nathan@acm.org>
10188 Remove TYPE_METHODS.
10189 * tree.h (TYPE_METHODS): Delete.
10190 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10191 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10192 (dbxout_type_methods): Scan TYPE_FIELDS.
10193 (dbxout_type): Don't check TYPE_METHODS here.
10194 * function.c (use_register_for_decl): Always ignore register for
10195 class types when not optimizing.
10196 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10197 * tree.c (free_lang_data_in_type): Stitch out member functions and
10198 templates from TYPE_FIELDS.
10199 (build_distinct_type_copy, verify_type_variant,
10200 verify_type): Member fns are on TYPE_FIELDS.
10201 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10202 * tree-pretty-print.c (dump_generic_node): Likewise.
10204 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10207 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10208 V2TImode and V4TImode.
10209 (ix86_expand_vector_extract): Likewise.
10210 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10211 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10212 (ssescalarmode): Handle V4TImode and V2TImode.
10213 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10214 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10215 (VEXTRACTI128_MODE): New mode iterator.
10216 (splitter for *vec_extractv?ti first element): New.
10217 (VEC_INIT_MODE): New mode iterator.
10218 (vec_init<mode>): Consolidate 3 expanders into one using
10219 VEC_INIT_MODE mode iterator.
10221 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10223 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10224 non_spilled_static_chain_regno_p.
10226 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10228 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10230 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10232 * bb-reorder.c (connect_traces): Allow copying of blocks within
10235 2017-07-20 Richard Biener <rguenther@suse.de>
10237 * gimple.h (gimple_phi_result): Add gphi * overload.
10238 (gimple_phi_result_ptr): Likewise.
10239 (gimple_phi_arg): Likewise. Adjust index assert to only
10240 allow actual argument accesses rather than all slots available
10242 (gimple_phi_arg_def): Add gphi * overload.
10243 * tree-phinodes.c (make_phi_node): Initialize only actual
10245 (resize_phi_node): Clear memory not covered by old node,
10246 do not initialize excess argument slots.
10247 (reserve_phi_args_for_new_edge): Initialize new argument slot
10250 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10252 PR tree-optimization/81388
10254 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10256 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10257 by removing computation of may_be_zero.
10259 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10260 Tom de Vries <tom@codesourcery.com>
10262 PR middle-end/81030
10263 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10264 when gimple level profile disagrees with what RTL expander did.
10266 2017-07-20 Richard Biener <rguenther@suse.de>
10268 PR tree-optimization/61171
10269 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10270 (vect_analyze_stmt): Add slp instance parameter.
10271 (vectorizable_reduction): Likewise.
10272 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10273 (vect_is_simple_reduction): Deal with chains not detected
10274 as SLP reduction chain, specifically not properly associated
10275 chains containing a mix of plus/minus.
10276 (get_reduction_op): Remove.
10277 (get_initial_defs_for_reduction): Simplify, pass in whether
10278 this is a reduction chain, pass in the SLP node for the PHIs.
10279 (vect_create_epilog_for_reduction): Get the SLP instance as
10281 (vectorizable_reduction): Get the SLP instance as arg.
10282 During analysis remember the SLP node with the PHIs in the
10283 instance. Simplify getting at the vectorized reduction PHIs.
10284 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10285 through SLP instance.
10286 (vect_slp_analyze_operations): Likewise.
10287 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10288 (vect_transform_stmt): Likewise.
10290 2017-07-20 Tom de Vries <tom@codesourcery.com>
10292 PR tree-optimization/81489
10293 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10294 read of phi arg location to before loop that modifies phi.
10296 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10298 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10301 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10303 PR middle-end/81331
10304 * except.c (execute): Fix ordering issue.
10306 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10308 PR rtl-optimization/81423
10309 * combine.c (make_compound_operation_int): Don't try to optimize
10310 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10312 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10314 PR rtl-optimization/81423
10315 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10316 with a constant that is -1 in the truncated to mode.
10318 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10320 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10321 (determine_unlikely_bbs): ... here.
10322 * predict.h (propagate_unlikely_bbs_forward): Declare.
10323 * cfgexpand.c (pass_expand::execute): Use it.
10324 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10326 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10327 propagate_unlikely_bbs_forward.
10329 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10331 PR middle-end/81331
10332 * except.c (maybe_add_nop_after_section_switch): New function.
10335 2017-07-19 Tom de Vries <tom@codesourcery.com>
10337 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10339 2017-07-19 Tom de Vries <tom@codesourcery.com>
10341 * gimple.h (gimple_phi_arg): Make assert more strict.
10343 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10345 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10346 * config/rs6000/mmintrin.h: New file.
10347 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10349 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10351 PR tree-optimization/81346
10352 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10354 2017-07-19 Tom de Vries <tom@codesourcery.com>
10356 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10358 2017-07-19 Tom de Vries <tom@codesourcery.com>
10360 * config/nvptx/nvptx-modes.def: Add V2DImode.
10361 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10362 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10363 (nvptx_output_mov_insn): Handle lack of mov.b128.
10364 (nvptx_print_operand): Handle 'H' and 'L' codes.
10365 (nvptx_vector_mode_supported): Allow V2DImode.
10366 (nvptx_preferred_simd_mode): New function.
10367 (nvptx_data_alignment): New function.
10368 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10369 nvptx_preferred_simd_mode.
10370 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10372 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10374 2017-07-19 Tom de Vries <tom@codesourcery.com>
10376 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10377 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10378 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10379 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10380 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10381 (mov<VECIM>_insn): New define_insn.
10382 (define_expand "mov<VECIM>): New define_expand.
10384 2017-07-19 Tom de Vries <tom@codesourcery.com>
10386 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10388 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10390 PR tree-optimization/81346
10391 * fold-const.h (fold_div_compare, range_check_type): Declare.
10392 * fold-const.c (range_check_type): New function.
10393 (build_range_check): Use range_check_type.
10394 (fold_div_compare): No longer static, rewritten into
10395 a match.pd helper function.
10396 (fold_comparison): Don't call fold_div_compare here.
10397 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10398 as helper function.
10400 2017-07-19 Nathan Sidwell <nathan@acm.org>
10402 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10403 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10404 * tree.c (find_decls_types_r, verify_type): Use
10405 TYPE_{MIN,MAX}_VALUE_RAW.
10406 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10407 (hash_tree): Likewise.
10408 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10410 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10413 2017-07-18 Tom de Vries <tom@codesourcery.com>
10415 PR middle-end/81464
10416 * omp-expand.c (expand_omp_for_static_chunk): Handle
10417 equal-argument loop exit phi.
10419 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10422 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10423 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10424 operand 2 predicate.
10425 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10426 operand 2 predicate.
10427 (ror,rol -> rorx splitters): Use const_int_operand as
10428 operand 2 predicate.
10430 2017-06-18 Richard Biener <rguenther@suse.de>
10432 PR tree-optimization/81410
10433 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10434 the gap in the ! slp_perm SLP case after each group.
10436 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10438 PR middle-end/81463
10439 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10442 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10444 PR middle-end/81462
10445 * predict.c (set_even_probabilities): Cleanup; do not affect
10446 probabilities that are already known.
10447 (combine_predictions_for_bb): Call even when count is set.
10449 2017-07-18 Nathan Sidwell <nathan@acm.org>
10451 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10454 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10457 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10458 optimization for loop niter analysis.
10460 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10463 * config/avr/avr.c (avr_optimize_casesi): Don't use
10464 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10466 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10468 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10469 body_cost_vec from _vect_peel_extended_info.
10470 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10471 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10474 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10476 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10478 2017-07-18 Richard Biener <rguenther@suse.de>
10480 PR tree-optimization/80620
10481 PR tree-optimization/81403
10482 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10483 info when re-using a VN table entry.
10485 2017-07-18 Richard Biener <rguenther@suse.de>
10487 PR tree-optimization/81418
10488 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10489 vectype_in. Verify that with lane-reducing reduction operations
10490 we have a single def-use cycle.
10492 2017-07-17 Carl Love <cel@us.ibm.com>
10494 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10496 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10497 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10498 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10500 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10501 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10502 VMULOSW): New enum "unspec" values.
10503 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10504 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10505 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10506 altivec_vmulosw): New patterns.
10507 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10508 VMULOSW): Add definitions.
10509 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10511 * config/alpha/alpha.c: Include predict.h.
10513 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10515 * tree-vrp.c (compare_assert_loc): Fix comparison function
10516 to return predictable results.
10518 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10520 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10522 (subdi3): Likewise.
10523 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10524 * doc/invoke.texi (mexpand-adddi): Update text.
10526 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10528 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10529 that also clobbers the CC register. The old expand code is moved
10531 (*arc_clzsi2): ... here.
10532 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10533 the CC register. The old expand code is moved to ...
10534 (arc_ctzsi2): ... here.
10536 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10538 * config/arc/arc.opt (mindexed-loads): Use initial value
10539 TARGET_INDEXED_LOADS_DEFAULT.
10540 (mauto-modify-reg): Use initial value
10541 TARGET_AUTO_MODIFY_REG_DEFAULT.
10542 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10543 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10544 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10545 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10547 2017-07-17 Martin Liska <mliska@suse.cz>
10550 * opts.c (finish_options): Do not allow -fgnu-tm
10551 w/ -fsanitize={kernel-,}address. Say sorry.
10553 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10556 * tree-loop-distribution.c (classify_partition): Only assert on
10557 numer of iterations.
10558 (merge_dep_scc_partitions): Delete prameter. Update function call.
10559 (distribute_loop): Remove code handling loop with unknown niters.
10560 (pass_loop_distribution::execute): Skip loop with unknown niters.
10562 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10565 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10566 function sort_partitions_by_post_order.
10568 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10570 PR tree-optimization/81374
10571 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10572 the max index of basic blocks, rather than number of basic blocks.
10574 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10576 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10578 (arc_legitimate_pic_operand_p): Likewise.
10579 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10581 (arc_needs_pcl_p): Likewise.
10582 (arc_legitimate_pc_offset_p): Likewise.
10583 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10584 function is also used in constrains.md.
10585 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10586 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10587 PLUS. Only return true/false in known cases, otherwise assert.
10588 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10589 is already called in arc_legitimate_constant_p.
10590 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10592 (LEGITIMATE_PIC_OPERAND_P): Use
10593 arc_raw_symbolic_reference_mentioned_p function.
10594 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10599 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10600 Andrew Burgess <andrew.burgess@embecosm.com>
10602 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10603 (arc_return_address_register): New function.
10604 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10605 (arc_handle_fndecl_attribute): Add naked attribute.
10606 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10607 (TARGET_WARN_FUNC_RETURN): Likewise.
10608 (arc_allocate_stack_slots_for_args): New function.
10609 (arc_warn_func_return): Likewise.
10610 (machine_function): Change type fn_type.
10611 (arc_compute_function_type): Consider new naked function type,
10612 change function return type.
10613 (arc_must_save_register): Adapt to handle new
10614 arc_compute_function_type's return type.
10615 (arc_expand_prologue): Likewise.
10616 (arc_expand_epilogue): Likewise.
10617 (arc_return_address_regs): Delete.
10618 (arc_return_address_register): New function.
10619 (arc_epilogue_uses): Use above function.
10620 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10621 (arc_function_type): Change encoding, add naked type.
10622 (ARC_INTERRUPT_P): Change to handle the new encoding.
10623 (ARC_FAST_INTERRUPT_P): Likewise.
10624 (ARC_NORMAL_P): Define.
10625 (ARC_NAKED_P): Likewise.
10626 (arc_compute_function_type): Delete prototype.
10627 * config/arc/arc.md (in_ret_delay_slot): Use
10628 arc_return_address_register function.
10629 (simple_return): Likewise.
10630 (p_return_i): Likewise.
10632 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10634 PR tree-optimization/81428
10635 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10636 can't be built for those types.
10638 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10640 Remove stuff dead since r239246.
10642 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10643 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10644 (avr_inform_devices): Remove dead stuff.
10646 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10648 * config/arm/arm_neon.h: Fix softp typo.
10650 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10652 PR tree-optimization/81365
10653 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10654 aggregate moves onto bb predecessor edges, make sure there are no
10655 loads that could alias the lhs in between the start of bb and the
10658 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10661 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10662 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10663 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10665 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10667 PR tree-optimization/81396
10668 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10669 (init_symbolic_number): Initialize it to 1.
10670 (perform_symbolic_merge): Add n_ops from both operands into the new
10672 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10673 without base_addr as useless if they need more than one operation.
10674 (bswap_replace): Handle !bswap case for NULL base_addr.
10676 2017-07-17 Tom de Vries <tom@codesourcery.com>
10679 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10682 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10684 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10685 conditional builtin define __FIX_LEON3FT_B2BST.
10687 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10689 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10690 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10693 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10695 PR rtl-optimization/81424
10696 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10697 to remove potential trapping from operands if -fnon-call-exceptions.
10699 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10701 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10702 profile_proability for scalling.
10703 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10705 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10707 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10709 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10711 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10712 fixpoint arithmetics.
10714 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10716 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10717 fixpoint arithmetics.
10719 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10721 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10722 fixpoint arithmetics.
10724 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10726 * profile-count.h (profile_probability::from_reg_br_prob_note,
10727 profile_probability::to_reg_br_prob_note): New functions.
10728 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10729 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10730 * predict.c (probability_reliable_p): Update.
10731 (edge_probability_reliable_p): Update.
10732 (br_prob_note_reliable_p): Update.
10733 (invert_br_probabilities): Update.
10734 (add_reg_br_prob_note): New function.
10735 (combine_predictions_for_insn): Update.
10736 * asan.c (asan_clear_shadow): Update.
10737 * cfgbuild.c (compute_outgoing_frequencies): Update.
10738 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10739 (update_br_prob_note): Update.
10740 (rtl_verify_edges): Update.
10741 (purge_dead_edges): Update.
10742 (fixup_reorder_chain): Update.
10743 * emit-rtl.c (try_split): Update.
10744 * ifcvt.c (cond_exec_process_insns): Update.
10745 (cond_exec_process_if_block): Update.
10746 (dead_or_predicable): Update.
10747 * internal-fn.c (expand_addsub_overflow): Update.
10748 (expand_neg_overflow): Update.
10749 (expand_mul_overflow): Update.
10750 * loop-doloop.c (doloop_modify): Update.
10751 * loop-unroll.c (compare_and_jump_seq): Update.
10752 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10753 * predict.h: Update.
10754 * reorg.c (mostly_true_jump): Update.
10756 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10757 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10758 * config/arc/arc.c: (emit_unlikely_jump): Update.
10759 * config/arm/arm.c: (emit_unlikely_jump): Update.
10760 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10761 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10762 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10763 (ix86_print_operand): Update.
10764 (ix86_split_fp_branch): Update.
10765 (predict_jump): Update.
10766 * config/ia64/ia64.c (ia64_print_operand): Update.
10767 * config/mmix/mmix.c (mmix_print_operand): Update.
10768 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10769 (rs6000_expand_split_stack_prologue): Update.
10770 * config/rs6000/rs6000.c: Update.
10771 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10772 (s390_expand_vec_movstr): Update.
10773 (s390_expand_cs_tdsi): Update.
10774 (s390_expand_split_stack_prologue): Update.
10775 * config/sh/sh.c (sh_print_operand): Update.
10776 (expand_cbranchsi4): Update.
10777 (expand_cbranchdi4): Update.
10778 * config/sparc/sparc.c (output_v9branch): Update.
10779 * config/spu/spu.c (get_branch_target): Update.
10780 (ea_load_store_inline): Update.
10781 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10782 * config/tilepro/tilepro.c: Update.
10784 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10786 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10787 (gimplify_save_expr): Likewise.
10789 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10791 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10793 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10795 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10797 * timevar.def (TV_IPA_FNSUMMARY): Define.
10799 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10801 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10802 to back store errata sensitive sequence from being generated.
10803 (sqrtdf2_fix): Likewise.
10805 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10807 * tree-ssa-threadupdate.c (compute_path_counts,
10808 update_joiner_offpath_counts): Use profile_probability.
10810 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10813 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10815 * config/arm/arm-c.c (arm_cpu_builtins): Define
10816 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10818 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10821 array entries to represent __ieee128 versions of the
10822 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10823 scalar_extract_sig, and scalar_insert_exp built-in functions.
10824 (altivec_resolve_overloaded_builtin): Add special case handling
10825 for the __builtin_scalar_insert_exp function, as represented by
10826 the P9V_BUILTIN_VEC_VSIEDP constant.
10827 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10828 exponent support for __ieee128 argument.
10829 (VSESQP): Add scalar extract signature support for __ieee128
10831 (VSTDCNQP): Add scalar test negative support for __ieee128
10833 (VSIEQP): Add scalar insert exponent support for __int128 argument
10834 with __ieee128 result.
10835 (VSIEQPF): Add scalar insert exponent support for __ieee128
10836 argument with __ieee128 result.
10837 (VSTDCQP): Add scalar test data class support for __ieee128
10839 (VSTDCNQP): Add overload support for scalar test negative with
10840 __ieee128 argument.
10841 (VSTDCQP): Add overload support for scalar test data class
10842 __ieee128 argument.
10843 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10844 UNSPEC_VSX_SXSIGDP.
10845 (UNSPEC_VSX_SIEXPQP): New constant.
10846 (xsxexpqp): New insn for VSX scalar extract exponent quad
10848 (xsxsigqp): New insn for VSX scalar extract significand quad
10850 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10851 precision with floating point argument.
10852 (xststdcqp): New expand for VSX scalar test data class quad
10854 (xststdcnegqp): New expand for VSX scalar test negative quad
10856 (xststdcqp): New insn to match expansions for VSX scalar test data
10857 class quad precision and VSX scalar test negative quad precision.
10858 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10859 special case operand checking to enforce that second operand of
10860 VSX scalar test data class with quad precision argument is a 7-bit
10862 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10863 prototypes and descriptions of __ieee128 versions of
10864 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10865 scalar_test_data_class, and scalar_test_neg built-in functions.
10867 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10869 PR tree-optimization/81162
10870 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10871 replace a negate with an add.
10873 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10875 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10877 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10879 * config/arm/arm-c.c (arm_cpu_builtins): Define
10880 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10882 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10884 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10885 (armv8-r): Set ARM Cortex-R52 as default CPU.
10886 * config/arm/arm-tables.opt: Regenerate.
10887 * config/arm/arm-tune.md: Regenerate.
10888 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10890 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10891 extension for -mcpu=cortex-r52.
10893 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10895 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10896 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10897 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10898 (fp-armv8): Define it as FP_ARMv8 only.
10899 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10900 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10902 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10903 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10904 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10905 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10906 than TARGET_FPU_ARMV8.
10907 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10908 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10909 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10911 * config/arm/neon.md (neon_vrint): Likewise.
10912 (neon_vcvt): Likewise.
10913 (neon_<fmaxmin_op><mode>): Likewise.
10914 (<fmaxmin><mode>3): Likewise.
10915 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10916 * config/arm/predicates.md (arm_cond_move_operator): Check against
10917 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10919 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10921 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10922 to top of function.
10924 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10926 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10927 loop in comment with memset.
10929 2017-07-14 Martin Liska <mliska@suse.cz>
10931 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10932 * dwarf2out.c (is_java): Remove the function.
10933 (output_pubname): Remove usage of the function.
10934 (lower_bound_default): Remove usage of DW_LANG_Java.
10935 (gen_compile_unit_die): Likewise.
10936 * gcc.c: Remove compiler defaults for .java and .zip files.
10937 * gimple-expr.c (remove_suffix): Change as there's no longer
10938 extension than 4-letter one.
10939 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10940 (gimplify_save_expr): Likewise.
10941 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10942 as it's possible even for other languages than Java.
10943 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10944 * lto-opts.c (lto_write_options): Remove reference to Java.
10945 * opts.c (strip_off_ending): Update file extension handling.
10946 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10947 * tree-eh.c (lower_resx): Likewise.
10948 * tree.c (free_lang_data_in_type): Remove dead code.
10949 (find_decls_types_r): Likewise.
10950 (build_common_builtin_nodes): Remove Java from a comment.
10951 (verify_type): Remove dead code.
10952 * varasm.c (assemble_external): Remove Java from a comment.
10954 2017-07-14 Martin Liska <mliska@suse.cz>
10956 * opts.c (finish_options): Add quotes.
10957 (common_handle_option): Likewise.
10959 2017-07-14 Martin Liska <mliska@suse.cz>
10961 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10962 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10963 Remove N_SO_PASCAL.
10964 * dwarf2out.c (lower_bound_default): Do not handle
10966 (gen_compile_unit_die): Likewise.
10967 * gcc.c: Remove default extension binding for GNU Pascal.
10968 * stmt.c: Remove Pascal language from a comment.
10969 * xcoffout.c: Likewise.
10971 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10974 * diagnostic-show-locus.c (fixit_cmp): New function.
10975 (layout::layout): Sort m_fixit_hints.
10976 (column_range::column_range): Assert that the values are valid.
10977 (struct char_span): New struct.
10978 (correction::overwrite): New method.
10979 (struct source_line): New struct.
10980 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10981 calls in terms of classes source_line and char_span, and
10982 correction::overwrite.
10983 (selftest::test_overlapped_fixit_printing_2): New function.
10984 (selftest::diagnostic_show_locus_c_tests): Call it.
10986 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10988 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10989 early if there is no lhs.
10991 2017-07-13 Martin Liska <mliska@suse.cz>
10993 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10994 (gen_reference_type_die): Likewise.
10995 * stor-layout.c: Remove Pascal-related comment.
10997 2017-07-13 Martin Liska <mliska@suse.cz>
10999 * opts.c (finish_options): Add quotes to error messages.
11000 (parse_sanitizer_options): Likewise.
11002 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11004 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11006 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11008 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11010 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11012 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11014 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11016 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11019 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11020 provides the hardware capability bits, define the macro
11021 __BUILTIN_CPU_SUPPORTS__.
11022 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11023 if GLIBC does not provide the hardware capability bits. Add a
11024 gcc_unreachable call if the built-in cpu function is neither
11025 __builtin_cpu_is nor __builtin_cpu_supports.
11026 (rs6000_get_function_versions_dispatcher): Change the warning
11027 that an old GLIBC is used which does not export the capability
11028 bits to be an error.
11029 * doc/extend.texi (target_clones attribute): Document the
11030 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11031 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11032 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11033 the macros defined by GCC if the newer GLIBC is available.
11035 2017-07-12 Jeff Law <law@redhat.com>
11037 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11038 remaining includes slightly.
11039 * config/riscv/riscv-builtins.c: Include profile-count.h.
11041 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11044 * config/avr/avr.c (avr_set_current_function): In diagnostic
11045 messages: Quote keywords and (parts of) identifiers.
11046 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11049 2017-07-12 Carl Love <cel@us.ibm.com>
11051 * config/rs6000/rs6000-c.c: Add support for built-in functions
11052 vector bool char vec_revb (vector bool char);
11053 vector bool short vec_revb (vector short char);
11054 vector bool int vec_revb (vector bool int);
11055 vector bool long long vec_revb (vector bool long long);
11056 * doc/extend.texi: Update the built-in documentation file for the
11057 new built-in functions.
11059 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11061 * config/s390/s390.md: Remove movcc splitter.
11063 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11065 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11066 load/store on condition.
11068 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11071 * config/avr/avr.c (avr_encode_section_info)
11072 [progmem && !TREE_READONLY]: Error if progmem object needs
11075 2017-07-11 Michael Collison <michael.collison@arm.com>
11077 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11080 2017-07-11 Carl Love <cel@us.ibm.com>
11082 * config/rs6000/rs6000-c.c: Add support for builtins
11083 vector unsigned int vec_parity_lsbb (vector signed int);
11084 vector unsigned int vec_parity_lsbb (vector unsigned int);
11085 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11086 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11087 vector unsigned long long vec_parity_lsbb (vector signed long long);
11088 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11089 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11090 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11091 * doc/extend.texi: Update the built-in documentation file for the
11092 new built-in functions.
11094 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11096 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11097 (layout::m_primary_loc): New field.
11098 (layout::layout): Initialize new field. Move location filtering
11099 logic from here to...
11100 (layout::maybe_add_location_range): ...this new method. Add
11101 support for filtering to just the lines already specified by other
11103 (layout::will_show_line_p): New method.
11104 (gcc_rich_location::add_location_if_nearby): New method.
11105 (selftest::test_add_location_if_nearby): New test function.
11106 (selftest::diagnostic_show_locus_c_tests): Call it.
11107 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11110 2017-07-11 Tom de Vries <tom@codesourcery.com>
11112 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11113 (bb_first_real_insn): New function.
11114 (nvptx_single): Add extra initialization of broadcasted condition
11117 2017-07-11 Nathan Sidwell <nathan@acm.org>
11119 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11121 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11123 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11124 Binutils doc as TEXI will mess them up.
11125 * doc/invoke.texi (AVR Options): Same here.
11127 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11129 * config/sparc/sparc.opt (mfix-ut700): New option.
11130 (mfix-gr712rc): Likewise.
11131 (sparc_fix_b2bst): New variable.
11132 * doc/invoke.texi (SPARC options): Document them.
11133 (ARM options): Fix warnings.
11134 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11135 instructions to prevent sequences that can trigger the store-store
11136 errata for certain LEON3FT processors.
11137 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11138 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11139 * config/sparc/sparc.md (fix_b2bst): New attribute.
11140 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11142 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11145 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11147 (*rsqrtsf2_sse): Ditto.
11149 (div<mode>3): Macroize insn from divdf3 and divsf3
11150 using MODEF mode iterator.
11152 2017-07-10 Martin Sebor <msebor@redhat.com>
11154 PR tree-optimization/80397
11155 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11156 instead of testing for equality to INTEGER_TYPE.
11158 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11160 * config.gcc: Remove uclibc from arc target spec.
11162 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11164 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11166 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11169 * lto-wrapper.c (remove_option): New function.
11170 (merge_and_complain): Merge PIC/PIE options more realistically.
11172 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11174 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11178 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11179 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11180 * config.in: Regenerate.
11181 * configure: Regenerate.
11182 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11183 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11184 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11185 (TARGET_GASISR_PROLOGUES): ...target mask.
11186 * common/config/avr/avr-common.c
11187 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11188 Set -mgas-isr-prologues.
11189 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11190 INSERT_PASS_BEFORE for it.
11191 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11192 * config/avr/avr.c (avr_option_override)
11193 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11194 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11195 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11196 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11197 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11199 (make_avr_pass_pre_proep): New function.
11200 (emit_push_sfr) <treg>: Add argument to function and use it
11201 instead of TMP_REG.
11202 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11203 and set machine->gasisr.yes.
11204 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11205 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11206 __gcc_isr.n_pushed to .L__stack_usage.
11207 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11208 (avr_asm_final_postscan_insn): ...this new static function.
11209 * config/avr/avr.h (machine_function)
11210 <is_no_gccisr, use_L__stack_usage>: New fields.
11211 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11212 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11213 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11214 (gasisr, *gasisr): New expander and insn.
11215 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11216 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11217 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11219 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11221 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11224 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11226 Move jump-tables out of .text again.
11229 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11230 (ASM_OUTPUT_ADDR_VEC): New function.
11231 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11232 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11233 INSN_ADDRESSes as asm comment.
11234 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11235 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11236 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11237 * config/avr/avr.md (*tablejump): Adjust comment.
11238 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11239 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11241 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11242 (avr_output_addr_vec): New proto.
11243 (avr_log_t) <insn_addresses>: New field.
11245 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11248 * config/i386/i386.c (ix86_function_arg_advance): Set
11249 outgoing_args_on_stack to true if there are outgoing arguments
11251 (ix86_function_arg): Likewise.
11252 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11253 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11254 * config/i386/i386.h (machine_function): Add
11255 outgoing_args_on_stack.
11257 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11259 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11260 supporting pthreds.
11261 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11263 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11265 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11266 (REAL_H): Remove $(MACHMODE_H).
11267 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11269 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11270 $(MACHMODE_H) and double-int.h.
11271 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11273 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11276 2017-07-07 Andrew Pinski <apinski@cavium.com>
11278 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11279 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11281 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11283 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11284 Add warning if GCC was not configured to link against a GLIBC that
11285 exports the hardware capability bits.
11286 (make_resolver_func): Make resolver function private and not a
11287 COMDAT function. Create the name with clone_function_name instead
11288 of make_unique_name.
11291 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11292 correct operand in doing the split.
11294 2017-07-07 Carl Love <cel@us.ibm.com>
11296 * config/rs6000/rs6000-c: Add support for built-in function
11297 vector unsigned short vec_pack_to_short_fp32 (vector float,
11299 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11300 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11301 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11302 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11303 (convert_4f32_8i16): Add define_expand.
11304 * doc/extend.texi: Update the built-in documentation file for the
11305 new built-in function.
11307 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11309 * config/sparc/m8.md: New file.
11310 * config/sparc/sparc.md: Include m8.md.
11312 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11314 * config/sparc/sparc.opt: New option -mvis4b.
11315 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11316 (sparc_option_override): Handle VIS4B.
11317 (enum sparc_builtins): Define
11318 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11319 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11320 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11321 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11322 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11323 (check_constant_argument): New function.
11324 (sparc_vis_init_builtins): Define builtins
11325 __builtin_vis_dictunpack{8,16,32},
11326 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11327 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11328 __builtin_vis_fpcmpde{8,16,32}shl and
11329 __builtin_vis_fpcmpur{8,16,32}shl.
11330 (sparc_expand_builtin): Check that the constant operands to
11331 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11332 constant and in range.
11333 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11335 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11336 (SPARC_IMM5_P): Likewise.
11337 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11338 (enabled): Handle vis4b.
11339 (UNSPEC_DICTUNPACK): New unspec.
11340 (UNSPEC_FPCMPSHL): Likewise.
11341 (UNSPEC_FPUCMPSHL): Likewise.
11342 (UNSPEC_FPCMPDESHL): Likewise.
11343 (UNSPEC_FPCMPURSHL): Likewise.
11344 (cpu_feature): New CPU feature `vis4b'.
11345 (dictunpack{8,16,32}): New insns.
11346 (FPCSMODE): New mode iterator.
11347 (fpcscond): New code iterator.
11348 (fpcsucond): Likewise.
11349 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11350 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11351 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11352 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11353 * config/sparc/constraints.md: Define constraints `q' for unsigned
11354 2-bit integer constants and `t' for unsigned 5-bit integer
11356 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11358 (imm5_operand_dictunpack16): Likewise.
11359 (imm5_operand_dictunpack32): Likewise.
11360 (imm2_operand): Likewise.
11361 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11362 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11363 ditunpack* and fpcmp*shl builtins.
11365 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11367 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11368 * config.in: Add HAVE_AS_SPARC6 define.
11369 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11371 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11373 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11374 (CPP_CPU_SPEC): Handle m8.
11375 (ASM_CPU_SPEC): Likewise.
11376 * config/sparc/sparc-opts.h (enum processor_type): Add
11378 * config/sparc/sparc.c (m8_costs): New struct.
11379 (sparc_option_override): Handle TARGET_CPU_m8.
11380 (sparc32_initialize_trampoline): Likewise.
11381 (sparc64_initialize_trampoline): Likewise.
11382 (sparc_issue_rate): Likewise.
11383 (sparc_register_move_cost): Likewise.
11384 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11385 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11386 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11387 (CPP_CPU_SPEC): Handle M8.
11388 (ASM_CPU_SPEC): Likewise.
11389 (AS_M8_FLAG): Define.
11390 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11391 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11392 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11394 * configure: Regenerate.
11395 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11398 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11400 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11402 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11403 ("*movdi_insn_sp32"): Do not set v3pipe.
11404 ("*movsi_insn"): Likewise.
11405 ("*movdi_insn_sp64"): Likewise.
11406 ("*movsf_insn"): Likewise.
11407 ("*movdf_insn_sp32"): Likewise.
11408 ("*movdf_insn_sp64"): Likewise.
11409 ("*zero_extendsidi2_insn_sp64"): Likewise.
11410 ("*sign_extendsidi2_insn"): Likewise.
11411 ("*mov<VM32:mode>_insn"): Likewise.
11412 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11413 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11414 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11415 ("<vlop:code><VL:mode>3"): Likewise.
11416 ("*not_<vlop:code><VL:mode>3"): Likewise.
11417 ("*nand<VL:mode>_vis"): Likewise.
11418 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11419 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11420 ("one_cmpl<VL:mode>2"): Likewise.
11421 ("faligndata<VM64:mode>_vis"): Likewise.
11422 ("alignaddrsi_vis"): Likewise.
11423 ("alignaddrdi_vis"): Likweise.
11424 ("alignaddrlsi_vis"): Likewise.
11425 ("alignaddrldi_vis"): Likewise.
11426 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11427 ("bmaskdi_vis"): Likewise.
11428 ("bmasksi_vis"): Likewise.
11429 ("bshuffle<VM64:mode>_vis"): Likewise.
11430 ("cmask8<P:mode>_vis"): Likewise.
11431 ("cmask16<P:mode>_vis"): Likewise.
11432 ("cmask32<P:mode>_vis"): Likewise.
11433 ("pdistn<P:mode>_vis"): Likewise.
11434 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11436 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11438 * config/sparc/sparc.md ("subtype"): New insn attribute.
11439 ("*wrgsr_sp64"): Set insn subtype.
11440 ("*rdgsr_sp64"): Likewise.
11441 ("alignaddrsi_vis"): Likewise.
11442 ("alignaddrdi_vis"): Likewise.
11443 ("alignaddrlsi_vis"): Likewise.
11444 ("alignaddrldi_vis"): Likewise.
11445 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11446 ("fexpand_vis"): Likewise.
11447 ("fpmerge_vis"): Likewise.
11448 ("faligndata<VM64:mode>_vis"): Likewise.
11449 ("bshuffle<VM64:mode>_vis"): Likewise.
11450 ("cmask8<P:mode>_vis"): Likewise.
11451 ("cmask16<P:mode>_vis"): Likewise.
11452 ("cmask32<P:mode>_vis"): Likewise.
11453 ("fchksm16_vis"): Likewise.
11454 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11455 ("fmean16_vis"): Likewise.
11456 ("fp<plusminus_insn>64_vis"): Likewise.
11457 ("<plusminus_insn>v8qi3"): Likewise.
11458 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11459 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11460 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11461 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11462 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11463 ("*movqi_insn"): Likewise.
11464 ("*movhi_insn"): Likewise.
11465 ("*movsi_insn"): Likewise.
11466 ("movsi_pic_gotdata_op"): Likewise.
11467 ("*movdi_insn_sp32"): Likewise.
11468 ("*movdi_insn_sp64"): Likewise.
11469 ("movdi_pic_gotdata_op"): Likewise.
11470 ("*movsf_insn"): Likewise.
11471 ("*movdf_insn_sp32"): Likewise.
11472 ("*movdf_insn_sp64"): Likewise.
11473 ("*zero_extendhisi2_insn"): Likewise.
11474 ("*zero_extendqihi2_insn"): Likewise.
11475 ("*zero_extendqisi2_insn"): Likewise.
11476 ("*zero_extendqidi2_insn"): Likewise.
11477 ("*zero_extendhidi2_insn"): Likewise.
11478 ("*zero_extendsidi2_insn_sp64"): Likewise.
11479 ("ldfsr"): Likewise.
11480 ("prefetch_64"): Likewise.
11481 ("prefetch_32"): Likewise.
11482 ("tie_ld32"): Likewise.
11483 ("tie_ld64"): Likewise.
11484 ("*tldo_ldub_sp32"): Likewise.
11485 ("*tldo_ldub1_sp32"): Likewise.
11486 ("*tldo_ldub2_sp32"): Likewise.
11487 ("*tldo_ldub_sp64"): Likewise.
11488 ("*tldo_ldub1_sp64"): Likewise.
11489 ("*tldo_ldub2_sp64"): Likewise.
11490 ("*tldo_ldub3_sp64"): Likewise.
11491 ("*tldo_lduh_sp32"): Likewise.
11492 ("*tldo_lduh1_sp32"): Likewise.
11493 ("*tldo_lduh_sp64"): Likewise.
11494 ("*tldo_lduh1_sp64"): Likewise.
11495 ("*tldo_lduh2_sp64"): Likewise.
11496 ("*tldo_lduw_sp32"): Likewise.
11497 ("*tldo_lduw_sp64"): Likewise.
11498 ("*tldo_lduw1_sp64"): Likewise.
11499 ("*tldo_ldx_sp64"): Likewise.
11500 ("*mov<VM32:mode>_insn"): Likewise.
11501 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11502 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11504 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11506 * config/sparc/sparc.md ("type"): New insn type viscmp.
11507 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11509 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11510 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11511 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11512 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11514 ("n7_vis_logical_11cycle"): Likewise.
11515 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11516 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11517 * config/sparc/niagara.md ("niag_vis"): Likewise.
11518 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11519 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11521 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11523 * config/sparc/sparc.md: New instruction type `bmask'.
11524 (bmaskdi_vis): Use the `bmask' type.
11525 (bmasksi_vis): Likewise.
11526 * config/sparc/ultra3.md (us3_array): Likewise.
11527 * config/sparc/niagara7.md (n7_array): Likewise.
11528 * config/sparc/niagara4.md (n4_array): Likewise.
11529 * config/sparc/niagara2.md (niag2_vis): Likewise.
11530 (niag3_vis): Likewise.
11531 * config/sparc/niagara.md (niag_vis): Likewise.
11533 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11535 * ipa-comdats.c: Remove optimize check from gate.
11536 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11537 for functions not optimized.
11538 (ipa_fn_summary_read): Skip optimize check.
11539 (ipa_fn_summary_write): Likewise.
11540 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11542 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11544 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11545 (check_callers): Give up on caller which is not optimized.
11546 (inline_small_functions): Likewise.
11547 (ipa_inline): Do not give up when not optimizing.
11548 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11549 away unoptimizes cdtors.
11550 (whole_program_function_and_variable_visibility): Do
11551 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11552 * ipa.c (process_references): Do not check optimize.
11553 (symbol_table::remove_unreachable_nodes): Update optimize check.
11554 (set_writeonly_bit): Update optimize check.
11555 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11556 (pass_ipa_single_use::gate): Remove.
11558 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11560 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11561 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11562 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11563 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11564 permute_load, permute_store, adjust_extract, adjust_splat,
11565 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11566 replace_swap_with_copy, dump_swap_insn_table,
11567 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11568 recombine_lvx_pattern, recombine_stvx_pattern,
11569 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11570 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11571 to file rs6000-p8swap.c.
11572 * config/rs6000/rs6000-p8swap.c: New file.
11573 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11574 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11575 and rs6000*-*-* targets.
11577 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11579 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11581 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11583 * lto-wrapper.c (merge_and_complain): Do not merge
11584 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11585 fsigned_zeros, ftrapping_math, fwrapv.
11586 (append_compiler_options): Do not track these options.
11587 (append_linker_options): Likewie
11589 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11591 * cgraphunit.c (cgraph_node::finalize_function): When
11592 !flag_toplevel_reorde set no_reorder flag.
11593 (varpool_node::finalize_decl): Likewise.
11594 (symbol_table::compile): Drop no toplevel reorder path.
11596 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11598 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11599 edges; zero probability is not better than uninitialized.
11601 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11603 * asan.h (asan_sanitize_allocas_p): Declare.
11604 * asan.c (asan_sanitize_allocas_p): New function.
11605 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11606 (handle_builtin_alloca): Likewise.
11607 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11608 if !asan_sanitize_allocas_p.
11609 * params.def (asan-instrument-allocas): Add new option.
11610 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11611 * opts.c (common_handle_option): Disable allocas sanitization for
11614 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11616 * asan.c: Include gimple-fold.h.
11617 (get_last_alloca_addr): New function.
11618 (handle_builtin_stackrestore): Likewise.
11619 (handle_builtin_alloca): Likewise.
11620 (asan_emit_allocas_unpoison): Likewise.
11621 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11622 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11623 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11624 (instrument_builtin_call): Pass gimple iterator to
11625 get_mem_refs_of_builtin_call.
11626 (last_alloca_addr): New global.
11627 * asan.h (asan_emit_allocas_unpoison): Declare.
11628 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11629 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11630 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11631 if function calls alloca.
11632 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11633 * gimple-fold.h (replace_call_with_value): Declare.
11634 * internal-fn.c: Include asan.h.
11635 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11636 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11638 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11640 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11641 (C_SELFTEST_FLAGS): New.
11642 (CPP_SELFTEST_FLAGS): New.
11643 (SELFTEST_DEPS): New, from deps of s-selftest.
11644 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11645 (CPP_SELFTEST_DEPS): New.
11646 (selftest): Add dependency on s-selftest-c++.
11647 (s-selftest): Rename to...
11648 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11649 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11650 than SELFTEST_FLAGS.
11651 (selftest-gdb): Rename to...
11652 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11654 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11655 (selftest-valgrind): Rename to...
11656 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11658 (selftest-valgrind): Reintroduce as an alias for
11659 selftest-c-valgrind.
11660 (s-selftest-c++): New.
11661 (selftest-c++-gdb): New.
11662 (selftest-c++-valgrind): New.
11664 2017-07-06 Olivier Hainque <hainque@adacore.com>
11666 * gcc.c (process_command): When deciding if undefined variables
11667 should be ignored when processing specs, accept "gcc -v" as well.
11669 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11671 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11672 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11674 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11676 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11677 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11678 * config/arm/arm-tables.opt: Regenerate.
11679 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11681 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11683 2017-07-06 Carl Love <cel@us.ibm.com>
11685 * ChangeLog: Clean up from mid air collision
11687 2017-07-06 Carl Love <cel@us.ibm.com>
11689 * config/rs6000/rs6000-c.c: Add support for built-in functions
11690 vector signed int vec_subc (vector signed int, vector signed int);
11691 vector signed __int128 vec_subc (vector signed __int128,
11692 vector signed __int128);
11693 vector unsigned __int128 vec_subc (vector unsigned __int128,
11694 vector unsigned __int128);
11695 vector signed int vec_sube (vector signed int, vector signed int,
11696 vector signed int);
11697 vector unsigned int vec_sube (vector unsigned int,
11698 vector unsigned int,
11699 vector unsigned int);
11700 vector signed __int128 vec_sube (vector signed __int128,
11701 vector signed __int128,
11702 vector signed__int128);
11703 vector unsigned __int128 vec_sube (vector unsigned __int128,
11704 vector unsigned __int128,
11705 vector unsigned __int128);
11706 vector signed int vec_subec (vector signed int, vector signed int,
11707 vector signed int);
11708 vector unsigned int vec_subec (vector unsigned int,
11709 vector unsigned int,
11710 vector unsigned int);
11711 vector signed __int128 vec_subec (vector signed __int128,
11712 vector signed __int128,
11713 vector signed__int128);
11714 vector unsigned __int128 vec_subec (vector unsigned __int128,
11715 vector unsigned __int128,
11716 vector unsigned __int128);
11717 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11718 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11719 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11720 BU_ALTIVEC_OVERLOAD_X definitions.
11721 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11722 * doc/extend.texi: Update the built-in documentation file for the new
11723 built-in functions.
11725 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11728 * diagnostic-show-locus.c (layout::layout): Use start and finish
11729 spelling location for the start and finish of each range.
11730 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11731 Add unused aspect param.
11732 * input.c (expand_location_1): Add "aspect" param, and use it
11733 to access the correct part of the location.
11734 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11736 (expand_location_to_spelling_point): Likewise.
11737 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11738 param, and pass it to expand_location_1.
11740 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11742 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11743 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11744 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11745 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11746 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11747 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11748 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11749 _mm_maskz_getmant_ss): New intrinsics.
11750 (__builtin_ia32_getexpss128_mask): Changed to ...
11751 __builtin_ia32_getexpss128_round ... this.
11752 (__builtin_ia32_getexpsd128_mask): Changed to ...
11753 __builtin_ia32_getexpsd128_round ... this.
11754 * config/i386/i386-builtin-types.def
11755 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11756 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11757 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11758 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11759 __builtin_ia32_getmantss_mask_round): New builtins.
11760 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11761 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11762 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11763 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11764 * config/i386/sse.md
11765 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11766 avx512f_sgetexp<mode><mask_scalar_name>
11767 <round_saeonly_scalar_name> ... this.
11768 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11769 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11770 vgetexp<ssescalarmodesuffix>
11771 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11772 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11773 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11774 avx512f_vgetmant<mode><mask_scalar_name>
11775 <round_saeonly_scalar_name> ... this.
11776 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11777 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11778 vgetmant<ssescalarmodesuffix>
11779 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11780 %0<mask_scalar_operand4>, %1, %2
11781 <round_saeonly_scalar_mask_op4>, %3} ... this.
11782 * config/i386/subst.md (mask_scalar_operand4,
11783 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11784 round_saeonly_scalar_nimm_predicate): New subst attributes.
11786 2017-07-06 Julia Koval <julia.koval@intel.com>
11788 * config/i386/i386.c (ix86_erase_embedded_rounding):
11789 Remove code for old rounding pattern.
11791 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11793 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11795 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11797 * doc/sourcebuild.texi (Test Directives, Variants of
11798 dg-require-support): Add documentation for dg-require-stack-check.
11800 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11802 * config/i386/subst.md (mask_scalar, round_scalar,
11803 round_saeonly_scalar): New meta-templates.
11804 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11805 round_scalar_mask_operand3, round_scalar_mask_op3,
11806 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11807 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11808 round_saeonly_scalar_constraint,
11809 round_saeonly_scalar_prefix): New subst attribute.
11810 * config/i386/sse.md
11811 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11812 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11813 <round_scalar_name> ... this.
11814 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11815 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11816 <round_scalar_name> ... this.
11817 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11818 <sse>_vm<code><mode>3<mask_scalar_name>
11819 <round_saeonly_scalar_name> ... this.
11820 (v<plusminus_mnemonic><ssescalarmodesuffix>
11821 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11822 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11823 v<plusminus_mnemonic><ssescalarmodesuffix>
11824 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11825 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11826 (v<multdiv_mnemonic><ssescalarmodesuffix>
11827 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11828 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11829 v<multdiv_mnemonic><ssescalarmodesuffix>
11830 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11831 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11832 (v<maxmin_float><ssescalarmodesuffix>
11833 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11834 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11835 v<maxmin_float><ssescalarmodesuffix>
11836 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11837 %0<mask_scalar_operand3>, %1, %<iptr>2
11838 <round_saeonly_scalar_mask_op3>} ... this.
11840 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11842 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11843 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11845 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11846 Alan Hayward <alan.hayward@arm.com>
11847 David Sherwood <david.sherwood@arm.com>
11849 * combine.c (simplify_if_then_else): Remove "enum" before
11851 * compare-elim.c (can_eliminate_compare): Likewise.
11852 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11854 (aarch64_lookup_simd_builtin_type): Likewise.
11855 (aarch64_simd_builtin_type): Likewise.
11856 (aarch64_init_simd_builtin_types): Likewise.
11857 (aarch64_simd_expand_args): Likewise.
11858 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11860 (aarch64_reverse_mask): Likewise.
11861 (aarch64_simd_emit_reg_reg_move): Likewise.
11862 (aarch64_gen_adjusted_ldpstp): Likewise.
11863 (aarch64_ccmp_mode_to_code): Likewise.
11864 (aarch64_operands_ok_for_ldpstp): Likewise.
11865 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11866 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11868 (aarch64_min_divisions_for_recip_mul): Likewise.
11869 (aarch64_reassociation_width): Likewise.
11870 (aarch64_get_condition_code_1): Likewise.
11871 (aarch64_simd_emit_reg_reg_move): Likewise.
11872 (aarch64_simd_attr_length_rglist): Likewise.
11873 (aarch64_reverse_mask): Likewise.
11874 (aarch64_operands_ok_for_ldpstp): Likewise.
11875 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11876 (aarch64_gen_adjusted_ldpstp): Likewise.
11877 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11879 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11880 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11881 (arm_lookup_simd_builtin_type): Likewise.
11882 (arm_simd_builtin_type): Likewise.
11883 (arm_init_simd_builtin_types): Likewise.
11884 (arm_expand_builtin_args): Likewise.
11885 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11886 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11887 (ft32_setup_incoming_varargs): Likewise.
11888 (ft32_function_arg): Likewise.
11889 (ft32_function_arg_advance): Likewise.
11890 (ft32_pass_by_reference): Likewise.
11891 (ft32_arg_partial_bytes): Likewise.
11892 (ft32_valid_pointer_mode): Likewise.
11893 (ft32_addr_space_pointer_mode): Likewise.
11894 (ft32_addr_space_legitimate_address_p): Likewise.
11895 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11897 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11898 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11899 (iamcu_alignment): Likewise.
11900 (canonicalize_vector_int_perm): Likewise.
11901 (ix86_noce_conversion_profitable_p): Likewise.
11902 (ix86_mpx_bound_mode): Likewise.
11903 (ix86_operands_ok_for_move_multiple): Likewise.
11904 * config/microblaze/microblaze-protos.h
11905 (microblaze_expand_conditional_branch_reg): Likewise.
11906 * config/microblaze/microblaze.c
11907 (microblaze_expand_conditional_branch_reg): Likewise.
11908 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11910 (rs6000_reassociation_width): Likewise.
11911 (rs6000_invalid_binary_op): Likewise.
11912 (fusion_p9_p): Likewise.
11913 (emit_fusion_p9_load): Likewise.
11914 (emit_fusion_p9_store): Likewise.
11915 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11917 (riscv_hard_regno_mode_ok_p): Likewise.
11918 (riscv_address_insns): Likewise.
11919 (riscv_split_symbol): Likewise.
11920 (riscv_legitimize_move): Likewise.
11921 (riscv_function_value): Likewise.
11922 (riscv_hard_regno_nregs): Likewise.
11923 (riscv_expand_builtin): Likewise.
11924 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11925 (riscv_build_integer): Likewise.
11926 (riscv_split_integer): Likewise.
11927 (riscv_legitimate_constant_p): Likewise.
11928 (riscv_cannot_force_const_mem): Likewise.
11929 (riscv_regno_mode_ok_for_base_p): Likewise.
11930 (riscv_valid_base_register_p): Likewise.
11931 (riscv_valid_offset_p): Likewise.
11932 (riscv_valid_lo_sum_p): Likewise.
11933 (riscv_classify_address): Likewise.
11934 (riscv_legitimate_address_p): Likewise.
11935 (riscv_address_insns): Likewise.
11936 (riscv_load_store_insns): Likewise.
11937 (riscv_force_binary): Likewise.
11938 (riscv_split_symbol): Likewise.
11939 (riscv_force_address): Likewise.
11940 (riscv_legitimize_address): Likewise.
11941 (riscv_move_integer): Likewise.
11942 (riscv_legitimize_const_move): Likewise.
11943 (riscv_legitimize_move): Likewise.
11944 (riscv_address_cost): Likewise.
11945 (riscv_subword): Likewise.
11946 (riscv_output_move): Likewise.
11947 (riscv_canonicalize_int_order_test): Likewise.
11948 (riscv_emit_int_order_test): Likewise.
11949 (riscv_function_arg_boundary): Likewise.
11950 (riscv_pass_mode_in_fpr_p): Likewise.
11951 (riscv_pass_fpr_single): Likewise.
11952 (riscv_pass_fpr_pair): Likewise.
11953 (riscv_get_arg_info): Likewise.
11954 (riscv_function_arg): Likewise.
11955 (riscv_function_arg_advance): Likewise.
11956 (riscv_arg_partial_bytes): Likewise.
11957 (riscv_function_value): Likewise.
11958 (riscv_pass_by_reference): Likewise.
11959 (riscv_setup_incoming_varargs): Likewise.
11960 (riscv_print_operand): Likewise.
11961 (riscv_elf_select_rtx_section): Likewise.
11962 (riscv_save_restore_reg): Likewise.
11963 (riscv_for_each_saved_reg): Likewise.
11964 (riscv_register_move_cost): Likewise.
11965 (riscv_hard_regno_mode_ok_p): Likewise.
11966 (riscv_hard_regno_nregs): Likewise.
11967 (riscv_class_max_nregs): Likewise.
11968 (riscv_memory_move_cost): Likewise.
11969 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11970 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11971 (rl78_addr_space_address_mode): Likewise.
11972 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11974 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11975 (rs6000_reassociation_width): Likewise.
11976 (rs6000_invalid_binary_op): Likewise.
11977 (fusion_p9_p): Likewise.
11978 (emit_fusion_p9_load): Likewise.
11979 (emit_fusion_p9_store): Likewise.
11980 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11981 (ok_for_simple_move_operands): Likewise.
11982 (ok_for_simple_move_strict_operands): Likewise.
11983 (ok_for_simple_arith_logic_operands): Likewise.
11984 (visium_legitimize_reload_address): Likewise.
11985 (visium_select_cc_mode): Likewise.
11986 (output_cbranch): Likewise.
11987 (visium_split_double_move): Likewise.
11988 (visium_expand_copysign): Likewise.
11989 (visium_expand_int_cstore): Likewise.
11990 (visium_expand_fp_cstore): Likewise.
11991 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11992 (visium_function_arg): Likewise.
11993 (visium_function_arg_advance): Likewise.
11994 (visium_libcall_value): Likewise.
11995 (visium_setup_incoming_varargs): Likewise.
11996 (visium_legitimate_constant_p): Likewise.
11997 (visium_legitimate_address_p): Likewise.
11998 (visium_legitimize_address): Likewise.
11999 (visium_secondary_reload): Likewise.
12000 (visium_register_move_cost): Likewise.
12001 (visium_memory_move_cost): Likewise.
12002 (prepare_move_operands): Likewise.
12003 (ok_for_simple_move_operands): Likewise.
12004 (ok_for_simple_move_strict_operands): Likewise.
12005 (ok_for_simple_arith_logic_operands): Likewise.
12006 (visium_function_value_1): Likewise.
12007 (rtx_ok_for_offset_p): Likewise.
12008 (visium_legitimize_reload_address): Likewise.
12009 (visium_split_double_move): Likewise.
12010 (visium_expand_copysign): Likewise.
12011 (visium_expand_int_cstore): Likewise.
12012 (visium_expand_fp_cstore): Likewise.
12013 (visium_split_cstore): Likewise.
12014 (visium_select_cc_mode): Likewise.
12015 (visium_split_cbranch): Likewise.
12016 (output_cbranch): Likewise.
12017 (visium_print_operand_address): Likewise.
12018 * expmed.c (flip_storage_order): Likewise.
12019 * expmed.h (emit_cstore): Likewise.
12020 (flip_storage_order): Likewise.
12021 * genrecog.c (validate_pattern): Likewise.
12022 * hsa-gen.c (gen_hsa_addr): Likewise.
12023 * internal-fn.c (expand_arith_overflow): Likewise.
12024 * ira-color.c (allocno_copy_cost_saving): Likewise.
12025 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12026 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12027 (process_invariant_for_inheritance): Likewise.
12028 * lra-eliminations.c (move_plus_up): Likewise.
12029 * omp-low.c (lower_oacc_reductions): Likewise.
12030 * simplify-rtx.c (simplify_subreg): Likewise.
12031 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12032 (TARGET_CHKP_BOUND_MODE): Likewise..
12033 * targhooks.c (default_chkp_bound_mode): Likewise.
12034 (default_setup_incoming_vararg_bounds): Likewise.
12035 * targhooks.h (default_chkp_bound_mode): Likewise.
12036 (default_setup_incoming_vararg_bounds): Likewise.
12037 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12038 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12039 (have_whole_vector_shift): Likewise.
12040 * tree-vect-stmts.c (vectorizable_load): Likewise.
12041 * doc/tm.texi: Regenerate.
12043 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12045 Graceful degrade if Binutils PR21472 is not available.
12048 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12049 .rodata in flash test fails.
12050 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12051 * confgure: Regenerate.
12052 * config.in: Regenerate.
12053 * config/avr/avr.c (avr_asm_named_section)
12054 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12055 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12056 (avr_asm_init_sections): Same.
12058 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12060 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12061 (fma<VH:mode>4_intrinsic): Likewise.
12062 (*fmsub<VCVTF:mode>4): Likewise.
12063 (*fmsub<VH:mode>4_intrinsic): Likewise.
12065 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12068 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12069 Don't depend on "optimize > 0".
12070 (out_movhi_r_mr, out_movqi_mr_r): Same.
12071 (out_movhi_mr_r, out_movqi_r_mr): Same.
12072 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12073 io_address_operand on "optimize > 0".
12075 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12077 * tree-loop-distribution.c: Add general explanantion on the pass.
12078 (generate_loops_for_partition): Mark distributed loop.
12079 (pg_add_dependence_edges): New parameter. Handle alias data
12080 dependence specially and record it in the parameter if asked.
12081 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12082 (init_partition_graph_vertices, add_partition_graph_edge): New.
12083 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12084 (free_partition_graph_vdata, build_partition_graph): New.
12085 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12086 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12087 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12088 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12089 (version_for_distribution_p, finalize_partitions): New.
12090 (distribute_loop): Handle alias data dependence specially. Factor
12091 out loop fusion code as functions and call these functions.
12093 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12095 * tree-loop-distribution.c (classify_partition): New parameter and
12096 better handle reduction statement.
12097 (rdg_build_partitions): Revise comment.
12098 (distribute_loop): Compute statements in all partitions and pass it
12099 to classify_partition.
12101 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12103 * tree-loop-distribution.c (enum partition_type): New.
12104 (struct partition): New field type.
12105 (partition_merge_into): Add parameter. Update partition type.
12106 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12107 (build_rdg_partition_for_vertex): Compute partition type.
12108 (rdg_build_partitions): Dump partition type.
12109 (distribute_loop): Update calls to partition_merge_into.
12111 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12113 * tree-loop-distribution.c (struct ddr_hasher): New.
12114 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12116 (classify_partition): Call get_data_dependence.
12117 (pg_add_dependence_edges): Ditto.
12118 (distribute_loop): Release data dependence hash table.
12120 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12122 * tree-loop-distribution.c (ref_base_address): Delete.
12123 (similar_memory_accesses): Rename ...
12124 (share_memory_accesses): ... to this. Check if partitions access
12125 the same memory reference.
12126 (distribute_loop): Call share_memory_accesses.
12128 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12130 * tree-loop-distribution.c (struct partition): New field recording
12131 its data reference.
12132 (partition_alloc, partition_free): Init and release data refs.
12133 (partition_merge_into): Merge data refs.
12134 (build_rdg_partition_for_vertex): Collect data refs for partition.
12135 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12137 (distribute_loop): Remve data refs from vertice data of partition
12140 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12142 * tree-loop-distribution.c (params.h): Include header file.
12143 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12144 (datarefs_vec): New global var.
12145 (create_rdg_vertices): Use datarefs_vec directly.
12146 (free_rdg): Don't free data references.
12147 (build_rdg): Update use. Don't free data references.
12148 (distribute_loop): Compute global variable for data references.
12149 Bail out if there are too many data references.
12151 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12153 * tree-loop-distribution.c (loop_nest): New global var.
12154 (build_rdg): Use loop directly, rather than loop nest.
12155 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12157 (distribute_loop): Compute global variable loop nest. Update use.
12159 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12161 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12162 (partition_merge_into): New parameter. Dump reason for fusion.
12163 (distribute_loop): Update use of partition_merge_into.
12165 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12167 * tree-loop-distribution.c (bb_top_order_index): New.
12168 (bb_top_order_index_size, bb_top_order_cmp): New.
12169 (stmts_from_loop): Use topological order.
12170 (pass_loop_distribution::execute): Compute and release topological
12171 order for basic blocks.
12173 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12175 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12178 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12180 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12181 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12182 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12183 * internal-fn.def (LOOP_DIST_ALIAS): New.
12184 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12185 (fold_loop_internal_call): ... this.
12186 (vect_loop_dist_alias_call): New function.
12187 (set_uid_loop_bbs): Call fold_loop_internal_call.
12188 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12191 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12194 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12195 Require dead FLAGS_REG at the beginning of a peephole.
12197 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12200 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12201 arguments in the call to __builtin_ia32_sbb_u32.
12202 (_subborrow_u64): Swap _X and _Y arguments in the call to
12203 __builtin_ia32_sbb_u64.
12205 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12208 * tree-vrp.c (compare_assert_loc): Turn into a function template
12209 with stable template parameter. Only test if a->e is NULL,
12210 !a->e == !b->e has been verified already. Use e == NULL or
12211 e != NULL instead of e or ! e tests. If stable is true, don't use
12212 iterative_hash_expr, on the other side allow a or b or both NULL
12213 and sort the NULLs last.
12214 (process_assert_insertions): Sort using compare_assert_loc<false>
12215 instead of compare_assert_loc, later sort using
12216 compare_assert_loc<true> before calling process_assert_insertions_for
12217 in a loop. Use break instead of continue once seen NULL pointer.
12219 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12221 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12222 Cortex-R7 and Cortex-R8 processors.
12224 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12226 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12227 uninitialized while src is not.
12229 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12231 * common/config/arm/arm-common.c: Adjust include path for
12233 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12234 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12235 (arm-cpu-data.h): Likewise.
12236 (arm-cpu-cdata.h): Likewise.
12237 * config/arm/arm-cpu.h: Delete.
12238 * config/arm/arm-cpu-cdata.h: Delete.
12239 * config/arm/arm-cpu-data.h: Delete.
12241 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12243 * config/arm/arm-cpus.in (cortex-a55): New.
12244 (cortex-a75): Likewise.
12245 (cortex-a75.cortex-a55): Likewise.
12246 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12248 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12249 * config/arm/arm-cpu-cdata.h: Regenerate.
12250 * config/arm/arm-cpu-data.h: Regenerate.
12251 * config/arm/arm-cpu.h: Regenerate.
12252 * config/arm/arm-tables.opt: Regenerate.
12253 * config/arm/arm-tune.md: Regenerate.
12255 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12257 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12259 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12261 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12264 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12266 PR tree-optimization/81292
12267 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12268 full_string_p, also call adjust_related_strinfos if the adjustment
12269 is simple, otherwise invalidate related strinfos.
12271 2017-07-04 Martin Liska <mliska@suse.cz>
12274 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12275 newly created variable as DECL_IGNORED_P.
12277 2017-07-04 Martin Liska <mliska@suse.cz>
12280 * ipa-inline.c (inline_small_functions):
12281 Use xstrdup_for_dump.
12283 2017-07-04 Tom de Vries <tom@codesourcery.com>
12285 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12287 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12290 * config/darwin.c (darwin_function_switched_text_sections):
12293 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12295 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12297 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12299 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12301 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12303 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12304 min_profitable_iters, and th as inclusive lower bounds.
12305 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12306 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12307 for min_profitable_iters and min_profitable_estimate.
12308 (vect_transform_loop): Treat th as an inclusive lower bound.
12309 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12311 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12314 * config/darwin.c (darwin_function_switched_text_sections):
12315 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12316 in two pieces, and suppress the use of buf.
12318 2017-07-03 Nathan Sidwell <nathan@acm.org>
12320 * hash-table.h (hash_table_mod1): Fix indentation.
12322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12324 PR middle-end/81290
12325 * predict.c (force_edge_cold): Be more careful about propagation
12327 * profile-count.h (profile_probability::guessed,
12328 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12330 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12332 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12334 * doc/invoke.texi (rcpc architecture extension): Document it.
12336 2017-07-03 Richard Biener <rguenther@suse.de>
12338 PR tree-optimization/60510
12339 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12340 the scalar reduction PHI and use it.
12341 (vectorizable_reduction): Properly guard the single_defuse_cycle
12342 path for non-SLP reduction chains where we cannot use it.
12343 Rework reduc_def/index and vector type deduction. Rework
12344 vector operand gathering during reduction op code-gen.
12345 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12346 chains dissolve the chain and leave it to non-SLP reduction
12349 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12351 * tree-data-ref.h (dr_alignment): Declare.
12352 * tree-data-ref.c (dr_alignment): New function.
12353 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12354 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12356 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12358 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12360 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12361 and base_misalignment fields.
12362 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12363 * tree-data-ref.c: Include builtins.h.
12364 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12365 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12366 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12367 * tree-vect-data-refs.c: Include tree-cfg.h.
12368 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12369 fields instead of calculating an alignment here.
12370 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12371 innermost_loop_behavior fields.
12373 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12375 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12377 (DR_STEP_ALIGNMENT): New macro.
12378 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12379 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12380 (create_data_ref): Print it.
12381 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12382 to tell whether the step preserves vector (mis)alignment.
12383 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12384 Move the check for an integer step and generalise to all INTEGER_CST.
12385 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12386 Print the outer step alignment.
12388 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12390 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12391 with offset_alignment.
12392 (DR_ALIGNED_TO): Delete.
12393 (DR_OFFSET_ALIGNMENT): New macro.
12394 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12395 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12396 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12397 (create_data_ref): Likewise.
12398 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12399 (vect_analyze_data_refs): Likewise.
12400 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12401 creating dummy innermost behavior.
12403 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12405 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12406 with a "innermost_loop_behavior *" and refeence tree.
12407 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12408 (create_data_ref): Update call accordingly.
12409 * tree-predcom.c (find_looparound_phi): Likewise.
12411 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12413 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12414 fields with dr_wrt_vec_loop.
12415 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12416 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12417 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12418 (vect_dr_behavior): New function.
12419 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12420 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12421 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12422 track whether the step preserves the misalignment.
12423 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12424 Use vect_dr_behavior.
12425 (vect_setup_realignment): Update call accordingly.
12426 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12427 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12428 call to vect_create_addr_base_for_vector_ref.
12429 (vect_create_cond_for_align_checks): Likewise.
12430 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12431 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12432 (vect_recog_mask_conversion_pattern): Likewise.
12433 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12434 (new_stmt_vec_info): Remove redundant zeroing.
12436 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12438 * common/config/arm/arm-common.c (arm_be8_option): New function.
12439 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12440 (ISA_ARMv6): Add isa_bit_be8.
12441 * config/arm/arm.h (arm_be8_option): Add prototype.
12442 (BE8_SPEC_FUNCTION): New define.
12443 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12444 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12445 (mlittle-endian): Similarly.
12446 (mbe8, mbe32): New options.
12447 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12448 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12450 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12452 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12454 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12456 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12457 (cleanup_tree_cfg_bb): Use it.
12458 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12460 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12462 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12465 * loop-doloop.c (add_test): Update profile.
12467 2017-07-03 Martin Liska <mliska@suse.cz>
12470 * sanopt.c (rewrite_usage_of_param): New function.
12471 (sanitize_rewrite_addressable_params): Likewise.
12472 (pass_sanopt::execute): Call rewrite_usage_of_param.
12474 2017-07-03 Richard Biener <rguenther@suse.de>
12476 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12477 back to using VIEW_CONVERT_EXPR.
12479 2017-07-03 Martin Liska <mliska@suse.cz>
12482 * doc/extend.texi: Document when a resolver function is
12483 generated for target_clones.
12485 2017-07-03 Martin Liska <mliska@suse.cz>
12487 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12488 and shadow memory of auto variables which are subject of
12489 use-after-scope sanitization.
12490 (asan_expand_mark_ifn): Add do set only when is_poison.
12492 2016-07-03 Richard Biener <rguenther@suse.de>
12494 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12496 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12497 (vectorizable_reduction): Perform reduction PHI creation when
12498 visiting a reduction PHI and adjust and simplify code generation
12499 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12500 (vect_transform_loop): Visit reduction PHIs.
12501 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12502 defs into the SLP tree.
12503 (vect_build_slp_tree): Reduction defs terminate the recursion.
12504 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12506 (vect_get_vec_defs_for_stmt_copy): Export.
12507 (vect_get_vec_defs): Likewise.
12508 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12510 (vect_get_vec_defs_for_stmt_copy): Declare.
12511 (vect_get_vec_defs): Likewise.
12513 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12515 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12516 parameter with a "loop" parameter and use it instead of the
12517 loop containing DR_STMT. Don't check simple_iv when doing
12518 BB analysis. Describe the two analysis modes in the comment.
12520 2017-07-03 Tom de Vries <tom@codesourcery.com>
12522 PR tree-optimization/69468
12523 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12524 (find_same_succ_bb): Handle ignore_edge_flags.
12526 2017-07-03 Tom de Vries <tom@codesourcery.com>
12528 PR tree-optimization/81192
12529 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12531 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12533 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12535 2017-07-03 Tom de Vries <tom@codesourcery.com>
12537 PR tree-optimization/81192
12538 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12539 BB_SAME_SUCC (bb) == NULL.
12541 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12543 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12546 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12548 * dumpfile.c: Include profile-count.h
12549 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12551 (insert_cond_bb): Update profile.
12552 * tree-cfg.h (insert_cond_bb): Update prototype.
12553 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12554 * tree-dump.c: Do not include tree-cfg.
12556 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12558 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12560 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12562 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12565 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12567 * tree-complex.c (expand_complex_div_wide): update profile.
12569 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12570 Alan Hayward <alan.hayward@arm.com>
12571 David Sherwood <david.sherwood@arm.com>
12573 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12574 (CORETYPES_H): New define.
12575 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12576 (insn-modes-inline.h, s-modes-inline-h): New rules.
12577 (generated_files): Add insn-modes-inline.h.
12578 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12579 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12580 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12581 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12582 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12583 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12584 (build/gencodes.o, build/genconditions.o): Likewise.
12585 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12586 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12587 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12588 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12589 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12590 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12591 * coretypes.h: Include everything up to real.h for generators.
12592 Include insn-modes.h first. Include wide-int-print.h after
12593 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12594 * machmode.h: Don't include insn-modes.h here.
12595 * function-tests.c: Remove includes of signop.h, machmode.h,
12596 double-int.h and wide-int.h.
12598 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12600 * optc-save-gen.awk: Likewise.
12601 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12602 * godump.c: Remove include of wide-int-print.h.
12603 * pretty-print.h: Likewise.
12604 * wide-int-print.cc: Likewise.
12605 * wide-int.cc: Likewise.
12606 * hash-map-tests.c: Remove include of signop.h.
12607 * hash-set-tests.c: Likewise.
12608 * rtl-tests.c: Likewise.
12609 * mkconfig.sh: Remove include of machmode.h.
12610 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12612 (emit_insn_modes_inline_h): ...this new function. Emit the code
12613 into an insn-modes-inline.h header file, adding appropriate
12614 include guards and end comments.
12615 (emit_insn_modes_c_header): Remove include of machmode.h.
12616 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12618 (main): Handle -i flag and call emit_insn_modes_inline_h when
12621 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12623 * tree-ssa-strlen.c (strinfo): Rename the length field to
12624 nonzero_chars. Add a full_string_p field.
12625 (compare_nonzero_chars, zero_length_string_p): New functions.
12626 (get_addr_stridx): Add an offset_out parameter.
12627 Use compare_nonzero_chars.
12628 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12629 (new_strinfo): Update after above changes to strinfo.
12630 (set_endptr_and_length): Set full_string_p.
12631 (get_string_length): Update after above changes to strinfo.
12632 (unshare_strinfo): Update call to new_strinfo.
12633 (maybe_invalidate): Likewise.
12634 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12635 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12636 as a uhwi instead of an shwi. Update after above changes to
12637 strinfo and new_strinfo.
12638 (zero_length_string): Assert that chainsi contains full strings.
12639 Use zero_length_string_p. Update call to new_strinfo.
12640 (adjust_related_strinfos): Update after above changes to strinfo.
12641 Copy full_string_p from origsi.
12642 (adjust_last_stmt): Use zero_length_string_p.
12643 (handle_builtin_strlen): Update after above changes to strinfo and
12644 new_strinfo. Install the lhs as the string length if the previous
12645 entry didn't describe a full string.
12646 (handle_builtin_strchr): Update after above changes to strinfo
12648 (handle_builtin_strcpy): Likewise.
12649 (handle_builtin_strcat): Likewise.
12650 (handle_builtin_malloc): Likewise.
12651 (handle_pointer_plus): Likewise.
12652 (handle_builtin_memcpy): Likewise. Track nonzero characters
12653 that aren't necessarily followed by a nul terminator.
12654 (handle_char_store): Likewise.
12656 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12658 PR tree-optimization/80769
12659 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12660 for malloc and calloc. Document the new invariant that all related
12661 strinfos have delayed lengths or none do.
12662 (verify_related_strinfos): Move earlier in file.
12663 (set_endptr_and_length): New function, split out from...
12664 (get_string_length): ...here. Also set the lengths of related
12666 (zero_length_string): Assert that chainsi has known (rather than
12668 (adjust_related_strinfos): Likewise.
12670 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12672 PR tree-optimization/81136
12673 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12674 assert that two references with the same misalignment have the same
12675 compile-time misalignment if those compile-time misalignments
12678 2017-07-01 Andi Kleen <ak@linux.intel.com>
12680 * print-tree.c (print_node): Print all attributes.
12682 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12684 * cfg.c (scale_bbs_frequencies): New function.
12685 * cfg.h (scale_bbs_frequencies): Declare it.
12686 * cfgloopanal.c (single_likely_exit): Cleanup.
12687 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12689 (scale_loop_profile): Likewise.
12690 (loop_version): Likewise.
12691 (create_empty_loop_on_edge): Update.
12692 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12693 scale_loop_frequencies, scale_loop_profile, loopify,
12694 loop_version): Update prototypes.
12695 * modulo-sched.c (sms_schedule): Update.
12696 * predict.c (unlikely_executed_edge_p): Also check probability.
12697 (probably_never_executed_edge_p): Fix typo.
12698 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12699 * tree-parloops.c (gen_parallel_loop): Update.
12700 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12701 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12702 * tree-ssa-loop-split.c (split_loop): Update.
12703 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12704 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12705 (vect_loop_versioning): Update.
12706 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12708 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12710 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12712 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12714 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12715 to keep profile consistent.
12717 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12719 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12720 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12721 * profile-count.h (max_safe_multiplier): Make unsigned.
12722 (profile_count::guessed_zero): New.
12724 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12726 * bb-reorder.c (fix_up_crossing_landing_pad,
12727 fix_crossing_conditional_branches): Use make_single_succ_edge
12728 to keep profile consistent.
12730 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12732 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12735 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12738 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12739 the right scopes, make sure cond_jump isn't preserved between multiple
12740 iterations. Search for fallthru edge whenever there are 3+ edges and
12741 use find_fallthru_edge for it.
12743 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12745 Patch by Alexander Monakov <amonakov@ispras.ru>
12746 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12747 probabilities consistently.
12749 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12751 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12752 emit_cmp_and_jump_insns.
12754 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12757 * tree-inline.c (expand_call_inline): Combine profile statuses.
12759 2017-06-30 Andrew Pinski <apinski@cavium.com>
12761 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12762 fold_stmt returned true.
12764 2017-06-30 Nathan Sidwell <nathan@acm.org>
12766 * ggc.h (empty_string): Delete.
12767 * cfgexpand.c (expand_asm_stmt): Use plain "".
12768 * optabs.c (expand_asm_memory_barrier): Likewise.
12769 * stringpool.c (empty_string): Delete.
12770 (digit_vector, digit_string): Delete.
12771 (ggc_alloc_string): Use plain "", don't optimize single digit
12772 strings. Use ggc_alloc_atomic.
12774 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12776 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12777 comparison set and one other set, use the cost of the non-comparison
12780 2017-06-30 Nathan Sidwell <nathan@acm.org>
12782 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12785 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12787 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12788 loops. Remove now unneeded calls to gimple_switch_set_label() that
12789 just set removed labels to NULL_TREE.
12791 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12793 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12794 (set_range_info): ...here. Only call set_range_info_raw if domain
12796 (set_nonzero_bits): Call set_range_info_raw.
12797 * tree-ssanames.h (set_range_info_raw): New.
12799 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12802 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12803 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12804 of nonimmediate_operand and <store_mask_constraint> instead of m
12805 for the input operand. For V8FI iterator, always split if input
12806 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12807 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12808 <store_mask_predicate> instead of register_operand and
12809 <store_mask_constraint> instead of v for the input operand. Make
12810 sure both operands aren't MEMs for if not <mask_applied>.
12812 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12814 * lto-wrapper.c (copy_file) Close both file descriptors before
12817 2017-06-30 Martin Liska <mliska@suse.cz>
12820 * multiple_target.c (create_dispatcher_calls): Make ifunc
12821 also for function that don't have calls or are not referenced.
12823 2017-06-30 Richard Biener <rguenther@suse.de>
12825 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12826 analyze the first scalar stmt. Move vector type computation
12827 for the BB case here from ...
12828 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12829 live operation processing in the SLP case properly.
12831 2017-06-30 Richard Biener <rguenther@suse.de>
12833 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12835 2017-06-30 Martin Liska <mliska@suse.cz>
12838 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12839 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12841 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12843 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12846 2017-06-30 Richard Biener <rguenther@suse.de>
12848 PR tree-optimization/81249
12849 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12850 condition reduction result to original scalar type.
12852 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12854 * profile-count.h (enum profile_quality): Fix typos and whitespace
12857 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12859 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12860 type for branch probabilities.
12862 2017-06-29 Julian Brown <julian@codesourcery.com>
12863 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12865 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12866 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12867 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12868 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12870 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12872 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12873 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12874 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12875 CC usage from generic code to here.
12876 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12877 CC usage into the target macros.
12879 2017-06-29 Maya Rashish <coypu@sdf.org>
12881 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12884 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12886 * arm/arm-builtins.c: Include profile-count.h
12887 * except.c (sjlj_emit_function_enter): Use
12888 profile_probability::unlikely.
12890 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12892 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12893 and tocrel_offset be pointer args rather than implicitly using
12895 (legitimate_constant_pool_address_p, rs6000_emit_move,
12896 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12897 tocrel_offset and use in toc_relative_expr_p call.
12898 (print_operand, print_operand_address): Use static tocrel_base_oac
12899 and tocrel_offset_oac.
12900 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12903 2017-06-29 Maya Rashish <coypu@sdf.org>
12905 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12907 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12909 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12910 objects, take into account only the alignment of 'op0' and 'mode1' if
12913 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12915 * ccmp.c (ccmp_tree_comparison_p): New function.
12916 (ccmp_candidate_p): Update to use above function.
12917 (get_compare_parts): New function.
12918 (expand_ccmp_next): Update to use new functions.
12919 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12921 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12922 take mode as argument.
12923 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12924 * expr.c (expand_expr_real_1): Pass mode as argument.
12926 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12928 * combine.c (combine_instructions): Print insns to dump_file, together
12931 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12933 * asan.c (asan_emit_stack_protection): Update.
12934 (create_cond_insert_point): Update.
12935 * auto-profile.c (afdo_propagate_circuit): Update.
12936 * basic-block.h (struct edge_def): Turn probability to
12937 profile_probability.
12938 (EDGE_FREQUENCY): Update.
12939 * bb-reorder.c (find_traces_1_round): Update.
12940 (better_edge_p): Update.
12941 (sanitize_hot_paths): Update.
12942 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12943 (make_single_succ_edge): Update.
12944 (check_bb_profile): Update.
12945 (dump_edge_info): Update.
12946 (update_bb_profile_for_threading): Update.
12947 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12948 probabilitycount to 0.
12949 * cfgbuild.c (compute_outgoing_frequencies): Update.
12950 * cfgcleanup.c (try_forward_edges): Update.
12951 (outgoing_edges_match): Update.
12952 (try_crossjump_to_edge): Update.
12953 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12954 (expand_gimple_tailcall): Update.
12955 (construct_init_block): Use make_single_succ_edge.
12956 (construct_exit_block): Use make_single_succ_edge.
12957 * cfghooks.c (verify_flow_info): Update.
12958 (redirect_edge_succ_nodup): Update.
12959 (split_edge): Update.
12960 (account_profile_record): Update.
12961 * cfgloopanal.c (single_likely_exit): Update.
12962 * cfgloopmanip.c (scale_loop_profile): Update.
12963 (set_zero_probability): Remove.
12964 (duplicate_loop_to_header_edge): Update.
12965 * cfgloopmanip.h (loop_version): Update prototype.
12966 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12967 (force_nonfallthru_and_redirect): Update.
12968 (update_br_prob_note): Update.
12969 (rtl_verify_edges): Update.
12970 (purge_dead_edges): Update.
12971 (rtl_lv_add_condition_to_bb): Update.
12972 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12973 * cgraphunit.c (init_lowered_empty_function): Update.
12974 (cgraph_node::expand_thunk): Update.
12975 * cilk-common.c: Include profile-count.h
12976 * dojump.c (inv): Remove.
12977 (jumpifnot): Update.
12978 (jumpifnot_1): Update.
12979 (do_jump_1): Update.
12981 (do_jump_by_parts_greater_rtx): Update.
12982 (do_compare_rtx_and_jump): Update.
12983 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12984 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12985 * dwarf2cfi.c: Include profile-count.h
12986 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12987 (sjlj_emit_dispatch_table): Likewise.
12988 * explow.c: Include profile-count.h
12989 * expmed.c (emit_store_flag_force): Update.
12990 (do_cmp_and_jump): Update.
12991 * expr.c (compare_by_pieces_d::generate): Update.
12992 (compare_by_pieces_d::finish_mode): Update.
12993 (emit_block_move_via_loop): Update.
12994 (store_expr_with_bounds): Update.
12995 (store_constructor): Update.
12996 (expand_expr_real_2): Update.
12997 (expand_expr_real_1): Update.
12998 * expr.h (try_casesi, try_tablejump): Update prototypes.
12999 * gimple-pretty-print.c (dump_probability): Update.
13000 (dump_profile): New.
13001 (dump_gimple_label): Update.
13002 (dump_gimple_bb_header): Update.
13003 * graph.c (draw_cfg_node_succ_edges): Update.
13004 * hsa-gen.c (convert_switch_statements): Update.
13005 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13006 (find_if_case_1): Update.
13007 (find_if_case_2): Update.
13008 * internal-fn.c (expand_arith_overflow_result_store): Update.
13009 (expand_addsub_overflow): Update.
13010 (expand_neg_overflow): Update.
13011 (expand_mul_overflow): Update.
13012 (expand_vector_ubsan_overflow): Update.
13013 * ipa-cp.c (good_cloning_opportunity_p): Update.
13014 * ipa-split.c (split_function): Use make_single_succ_edge.
13015 * ipa-utils.c (ipa_merge_profiles): Update.
13016 * loop-doloop.c (add_test): Update.
13017 (doloop_modify): Update.
13018 * loop-unroll.c (compare_and_jump_seq): Update.
13019 (unroll_loop_runtime_iterations): Update.
13020 * lra-constraints.c (lra_inheritance): Update.
13021 * lto-streamer-in.c (input_cfg): Update.
13022 * lto-streamer-out.c (output_cfg): Update.
13023 * mcf.c (adjust_cfg_counts): Update.
13024 * modulo-sched.c (sms_schedule): Update.
13025 * omp-expand.c (expand_omp_for_init_counts): Update.
13026 (extract_omp_for_update_vars): Update.
13027 (expand_omp_ordered_sink): Update.
13028 (expand_omp_for_ordered_loops): Update.
13029 (expand_omp_for_generic): Update.
13030 (expand_omp_for_static_nochunk): Update.
13031 (expand_omp_for_static_chunk): Update.
13032 (expand_cilk_for): Update.
13033 (expand_omp_simd): Update.
13034 (expand_omp_taskloop_for_outer): Update.
13035 (expand_omp_taskloop_for_inner): Update.
13036 * omp-simd-clone.c (simd_clone_adjust): Update.
13037 * optabs.c (expand_doubleword_shift): Update.
13038 (expand_abs): Update.
13039 (emit_cmp_and_jump_insn_1): Update.
13040 (expand_compare_and_swap_loop): Update.
13041 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13042 * predict.c (predictable_edge_p): Update.
13043 (edge_probability_reliable_p): Update.
13044 (set_even_probabilities): Update.
13045 (combine_predictions_for_insn): Update.
13046 (combine_predictions_for_bb): Update.
13047 (propagate_freq): Update.
13048 (estimate_bb_frequencies): Update.
13049 (force_edge_cold): Update.
13050 * profile-count.c (profile_count::dump): Add missing space into dump.
13051 (profile_count::debug): Add newline.
13052 (profile_count::differs_from_p): Explicitly convert to unsigned.
13053 (profile_count::stream_in): Update.
13054 (profile_probability::dump): New member function.
13055 (profile_probability::debug): New member function.
13056 (profile_probability::differs_from_p): New member function.
13057 (profile_probability::differs_lot_from_p): New member function.
13058 (profile_probability::stream_in): New member function.
13059 (profile_probability::stream_out): New member function.
13060 * profile-count.h (profile_count_quality): Rename to ...
13061 (profile_quality): ... this one.
13062 (profile_probability): New.
13063 (profile_count): Update.
13064 * profile.c (compute_branch_probabilities): Update.
13065 * recog.c (peep2_attempt): Update.
13066 * sched-ebb.c (schedule_ebbs): Update.
13067 * sched-rgn.c (find_single_block_region): Update.
13068 (compute_dom_prob_ps): Update.
13069 (schedule_region): Update.
13070 * sel-sched-ir.c (compute_succs_info): Update.
13071 * stmt.c (struct case_node): Update.
13072 (do_jump_if_equal): Update.
13073 (get_outgoing_edge_probs): Update.
13074 (conditional_probability): Update.
13075 (emit_case_dispatch_table): Update.
13076 (expand_case): Update.
13077 (expand_sjlj_dispatch_table): Update.
13078 (emit_case_nodes): Update.
13079 * targhooks.c: Update.
13080 * tracer.c (better_p): Update.
13081 (find_best_successor): Update.
13082 * trans-mem.c (expand_transaction): Update.
13083 * tree-call-cdce.c: Update.
13084 * tree-cfg.c (gimple_split_edge): Upate.
13085 (move_sese_region_to_fn): Upate.
13086 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13087 * tree-eh.c (lower_resx): Upate.
13088 (cleanup_empty_eh_move_lp): Upate.
13089 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13090 * tree-inline.c (copy_edges_for_bb): Update.
13091 (copy_cfg_body): Update.
13092 * tree-parloops.c (gen_parallel_loop): Update.
13093 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13094 (gimple_gen_time_profiler): Update.
13095 * tree-ssa-dce.c (remove_dead_stmt): Update.
13096 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13097 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13098 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13099 (unloop_loops): Update.
13100 (try_peel_loop): Update.
13101 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13102 * tree-ssa-loop-split.c (connect_loops): Update.
13103 (split_loop): Update.
13104 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13105 (hoist_guard): Update.
13106 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13107 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13108 (value_replacement): Update.
13109 * tree-ssa-reassoc.c (branch_fixup): Update.
13110 * tree-ssa-tail-merge.c (replace_block_by): Update.
13111 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13112 (create_edge_and_update_destination_phis): Update.
13113 (compute_path_counts): Update.
13114 (recompute_probabilities): Update.
13115 (update_joiner_offpath_counts): Update.
13116 (freqs_to_counts_path): Update.
13117 (duplicate_thread_path): Update.
13118 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13119 (struct switch_conv_info): Update.
13120 (gen_inbound_check): Update.
13121 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13122 (vect_do_peeling): Update.
13123 (vect_loop_versioning): Update.
13124 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13125 (optimize_mask_stores): Update.
13126 * ubsan.c (ubsan_expand_null_ifn): Update.
13127 * value-prof.c (gimple_divmod_fixed_value): Update.
13128 (gimple_divmod_fixed_value_transform): Update.
13129 (gimple_mod_pow2): Update.
13130 (gimple_mod_pow2_value_transform): Update.
13131 (gimple_mod_subtract): Update.
13132 (gimple_mod_subtract_transform): Update.
13133 (gimple_ic): Update.
13134 (gimple_stringop_fixed_value): Update.
13135 (gimple_stringops_transform): Update.
13136 * value-prof.h: Update.
13138 2017-06-29 Carl Love <cel@us.ibm.com>
13140 * config/rs6000/rs6000-c.c: Add support for built-in functions
13141 vector signed int vec_signed (vector float);
13142 vector signed long long vec_signed (vector double);
13143 vector signed int vec_signed2 (vector double, vector double);
13144 vector signed int vec_signede (vector double);
13145 vector signed int vec_signedo (vector double);
13146 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13147 instruction generator.
13148 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13149 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13150 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13152 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13153 vunsignede_v2df): Add define_expands.
13154 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13155 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13156 VEC_UNSIGNEDO): Add definitions.
13157 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13158 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13159 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13160 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13161 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13162 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13163 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13164 vec_unsignede, vec_unsignedo): Add builtin defines.
13165 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13167 * doc/extend.texi: Update the built-in documentation file for the
13168 new built-in functions.
13170 2017-06-29 Richard Biener <rguenther@suse.de>
13172 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13173 reduction chains to LOOP_VINFO_REDUCTIONS.
13174 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13175 SLP reductions after processing reduction chains.
13177 2017-06-29 Nathan Sidwell <nathan@acm.org>
13179 * builtins.c (fold_builtin_FUNCTION): Use
13180 lang_hooks.decl_printable_name.
13182 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13184 PR middle-end/81194
13185 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13186 with only one label.
13187 * stmt.c (expand_case): Assert NCASES is greater than one.
13189 2017-06-29 Richard Biener <rguenther@suse.de>
13191 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13193 (group_case_labels): Likewise.
13194 (find_taken_edge): Push sanity checking on val to workers...
13195 (find_taken_edge_cond_expr): ... here
13196 (find_taken_edge_switch_expr): ... and here, handle cases
13197 with just a default label.
13198 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13199 (group_case_labels): Likewise.
13200 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13201 group_case_labels does anything cleanup the CFG again.
13203 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13205 PR tree-optimization/81196
13206 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13207 exit condition comparing two IVs.
13209 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13211 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13212 profile to the dummy entry at the end of the list of architectures.
13213 * config/arm/arm-cpu-cdata.h: Regenerated.
13215 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13216 Michael Collison <michael.collison@arm.com>
13219 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13221 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13222 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13223 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13224 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13225 of shift when the shift amount is masked with constant equal to
13226 the size of the mode.
13227 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13230 2017-06-29 Martin Liska <mliska@suse.cz>
13232 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13234 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13237 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13240 (edge_growth_cache_entry::edge_growth_cache_entry): New
13242 (reset_edge_growth_cache): Update to use constructor.
13244 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13246 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13247 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13248 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13250 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13252 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13253 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13255 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13257 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13258 (*-linux-uclibc*): Add t-uclibc tmake_file.
13259 * config/t-musl: New.
13260 * config/t-uclibc: New.
13262 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13264 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13266 (gen_comm_data): Emit architectural setting of arch_prof.
13267 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13269 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13270 (armv8-m.base, armv8-m.main): Likewise.
13271 * arm-protos.h (arm_build_target): Add profile field.
13272 (arch_option): Likewise.
13273 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13275 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13276 arm_active_target.profile.
13278 2017-06-28 Richard Biener <rguenther@suse.de>
13280 PR middle-end/81227
13281 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13282 TYPE_OVERFLOW_WRAPS.
13283 * match.pd (negate_expr_p): Likewise.
13284 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13285 fold_build2, not fold_binary.
13287 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13289 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13290 Convert memory address to Pmode.
13291 (aarch64_print_operand): Assert MEM operands are always Pmode.
13293 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13296 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13297 Remove redundant if.
13298 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13299 * config/arm/aarch-common-protos.h
13300 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13301 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13303 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13306 * multiple_target.c (create_dispatcher_calls): Set the default
13307 clone to be static, not public.
13309 2017-06-28 Richard Biener <rguenther@suse.de>
13311 * tree-vect-loop.c (vectorizable_reduction): Move special
13312 cond reduction IV var creation ...
13313 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13314 parameter. Use STMT_VINFO_VECTYPE.
13315 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13318 2017-06-28 Martin Liska <mliska@suse.cz>
13321 * ipa-visibility.c (non_local_p): Handle visibility.
13323 2017-06-28 Martin Liska <mliska@suse.cz>
13326 * common.opt: Add IntegerRange to various options.
13327 * opt-functions.awk (integer_range_info): New function.
13328 * optc-gen.awk: Add integer_range_info to cl_options struct.
13329 * opts-common.c (decode_cmdline_option): Handle
13330 CL_ERR_INT_RANGE_ARG.
13331 (cmdline_handle_error): Likewise.
13332 * opts.c (print_filtered_help): Show valid interval in
13333 when --help is provided.
13334 * opts.h (struct cl_option): Add range_min and range_max fields.
13335 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13337 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13339 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13340 (x * C EQ/NE y * C): New transformation.
13342 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13344 * genmultilib (combination_space): Accept '+' in option names.
13346 2017-06-28 Martin Liska <mliska@suse.cz>
13349 * asan.c (instrument_derefs): Bail out inner references
13350 that are hard register variables.
13352 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13355 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13356 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13358 2017-06-28 Richard Biener <rguenther@suse.de>
13360 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13361 (vect_get_slp_defs): Adjust.
13362 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13364 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13366 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13367 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13368 (vectorizable_reduction): Adjust.
13369 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13371 (vect_get_slp_defs): Likewise.
13372 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13373 (vectorizable_bswap): Adjust.
13374 (vectorizable_call): Likewise.
13375 (vectorizable_conversion): Likewise.
13376 (vectorizable_assignment): Likewise.
13377 (vectorizable_shift): Likewise.
13378 (vectorizable_operation): Likewise.
13379 (vectorizable_store): Likewise.
13380 (vectorizable_condition): Likewise.
13381 (vectorizable_comparison): Likewise.
13383 2017-06-28 Michael Collison <michael.collison@arm.com>
13386 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13388 (arm_gen_movmemqi): Removed unused variable 'i'.
13389 Convert 'for' loop into 'while' loop.
13390 (arm_expand_prologue): Remove last unnecessary set of insn.
13391 (thumb_pop): Remove unused variable 'pushed_words'.
13392 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13394 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13396 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13397 * config/s390/s390-protos.h: Add prototype of
13398 s390_rel_address_ok_p.
13399 * config/s390/s390.c (s390_got_symbol): New function.
13400 (s390_rel_address_ok_p): New function.
13401 (legitimize_pic_address): Use s390_rel_address_ok_p.
13402 (s390_load_got): Use s390_got_symbol.
13403 (s390_option_override): Issue error if
13404 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13405 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13407 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13409 2017-06-27 Andrew Pinski <apinski@cavium.com>
13411 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13412 (X * copysign (1.0, X)): New pattern.
13413 (X * copysign (1.0, -X)): New pattern.
13414 (copysign (-1.0, CST)): New pattern.
13416 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13418 * genmultilib (combination_space): Remove variable.
13419 Validate reuse rules against regular expression for any sequence
13420 of multilib options in any order.
13422 2017-06-27 Michael Collison <michael.collison@arm.com>
13424 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13425 call aarch64_split_simd_combine.
13426 * (aarch64_combine_internal<mode>): Delete pattern.
13427 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13428 Allow register and subreg operands.
13430 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13432 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13433 specific need, just fallback on defaults.
13434 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13436 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13437 Olivier Hainque <hainque@adacore.com>
13439 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13441 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13442 targets. Pick a default if no particular attempt applied.
13443 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13446 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13448 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13449 (x86_64-wrs-vxworks7): Likewise.
13451 2017-06-27 Marek Polacek <polacek@redhat.com>
13454 * ubsan.c (instrument_null): Check get_base_address's result for null.
13456 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13458 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13460 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13462 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13463 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13464 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13465 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13466 New function types.
13467 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13468 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13469 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13470 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13471 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13472 BUILT_IN_FEUPDATEENV): New builtins.
13473 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13474 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13475 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13476 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13478 (builtin_structptr_types): Adjust size.
13479 * tree.c (builtin_structptr_types): Add four entries.
13481 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13482 Olivier Hainque <hainque@adacore.com>
13484 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13485 (TLS_SYM): New local macro, forcing reference to __tls__ on
13486 link command lines for VxWorks 7 RTPs, triggering initialization
13488 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13489 OS features TLS support, true for RTPs on VxWorks 7.
13490 * config/vxworks.c (vxworks_override_options): Setup emutls
13493 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13495 * predict.c (test_prediction_value_range): Use -1U instead of -1
13496 to avoid narrowing conversion warning.
13497 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13498 to avoid narrowing conversion warning.
13499 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13501 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13503 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13505 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13506 64bit configurations.
13507 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13508 (SIZE_TYPE): Likewise.
13509 * config/vxworks.c (vxworks_emutls_var_fields): Use
13510 long_unsigned_type_node instead of unsigned_type_node as the offset
13511 field type, which is "pointer" mode in emutls.c.
13513 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13516 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13518 PR middle-end/81207
13519 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13520 gimple_vuse copying separately from gimple_vdef copying.
13522 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13524 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13527 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13528 Olivier Hainque <hainque@adacore.com>
13530 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13531 for all vxworks7 targets.
13532 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13533 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13534 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13535 variations for VX6/VX7 and 32/64bits later on in ...
13536 (VXWORKS_LIB_SPEC): Leverage new macros.
13537 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13538 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13540 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13542 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13543 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13545 2017-06-26 Carl Love <cel@us.ibm.com>
13547 * config/rs6000/rs6000-c.c: Add support for built-in functions
13548 vector bool char vec_reve (vector bool char);
13549 vector signed char vec_reve (vector signed char);
13550 vector unsigned char vec_reve (vector unsigned char);
13551 vector bool int vec_reve (vector bool int);
13552 vector signed int vec_reve (vector signed int);
13553 vector unsigned int vec_reve (vector unsigned int);
13554 vector bool long long vec_reve (vector bool long long);
13555 vector signed long long vec_reve (vector signed long long);
13556 vector unsigned long long vec_reve (vector unsigned long long);
13557 vector bool short vec_reve (vector bool short);
13558 vector signed short vec_reve (vector signed short);
13559 vector double vec_reve (vector double);
13560 vector float vec_reve (vector float);
13561 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13562 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13563 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13564 (altivec_vreve): New pattern.
13565 * config/rs6000/altivec.h (vec_reve): New define.
13566 * doc/extend.texi (vec_rev): Update the built-in documentation file
13567 for the new built-in functions.
13569 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13571 PR tree-optimization/71815
13572 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13574 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13576 (slsr_process_phi): Likewise.
13577 (replace_uncond_cands_and_profitable_phis): Don't replace a
13578 multiply candidate with a stride of 1 (copy or cast).
13579 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13581 (lowest_cost_path): Likewise.
13582 (total_savings): Likewise.
13584 2017-06-26 Richard Biener <rguenther@suse.de>
13587 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13588 Use def_builtin_pure for all gather builtins.
13590 2017-06-26 Richard Biener <rguenther@suse.de>
13592 PR tree-optimization/81203
13593 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13594 non-dominating BBs.
13596 2017-06-26 Marek Polacek <polacek@redhat.com>
13599 * doc/invoke.texi: Document -Wmultistatement-macros.
13601 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13603 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13604 arm_neon_ok_no_float_abi effective target.
13606 2017-06-26 Richard Biener <rguenther@suse.de>
13608 PR tree-optimization/80928
13609 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13610 (copy_bbs): Set BB_DUPLICATED flag early.
13611 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13613 (execute_on_shrinking_pred): Likewise.
13614 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13615 BB_DUPLICATED blocks.
13616 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13617 iterate over all PHIs considering removal of *gsi.
13619 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13621 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13624 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13626 * config/rs6000/rs6000-string.c: (expand_block_clear,
13627 do_load_for_compare, select_block_compare_mode,
13628 compute_current_alignment, expand_block_compare,
13629 expand_strncmp_align_check, expand_strn_compare,
13630 expand_block_move, rs6000_output_load_multiple)
13631 Move functions related to string/block move/compare
13632 to a separate file.
13633 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13634 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13635 for this function which is now used in two files.
13636 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13637 * config.gcc: Add rs6000-string.o to extra_objs for
13638 targets powerpc*-*-* and rs6000*-*-*.
13640 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13643 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13644 32-bit, since indexed is not valid for DImode.
13645 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13646 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13647 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13648 (define_peephole2 for Altivec d-form store): Likewise.
13651 * multiple_target.c (create_dispatcher_calls): Only create the
13652 dispatcher call if the function is the default clone of a
13653 versioned function.
13655 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13657 PR middle-end/80902
13658 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13659 a call, force the call to not be a tail call.
13661 2017-06-23 Jeff Law <law@redhat.com>
13663 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13666 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13668 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13669 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13670 handling for early expansion of vector shifts (sl,sr,sra,rl).
13671 (builtin_function_type): Add vector shift right instructions
13672 to the unsigned argument list.
13674 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13676 rtl-optimizatoin/79286
13677 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13678 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13679 trap. PIC register plus a const unspec without offset can never trap.
13681 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13683 * tree.h (builtin_structptr_type): New type.
13684 (builtin_structptr_types): Declare new array.
13685 * tree.c (builtin_structptr_types): New array.
13686 (free_lang_data, build_common_tree_nodes): Use it.
13688 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13691 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13694 2017-06-22 Matt Turner <mattst88@gmail.com>
13696 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13697 Lake models to skylake case. Assume skylake for unknown
13698 models with clflushopt.
13700 2017-06-22 Jeff Law <law@redhat.com>
13702 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13703 frame sizes that do not satisfy aarch64_uimm12_shift.
13705 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13707 * profile-count.h (apply_probability,
13708 apply_scale, probability_in): Fix checks for zero.
13710 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13712 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13713 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13715 2016-06-22 Richard Biener <rguenther@suse.de>
13717 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13718 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13719 REDUC_MAX_EXPR support.
13720 (vectorizable_reduction): Likewise.
13721 (vect_create_epilog_for_reduction): Likewise.
13723 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13725 * match.pd (A / (1 << B) -> A >> B): New.
13726 * generic-match-head.c: Include optabs-tree.h.
13727 * gimple-match-head.c: Likewise.
13728 * optabs-tree.h (target_supports_op_p): New.
13729 * optabs-tree.c (target_supports_op_p): New.
13731 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13733 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13734 $gcc_cv_ld --help output.
13735 (gcc_cv_ld_demangle): Likewise.
13736 (gcc_cv_ld_eh_frame_hdr): Likewise.
13737 (gcc_cv_ld_pie): Likewise.
13738 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13739 (gcc_cv_ld_buildid): Likewise.
13740 (gcc_cv_ld_sysroot): Likewise.
13741 (ld_bndplt_support): Likewise.
13742 (ld_pushpopstate_support): Likewise.
13743 * configure: Regenerate.
13744 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13746 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13749 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13750 operands indexes to avoid gap between operands and match_dups.
13752 2017-06-21 Andrew Pinski <apinski@cavium.com>
13754 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13755 Increment Arith_shift and Arith_shift_reg by 1.
13756 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13758 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13759 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13760 (aarch64_strip_extend): Add new argument and test for it.
13761 (aarch64_cheap_mult_shift_p): New function.
13762 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13763 add a cost if it is true.
13764 Update calls to aarch64_strip_extend.
13765 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13767 2017-06-21 Andrew Pinski <apinski@cavium.com>
13769 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13771 (thunderxt88): Likewise.
13772 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13773 (thunderx_prefetch_tune): New variable.
13774 (thunderx2t99_prefetch_tune): Update for the correct values.
13775 (thunderxt88_tunings): New variable.
13776 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13777 generic_prefetch_tune.
13778 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13780 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13782 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13783 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13784 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13785 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13786 (aarch64_atomic_cas<mode>, GPI): Likewise.
13788 2017-06-21 Martin Liska <mliska@suse.cz>
13790 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13791 statements on cold and hot labels.
13792 * predict.c (tree_estimate_probability_bb): Remove the
13793 prediction from this place.
13795 2017-06-21 Martin Liska <mliska@suse.cz>
13797 PR tree-optimization/79489
13798 * gimplify.c (maybe_add_early_return_predict_stmt): New
13800 (gimplify_return_expr): Call the function.
13801 * predict.c (tree_estimate_probability_bb): Remove handling
13803 * predict.def: Update comment about early return predictor.
13804 * gimple-predict.h (is_gimple_predict): New function.
13805 * predict.def: Change default value of early return to 66.
13806 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13808 * passes.def: Put pass_strip_predict_hints to the beginning of
13811 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13813 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13814 FUNCTION_DECL declarations.
13815 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13817 (dwaf2out_decl): Likewise.
13818 * godump.c (go_early_global_decl): Skip call to the real debug hook
13819 for FUNCTION_DECL declarations.
13820 * passes.c (rest_of_decl_compilation): Skip call to the
13821 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13822 -fdump-go-spec is passed.
13824 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13826 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13827 Reorder for compactness.
13828 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13829 (def_builtin_pure, def_builtin_pure2): New functions.
13830 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13832 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13834 * match.pd (nop_convert): New predicate.
13835 ((A +- CST1) +- CST2): Allow some NOP conversions.
13837 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13840 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13841 with ctors/dtors if GOVD_SHARED is set.
13843 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13845 * config/aarch64/aarch64.md (movti_aarch64):
13846 Emit mov rather than orr.
13847 (movtf_aarch64): Likewise.
13848 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13849 Emit mov rather than orr.
13851 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13853 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13854 Swap alternatives, make integer dup more expensive.
13856 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13858 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13859 Return true for non-tls symbols.
13861 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13863 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13864 (cortex-a75): Likewise.
13865 (cortex-a75.cortex-a55): Likewise.
13866 * config/aarch64/aarch64-tune.md: Regenerate.
13867 * doc/invoke.texi (-mtune): Document new values for -mtune.
13869 2017-06-21 Tom de Vries <tom@codesourcery.com>
13871 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13872 stack_size feature.
13873 (Effective-Target Keywords, Other attributes): Suggest using
13874 dg-add-options stack_size feature to get stack limit in stack_size
13875 effective target documentation.
13877 2017-06-21 Julian Brown <julian@codesourcery.com>
13878 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13880 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13881 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13882 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13884 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13885 attribute type list for neon_multiply.
13886 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13887 attribute type list for neon_multiply.
13888 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13889 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13890 attribute type list for neon_multiply.
13891 * config/arm/types.md (crypto_pmull): Add.
13892 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13893 attribute type list.
13895 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13897 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13900 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13902 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13903 to make sure not to dereference a NULL cost_classes_ptr pointer.
13905 2017-06-20 Carl Love <cel@us.ibm.com>
13907 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13908 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13909 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13910 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13911 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13912 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13913 VMULOSW): New enum "unspec" values.
13914 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13915 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13916 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13917 altivec_vmulosw): New patterns.
13918 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13919 VMULOSW): Add definitions.
13921 2017-06-20 Julia Koval <julia.koval@intel.com>
13923 * config/i386/i386.c: Fix rounding expand for new pattern.
13924 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13926 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13928 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13929 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13931 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13933 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13936 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13938 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13939 architecture, then by implementer ID.
13940 * config/aarch64/aarch64-tune.md: Regenerate.
13942 2017-06-20 Richard Biener <rguenther@suse.de>
13944 PR middle-end/81097
13945 * fold-const.c (split_tree): Fold to type before negating.
13947 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13949 * diagnostic-show-locus.c
13950 (selftest::test_fixit_deletion_affecting_newline): New function.
13951 (selftest::diagnostic_show_locus_c_tests): Call it.
13953 2017-06-20 Andreas Schwab <schwab@suse.de>
13956 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13959 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13961 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13962 __ARM_FEATURE_COPROC according to support.
13964 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13966 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13967 Rewritten to avoid overflow for > 32-bit pointers.
13970 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13971 by removing enum keyword.
13972 (ubsan_type_descriptor): Likewise. Formatting fix.
13975 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13976 splitter): Require TARGET_SSE2 in the condition.
13978 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13981 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13982 for doing vector set of SFmode on ISA 3.0.
13983 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13984 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13986 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13987 SFmode value into a V4SF variable that was extracted from another
13988 V4SF variable without converting the element to double precision
13989 and back to single precision vector format.
13990 (vsx_insert_extract_v4sf_p9_2): Likewise.
13992 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13994 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13995 in UWHI to avoid undefined overflow.
13998 * ubsan.h (enum ubsan_encode_value_phase): New.
13999 (ubsan_encode_value): Change second argument to
14000 enum ubsan_encode_value_phase with default value of
14001 UBSAN_ENCODE_VALUE_GENERIC.
14002 * ubsan.c (ubsan_encode_value): Change second argument to
14003 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14004 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14005 create_tmp_var_raw instead of create_tmp_var and use a
14007 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14008 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14009 ubsan_encode_value callers.
14012 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14013 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14014 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14016 2017-06-19 Richard Biener <rguenther@suse.de>
14018 PR middle-end/81118
14019 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14020 estimates if we changed anything.
14022 2017-06-19 Richard Biener <rguenther@suse.de>
14024 PR tree-optimization/80887
14025 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14026 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14027 simplified lookups, then reset mprts_hook.
14028 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14030 (try_to_simplify): Likewise.
14032 2017-06-19 Martin Liska <mliska@suse.cz>
14035 * gimplify.c (gimplify_switch_expr):
14036 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14038 2017-06-19 Martin Liska <mliska@suse.cz>
14040 * doc/install.texi: Document that PGO runs in 4 stages.
14042 2017-06-19 Martin Liska <mliska@suse.cz>
14045 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14046 to dispatcher function name.
14047 * multiple_target.c (replace_function_decl): New function.
14048 (create_dispatcher_calls): Redirect both edges and references.
14050 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14052 * profile-count.c (profile_count::dump): Dump quality.
14053 (profile_count::differs_from_p): Update for unsigned val.
14054 * profile-count.h (profile_count_quality): New enum.
14055 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14057 2017-06-19 Richard Biener <rguenther@suse.de>
14059 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14060 struct function as arg.
14061 (estimate_numbers_of_iterations): Export overload with loop arg.
14062 (free_numbers_of_iterations_estimates_loop): Use an overload of
14063 free_numbers_of_iterations_estimates instead.
14064 * tree-cfg.c (remove_bb): Adjust.
14065 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14066 * tree-parloops.c (gen_parallel_loop): Likewise.
14067 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14069 (tree_unroll_loops_completely): Likewise.
14070 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14071 Use an overload instead and export.
14072 (estimated_loop_iterations): Adjust.
14073 (max_loop_iterations): Likewise.
14074 (likely_max_loop_iterations): Likewise.
14075 (estimate_numbers_of_iterations): Take struct function as arg
14077 (loop_exits_before_overflow): Adjust.
14078 (free_numbers_of_iterations_estimates_loop): Use an overload.
14079 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14080 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14082 2017-06-19 Richard Biener <rguenther@suse.de>
14085 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14086 RANGE_EXPR conservatively.
14088 2017-06-16 Carl Love <cel@us.ibm.com>
14090 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14091 definitions for vec_float, vec_float2, vec_floato,
14092 vec_floate built-ins.
14093 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14094 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14096 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14097 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14098 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14099 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14100 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14101 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14102 vec_floato): Add builtin defines.
14103 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14104 Update the built-in documentation file for the new built-in
14107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14109 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14110 (mthumb): Mark as the negative of -marm.
14112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14114 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14116 (ARM Options, -mtune): Document that this accepts the same
14117 extension options as -mcpu.
14118 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14120 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14122 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14123 permitted extensions.
14125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14127 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14128 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14129 (armv8-m.main): Add option +nodsp.
14130 * config/arm/arm-cpu-cdata.h: Regenerated.
14132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14134 * config/arm/t-fuchsia: New file.
14135 * config.gcc (arm*-*-fuchsia*): Use it.
14137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14139 * config/arm/t-symbian: Rewrite for new option infrastructure.
14141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14143 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14144 (MULTILIB_REQUIRED): Likewise.
14146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14148 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14149 (MULTILIB_RESUE): Likewise.
14150 (MULTILIB_MATCHES): Likewise.
14151 (MULTLIB_REQUIRED): Likewise.
14153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14155 * config/arm/t-rtems: Rewrite for new option framework.
14157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14159 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14160 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14161 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14162 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14163 * config/arm/t-multilib: ... here.
14164 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14165 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14166 armv7-a and armv8*-a when A-profile libraries have not been built.
14167 * config/arm/t-rmprofile: Rewrite.
14169 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14171 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14172 with a backslash. Remove the backslash after substituting unescaped
14174 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14178 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14179 the driver to pass through the default setting of -mfloat-abi.
14180 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14182 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14183 (all_feat_combs): New rule.
14184 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14186 * config/arm/t-aprofile: Rewrite.
14188 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14190 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14191 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14192 fpu is not specified by the user/command-line.
14193 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14194 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14195 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14196 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14197 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14198 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14202 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14203 * config/arm/t-arm-elf: Rewritten.
14205 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14207 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14208 have some floating-point instructions.
14209 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14210 (TARGET_MAYBE_HARD_FLOAT): New macro.
14211 * config/arm/arm-builtins.c (arm_init_builtins): Use
14212 TARGET_MAYBE_HARD_FLOAT.
14213 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14217 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14218 (configargs.h): Include it.
14219 (arm_print_hint_for_fpu_option): New function.
14220 (arm_parse_fpu_option): New function.
14221 (candidate_extension): New class.
14222 (arm_canon_for_multilib): New function.
14223 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14224 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14225 (ARCH_CANONICAL_SPECS): New macro.
14226 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14230 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14231 are set after handling multilib fragments. Set target_cpu_default2
14234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14236 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14238 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14239 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14241 (check_arch): Likewise.
14242 * config/arm/arm.c (arm_configure_build_target): Handle
14243 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14244 options in the default.
14246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14248 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14249 when an option is an alias of another.
14250 * config/arm/parsecpu.awk (optalias): New parser token.
14251 (gen_comm_data): Mark non-alias options as such. Emit entries
14252 for extension aliases.
14253 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14254 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14255 (armv6kz, armv6zk, armv6t2): Likewise.
14256 (armv7): Make vfpv3-d16 an alias.
14257 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14259 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14260 Sort in canonical order.
14261 (armv8-a): Sort in canonical order.
14262 (armv8.1-a, armv8.2-a): Likewise.
14263 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14265 (cortex-a9): Sort in canonical order.
14266 * config/arm/arm.c (selftests.h): Include it.
14267 (arm_test_cpu_arch_data): New function.
14268 (arm_run_self_tests): New function.
14269 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14270 (targetm): Move declaration to the end of the file.
14271 * arm-cpu-cdata.h: Regenerated.
14273 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14275 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14276 call to target_mode_check describing the type of option passed.
14277 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14278 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14279 arm_parse_cpu_option_name to match parameters against list of
14281 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14282 arm_arch_core_flags data structure.
14283 * config/arm/arm-cpu_cdata.h: Regenerated.
14285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14287 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14289 (arm_print_hint_for_cpu_option): Likewise.
14290 (arm_print_hint_for_arch_option): Likewise.
14291 (arm_parse_cpu_option_name): Likewise.
14292 (arm_parse_arch_option_name): Likewise.
14293 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14294 of entries in the all_fpus list.
14295 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14296 (arm_parse_cpu_option_name): Declare.
14297 (arm_parse_arch_option_name): Declare.
14298 (arm_parse_option_features): Declare.
14299 (arm_intialize_isa): Declare.
14300 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14302 (gen_comm_data): ... here. Make definitions non-static.
14303 * config/arm/arm-cpu-data.h: Regenerated.
14304 * config/arm/arm-cpu-cdata.h: Regenerated.
14306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14308 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14309 (cpu_arch_extension): New structure.
14310 (cpu_arch_option, arch_option, cpu_option): New structures.
14311 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14312 architecture types.
14313 (gen_data): Generate new format data tables.
14314 * config/arm/arm.c (cpu_tune): New structure.
14315 (cpu_option, processors): Delete.
14316 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14317 (arm_print_hint_for_cpu_option): ... this and ...
14318 (arm_print_hint_for_arch_option): ... this.
14319 (arm_parse_arch_cpu_name): Delete. Replace with ...
14320 (arm_parse_cpu_option_name): ... this and ...
14321 (arm_parse_arch_option_name): ... this.
14322 (arm_unrecognized_feature): Change type of target parameter to
14324 (arm_parse_arch_cpu_features): Delete. Replace with ...
14325 (arm_parse_option_features): ... this.
14326 (arm_configure_build_target): Rework to use new configuration data
14328 (arm_print_tune_info): Rework for new configuration data tables.
14329 * config/arm/arm-cpu-data.h: Regenerated.
14330 * config/arm/arm-cpu.h: Regenerated.
14332 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14334 * Makefile.in (OBJS): Move sbitmap.o from here ...
14335 (OBJS-libcommon): ... to here.
14337 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14339 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14340 (ISA_ALL_CRYPTO): New macro.
14341 (ISA_ALL_SIMD): New macro
14342 (ISA_ALL_FP): New macro.
14343 * config/arm/arm.c (fpu_bitlist): Update initializer.
14344 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14346 (arm9e): Add fpu. Add option for nofp
14347 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14348 (arm926ej-s, arm1026ej-s): Likewise.
14349 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14350 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14351 neon-fp16, neon-vfpv4, nofp and nosimd.
14352 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14353 (cortex-a8): Add fpu. Add option for nofp.
14354 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14355 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14356 (cortex-r4f): Add fpu.
14357 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14358 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14360 (cortex-r8): Likewise.
14361 (cortex-m4): Add fpu. Add option for nofp.
14362 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14363 (cortex-a17.cortex-a7): Likewise.
14364 (cortex-a32): Add fpu. Add options for crypto and nofp.
14365 (cortex-a35, cortex-a53): Likewise.
14366 (cortex-a57): Add fpu. Add option for crypto.
14367 (cortex-a72, cortex-a73): Likewise.
14368 (exynos-m1): Likewise.
14369 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14370 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14371 (cortex-m33): Add fpu. Add option for nofp.
14372 * config/arm/arm-cpu-cdata.h: Regenerated
14373 * config/arm/arm-cpu-data.h: Regenerated.
14375 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14377 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14378 (armv5te, armv5tej): Likewise.
14379 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14380 (armv7): Add options fp and vfpv3-d16.
14381 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14382 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14384 (armv7ve): Likewise.
14385 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14386 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14387 (armv8-a): Add nocrypto option.
14388 (armv8.1-a, armv8.2-a): Likewise.
14389 (armv8-m.main): add options fp, fp.dp and nofp.
14391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14393 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14395 (armv8-a+crc): Delete.
14396 (armv8.1-a): Add options simd, crypto and nofp.
14397 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14398 (armv8.2-a+fp16): Delete.
14399 (armv8-m.main): Add option dsp.
14400 (armv8-m.main+dsp): Delete.
14401 (cortex-a8): Add fpu. Add nofp option.
14402 (cortex-a9): Add fpu. Add nofp and nosimd options.
14403 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14404 link to main cpu and architecture data structures.
14405 (gen_comm_data): Only put isa attributes from the main architecture
14407 (option): New statement for architecture and CPU entries.
14408 * arm.c (struct cpu_option): New structure.
14409 (struct processors): Add entry for options.
14410 (arm_unrecognized_feature): New function.
14411 (arm_parse_arch_cpu_name): Ignore any characters after the first
14413 (arm_parse_arch_cpu_feature): New function.
14414 (arm_configure_build_target): Separate out any CPU and architecture
14415 features and parse separately. Don't error out if -mfpu=auto is
14416 used with only an architecture string.
14417 (arm_print_asm_arch_directives): New function.
14418 (arm_file_start): Call it.
14419 * config/arm/arm-cpu-cdata.h: Regenerated.
14420 * config/arm/arm-cpu-data.h: Likewise.
14421 * config/arm/arm-tables.opt: Likewise.
14423 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14425 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14426 assembler when it is not -mfpu=auto.
14428 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14430 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14431 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14432 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14433 (ASM_CPU_SPEC): Rewrite.
14434 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14435 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14436 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14437 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14438 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14439 copied string is NUL-terminated. Also strip any characters prefixed
14441 (arm_rewrite_selected_arch): New function.
14442 (arm_rewrite_march): New function.
14444 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14446 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14447 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14448 (march, mcpu, mtune): Convert to string-based options.
14449 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14450 (arm_parse_arch_cpu_name): New function.
14451 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14452 identify selected architecture or CPU.
14453 (arm_option_save): New function.
14454 (TARGET_OPTION_SAVE): Redefine.
14455 (arm_option_restore): Restore string options.
14456 (arm_option_print): Print string options.
14458 2017-06-16 Martin Sebor <msebor@redhat.com>
14460 PR tree-optimization/80933
14461 PR tree-optimization/80934
14462 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14463 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14464 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14465 (gimple_fold_builtin): Call them.
14467 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14469 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14470 as unlikely; update profile.
14472 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14474 * predict.c (force_edge_cold): Handle declaring edges impossible
14477 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14479 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14481 (try_unroll_loop_completely): Fix reporting.
14483 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14485 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14487 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14490 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14491 if given a non-constant argument for an intrinsic which requires a
14494 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14496 * profile.c (compare_freqs): New function.
14497 (branch_prob): Sort edge list.
14498 (find_spanning_tree): Assume that the list is priority sorted.
14500 2017-06-16 Richard Biener <rguenther@suse.de>
14502 PR tree-optimization/81090
14503 * passes.def (pass_record_bounds): Remove.
14504 * tree-pass.h (make_pass_record_bounds): Likewise.
14505 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14506 make_pass_record_bounds): Likewise.
14507 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14508 not free niter estimates at the beginning but at the end.
14509 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14511 2017-06-16 Richard Biener <rguenther@suse.de>
14513 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14514 initializer to workaround ICE in host GCC 4.8.
14516 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14518 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14520 (clone_inlined_nodes): Update.
14522 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14524 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14525 prefetch settings, and enable prefetching by default at -O3.
14527 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14529 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14530 Set flag_prefetch_loop_arrays according to tuning data.
14532 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14534 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14535 New tune structure.
14536 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14537 [Unrelated to main purpose of the patch] Place the pointer field last
14538 to enable type checking errors when tune structure are wrongly merged.
14539 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14540 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14541 (thunderx2t99_prefetch_tune): New tune constants.
14542 (tune_params *_tunings): Update all tunings (no functional change).
14543 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14544 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14545 from tunings structures.
14547 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14550 * ubsan.c (instrument_null): Add T argument, use it instead
14551 of computing it based on IS_LHS.
14552 (instrument_object_size): Likewise.
14553 (pass_ubsan::execute): Adjust instrument_null and
14554 instrument_object_size callers to pass gimple_get_lhs or
14555 gimple_assign_rhs1 result to it. Use instrument_null instead of
14556 calling get_base_address and instrument_mem_ref. Handle
14557 aggregate call arguments for object-size sanitization.
14559 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14561 PR tree-optimization/81089
14562 * tree-vrp.c (is_masked_range_test): Validate operands of
14565 2017-06-15 Martin Sebor <msebor@redhat.com>
14568 * dumpfile.c (dump_register): Avoid calling memset to initialize
14569 a class with a default ctor.
14570 * gcc.c (struct compiler): Remove const qualification.
14571 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14572 * hash-table.h: Ditto.
14573 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14575 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14576 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14578 * params.h (struct param_info): Make struct members non-const.
14579 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14580 with default initialization.
14581 * vec.h (vec_copy_construct, vec_default_construct): New helper
14583 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14584 with vec_copy_construct.
14585 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14586 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14587 * doc/invoke.texi (-Wclass-memaccess): Document.
14589 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14591 * emit-rtl.h (is_leaf): Update comment about local
14592 register allocator.
14594 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14597 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14598 for a variable to have a section before checking if the section has a
14600 Set section to.persistent if persistent attribute is set.
14601 Warn if .persistent attribute is used on an automatic variable.
14603 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14605 PR rtl-optimization/80474
14606 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14608 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14610 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14613 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14615 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14616 (casesi): Emit bounds checking as RTL.
14617 (casesi_internal_mips16_<mode>): Remove bounds checking.
14619 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14621 * config/xtensa/xtensa.c (xtensa_option_override): Append
14622 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14623 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14624 xtensa_doloop_hooks): Define unconditionally.
14625 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14627 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14628 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14629 for it in xtensa_option_override.
14630 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14631 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14633 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14635 * doc/cppopts.texi: Document '-' special value to -MF.
14637 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14639 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14640 (cortex_a53_fconst): Likewise.
14641 (cortex_a53_fpmul): Likewise.
14642 (cortex_a53_f_load_64): Likewise.
14643 (cortex_a53_f_load_many): Likewise.
14644 (cortex_a53_advsimd_alu): Likewise.
14645 (cortex_a53_advsimd_alu_q): Likewise.
14646 (cortex_a53_advsimd_mul): Likewise.
14647 (cortex_a53_advsimd_mul_q): Likewise.
14648 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14649 Add missing fmul, r2f_cvt and fconst cases.
14651 2017-06-14 Richard Biener <rguenther@suse.de>
14653 PR middle-end/81088
14654 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14656 (fold_binary_loc): When associating do not treat pre-existing
14657 TREE_OVERFLOW on literal constants as a reason to allow
14658 TREE_OVERFLOW on associated literal constants.
14660 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14662 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14663 (MASK_FEATURES): New macro.
14664 * config/sparc/sparc.c (sparc_option_override): Remove the special
14665 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14667 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14669 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14670 a division of 0 if non-call exceptions are enabled.
14672 2017-06-14 Andrew Pinski <apinski@cavium.com>
14673 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14676 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14677 Improve vector initialization code gen for only variable case.
14679 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14681 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14683 2017-06-14 Richard Biener <rguenther@suse.de>
14685 PR tree-optimization/81083
14686 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14689 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14691 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14692 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14693 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14694 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14695 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14696 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14698 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14700 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14701 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14703 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14705 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14707 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14709 * config/rs6000/t-rtems: Don't handle SPE.
14711 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14713 * config/rs6000/t-linux: Don't handle SPE.
14715 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14717 * config/rs6000/eabispe.h: Delete file.
14719 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14721 * config/rs6000/t-spe: Delete file.
14723 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14725 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14726 (rs6000_legitimate_offset_address_p): Return false for anything in
14727 V2SImode or V2SFmode.
14729 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14731 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14732 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14733 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14735 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14736 (rs6000_legitimate_offset_address_p): Ditto.
14737 (rs6000_emit_move): Ditto.
14738 (rs6000_init_builtins): Remove V4HI_type_node.
14740 2017-06-13 Martin Liska <mliska@suse.cz>
14743 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14744 (gate_asan): Likewise.
14745 * asan.h (asan_no_sanitize_address_p): Remove the function.
14746 (sanitize_flags_p): New function.
14747 * builtins.def: Fix coding style.
14748 * common.opt: Use renamed enum value.
14749 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14750 * doc/extend.texi: Document no_sanitize attribute.
14751 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14752 to SANITIZE_UNDEFINED_NONDEFAULT.
14753 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14754 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14755 Use sanitize_flags_p.
14756 * gimplify.c (gimplify_function_tree): Likewise.
14757 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14758 * opts.c (parse_no_sanitize_attribute): New function.
14759 (common_handle_option): Use renamed enum value.
14760 * opts.h (parse_no_sanitize_attribute): Declare.
14761 * tree.c (sanitize_flags_p): New function.
14762 * tree.h: Declared here.
14763 * tsan.c: Use sanitize_flags_p.
14764 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14765 (instrument_mem_ref): Likewise.
14766 (instrument_bool_enum_load): Likewise.
14767 (do_ubsan_in_current_function): Remove the function.
14768 (pass_ubsan::execute): Use sanitize_flags_p.
14769 * ubsan.h: Remove do_ubsan_in_current_function
14770 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14771 (dump_function_to_file): Use it here.
14773 2017-06-13 Martin Jambor <mjambor@suse.cz>
14775 PR tree-optimization/80803
14776 PR tree-optimization/81063
14777 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14778 (propagate_subaccesses_across_link): Enqueue subtree whenever
14779 necessary instead of relying on the caller.
14781 2017-06-13 Martin Jambor <mjambor@suse.cz>
14783 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14784 that have a first_link.
14785 (sort_and_splice_var_accesses): Do not check first_link before
14787 (subtree_mark_written_and_enqueue): Likewise.
14788 (propagate_all_subaccesses): Likewise and do not stop at first
14789 parent with a first_link.
14791 2017-06-13 Martin Jambor <mjambor@suse.cz>
14793 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14796 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14798 * match.pd: New pattern.
14800 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14802 * tree-vrp.c (is_masked_range_test): New function.
14803 (register_edge_assert_for): Determine ranges for
14806 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14808 PR tree-optimization/67328
14809 * fold-const.c (maskable_range_p): New function.
14810 (build_range_check): Generate bittests if possible.
14812 2017-06-13 Martin Liska <mliska@suse.cz>
14814 * gimple-pretty-print.c (dump_probability): Add new argument.
14815 (dump_edge_probability): Dump both probability and count.
14816 (dump_gimple_label): Likewise.
14817 (dump_gimple_bb_header): Likewise.
14819 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14822 * config/avr/avr-devices.c: Fix indentation.
14823 * config/avr/gen-avr-mmcu-specs.c: Dito.
14825 2017-06-13 Richard Biener <rguenther@suse.de>
14827 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14828 instead get vector type from stmt_info.
14829 (vectorizable_reduction): Adjust. Remove dead code.
14831 2017-06-13 Richard Biener <rguenther@suse.de>
14833 PR middle-end/81065
14834 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14835 case of C * (x * C2 + C3).
14836 (fold_addr_of_array_ref_difference): Properly fold index difference.
14838 2017-06-12 David S. Miller <davem@davemloft.net>
14841 * config/sparc/sparc.md (return expander): Emit frame blockage if
14842 function uses alloca.
14844 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14846 * combine.c (make_field_assignment): Check len rather than the mode
14847 precision when calling force_to_mode.
14849 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14851 Support multilibs and devices that see flash in RAM address range.
14854 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14855 (avr_mcu_t) <flash_pm_offset>: New field.
14856 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14857 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14858 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14859 (AVR_TINY_PM_OFFSET): Remove macro.
14860 * config/avr/avr.opt (-mshort-calls): New option.
14861 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14862 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14863 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14864 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14865 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14866 instead of avr_arch->have_jmp_call.
14867 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14868 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14869 avr_arch->flash_pm_offset to define.
14870 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14871 new field flash_pm_offset. Add entry for avrxmega3.
14872 (avr_texinfo): Add entry for avrxmega3.
14873 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14874 attiny212, attiny214,
14875 attiny412, attiny414, attiny416, attiny417,
14876 attiny814, attiny816, attiny817,
14877 attiny1614, attiny1616, attiny1617,
14878 attiny3214, attiny3216, attiny3217.
14879 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14880 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14881 (avr_print_operand_address) [AVR_TINY]: Same.
14882 (avr_asm_init_sections) <readonly_data_section>: Only patch
14883 callback if avr_arch->flash_pm_offset = 0.
14884 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14885 for rodata if avr_arch->flash_pm_offset != 0.
14886 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14887 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14888 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14889 (m_options): Append opt_rcall.
14890 (m_dirnames): Append dir_rcall.
14891 * config/avr/t-multilib: Regenerate.
14893 * configure.ac [target=avr]: Check whether avrxmega3 default
14894 linker description file works as needed.
14895 * configure: Regenerate.
14896 * doc/avr-mmcu.texi: Regenerate.
14897 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14898 <__AVR_ARCH__>: Document avrxmega3 and 103.
14899 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14900 <__AVR_SHORT_CALLS__>: Document it.
14901 <__AVR_PM_BASE_ADDRESS__>: Document it.
14902 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14903 (AVR Variable Attributes) <progmem>: Document this is
14904 not needed for avrxmega3.
14905 (AVR Named Address Spaces) <__flash>: Dito.
14907 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14909 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14911 2017-06-12 Doug Rupp <rupp@adacore.com>
14913 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14914 Append vxworks-stdint.h to the tm_file list.
14915 * config/vxworks-stdint.h: New file.
14917 2017-06-12 Martin Liska <mliska@suse.cz>
14919 PR tree-optimization/81041
14920 * tree-profile.c (gimple_gen_ic_func_profiler):
14921 Create an extra BB in profile-generate
14922 (gimple_gen_time_profiler): Likewise.
14924 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14926 PR tree-optimization/81003
14927 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14928 (update_range_test): Use it instead of force_gimple_operand_gsi.
14930 2017-06-12 Richard Biener <rguenther@suse.de>
14932 PR tree-optimization/81053
14933 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14934 with backedge value not defined in loop. Simplify def stmt
14937 2017-06-11 Tom de Vries <tom@codesourcery.com>
14940 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14942 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14943 nvptx_cannot_force_const_mem.
14945 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14947 * opts.c (finish_options): Move test for flag_split_stack after
14948 it has been initialized.
14950 2017-06-11 Jason Merrill <jason@redhat.com>
14952 * tree.h (id_equal): New.
14953 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14954 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14955 instead of strcmp of IDENTIFIER_POINTER.
14957 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14959 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14960 (mark_all_inlined_calls_cdtor): Fix formating.
14961 (inline_transform): Rescale profile before inlining.
14963 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14965 * cgraph.h (cgraph_edge::clone): Update prototype.
14966 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14967 (cgraph_node::create_clone): Update.
14968 (cgraph_node::create_version_clone): Update.
14969 * tree-inline.c (copy_bb): Update.
14970 (expand_call_inline): Update.
14972 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14974 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14975 factored out from ...
14976 (rs6000_emit_prologue): ... here.
14978 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14980 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14981 factored out from ...
14982 (rs6000_emit_prologue): ... here.
14984 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14986 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14988 (handle_missing_profiles): Fix computation of tp_first_run.
14989 (counts_to_freqs): Do not touch freqs when count is 0.
14991 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14993 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14996 2017-06-10 Tom de Vries <tom@codesourcery.com>
14998 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14999 attributes): Document signal effective target.
15001 2017-06-10 Tom de Vries <tom@codesourcery.com>
15003 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15004 Document effective target stack_size.
15006 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15008 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15009 to the edit_context if they can be auto-applied.
15011 2017-06-9 Ian Lance Taylor <iant@golang.org>
15013 * opts.c (finish_options): If -fsplit-stack, disable implicit
15014 -forder-blocks-and-partition.
15015 * doc/invoke.texi (Optimize Options): Document that when using
15016 -fsplit-stack -forder-blocks-and-partition is not implicitly
15019 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15021 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15022 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15023 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15024 * builtins.def (abort, trap, unreachable): Declare cold.
15025 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15026 * tree-core.h (ECF_COLD): New.
15027 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15028 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15030 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15032 * predict.c (unlikely_executed_stmt_p): Cleanup.
15034 2017-06-09 Richard Biener <rguenther@suse.de>
15036 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15037 model if the ref is always written to.
15039 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15041 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15043 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15045 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15048 2017-06-09 Tom de Vries <tom@codesourcery.com>
15051 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15052 "target cannot support label values" when encountering LABEL_REF.
15054 2017-06-09 Martin Liska <mliska@suse.cz>
15056 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15057 (gimple_gen_ic_func_profiler): Emit direct comparison
15058 of __gcov_indirect_call_callee with NULL.
15059 (gimple_gen_time_profiler): Change probability from
15060 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15062 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15064 * profile.c (edge_gcov_counts): Turn to pointer.
15065 (compute_branch_probabilities, compute_branch_probabilities): Update.
15066 (branch_prob): Do not clear edge_gcov_count.
15067 * profile.h (edge_gcov_counts): Turn to pointer.
15068 (edge_gcov_count): Update.
15070 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15072 * gimple.h (gimple_check_failed): Mark cold.
15074 2017-06-09 Richard Biener <rguenther@suse.de>
15076 PR tree-optimization/66623
15077 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15078 refactor check_reduction into two parts, properly computing
15079 whether we have to check reduction validity for outer loop
15082 2017-06-09 Richard Biener <rguenther@suse.de>
15084 PR tree-optimization/79483
15085 * graphite-scop-detection.c (order): New global.
15086 (get_order): Compute bb to order mapping that satisfies code
15087 generation constraints.
15088 (cmp_pbbs): New helper.
15089 (build_scops): Start domwalk at entry block, sort generated
15092 2017-06-09 Richard Biener <rguenther@suse.de>
15094 PR middle-end/81007
15095 * ipa-polymorphic-call.c
15096 (ipa_polymorphic_call_context::restrict_to_inner_class):
15097 Skip FIELD_DECLs with error_mark_node type.
15098 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15101 2017-06-09 Martin Liska <mliska@suse.cz>
15103 * predict.c (struct branch_predictor): New struct.
15104 (test_prediction_value_range): New test.
15105 (predict_c_tests): New function.
15106 * selftest-run-tests.c (selftest::run_tests): Run the function.
15107 * selftest.h: Declare new tests.
15109 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15112 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15113 gen_add3_insn did not fail.
15114 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15115 r0, construct that number in a temporary reg and add that reg to r0.
15116 If asked to put the result in r0 as well, fail.
15118 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15120 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15121 for early expansion of vec_eqv.
15123 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15125 PR middle-end/81005
15126 * ubsan.c (instrument_null): Avoid pointless code temporary.
15127 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15129 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15133 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15136 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15141 * predict.c (unlikely_executed_edge_p): Move ahead.
15142 (probably_never_executed_edge_p): Use it.
15144 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15146 PR middle-end/79988
15147 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15148 gimple_call_builtin_p call.
15150 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15152 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15153 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15154 rtl_check_failed_type2, rtl_check_failed_code1,
15155 rtl_check_failed_code2, rtl_check_failed_code_mode,
15156 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15157 rtvec_check_failed_bounds, rtl_check_failed_flag,
15158 _fatal_insn_not_found, _fatal_insn): Likewise.
15159 * tree.h (tree_contains_struct_check_failed,
15160 tree_check_failed, tree_not_check_failed,
15161 tree_class_check_failed, tree_range_check_failed,
15162 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15163 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15164 tree_operand_check_failed, omp_clause_check_failed,
15165 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15168 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15170 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15171 flag_branch_probabilities.
15172 * ipa-inline.c (edge_badness): Likewise.
15173 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15174 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15175 * predict.c (maybe_hot_frequency_p): Likewise.
15176 (probably_never_executed): Likewise.
15177 * sched-ebb.c (schedule_ebbs): Likewise.
15178 * sched-rgn.c (find_single_block_region): Likewise.
15179 * tracer.c (tail_duplicate): Likewise.
15181 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15183 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15184 longer requires x_flag_profile_use.
15186 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15188 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15189 instead of flag_reorder_blocks_and_partition.
15190 * dbxout.c (dbxout_function_end): Likewise.
15191 * dwarf2out.c (gen_subprogram_die): Likewise.
15192 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15193 * hw-doloop.c (reorg_loops): Likewise.
15194 * varasm.c (assemble_start_function,
15195 assemble_end_function): Likewise.
15196 (decide_function_section): Do not check for
15197 flag_reorder_blocks_and_partition.
15199 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15201 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15203 (chkp_get_hard_register_fake_addr_expr): Ditto.
15204 (chkp_build_addr_expr): Add check for hard reg case.
15205 (chkp_parse_array_and_component_ref): Ditto.
15206 (chkp_find_bounds_1): Ditto.
15207 (chkp_process_stmt): Don't generate bounds store for
15210 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15212 * predict.c (maybe_hot_bb_p): Do not check profile status.
15213 (maybe_hot_edge_p): Likewise.
15214 (probably_never_executed): Check for zero counts even if profile
15216 (unlikely_executed_edge_p): New function.
15217 (unlikely_executed_stmt_p): New function.
15218 (unlikely_executed_bb_p): New function.
15219 (set_even_probabilities): Use unlikely predicates.
15220 (combine_predictions_for_bb): Likewise.
15221 (predict_paths_for_bb): Likewise.
15222 (predict_paths_leading_to_edge): Likewise.
15223 (determine_unlikely_bbs): New function.
15224 (estimate_bb_frequencies): Use it.
15225 (compute_function_frequency): Use zero counts even if profile is
15227 * profile-count.h: Fix typo.
15229 2017-08-08 Julia Koval <julia.koval@intel.com>
15231 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15232 _mm512_mask_cvtsepi16_storeu_epi8,
15233 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15234 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15235 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15236 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15237 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15238 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15239 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15240 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15241 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15242 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15243 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15244 __builtin_ia32_pmovuswb256mem_mask,
15245 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15246 __builtin_ia32_pmovwb512mem_mask): New builtins.
15248 2017-08-08 Julia Koval <julia.koval@intel.com>
15250 PR target/73350,80862
15251 * config/i386/subst.md (round): Fix round pattern.
15252 * config/i386/i386.c (ix86_erase_embedded_rounding):
15253 Fix erasing rounding for the fixed pattern.
15255 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15257 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15259 2017-06-08 Martin Liska <mliska@suse.cz>
15261 PR gcov-profile/80911
15262 * gcov.c (block_info::block_info): New constructor.
15264 2017-06-07 Carl Love <cel@us.ibm.com>
15266 * config/rs6000/rs6000-c: The return type of the following
15267 built-in functions was implemented as int not long long. Fix sign
15268 of return value for the unsigned version of vec_mulo and vec_mule.
15269 vector unsigned long long vec_bperm (vector unsigned long long,
15270 vector unsigned char)
15271 vector signed long long vec_mule (vector signed int,
15273 vector unsigned long long vec_mule (vector unsigned int,
15274 vector unsigned int)
15275 vector signed long long vec_mulo (vector signed int,
15277 vector unsigned long long vec_mulo (vector unsigned int,
15278 vector unsigned int)
15279 * doc/extend.texi: Fix the documentation for the built-in
15282 2017-06-07 Carl Love <cel@us.ibm.com>
15285 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15288 2017-06-07 Carl Love <cel@us.ibm.com>
15290 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15291 support, Generate doublehv for signed int/float for BE case only.
15293 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15295 * doc/invoke.texi (mcx16): Rewrite.
15297 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15299 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15300 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15301 *mov<mode>_softfloat, and an anonymous splitter): Use
15302 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15306 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15308 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15309 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15310 (rs6000_debug_reg_global): Adjust.
15311 (rs6000_init_hard_regno_mode_ok): Adjust.
15312 (rs6000_dbx_register_number): Adjust.
15313 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15314 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15315 Remove SPE_ACC and SPEFSCR.
15316 (REG_ALLOC_ORDER): Ditto.
15317 (FRAME_POINTER_REGNUM): Change to 111.
15318 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15319 (REG_CLASS_NAMES): Ditto.
15320 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15321 (REGISTER_NAMES): Ditto.
15322 (ADDITIONAL_REG_NAMES): Ditto.
15323 (rs6000_reg_names): Ditto.
15324 * config/rs6000/rs6000.md: Renumber some register number
15327 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15329 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15331 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15332 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15334 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15335 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15336 Delete the SPE high registers.
15337 (REG_ALLOC_ORDER): Ditto.
15338 (enum reg_class): Remove SPE_HIGH_REGS.
15339 (REG_CLASS_NAMES): Ditto.
15340 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15341 (REGISTER_NAMES): Ditto.
15342 (rs6000_reg_names): Ditto.
15343 * doc/tm.texi.in: Remove SPE as example.
15344 * doc/tm.texi: Regenerate.
15346 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15348 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15349 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15350 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15351 * config/rs6000/rs6000.md (type): Remove "brinc".
15353 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15355 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15356 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15357 * config/rs6000/linuxspe.h: Delete file.
15358 * config/rs6000/rs6000.md: Don't include spe.md.
15359 * config/rs6000/spe.h: Delete file.
15360 * config/rs6000/spe.md: Delete file.
15361 * config/rs6000/t-rs6000: Remove spe.md.
15363 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15365 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15366 (reg_or_none500mem_operand): Delete.
15367 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15368 instead of reg_or_none500mem_operand.
15370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15372 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15373 handling of SPE flags.
15374 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15376 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15378 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15380 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15381 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15382 paired_divv2sf3): Similar.
15383 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15384 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15385 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15387 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15388 Rename the paired_* instruction patterns.
15389 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15391 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15392 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15393 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15394 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15395 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15396 PAIRED_VECTOR_MODE.
15397 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15398 (spe_func_has_64bit_regs_p): Delete.
15399 (spe_expand_predicate_builtin): Delete.
15400 (spe_expand_evsel_builtin): Delete.
15401 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15402 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15403 (invalid_e500_subreg): Delete.
15404 (rs6000_legitimize_address): Always force_reg op2 as well, for
15405 paired single memory accesses.
15406 (rs6000_member_type_forces_blk): Delete.
15407 (rs6000_spe_function_arg): Delete.
15408 (rs6000_expand_unop_builtin): Delete SPE handling.
15409 (rs6000_expand_binop_builtin): Ditto.
15410 (spe_expand_stv_builtin): Delete.
15411 (bdesc_2arg_spe): Delete.
15412 (spe_expand_builtin): Delete.
15413 (spe_expand_predicate_builtin): Delete.
15414 (spe_expand_evsel_builtin): Delete.
15415 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15416 (spe_init_builtins): Delete.
15417 (spe_func_has_64bit_regs_p): Delete.
15418 (savres_routine_name): Delete "info" parameter. Adjust callers.
15419 (rs6000_emit_stack_reset): Ditto.
15420 (rs6000_dwarf_register_span): Delete.
15421 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15422 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15423 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15425 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15427 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15428 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15429 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15430 mulv2sf3, divv2sf3): Delete expanders.
15432 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15434 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15436 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15438 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15439 * config/rs6000/rs6000.c: Ditto.
15441 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15443 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15444 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15445 comparison_operator.
15447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15449 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15450 * config/rs6000/rs6000.opt: Ditto.
15451 * config/rs6000/t-rtems: Ditto.
15453 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15455 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15456 TARGET_E500_SINGLE by 0, simplify.
15457 * config/rs6000/rs6000.c: Ditto.
15458 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15459 (spe_build_register_parallel): Delete.
15460 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15461 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15462 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15463 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15464 (E500_CONVERT): Delete.
15465 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15467 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15469 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15470 * config/rs6000/dfp.md: Ditto.
15471 (negdd2, *negdd2_fpr): Merge.
15472 (absdd2, *absdd2_fpr): Merge.
15473 (negtd2, *negtd2_fpr): Merge.
15474 (abstd2, *abstd2_fpr): Merge.
15475 * config/rs6000/e500.h: Delete file.
15476 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15477 TARGET_FPRS by 1 and simplify.
15478 * config/rs6000/rs6000-c.c: Ditto.
15479 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15480 TARGET_DF_SPE by 0.
15481 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15483 * config/rs6000/rs6000.md: Ditto.
15484 (floatdidf2, *floatdidf2_fpr): Merge.
15485 (move_from_CR_gt_bit): Delete.
15486 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15487 (E500_CR_IOR_COMPARE): Delete.
15488 (All patterns that require !TARGET_FPRS): Delete.
15489 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15491 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15493 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15495 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15497 * graphds.c (add_edge): Intitialize edge's attached data.
15498 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15499 pointer parameter. Call pointed function on each edge during
15500 graph traversing. Skip traversing the edge when the function
15502 (graphds_dfs, graphds_scc): Ditto.
15503 (for_each_edge): New parameter. Pass the new parameter to callback
15505 * graphds.h (skip_edge_callback): New function pointer type.
15506 (graphds_dfs, graphds_scc): New function pointer parameter.
15507 (graphds_edge_callback, for_each_edge): New parameter.
15509 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15511 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15512 out code checking if runtime alias check is possible to below ...
15513 Call the new function.
15514 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15515 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15517 2017-06-07 Marek Polacek <polacek@redhat.com>
15520 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15521 TYPE_OVERFLOW_WRAPS check.
15523 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15525 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15526 if versioning is required.
15527 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15528 peeling with the check for versioning.
15530 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15532 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15533 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15534 Set true to new parameter if new ssa variable is defined.
15535 (vect_gen_vector_loop_niters): Refactor. Set range information
15536 for the new vector loop bound variable.
15537 (vect_do_peeling): Ditto.
15539 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15541 * tree-affine.c (ssa.h): Include header file.
15542 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15543 has wrapping overflow behavior.
15545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15547 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15549 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15551 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15552 (tree_to_aff_combination): ... here.
15554 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15556 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15557 reg_pressure model function.
15558 (ivopts_global_cost_for_size): Delete.
15559 (determine_set_costs, iv_ca_recount_cost): Call new model function
15560 ivopts_estimate_reg_pressure.
15562 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15564 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15565 expensive than udiv. Remove floating point cases from mod.
15567 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15569 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15570 Increase idiv cost.
15572 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15574 * config/aarch64/aarch64.md
15575 (copysignsf3): Fix mask generation.
15577 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15579 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15581 (class dump_manager): Add register_dumps method.
15582 * dumpfile.c: Include langhooks.h.
15583 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15584 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15585 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15586 (dump_manager::dump_register): Start with 512 entries instead of 32.
15587 (dump_manager::register_dumps): New method.
15588 * toplev.c (general_init): Instead of invoking register_dumps
15589 langhook, invoke register_dumps method on the dump manager.
15590 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15593 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15595 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15596 Say that # requires an associated define_split to exist, and that
15597 the define_split must be suitable for use after register allocation.
15599 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15601 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15602 (compute_outgoing_frequencies): Also initialize zero counts.
15603 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15604 around loops; preserve more of profile when nothing changes.
15606 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15608 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15610 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15611 * config/arm/arm-cpu-cdata.h: Regenerate.
15612 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15613 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15614 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15615 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15616 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15618 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15620 * config/arm/t-rmprofile: Likewise.
15621 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15623 2017-06-06 David S. Miller <davem@davemloft.net>
15626 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15627 blockage if function uses alloca.
15629 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15631 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15632 New "uid" fields to hold pretty-print IDs of group and ref.
15633 Memory references are now identified as <group_id>:<ref_id>
15634 instead of using [random] addresses.
15635 (dump_mem_details): Simplify, no functional change.
15636 (dump_mem_ref): Simplify and make output more concise.
15637 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15638 (find_or_create_group): Initialize group uid.
15639 (record_ref): Initialize ref uid. Improve debug output.
15640 (prune_group_by_reuse, should_issue_prefetch_p,)
15641 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15642 (mark_nontemporal_store, determine_loop_nest_reuse):
15643 Improve debug output.
15645 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15647 * dbgcnt.def (prefetch): New debug counter.
15648 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15649 (schedule_prefetches): Stop issueing prefetches if debug counter
15652 2017-06-06 Tom de Vries <tom@codesourcery.com>
15654 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15655 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15657 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15659 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15660 Use aarch64_reg_or_zero predicate for operand 4.
15661 (aarch64_compare_and_swap<mode> define_insn_and_split):
15662 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15663 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15665 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15667 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15668 (arm_compute_save_core_reg_mask): This.
15669 (thumb1_compute_save_reg_mask): Rename into ...
15670 (thumb1_compute_save_core_reg_mask): This.
15671 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15672 (arm_compute_frame_layout): Likewise.
15674 2017-06-06 Richard Biener <rguenther@suse.de>
15676 PR tree-optimization/80974
15677 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15678 keep or clear leaders SSA info.
15680 2017-06-06 Tom de Vries <tom@codesourcery.com>
15682 * config/nvptx/nvptx.c (split_mode_p): New function.
15683 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15685 2017-06-06 Tom de Vries <tom@codesourcery.com>
15687 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15689 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15692 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15695 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15697 * shrink-wrap.c (handle_simple_exit): Update profile.
15698 (try_shrink_wrapping): Upate profile.
15700 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15702 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15703 (tree_guess_outgoing_edge_probabilities): New.
15704 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15705 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15707 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15709 * ipa-split.c (split_function): Initialize return bb profile.
15711 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15713 * profile.c (compute_branch_probabilities): Also initialize
15714 EXIT_BLOCK profile.
15716 2017-06-06 Richard Biener <rguenther@suse.de>
15718 PR tree-optimization/80928
15719 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15720 (vect_analyze_loop_operations): Properly guard analysis for
15722 (vect_transform_loop): Likewise.
15723 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15724 (vect_model_induction_cost): Do not cost for pure SLP.
15725 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15726 of induction in inner loop vectorization.
15727 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15728 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15729 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15731 (vect_analyze_slp_cost_1): Cost induction.
15732 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15733 (vect_get_slp_vect_defs): Likewise.
15734 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15735 (vect_transform_stmt): Handle SLP reductions.
15736 * tree-vectorizer.h (vectorizable_induction): Adjust.
15738 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15740 * config/rs6000/rs6000.c (make_resolver_func): Update
15741 init_lowered_empty_function call.
15743 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15745 * doc/invoke.texi: Document the -fprofile-abs-path option.
15746 * common.opt (fprofile-abs-path): New option.
15747 * gcov-io.h (gcov_write_filename): Declare.
15748 * gcov-io.c (gcov_write_filename): New function.
15749 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15750 * profile.c (output_location): Likewise.
15752 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15754 * shring-wrap.c: Revert accidental commit.
15756 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15758 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15760 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15762 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15764 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15765 profile in callgraph edge.
15766 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15767 (apply_scale): Likewise.
15768 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15769 Also scale profile when inlining function with zero profile.
15770 (initialize_cfun): Update exit block profile even when it is zero.
15771 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15772 when profile is read.
15774 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15776 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15777 (CLONE_*): New constants to define the processors we can generate
15778 code for with the target_clone attribute.
15779 (rs6000_clone_map): New array to identify which clone processors
15780 the current program is running on.
15781 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15782 target_clone attribute.
15783 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15784 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15785 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15786 (cpu_expand_builtin): Add support for target_clone attribute.
15787 (rs6000_valid_attribute_p): Allow "default" attribute.
15788 (get_decl_name): New debug function to simplify printing the
15789 current function name in debugging statements.
15790 (rs6000_clone_priority): New functions to support the target_clone
15791 attribute, and be able to generate code to switch between ISA 2.05
15792 through ISA 3.0 (power6 through power9).
15793 (rs6000_compare_version_priority): Likewise.
15794 (rs6000_get_function_versions_dispatcher): Likewise.
15795 (make_resolver_func): Likewise.
15796 (add_condition_to_bb): Likewise.
15797 (dispatch_function_versions): Likewise.
15798 (rs6000_generate_version_dispatcher_body): Likewise.
15799 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15800 (fusion_gpr_load_p): Fix a spacing issue.
15801 * doc/extend.texi (Common Function Attributes): Document that the
15802 PowerPC supports the target_clone attribute.
15804 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15806 * config/arm/arm.h: explain F symbol found in description of ARM
15807 register allocation in its legend.
15809 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15811 * config/mips/frame-header-opt.c: Include profile-count.h.
15812 * config/riscv/riscv.c: Include profile-count.h
15814 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15816 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15818 (sm_set_flag_if_changed): Add bbs field.
15819 (execute_sm_if_changed_flag_set): Pass BBS.
15820 (execute_sm): Update.
15822 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15824 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15827 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15829 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15830 (peephole2): New peephole2 to emit the above.
15831 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15833 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15835 * config/aarch64/aarch64.c (define_peephole2 above
15836 *sub_<shift>_<mode>): New peephole.
15838 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15840 * config/i386/i386.c (make_resolver_func): Update.
15841 * Makefile.in: Add profile-count.h and profile-count.o
15842 * auto-profile.c (afdo_indirect_call): Update to new API.
15843 (afdo_set_bb_count): Update.
15844 (afdo_propagate_edge): Update.
15845 (afdo_propagate_circuit): Update.
15846 (afdo_calculate_branch_prob): Update.
15847 (afdo_annotate_cfg): Update.
15848 * basic-block.h: Include profile-count.h
15849 (struct edge_def): Turn count to profile_count.
15850 (struct basic_block_def): Likewie.
15851 (REG_BR_PROB_BASE): Move to profile-count.h
15852 (RDIV): Move to profile-count.h
15853 * bb-reorder.c (max_entry_count): Turn to profile_count.
15854 (find_traces): Update.
15855 (rotate_loop):Update.
15856 (connect_traces):Update.
15857 (sanitize_hot_paths):Update.
15858 * bt-load.c (migrate_btr_defs): Update.
15859 * cfg.c (RDIV): Remove.
15860 (init_flow): Use alloc_block.
15861 (alloc_block): Uninitialize count.
15862 (unchecked_make_edge): Uninitialize count.
15863 (check_bb_profile): Update.
15864 (dump_edge_info): Update.
15865 (dump_bb_info): Update.
15866 (update_bb_profile_for_threading): Update.
15867 (scale_bbs_frequencies_int): Update.
15868 (scale_bbs_frequencies_gcov_type): Update.
15869 (scale_bbs_frequencies_profile_count): New.
15870 * cfg.h (update_bb_profile_for_threading): Update.
15871 (scale_bbs_frequencies_profile_count): Declare.
15872 * cfgbuild.c (compute_outgoing_frequencies): Update.
15873 (find_many_sub_basic_blocks): Update.
15874 * cfgcleanup.c (try_forward_edges): Update.
15875 (try_crossjump_to_edge): Update.
15876 * cfgexpand.c (expand_gimple_tailcall): Update.
15877 (construct_exit_block): Update.
15878 * cfghooks.c (verify_flow_info): Update.
15879 (dump_bb_for_graph): Update.
15880 (split_edge): Update.
15881 (make_forwarder_block): Update.
15882 (duplicate_block): Update.
15883 (account_profile_record): Update.
15884 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15885 (get_estimated_loop_iterations): Update.
15886 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15887 (single_likely_exit): Update.
15888 * cfgloopmanip.c (scale_loop_profile): Update.
15890 (set_zero_probability): Update.
15891 (lv_adjust_loop_entry_edge): Update.
15892 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15893 (purge_dead_edges): Update.
15894 (rtl_account_profile_record): Update.
15895 * cgraph.c (cgraph_node::create): Uninitialize count.
15896 (symbol_table::create_edge): Uninitialize count.
15897 (cgraph_update_edges_for_call_stmt_node): Update.
15898 (cgraph_edge::dump_edge_flags): Update.
15899 (cgraph_node::dump): Update.
15900 (cgraph_edge::maybe_hot_p): Update.
15901 * cgraph.h: Include profile-count.h
15902 (create_clone), create_edge, create_indirect_edge): Update.
15903 (cgraph_node): Turn count to profile_count.
15904 (cgraph_edge0: Likewise.
15905 (make_speculative, clone): Update.
15906 (create_edge): Update.
15907 (init_lowered_empty_function): Update.
15908 * cgraphclones.c (cgraph_edge::clone): Update.
15909 (duplicate_thunk_for_node): Update.
15910 (cgraph_node::create_clone): Update.
15911 * cgraphunit.c (cgraph_node::analyze): Update.
15912 (cgraph_node::expand_thunk): Update.
15913 * final.c (dump_basic_block_info): Update.
15914 * gimple-streamer-in.c (input_bb): Update.
15915 * gimple-streamer-out.c (output_bb): Update.
15916 * graphite.c (print_global_statistics): Update.
15917 (print_graphite_scop_statistics): Update.
15918 * hsa-brig.c: Include basic-block.h.
15919 * hsa-dump.c: Include basic-block.h.
15920 * hsa-gen.c (T sum_slice): Update.
15921 (convert_switch_statements):Update.
15922 * hsa-regalloc.c: Include basic-block.h.
15923 * ipa-chkp.c (chkp_produce_thunks): Update.
15924 * ipa-cp.c (struct caller_statistics): Update.
15925 (init_caller_stats): Update.
15926 (gather_caller_stats): Update.
15927 (ipcp_cloning_candidate_p): Update.
15928 (good_cloning_opportunity_p): Update.
15929 (get_info_about_necessary_edges): Update.
15930 (dump_profile_updates): Update.
15931 (update_profiling_info): Update.
15932 (update_specialized_profile): Update.
15933 (perhaps_add_new_callers): Update.
15934 (decide_about_value): Update.
15935 (ipa_cp_c_finalize): Update.
15936 * ipa-devirt.c (struct odr_type_warn_count): Update.
15937 (struct decl_warn_count): Update.
15938 (struct final_warning_record): Update.
15939 (possible_polymorphic_call_targets): Update.
15940 (ipa_devirt): Update.
15941 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15942 * ipa-icf.c (sem_function::merge): Update.
15943 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15944 * ipa-inline.c (compute_uninlined_call_time): Update.
15945 (compute_inlined_call_time): Update.
15946 (want_inline_small_function_p): Update.
15947 (want_inline_self_recursive_call_p): Update.
15948 (edge_badness): Update.
15949 (lookup_recursive_calls): Update.
15950 (recursive_inlining): Update.
15951 (inline_small_functions): Update.
15952 (dump_overall_stats): Update.
15953 (dump_inline_stats): Update.
15954 * ipa-profile.c (ipa_profile_generate_summary): Update.
15955 (ipa_propagate_frequency): Update.
15956 (ipa_profile): Update.
15957 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15958 * ipa-utils.c (ipa_merge_profiles): Update.
15959 * loop-doloop.c (doloop_modify): Update.
15960 * loop-unroll.c (report_unroll): Update.
15961 (unroll_loop_runtime_iterations): Update.
15962 * lto-cgraph.c (lto_output_edge): Update.
15963 (lto_output_node): Update.
15964 (input_node): Update.
15965 (input_edge): Update.
15966 (merge_profile_summaries): Update.
15967 * lto-streamer-in.c (input_cfg): Update.
15968 * lto-streamer-out.c (output_cfg): Update.
15969 * mcf.c (create_fixup_graph): Update.
15970 (adjust_cfg_counts): Update.
15971 (sum_edge_counts): Update.
15972 * modulo-sched.c (sms_schedule): Update.
15973 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15974 * predict.c (maybe_hot_count_p): Update.
15975 (probably_never_executed): Update.
15976 (dump_prediction): Update.
15977 (combine_predictions_for_bb): Update.
15978 (propagate_freq): Update.
15979 (handle_missing_profiles): Update.
15980 (counts_to_freqs): Update.
15981 (rebuild_frequencies): Update.
15982 (force_edge_cold): Update.
15983 * predict.h: Include profile-count.h
15984 (maybe_hot_count_p, counts_to_freqs): UPdate.
15985 * print-rtl-function.c: Do not include cfg.h
15986 * print-rtl.c: Include basic-block.h
15987 * profile-count.c: New file.
15988 * profile-count.h: New file.
15989 * profile.c (is_edge_inconsistent): Update.
15990 (correct_negative_edge_counts): Update.
15991 (is_inconsistent): Update.
15992 (set_bb_counts): Update.
15993 (read_profile_edge_counts): Update.
15994 (compute_frequency_overlap): Update.
15995 (compute_branch_probabilities): Update; Initialize and deinitialize
15997 (branch_prob): Update.
15998 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15999 (edge_gcov_count): New.
16000 (bb_gcov_count): New.
16001 * shrink-wrap.c (try_shrink_wrapping): Update.
16002 * tracer.c (better_p): Update.
16003 * trans-mem.c (expand_transaction): Update.
16004 (ipa_tm_insert_irr_call): Update.
16005 (ipa_tm_insert_gettmclone_call): Update.
16006 * tree-call-cdce.c: Update.
16007 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16008 (gimple_duplicate_sese_tail): Update.
16009 (gimple_account_profile_record): Update.
16010 (execute_fixup_cfg): Update.
16011 * tree-inline.c (copy_bb): Update.
16012 (copy_edges_for_bb): Update.
16013 (initialize_cfun): Update.
16014 (freqs_to_counts): Update.
16015 (copy_cfg_body): Update.
16016 (expand_call_inline): Update.
16017 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16018 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16019 (try_unroll_loop_completely): Update.
16020 (try_peel_loop): Update.
16021 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16022 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16023 * tree-ssa-loop-split.c (connect_loops): Update.
16024 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16025 * tree-ssa-reassoc.c (branch_fixup): Update.
16026 * tree-ssa-tail-merge.c (replace_block_by): Update.
16027 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16028 (compute_path_counts): Update.
16029 (update_profile): Update.
16030 (recompute_probabilities): Update.
16031 (update_joiner_offpath_counts): Update.
16032 (estimated_freqs_path): Update.
16033 (freqs_to_counts_path): Update.
16034 (clear_counts_path): Update.
16035 (ssa_fix_duplicate_block_edges): Update.
16036 (duplicate_thread_path): Update.
16037 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16038 (struct switch_conv_info): Update.
16039 * tree-tailcall.c (decrease_profile): Update.
16040 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16041 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16042 * value-prof.c (check_counter): Update.
16043 (gimple_divmod_fixed_value): Update.
16044 (gimple_mod_pow2): Update.
16045 (gimple_mod_subtract): Update.
16046 (gimple_ic_transform): Update.
16047 (gimple_stringop_fixed_value): Update.
16048 * value-prof.h (gimple_ic): Update.
16050 2017-06-02 Carl Love <cel@us.ibm.com>
16052 * config/rs6000/rs6000-c: Add support for built-in functions
16053 vector double vec_doublee (vector signed int);
16054 vector double vec_doublee (vector unsigned int);
16055 vector double vec_doublee (vector float);
16056 vector double vec_doubleh (vector signed int);
16057 vector double vec_doubleh (vector unsigned int);
16058 vector double vec_doubleh (vector float);
16059 vector double vec_doublel (vector signed int);
16060 vector double vec_doublel (vector unsigned int);
16061 vector double vec_doublel (vector float);
16062 vector double vec_doubleo (vector signed int);
16063 vector double vec_doubleo (vector unsigned int);
16064 vector double vec_doubleo (vector float);.
16065 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16066 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16068 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16069 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16070 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16072 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16073 vec_doublel, vec_doubleh.
16074 * doc/extend.texi: Update the built-in documentation file for the
16075 new built-in functions.
16077 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16080 * ipa-inline-analysis.c (free_growth_caches): Set
16081 edge_removal_hook_holder to NULL after removing it.
16083 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16085 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16086 comparision with zero.
16088 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16089 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16090 for early expansion of vec_min and vec_max builtins.
16091 (builtin_function_type): Add min/max unsigned variants to those
16092 identified as having unsigned arguments.
16094 2017-06-02 Olivier Hainque <hainque@adacore.com>
16096 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16098 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16100 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16101 Use VALL_F16 iterator rather than VALL.
16103 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16105 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16106 Emit CBNZ inside loop when doing a strong exchange and comparing
16107 against zero. Generate the CC flags after the loop.
16109 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16111 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16112 (dl_section_ref): New.
16113 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16114 On AIX, append an expression to subtract the size of the
16115 section length to dl_section_ref.
16117 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16119 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16120 for early expansion of vector absolute builtins.
16122 2017-06-02 Richard Biener <rguenther@suse.de>
16124 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16125 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16127 2017-06-02 Richard Biener <rguenther@suse.de>
16129 PR tree-optimization/80948
16130 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16131 stmt order as well.
16133 2017-06-02 Richard Biener <rguenther@suse.de>
16135 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16137 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16138 for inductions as relevant.
16140 2017-06-02 Richard Biener <rguenther@suse.de>
16142 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16143 (vectorizable_induction): ... this. Remove dead code.
16145 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16147 * builtins. (expand_builtin_alloca): Remove second parameter and
16148 infer its value from the first parameter instead.
16149 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16151 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16153 PR rtl-optimization/80903
16154 * loop-doloop.c (add_test): Unshare sequence.
16156 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16158 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16160 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16162 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16164 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16165 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16166 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16167 (xlogue_layout::get_stub_rtx): Make static.
16168 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16169 (xlogue_layout::compute_stub_managed_regs): Rename to...
16170 (xlogue_layout::count_stub_managed_regs): ...this.
16171 (xlogue_layout::is_stub_managed_reg): New function.
16172 (xlogue_layout::m_stub_names): Rename to...
16173 (xlogue_layout::s_stub_names): ...this, make static.
16174 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16175 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16176 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16177 xlogue_layout::s_stub_names): Instantiate statics.
16178 (stub_managed_regs): Remove.
16179 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16180 (disable_call_ms2sysv_xlogues): Rename to...
16181 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16182 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16184 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16185 change after reload_completed.
16186 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16188 (ix86_expand_prologue): Likewise.
16189 (ix86_expand_epilogue): Likewise.
16190 (ix86_expand_split_stack_prologue): Likewise.
16191 (ix86_compute_frame_layout): Remove frame parameter ...
16192 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16193 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16195 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16196 (ix86_frame): Move from here ...
16197 * config/i386/i386.h (ix86_frame): ... to here.
16198 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16199 complete ix86_frame data structure instead. Remove some_ld_name.
16201 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16203 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16204 symbols that hold a DECL_VALUE_EXPR.
16206 2017-06-01 Martin Jambor <mjambor@suse.cz>
16208 PR tree-optimization/80898
16209 * tree-sra.c (process_subtree_disqualification): Removed.
16210 (disqualify_candidate): Do not acll
16211 process_subtree_disqualification.
16212 (subtree_mark_written_and_enqueue): New function.
16213 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16214 RHS has been disqualified and re-queue LHS if necessary. Apart
16215 from that, ignore disqualified RHS.
16217 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16219 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16220 address fetch for z10 or later.
16222 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16224 * config/arc/arc.md (tst_movb): Add guard when splitting.
16226 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16228 * config/arc/arc.c (arc_can_eliminate): Test against
16229 arc_frame_pointer_needed.
16231 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16233 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16234 to prevent store reordering.
16235 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16236 (type): Add block type.
16237 (stack_tie): Define special instruction to be used in
16240 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16242 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16243 constraint. It is not valid for the pattern.
16244 (noncommutative_binary_comparison): Likewise.
16246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16248 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16251 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16253 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16254 be used by the reg-alloc.
16256 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16258 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16259 reg-alloc when having mul64 or mul32x16 instructions.
16260 (mulsidi3): Likewise.
16261 (umulsidi3): Likewise.
16262 (mulsi32x16): New pattern.
16263 (mulsi64): Likewise.
16264 (mulsidi64): Likewise.
16265 (umulsidi64): Likewise.
16266 (MUL32x16_REG): Define.
16267 (mul64_600): Use MUL32x16_REG.
16268 (mac64_600): Likewise.
16269 (umul64_600): Likewise.
16270 (umac64_600): Likewise.
16272 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16274 * config/arc/arc.md (mulsi3_700): Make it commutative.
16276 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16278 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16280 (*sign_extendsidi2_insn): Likewise for movstosw.
16282 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16284 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16285 the type of the input discriminant value. Convert the
16286 discriminant value of signedness vary.
16288 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16290 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16291 Add to -Wall section.
16293 2017-06-01 Richard Biener <rguenther@suse.de>
16295 PR middle-end/66313
16296 * fold-const.c (fold_plusminus_mult_expr): If the factored
16297 factor may be zero use a wrapping type for the inner operation.
16298 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16299 and handle moved defs.
16300 (process_assignment): Properly guard the unary op case. Return a
16301 tri-state indicating that moving the stmt before the call may allow
16302 to continue. Pass through to_move.
16303 (find_tail_calls): Handle moving unrelated defs before
16306 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16309 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16310 splitter result in the canonical way.
16312 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16314 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16315 also for 32bit target. Update insn attributes.
16316 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16318 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16320 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16321 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16322 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16323 (_mm_maskz_min_ss): New intrinsics.
16325 2017-05-31 Martin Liska <mliska@suse.cz>
16327 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16328 Change comment style to one we normally use.
16329 (vectorizable_reduction): Likewise.
16330 (vectorizable_induction): Likewise.
16331 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16332 (vectorizable_call): Likewise.
16333 (vectorizable_simd_clone_call): Likewise.
16334 (vectorizable_conversion): Likewise.
16335 (vectorizable_assignment): Likewise.
16336 (vectorizable_shift): Likewise.
16337 (vectorizable_operation): Likewise.
16338 (vectorizable_store): Likewise.
16339 (vectorizable_load): Likewise.
16340 * tree-vectorizer.h: Likewise.
16342 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16344 * passes.c (emergency_dump_function): New.
16345 * tree-pass.h (emergency_dump_function): Declare.
16346 * plugin.c (plugins_internal_error_function): Remove.
16347 * plugin.h (plugins_internal_error_function): Remove declaration.
16348 * toplev.c (internal_error_function): New static function. Use it...
16349 (general_init): ...here.
16351 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16353 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16354 (arc_rtx_costs): Add costs for new patterns.
16355 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16356 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16358 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16360 * tree-ssa-strlen.c (get_next_strinfo): New function.
16361 (get_stridx_plus_constant): Use it.
16362 (zero_length_string): Likewise.
16363 (adjust_related_strinfos): Likewise.
16364 (adjust_last_stmt): Likewise.
16366 2017-05-31 Richard Biener <rguenther@suse.de>
16369 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16370 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16372 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16374 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16375 loop_vinfo argument and use of dependence distance vectors.
16376 Check instead whether the two references differ only in their
16377 initial value and assume that they have the same alignment if the
16378 difference is a multiple of the vector alignment.
16379 (vect_analyze_data_refs_alignment): Update call accordingly.
16381 2017-05-31 Martin Liska <mliska@suse.cz>
16384 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16386 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16388 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16389 (create_intersect_range_checks): Move from ...
16390 * tree-data-ref.c (create_intersect_range_checks_index)
16391 (create_intersect_range_checks): ... to here.
16392 (create_runtime_alias_checks): New function factored from ...
16393 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16394 here. Call above function.
16395 * tree-data-ref.h (create_runtime_alias_checks): New function.
16397 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16399 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16400 segment length for dr_b and compute it in wide_int.
16402 2017-05-31 Richard Biener <rguenther@suse.de>
16404 PR tree-optimization/80906
16405 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16406 and pass through iv_map.
16407 (copy_bb_and_scalar_dependences): Adjust.
16408 (translate_pending_phi_nodes): Likewise.
16409 (copy_loop_close_phi_args): Handle code-generating IVs instead
16412 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16414 * diagnostic-color.c (color_dict): Add "type-diff".
16415 (parse_gcc_colors): Update comment.
16416 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16417 -fdiagnostics-show-template-tree and -fno-elide-type.
16418 (GCC_COLORS): Add type-diff to example.
16420 (-fdiagnostics-show-template-tree): New.
16421 (-fno-elide-type): New.
16422 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16423 the pp_format_decoder callback. Call any m_format_postprocessor's
16425 (pretty_printer::pretty_printer): Initialize
16426 m_format_postprocessor.
16427 (pretty_printer::~pretty_printer): Delete any
16428 m_format_postprocessor.
16429 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16430 (class format_postprocessor): New class.
16431 (struct pretty_printer::format_decoder): Document the new parameters.
16432 (struct pretty_printer::m_format_postprocessor): New field.
16433 * tree-diagnostic.c (default_tree_printer): Update for new
16434 bool and const char ** params.
16435 * tree-diagnostic.h (default_tree_printer): Likewise.
16437 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16439 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16440 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16441 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16442 rs6000_gen_cell_microcode code.
16443 (rs6000_final_prescan_insn): Delete.
16444 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16445 "warn-cell-microcode" entries.
16446 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16447 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16448 throughout. Change cc_reg_not_micro_cr0_operand to
16449 cc_reg_not_cr0_operand throughout.
16450 (*extendhi<mode>2_noload): Delete.
16451 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16452 (mwarn-cell-microcode): Delete.
16453 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16454 -mgen-cell-microcode and -mwarn-cell-microcode.
16456 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16459 * config/i386/constraints.md (Yd): New constraint.
16461 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16462 and (?Yd, r) alternatives. Update insn attributes.
16463 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16464 and (?*Yd, r) alternatives. Update insn attributes.
16465 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16467 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16469 * gimplify.c (gimplify_modify_expr): Don't create a
16470 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16473 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16475 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16477 2017-05-30 Richard Biener <rguenther@suse.de>
16479 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16480 and reduc_def fields.
16481 (STMT_VINFO_REDUC_TYPE): New define.
16482 (STMT_VINFO_REDUC_DEF): Likewise.
16483 (vect_force_simple_reduction): Adjust prototype.
16484 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16485 (vect_is_simple_reduction): Remove check_reduction argument.
16486 (vect_force_simple_reduction): Adjust and set
16487 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16488 (vectorizable_reduction): Do not re-do reduction analysis
16489 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16490 * tree-parloops.c (gather_scalar_reductions): Adjust.
16492 2017-05-30 Richard Biener <rguenther@suse.de>
16494 PR middle-end/80901
16495 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16498 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16500 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16501 Introduce unknown_misalignment parameter and remove vf.
16502 (vect_peeling_hash_get_lowest_cost):
16503 Pass unknown_misalignment parameter.
16504 (vect_enhance_data_refs_alignment):
16505 Fix unsupportable data ref treatment.
16507 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16509 * tree-vect-data-refs.c (vect_get_data_access_cost):
16510 Workaround for SLP handling.
16511 (vect_enhance_data_refs_alignment):
16512 Compute costs for doing no peeling at all, compare to the best
16513 peeling costs so far and avoid peeling if cheaper.
16515 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16517 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16518 Return peeling info and set costs to zero for unlimited cost
16520 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16521 with unknown misalignment. Compute and costs for unknown
16522 misalignment, compare them to the costs for known misalignment
16523 and choose the cheapest for peeling.
16525 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16527 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16528 (vect_get_peeling_costs_all_drs): Create function.
16529 (vect_peeling_hash_get_lowest_cost):
16530 Use vect_get_peeling_costs_all_drs.
16531 (vect_peeling_supportable): Create function.
16532 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16534 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16536 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16537 DR_HAS_NEGATIVE_STEP.
16538 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16539 (vect_enhance_data_refs_alignment): Use.
16540 (vect_duplicate_ssa_name_ptr_info): Use.
16541 * tree-vectorizer.h (dr_misalignment): Use.
16542 (known_alignment_for_access_p): Use.
16544 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16547 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16549 (has_section_name): New function.
16551 2017-05-30 Martin Liska <mliska@suse.cz>
16554 * auto-profile.c (get_function_decl_from_block): Fix
16557 2017-05-30 Richard Biener <rguenther@suse.de>
16559 PR middle-end/80876
16560 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16562 2017-05-30 Martin Liska <mliska@suse.cz>
16564 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16565 * dumpfile.h (struct dump_file_info): Remove ctors.
16567 2017-05-30 Martin Liska <mliska@suse.cz>
16569 * predict.def: Fix GNU coding style.
16571 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16573 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16574 Mark 'to' argument with ATTRIBUTE_UNUSED.
16576 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16578 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16579 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16580 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16583 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16585 * doc/install.texi (Options specification): Restore entry of
16586 --enable-sjlj-exceptions.
16588 2017-05-27 Michael Eager <eager@eagercon.com>
16591 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16593 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16595 * config/microblaze/microblaze.h
16596 (FIXED_REGISTERS): Update in macro.
16597 (CALL_USED_REGISTERS): Update in macro.
16599 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16601 * doc/install.texi: Add links to macOS binary distributions.
16603 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16607 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16609 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16611 2017-05-26 Martin Liska <mliska@suse.cz>
16613 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16615 2017-05-26 Martin Liska <mliska@suse.cz>
16617 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16618 always leading ';; '.
16619 (dump_bb_info): Likewise.
16620 (brief_dump_cfg): Likewise.
16621 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16622 * dumpfile.c: Remove usage of TDF_VERBOSE.
16623 * dumpfile.h (enum dump_kind): Likewise.
16624 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16625 * print-tree.c (debug_verbose): Remove.
16626 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16627 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16628 * tree-diagnostic.c (default_tree_printer): Replace
16629 TDF_DIAGNOSTIC with TDF_SLIM.
16631 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16633 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16634 in parameter loop, rather than loop_vinfo.
16635 (create_intersect_range_checks): Ditto.
16636 (vect_create_cond_for_alias_checks): Update call to above functions.
16638 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16640 PR tree-optimization/80815
16641 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16642 for merging runtime alias checks. Handle negative DR_STEPs.
16644 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16646 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16648 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16649 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16650 out code pruning runtime alias checks.
16651 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16652 factored out from above.
16653 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16655 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16657 (prune_runtime_alias_test_list): New decalaration.
16659 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16661 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16662 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16663 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16664 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16665 (operator==, comp_dr_with_seg_len_pair): Ditto.
16666 (vect_prune_runtime_alias_test_list): Ditto.
16668 2017-05-26 Martin Liska <mliska@suse.cz>
16671 * params.def: Bound partial-inlining-entry-probability param.
16673 2017-05-26 Marek Polacek <polacek@redhat.com>
16676 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16679 2017-05-26 Richard Biener <rguenther@suse.de>
16681 PR tree-optimization/80842
16682 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16685 2017-05-26 Richard Biener <rguenther@suse.de>
16687 PR tree-optimization/80844
16688 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16690 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16692 * doc/md.texi (Machine Constraints): Update x86 family
16693 machine constraints section to match 'config/i386/constraints.md'.
16695 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16697 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16699 2017-05-25 Nathan Sidwell <nathan@acm.org>
16701 * doc/invoke.texi (--enable-languages): Update documentation.
16703 2017-05-25 Martin Liska <mliska@suse.cz>
16705 * dumpfile.c: Add TDF_FOLDING.
16706 * dumpfile.h (enum dump_kind): Likewise.
16707 * genmatch.c (dt_simplify::gen_1): Use it.
16709 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16711 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16713 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16715 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16716 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16718 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16720 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16721 * match.pd (X == C): Rewrite it here.
16722 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16723 with_certain_nonzero_bits2): New predicates.
16724 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16726 2017-05-24 Nathan Sidwell <nathan@acm.org>
16728 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16731 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16734 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16736 * config/powerpcspe: New port. Files are copied from the rs6000
16737 port, with "rs6000" in filenames replaced by "powerpcspe".
16739 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16741 PR rtl-optimization/80754
16742 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16744 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16746 * config/sparc/sparc.md (length): Return the correct value for -mflat
16747 sibcalls to match output_sibcall.
16749 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16753 * config/rs6000/rs6000.c (struct machine_function): Add new field
16755 (rs6000_get_separate_components): Init that field, use it.
16756 (rs6000_components_for_bb): Use the field.
16758 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16760 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16762 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16764 PR middle-end/80823
16765 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16767 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16770 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16771 address against address_operand predicate.
16772 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16774 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16776 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16777 parameters passed indirectly.
16779 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16781 * config/i386/i386.md (*movdi_internal): Remove SSE4
16782 alternative 18 (?r, *v). Update insn attributes.
16783 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16784 Update insn attributes.
16785 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16786 Update insn attributes.
16787 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16788 alternative 1 (r, v). Remove isa attribute.
16789 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16790 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16791 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16793 2017-05-23 Tom de Vries <tom@codesourcery.com>
16795 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16798 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16800 * cgraphunit.c (symbol_table::process_new_functions): Update.
16801 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16802 (inline_generate_summary): Rename to ...
16803 (ipa_fn_summary_generate): ... this one.
16804 (inline_read_summary): Rename to ...
16805 (ipa_fn_summary_read): ... this one.
16806 (inline_write_summary): Rename to ...
16807 (ipa_fn_summary_write): ... this one.
16808 (inline_free_summary): Rename to ...
16809 (ipa_free_fn_summary): ... this one.
16810 (pass_data_local_fn_summary, pass_local_fn_summary,
16811 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16812 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16813 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16814 make_pass_ipa_fn_summary): New.
16815 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16816 inline_write_summary, inline_free_summary): Remove.
16817 (ipa_free_fn_summary) : New.
16818 * ipa-inline.c (ipa_inline): Update.
16819 (pass_ipa_inline): Do not generate summaries.
16820 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16822 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16823 and add pass_ipa_fn_summary.
16824 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16826 (make_pass_inline_parameters): Remove.
16828 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16830 * omp-low.c (struct omp_context): Remove "default_kind" member.
16833 * omp-offload.c (execute_oacc_device_lower): Remove the
16834 parallelism dimensions function attributes for unparallelized
16835 OpenACC kernels constructs.
16837 2017-05-23 Martin Liska <mliska@suse.cz>
16839 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16841 (cgraph_edge::make_speculative): Likewise.
16842 (cgraph_edge::resolve_speculation): Likewise.
16843 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16844 (cgraph_node::dump): Likewise.
16845 * cgraph.h: Likewise.
16846 * cgraphunit.c (analyze_functions): Likewise.
16847 (symbol_table::compile): Likewise.
16848 * ipa-cp.c (print_all_lattices): Likewise.
16849 (determine_versionability): Likewise.
16850 (initialize_node_lattices): Likewise.
16851 (ipcp_verify_propagated_values): Likewise.
16852 (estimate_local_effects): Likewise.
16853 (update_profiling_info): Likewise.
16854 (create_specialized_node): Likewise.
16855 (perhaps_add_new_callers): Likewise.
16856 (decide_about_value): Likewise.
16857 (decide_whether_version_node): Likewise.
16858 (identify_dead_nodes): Likewise.
16859 (ipcp_store_bits_results): Likewise.
16860 * ipa-devirt.c (dump_targets): Likewise.
16861 (ipa_devirt): Likewise.
16862 * ipa-icf.c (sem_item::dump): Likewise.
16863 (sem_function::equals): Likewise.
16864 (sem_variable::equals): Likewise.
16865 (sem_item_optimizer::read_section): Likewise.
16866 (sem_item_optimizer::execute): Likewise.
16867 (congruence_class::dump): Likewise.
16868 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16869 (dump_inline_summary): Likewise.
16870 (estimate_node_size_and_time): Likewise.
16871 (inline_analyze_function): Likewise.
16872 * ipa-inline-transform.c (inline_call): Likewise.
16873 * ipa-inline.c (report_inline_failed_reason): Likewise.
16874 (want_early_inline_function_p): Likewise.
16875 (edge_badness): Likewise.
16876 (update_edge_key): Likewise.
16877 (inline_small_functions): Likewise.
16878 * ipa-profile.c (ipa_profile): Likewise.
16879 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16880 (ipa_make_edge_direct_to_target): Likewise.
16881 (remove_described_reference): Likewise.
16882 (ipa_impossible_devirt_target): Likewise.
16883 (propagate_controlled_uses): Likewise.
16884 (ipa_print_node_params): Likewise.
16885 (ipcp_transform_function): Likewise.
16886 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16887 (propagate_pure_const): Likewise.
16888 * ipa-reference.c (generate_summary): Likewise.
16889 (read_write_all_from_decl): Likewise.
16890 (propagate): Likewise.
16891 (ipa_reference_read_optimization_summary): Likewise.
16892 * ipa-utils.c (ipa_merge_profiles): Likewise.
16893 * ipa.c (walk_polymorphic_call_targets): Likewise.
16894 (symbol_table::remove_unreachable_nodes): Likewise.
16895 (ipa_single_use): Likewise.
16896 * passes.c (execute_todo): Likewise.
16897 * predict.c (drop_profile): Likewise.
16898 * symtab.c (symtab_node::get_dump_name): New function.
16899 (symtab_node::dump_name): Likewise.
16900 (symtab_node::dump_asm_name): Likewise.
16901 (symtab_node::dump_references): Likewise.
16902 (symtab_node::dump_referring): Likewise.
16903 (symtab_node::dump_base): Likewise.
16904 (symtab_node::debug_symtab): Likewise.
16905 * tree-sra.c (convert_callers_for_node): Likewise.
16906 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16907 * value-prof.c (init_node_map): Likewise.
16909 2017-05-23 Martin Liska <mliska@suse.cz>
16911 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16912 and symtab_node::debug_symtab to symbol_table::debug.
16913 * cgraphunit.c (analyze_functions): Use the renamed function.
16914 (symbol_table::compile): Likewise.
16915 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16916 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16917 * passes.c (execute_todo): Likewise.
16918 * symtab.c (symbol_table::dump): New function.
16919 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16921 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16923 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16924 that nonconst implies exec.
16926 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16928 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16929 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16930 (inline_edge_summary_vec): Turn into ...
16931 (ipa_call_summaries): ... this one.
16932 (redirect_to_unreachable, edge_set_predicate,
16933 evaluate_properties_for_edge, inline_summary_alloc,
16934 reset_ipa_call_summary, reset_inline_summary,
16935 inline_summary_t::duplicate): Update.
16936 (inline_edge_duplication_hook): Turn to ...
16937 (ipa_call_summary_t::duplicate): ... this one.
16938 (inline_edge_removal_hook): Turn to ...
16939 (ipa_call_summary_t::remove): ... this one.
16940 (dump_inline_edge_summary): Turn to ...
16941 (dump_ipa_call_summary): ... this one.
16942 (estimate_function_body_sizes): Update.
16943 (inline_update_callee_summaries): Update.
16944 (remap_edge_change_prob): Update.
16945 (remap_edge_summaries): Update.
16946 (inline_merge_summary): Update.
16947 (do_estimate_edge_time): Update.
16948 (inline_generate_summary): Update.
16949 (inline_read_section): Update.
16950 (inline_read_summary): Update.
16951 (inline_free_summary): Update.
16952 * ipa-inline.c (can_inline_edge_p): Update.
16953 (compute_inlined_call_time): Update.
16954 (want_inline_small_function_p): Update.
16955 (edge_badness): Update.
16956 (early_inliner): Update.
16957 * ipa-inline.h (inline_edge_summary): Turn to ...
16958 (ipa_call_summary): ... this one.
16959 (ipa_call_summary_t): New class.
16960 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16961 (ipa_call_summaries): New.
16962 (inline_edge_summary): Remove.
16963 (estimate_edge_growth): Update.
16964 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16965 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16966 * ipa-split.c (execute_split_functions): Update.
16967 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16969 2017-05-23 Tom de Vries <tom@codesourcery.com>
16971 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16972 attributes): Document rdrand effective target.
16974 2017-05-23 Tom de Vries <tom@codesourcery.com>
16976 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16977 attributes): Sort alphabetically.
16979 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16981 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16983 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16986 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16987 V2DF/V2DI splat into two separate patterns, one that handles
16988 registers, and the other that only handles memory. Drop support
16989 for splatting from a GPR on ISA 2.07 and then splitting the
16990 splat into direct move and splat.
16991 (vsx_splat_<mode>_reg): Likewise.
16992 (vsx_splat_<mode>_mem): Likewise.
16994 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16996 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16998 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17000 PR middle-end/80809
17001 * omp-low.c (finish_taskreg_remap): New function.
17002 (finish_taskreg_scan): If unit size of ctx->record_type
17003 is non-constant, unshare the size expression and replace
17004 decls in it with possible outer var refs.
17006 PR middle-end/80809
17007 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17008 GOVD_SHARED rather than GOVD_PRIVATE with it.
17009 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17010 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17012 PR middle-end/80853
17013 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17014 as last argument to build_outer_var_ref for pointer bases of array
17015 section reductions.
17017 2017-05-19 Martin Sebor <msebor@redhat.com>
17019 * print-tree.c (print_node): Print DECL_READ_P flag.
17021 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17023 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17024 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17025 * cgraph.c: Likewise.
17026 * cgraphunit.c: Likewise.
17027 * gengtype.c: Likewise.
17028 * ipa-cp.c: Likewise.
17029 * ipa-devirt.c: Likewise.
17030 * ipa-icf.c: Likewise.
17031 * ipa-predicate.c: Likewise.
17032 * ipa-profile.c: Likewise.
17033 * ipa-prop.c: Likewise.
17034 * ipa-split.c: Likewise.
17036 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17037 edge_predicate_pool, dump_inline_hints,
17038 inline_summary::account_size_time, redirect_to_unreachable,
17039 edge_set_predicate, set_hint_predicate,
17040 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17041 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17042 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17043 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17044 ipa_call_summary_t::remove, initialize_growth_caches,
17045 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17046 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17047 mark_modified, unmodified_parm_1, unmodified_parm,
17048 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17049 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17050 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17051 will_be_nonconstant_predicate, record_modified_bb_info,
17052 get_minimal_bb, record_modified, param_change_prob,
17053 phi_result_unknown_predicate, predicate_for_phi_result,
17054 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17055 estimate_function_body_sizes, compute_inline_parameters,
17056 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17057 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17058 inline_update_callee_summaries, remap_edge_change_prob,
17059 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17060 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17061 inline_analyze_function, inline_summary_t::insert,
17062 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17063 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17064 inline_free_summary): Move to ipa-fnsummary.h
17065 (predicate_t): Remove.
17066 * ipa-fnsummary.c: New file.
17067 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17068 (enum inline_hints_vals, inline_hints, agg_position_info,
17069 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17070 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17071 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17072 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17073 inline_read_summary, inline_write_summary, inline_free_summary,
17074 inline_analyze_function, initialize_inline_failed,
17075 inline_merge_summary, inline_update_overall_summary,
17076 compute_inline_parameters): Move to ipa-fnsummary.h
17077 * ipa-fnsummary.h: New file.
17078 * ipa-inline-transform.h: Include ipa-inline.h.
17079 * ipa-inline.c: LIkewise.
17081 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17083 * ipa-inline.c (edge_badness): Use inlined_time instead of
17084 inline_summaries->get.
17086 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17088 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17090 2017-05-22 Nathan Sidwell <nathan@acm.org>
17092 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17093 (fdump-lang): Document 'raw' option.
17094 * dumpfile.h (TDI_tu): Delete.
17095 * dumpfile.c (dump_files): Remove translation-unit.
17096 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17098 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17100 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17101 command option from $(AWK) call.
17102 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17103 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17104 [FORMAT]: Remove handling of variable.
17105 * config/avr/t-multilib: Regenerate.
17107 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17109 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17111 (dump_inline_summary): Do not print self_time.
17112 (estimate_function_body_sizes): Do not set self_time.
17113 (compute_inline_parameters): Likewise.
17114 (inline_read_section, inline_write_summary): Do not stream self_time.
17115 * ipa-inline.h (inline_summary): Drop self_time.
17117 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17119 * ipa-inline-analysis.c (account_size_time): Rename to ...
17120 (inline_summary::account_size_time): ... this one.
17121 (reset_ipa_call_summary): Turn to ...
17122 (ipa_call_summary::reset): ... this one.
17123 (reset_inline_summary): Turn to ...
17124 (inline_summary::reset): ... this one.
17125 (inline_summary_t::remove): Update.
17126 (inline_summary_t::duplicate): Update.
17127 (ipa_call_summary_t::remove): Update.
17128 (dump_inline_summary): Update.
17129 (estimate_function_body_sizes): Update.
17130 (compute_inline_parameters): Update.
17131 (estimate_node_size_and_time): Update.
17132 (inline_merge_summary): Update.
17133 (inline_update_overall_summary): Update.
17134 (inline_read_section): Update.
17135 (inline_write_summary): Update.
17136 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17137 add account_size_time and reset member functions.
17138 (ipa_call_summary): Add reset function.
17139 * ipa-predicate.h (predicate::operator &): Constify.
17141 2017-05-22 Richard Biener <rguenther@suse.de>
17143 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17145 2017-05-19 Jason Merrill <jason@redhat.com>
17147 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17149 2017-05-19 Marek Polacek <polacek@redhat.com>
17152 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17153 TYPE_OVERFLOW_WRAPS checks.
17155 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17157 * tree-core.h (enum omp_clause_default_kind): Add
17158 "OMP_CLAUSE_DEFAULT_PRESENT".
17159 * tree-pretty-print.c (dump_omp_clause): Handle it.
17160 * gimplify.c (enum gimplify_omp_var_data): Add
17161 "GOVD_MAP_FORCE_PRESENT".
17162 (gimplify_adjust_omp_clauses_1): Map it to
17163 "GOMP_MAP_FORCE_PRESENT".
17164 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17166 * gimplify.c (oacc_default_clause): Clarify.
17168 2017-05-19 Nathan Sidwell <nathan@acm.org>
17170 LANG_HOOK_REGISTER_DUMPS
17171 * toplev.c (general_init): Call register dump lang hook.
17172 * doc/invoke.texi: Document -fdump-lang option family.
17173 * dumpfile.c (dump_files): Remove class dump here.
17174 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17175 * dumpfile.h (tree_dump_index): Remove TDI_class.
17176 * langhooks-def.h (lhd_register_dumps): Declare.
17177 (LANG_HOOKS_REGISTER_DUMPS): Define.
17178 (LANG_HOOKS_INITIALIZER): Add it.
17179 * langhooks.c (lhd_register_dumps): Define.
17180 * langhooks.h (struct lang_hooks): Add register_dumps.
17182 2017-05-19 Nathan Sidwell <nathan@acm.org>
17184 * context.h (context::set_passes): New.
17185 * context.c (context::context): Do not create pass manager.
17186 * toplev.c (general_init): Create pass manager here.
17188 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17190 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17191 use this splitter if two add or or instructions would also work for
17192 the constant we want to generate.
17194 2017-05-19 Richard Biener <rguenther@suse.de>
17197 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17198 predicate evaluation.
17200 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17202 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17204 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17207 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17209 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17210 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17211 (inline_edge_summary_vec): Turn into ...
17212 (ipa_call_summaries): ... this one.
17213 (redirect_to_unreachable, edge_set_predicate,
17214 evaluate_properties_for_edge, inline_summary_alloc,
17215 reset_ipa_call_summary, reset_inline_summary,
17216 inline_summary_t::duplicate): Update.
17217 (inline_edge_duplication_hook): Turn to ...
17218 (ipa_call_summary_t::duplicate): ... this one.
17219 (inline_edge_removal_hook): Turn to ...
17220 (ipa_call_summary_t::remove): ... this one.
17221 (dump_inline_edge_summary): Turn to ...
17222 (dump_ipa_call_summary): ... this one.
17223 (estimate_function_body_sizes): Update.
17224 (inline_update_callee_summaries): Update.
17225 (remap_edge_change_prob): Update.
17226 (remap_edge_summaries): Update.
17227 (inline_merge_summary): Update.
17228 (do_estimate_edge_time): Update.
17229 (inline_generate_summary): Update.
17230 (inline_read_section): Update.
17231 (inline_read_summary): Update.
17232 (inline_free_summary): Update.
17233 * ipa-inline.c (can_inline_edge_p): Update.
17234 (compute_inlined_call_time): Update.
17235 (want_inline_small_function_p): Update.
17236 (edge_badness): Update.
17237 (early_inliner): Update.
17238 * ipa-inline.h (inline_edge_summary): Turn to ...
17239 (ipa_call_summary): ... this one.
17240 (ipa_call_summary_t): New class.
17241 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17242 (ipa_call_summaries): New.
17243 (inline_edge_summary): Remove.
17244 (estimate_edge_growth): Update.
17245 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17246 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17247 * ipa-split.c (execute_split_functions): Update.
17248 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17250 2017-05-19 Richard Biener <rguenther@suse.de>
17252 PR middle-end/80764
17253 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17255 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17257 * config/rs6000/rs6000.c (struct machine_function): Add field
17258 fpr_is_wrapped_separately.
17259 (rs6000_get_separate_components): Use 64 components. Handle the
17260 new FPR components.
17261 (rs6000_components_for_bb): Handle the FPR components.
17262 (rs6000_emit_prologue_components): Handle the FPR components.
17263 (rs6000_emit_epilogue_components): Handle the FPR components.
17264 (rs6000_set_handled_components): Handle the FPR components.
17265 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17266 that are already separately shrink-wrapped.
17267 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17268 that are already separately shrink-wrapped.
17270 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17273 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17276 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17277 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17278 cases where the register allocator uses a move and an offsettable
17279 memory operation to/from a FPR register on ISA 2.06/2.07.
17280 (define_peephole2 for Altivec d-form store): Likewise.
17282 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17285 * config/i386/mmx.md (*mov<mode>_internal): Enable
17286 alternatives 11, 12, 13 and 14 also for 32bit targets.
17287 Remove alternatives 15, 16, 17 and 18.
17288 * config/i386/sse.md (vec_concatv2di): Change
17289 alternative (!x, *y) to (x, ?!*Yn).
17291 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17293 * dumpfile.h (enum dump_kind): Remove stray comma.
17295 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17297 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17298 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17299 predicate::num_conditions
17300 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17301 (CHANGED): turn into predicate::changed.
17302 (agg_position_info): Move to ipa-predicate.h
17303 (add_condition, predicate::add_clause, predicate::operator &=,
17304 predicate::or_with, predicate::evaluate, predicate::probability,
17305 dump_condition, dump_clause, predicate::dump,
17306 predicate::remap_after_duplication, predicate::remap_after_inlining,
17307 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17308 (evaluate_conditions_for_known_args): Update.
17309 (set_cond_stmt_execution_predicate): Update.
17310 * ipa-inline.h: Include ipa-predicate.h
17311 (condition, inline_param_summary, conditions, agg_position_info,
17312 predicate): Move to ipa-predicate.h
17313 * ipa-predicate.c: New file.
17314 * ipa-predicate.h: New file.
17316 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17318 * final.c (leaf_function_p): Check we are not in a sequence.
17320 2017-05-18 Martin Liska <mliska@suse.cz>
17322 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17323 * dumpfile.c (dump_register): Use new enum dump_kind.
17324 (get_dump_file_name): Likewise.
17325 (dump_enable_all): Likewise.
17326 (dump_switch_p_1): Likewise.
17327 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17328 * dumpfile.h (enum dump_kind): New enum type.
17329 (struct dump_file_info): Create constructor and
17330 format fields and comments.
17331 * passes.c (pass_manager::register_one_dump_file):
17333 * statistics.c (statistics_early_init): Likewise.
17334 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17335 TDF_TREE with TDF_SLIM.
17336 (gather_memory_references_ref): Likewise.
17338 2017-05-18 Martin Liska <mliska@suse.cz>
17340 * vec.h (struct vnull): Use it.
17342 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17344 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17345 (true_predicate, false_predicate, true_predicate_p,
17346 false_predicate_p): Remove.
17347 (single_cond_predicate, not_inlined_predicate): Turn to member function
17349 (add_condition): Update.
17350 (add_clause): Turn to...
17351 (predicate::add_clause): ... this one; update; allow passing NULL
17353 (and_predicates): Turn to ...
17354 (predicate::operator &=): ... this one.
17355 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17356 (or_predicates): Turn to ...
17357 (predicate::or_with): ... this one.
17358 (evaluate_predicate): Turn to ...
17359 (predicate::evaluate): ... this one.
17360 (predicate_probability): Turn to ...
17361 (predicate::probability): ... this one.
17362 (dump_condition): Update.
17363 (dump_predicate): Turn to ...
17364 (predicate::dump): ... this one.
17365 (account_size_time): Update.
17366 (edge_set_predicate): Update.
17367 (set_hint_predicate): UPdate.
17368 (evaluate_conditions_for_known_args): Update.
17369 (evaluate_properties_for_edge): Update.
17370 (remap_predicate_after_duplication): Turn to...
17371 (predicate::remap_after_duplication): ... this one.
17372 (remap_hint_predicate_after_duplication): Update.
17373 (inline_summary_t::duplicate): UPdate.
17374 (dump_inline_edge_summary): Update.
17375 (dump_inline_summary): Update.
17376 (set_cond_stmt_execution_predicate): Update.
17377 (set_switch_stmt_execution_predicate): Update.
17378 (compute_bb_predicates): Update.
17379 (will_be_nonconstant_expr_predicate): Update.
17380 (will_be_nonconstant_predicate): Update.
17381 (phi_result_unknown_predicate): Update.
17382 (predicate_for_phi_result): Update.
17383 (array_index_predicate): Update.
17384 (estimate_function_body_sizes): Update.
17385 (estimate_node_size_and_time): Update.
17386 (estimate_ipcp_clone_size_and_time): Update.
17387 (remap_predicate): Rename to ...
17388 (predicate::remap_after_inlining): ... this one.
17389 (remap_hint_predicate): Update.
17390 (inline_merge_summary): Update.
17391 (inline_update_overall_summary): Update.
17392 (estimate_size_after_inlining): Update.
17393 (read_predicate): Rename to ...
17394 (predicate::stream_in): ... this one.
17395 (read_inline_edge_summary): Update.
17396 (write_predicate): Rename to ...
17397 (predicate::stream_out): ... this one.
17398 (write_inline_edge_summary): Update.
17399 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17400 (clause_t): Turn to uint32_t
17401 (predicate): Turn to class; implement constructor and operators
17403 (size_time_entry): Update.
17404 (inline_summary): Update.
17405 (inline_edge_summary): Update.
17407 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17409 * fold-const.c (fold_binary_loc): Move transformation...
17410 * match.pd (C - X CMP X): ... here.
17412 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17414 * config/sparc/sparc.c (sparc_option_override): Set function
17415 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17416 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17418 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17420 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17422 2017-05-18 Marek Polacek <polacek@redhat.com>
17425 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17426 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17427 gimple_assign_load_p.
17429 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17431 PR middle-end/80692
17432 * real.c (do_compare): Give decimal_do_compare preference over
17433 comparing just the signs.
17435 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17437 * doc/md.texi (Canonicalization of Instructions): Describe the
17438 canonical form of instructions that inherently set a condition
17441 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17443 PR middle-end/80775
17444 * tree-cfg.c: Move deletion of unreachable case statements to after
17445 the merging of consecutive case labels.
17447 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17449 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17450 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17451 restoring of callee-saved registers.
17453 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17455 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17456 * config/visium/visium.c (single_set_and_flags): Likewise.
17457 * config/visium/visium.md (Substitutions): Likewise.
17459 2017-05-17 Martin Liska <mliska@suse.cz>
17461 * cfg.c: Introduce dump_flags_t type and
17462 use it instead of int type.
17464 * cfghooks.c: Likewise.
17465 * cfghooks.h (struct cfg_hooks): Likewise.
17466 * cfgrtl.c: Likewise.
17467 * cfgrtl.h: Likewise.
17468 * cgraph.c (cgraph_node::get_body): Likewise.
17469 * coretypes.h: Likewise.
17470 * domwalk.c: Likewise.
17471 * domwalk.h: Likewise.
17472 * dumpfile.c (struct dump_option_value_info): Likewise.
17473 (dump_enable_all): Likewise.
17474 (dump_switch_p_1): Likewise.
17475 (opt_info_switch_p): Likewise.
17476 * dumpfile.h (enum tree_dump_index): Likewise.
17477 (struct dump_file_info): Likewise.
17478 * genemit.c: Likewise.
17479 * generic-match-head.c: Likewise.
17480 * gengtype.c (open_base_files): Likewise.
17481 * gimple-pretty-print.c: Likewise.
17482 * gimple-pretty-print.h: Likewise.
17483 * graph.c (print_graph_cfg): Likewise.
17484 * graphite-scop-detection.c (dot_all_sese): Likewise.
17485 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17486 * loop-unroll.c (report_unroll): Likewise.
17487 * passes.c (pass_manager::register_one_dump_file): Likewise.
17488 * print-tree.c: Likewise.
17489 * statistics.c: Likewise.
17490 * tree-cfg.c: Likewise.
17491 * tree-cfg.h: Likewise.
17492 * tree-dfa.c: Likewise.
17493 * tree-dfa.h: Likewise.
17494 * tree-dump.c (dump_function): Likewise.
17495 * tree-dump.h (struct dump_info): Likewise.
17496 * tree-pretty-print.c: Likewise.
17497 * tree-pretty-print.h: Likewise.
17498 * tree-ssa-live.c: Likewise.
17499 * tree-ssa-live.h: Likewise.
17500 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17501 * tree-vect-loop.c: Likewise.
17502 * tree-vect-slp.c: Likewise.
17504 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17505 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17507 PR tree-optimization/80457
17508 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17509 of all arguments to a statement as scalar_to_vec operations.
17510 (vectorizable_call): Adjust call to vect_model_simple_cost for
17512 (vectorizable_conversion): Likewise.
17513 (vectorizable_assignment): Likewise.
17514 (vectorizable_shift): Likewise.
17515 (vectorizable_operation): Likewise.
17516 (vectorizable_comparison): Likewise.
17517 (vect_is_simple_cond): Record the def types for operands.
17518 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17519 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17520 for statement argument count.
17522 2017-05-16 Carl Love <cel@us.ibm.com>
17524 * config/rs6000/rs6000-c: Add support for built-in functions
17525 vector unsigned long long vec_bperm (vector unsigned long long,
17526 vector unsigned char)
17527 vector signed long long vec_mule (vector signed int,
17529 vector unsigned long long vec_mule (vector unsigned int,
17530 vector unsigned int)
17531 vector signed long long vec_mulo (vector signed int,
17533 vector unsigned long long vec_mulo (vector unsigned int,
17534 vector unsigned int)
17535 vector signed char vec_sldw (vector signed char,
17536 vector signed char,
17538 vector unsigned char vec_sldw (vector unsigned char,
17539 vector unsigned char,
17541 vector signed short vec_sldw (vector signed short,
17542 vector signed short,
17544 vector unsigned short vec_sldw (vector unsigned short,
17545 vector unsigned short,
17547 vector signed int vec_sldw (vector signed int,
17550 vector unsigned int vec_sldw (vector unsigned int,
17551 vector unsigned int,
17553 vector signed long long vec_sldw (vector signed long long,
17554 vector signed long long,
17556 vector unsigned long long vec_sldw (vector unsigned long long,
17557 vector unsigned long long,
17559 * config/rs6000/rs6000-c: Add support for built-in functions
17560 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17561 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17562 * doc/extend.texi: Update the built-in documentation for the
17563 new built-in functions.
17565 2017-05-16 Marek Polacek <polacek@redhat.com>
17569 * tree.c (save_expr): Don't fold the expression.
17571 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17573 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17574 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17575 and (?*y,m). Update insn attributes.
17577 2017-05-16 Martin Liska <mliska@suse.cz>
17579 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17580 flags argument of print_gimple_stmt, print_gimple_expr,
17581 print_generic_stmt and print_generic_expr.
17582 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17583 * coretypes.h: Likewise.
17584 * except.c (dump_eh_tree): Likewise.
17585 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17586 * gimple-pretty-print.h: Likewise.
17587 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17588 (backprop::push_to_worklist): Likewise.
17589 (backprop::pop_from_worklist): Likewise.
17590 (backprop::process_use): Likewise.
17591 (backprop::intersect_uses): Likewise.
17592 (note_replacement): Likewise.
17593 * gimple-ssa-store-merging.c
17594 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17595 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17596 (pass_store_merging::execute): Likewise.
17597 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17598 (ssa_base_cand_dump_callback): Likewise.
17599 (dump_incr_vec): Likewise.
17600 (replace_refs): Likewise.
17601 (replace_mult_candidate): Likewise.
17602 (create_add_on_incoming_edge): Likewise.
17603 (create_phi_basis): Likewise.
17604 (insert_initializers): Likewise.
17605 (all_phi_incrs_profitable): Likewise.
17606 (introduce_cast_before_cand): Likewise.
17607 (replace_one_candidate): Likewise.
17608 * gimplify.c (gimplify_expr): Likewise.
17609 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17610 (set_rename): Likewise.
17611 (rename_uses): Likewise.
17612 (copy_loop_phi_nodes): Likewise.
17613 (add_close_phis_to_merge_points): Likewise.
17614 (copy_loop_close_phi_args): Likewise.
17615 (copy_cond_phi_args): Likewise.
17616 (graphite_copy_stmts_from_block): Likewise.
17617 (translate_pending_phi_nodes): Likewise.
17618 * graphite-poly.c (print_pdr): Likewise.
17619 (dump_gbb_cases): Likewise.
17620 (dump_gbb_conditions): Likewise.
17621 (print_scop_params): Likewise.
17622 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17623 (build_cross_bb_scalars_use): Likewise.
17624 (gather_bbs::before_dom_children): Likewise.
17625 * hsa-dump.c (dump_hsa_immed): Likewise.
17626 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17627 (get_replacement_map): Likewise.
17628 * ipa-inline-analysis.c (dump_condition): Likewise.
17629 (estimate_function_body_sizes): Likewise.
17630 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17631 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17632 * ipa-prop.c (ipa_dump_param): Likewise.
17633 (ipa_print_node_jump_functions_for_edge): Likewise.
17634 (ipa_modify_call_arguments): Likewise.
17635 (ipa_modify_expr): Likewise.
17636 (ipa_dump_param_adjustments): Likewise.
17637 (ipa_dump_agg_replacement_values): Likewise.
17638 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17639 * ipa-pure-const.c (check_stmt): Likewise.
17640 (pass_nothrow::execute): Likewise.
17641 * ipa-split.c (execute_split_functions): Likewise.
17642 * omp-offload.c (dump_oacc_loop_part): Likewise.
17643 (dump_oacc_loop): Likewise.
17644 * trans-mem.c (tm_log_emit): Likewise.
17645 (tm_memopt_accumulate_memops): Likewise.
17646 (dump_tm_memopt_set): Likewise.
17647 (dump_tm_memopt_transform): Likewise.
17648 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17649 (print_loop): Likewise.
17650 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17651 (chkp_gather_checks_info): Likewise.
17652 (chkp_get_check_result): Likewise.
17653 (chkp_remove_check_if_pass): Likewise.
17654 (chkp_use_outer_bounds_if_possible): Likewise.
17655 (chkp_reduce_bounds_lifetime): Likewise.
17656 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17657 (chkp_mark_completed_bounds): Likewise.
17658 (chkp_register_incomplete_bounds): Likewise.
17659 (chkp_mark_invalid_bounds): Likewise.
17660 (chkp_maybe_copy_and_register_bounds): Likewise.
17661 (chkp_build_returned_bound): Likewise.
17662 (chkp_get_bound_for_parm): Likewise.
17663 (chkp_build_bndldx): Likewise.
17664 (chkp_get_bounds_by_definition): Likewise.
17665 (chkp_generate_extern_var_bounds): Likewise.
17666 (chkp_get_bounds_for_decl_addr): Likewise.
17667 * tree-chrec.c (chrec_apply): Likewise.
17668 * tree-data-ref.c (dump_data_reference): Likewise.
17669 (dump_subscript): Likewise.
17670 (dump_data_dependence_relation): Likewise.
17671 (analyze_overlapping_iterations): Likewise.
17672 * tree-inline.c (expand_call_inline): Likewise.
17673 (tree_function_versioning): Likewise.
17674 * tree-into-ssa.c (dump_defs_stack): Likewise.
17675 (dump_currdefs): Likewise.
17676 (dump_names_replaced_by): Likewise.
17677 (dump_update_ssa): Likewise.
17678 (update_ssa): Likewise.
17679 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17680 * tree-parloops.c (build_new_reduction): Likewise.
17681 (try_create_reduction_list): Likewise.
17682 (ref_conflicts_with_region): Likewise.
17683 (oacc_entry_exit_ok_1): Likewise.
17684 (oacc_entry_exit_single_gang): Likewise.
17685 * tree-pretty-print.h: Likewise.
17686 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17687 (get_scalar_evolution): Likewise.
17688 (add_to_evolution): Likewise.
17689 (get_loop_exit_condition): Likewise.
17690 (analyze_evolution_in_loop): Likewise.
17691 (analyze_initial_condition): Likewise.
17692 (analyze_scalar_evolution): Likewise.
17693 (instantiate_scev): Likewise.
17694 (number_of_latch_executions): Likewise.
17695 (gather_chrec_stats): Likewise.
17696 (final_value_replacement_loop): Likewise.
17697 (scev_const_prop): Likewise.
17698 * tree-sra.c (dump_access): Likewise.
17699 (disqualify_candidate): Likewise.
17700 (create_access): Likewise.
17701 (reject): Likewise.
17702 (maybe_add_sra_candidate): Likewise.
17703 (create_access_replacement): Likewise.
17704 (analyze_access_subtree): Likewise.
17705 (analyze_all_variable_accesses): Likewise.
17706 (sra_modify_assign): Likewise.
17707 (initialize_constant_pool_replacements): Likewise.
17708 (find_param_candidates): Likewise.
17709 (decide_one_param_reduction): Likewise.
17710 (replace_removed_params_ssa_names): Likewise.
17711 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17712 * tree-ssa-copy.c (dump_copy_of): Likewise.
17713 (copy_prop_visit_cond_stmt): Likewise.
17714 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17715 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17716 (record_equivalences_from_stmt): Likewise.
17717 * tree-ssa-dse.c (compute_trims): Likewise.
17718 (delete_dead_call): Likewise.
17719 (delete_dead_assignment): Likewise.
17720 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17721 (forward_propagate_into_cond): Likewise.
17722 (pass_forwprop::execute): Likewise.
17723 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17724 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17726 (move_computations_worker): Likewise.
17727 (execute_sm): Likewise.
17728 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17729 (remove_exits_and_undefined_stmts): Likewise.
17730 (remove_redundant_iv_tests): Likewise.
17731 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17732 (adjust_iv_update_pos): Likewise.
17733 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17734 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17735 (value_replacement): Likewise.
17736 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17737 * tree-ssa-pre.c (print_pre_expr): Likewise.
17738 (get_representative_for): Likewise.
17739 (create_expression_by_pieces): Likewise.
17740 (insert_into_preds_of_block): Likewise.
17741 (eliminate_insert): Likewise.
17742 (eliminate_dom_walker::before_dom_children): Likewise.
17743 (eliminate): Likewise.
17744 (remove_dead_inserted_code): Likewise.
17745 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17746 * tree-ssa-reassoc.c (get_rank): Likewise.
17747 (eliminate_duplicate_pair): Likewise.
17748 (eliminate_plus_minus_pair): Likewise.
17749 (eliminate_not_pairs): Likewise.
17750 (undistribute_ops_list): Likewise.
17751 (eliminate_redundant_comparison): Likewise.
17752 (update_range_test): Likewise.
17753 (optimize_range_tests_var_bound): Likewise.
17754 (optimize_vec_cond_expr): Likewise.
17755 (rewrite_expr_tree): Likewise.
17756 (rewrite_expr_tree_parallel): Likewise.
17757 (linearize_expr): Likewise.
17758 (break_up_subtract): Likewise.
17759 (linearize_expr_tree): Likewise.
17760 (attempt_builtin_powi): Likewise.
17761 (attempt_builtin_copysign): Likewise.
17762 (transform_stmt_to_copy): Likewise.
17763 (transform_stmt_to_multiply): Likewise.
17764 (dump_ops_vector): Likewise.
17765 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17766 (print_scc): Likewise.
17767 (set_ssa_val_to): Likewise.
17768 (visit_reference_op_store): Likewise.
17769 (visit_use): Likewise.
17770 (sccvn_dom_walker::before_dom_children): Likewise.
17771 (run_scc_vn): Likewise.
17772 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17774 (expr_hash_elt::print): Likewise.
17775 (const_and_copies::pop_to_marker): Likewise.
17776 (const_and_copies::record_const_or_copy_raw): Likewise.
17777 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17778 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17779 (dump_predicates): Likewise.
17780 (find_uninit_use): Likewise.
17781 (warn_uninitialized_phi): Likewise.
17782 (pass_late_warn_uninitialized::execute): Likewise.
17783 * tree-ssa.c (verify_vssa): Likewise.
17784 (verify_ssa): Likewise.
17785 (maybe_optimize_var): Likewise.
17786 * tree-vrp.c (dump_value_range): Likewise.
17787 (dump_all_value_ranges): Likewise.
17788 (dump_asserts_for): Likewise.
17789 (register_edge_assert_for_2): Likewise.
17790 (vrp_visit_cond_stmt): Likewise.
17791 (vrp_visit_switch_stmt): Likewise.
17792 (vrp_visit_stmt): Likewise.
17793 (vrp_visit_phi_node): Likewise.
17794 (simplify_cond_using_ranges_1): Likewise.
17795 (fold_predicate_in): Likewise.
17796 (evrp_dom_walker::before_dom_children): Likewise.
17797 (evrp_dom_walker::push_value_range): Likewise.
17798 (evrp_dom_walker::pop_value_range): Likewise.
17799 (execute_early_vrp): Likewise.
17801 2017-05-16 Richard Biener <rguenther@suse.de>
17803 * dwarf2out.c (loc_list_from_tree_1): Do not create
17804 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17806 2017-05-16 Richard Biener <rguenther@suse.de>
17808 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17810 (note_variable_value_in_expr): If we resolved the decl ref
17811 do not push to the stack.
17813 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17815 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17816 operations in fast-math mode.
17817 (vaddq_f16): Likewise.
17818 (vmul_f16): Likewise.
17819 (vmulq_f16): Likewise.
17820 (vsub_f16): Likewise.
17821 (vsubq_f16): Likewise.
17822 * config/arm/neon.md (add<mode>3): New.
17824 (fma:<VH:mode>3): New. Also remove outdated comment.
17827 2017-05-16 Martin Liska <mliska@suse.cz>
17831 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17832 (odr_types_equivalent_p): Likewise.
17834 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17836 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17838 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17841 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17842 non-interunit SSE move alternatives with '?'.
17843 (zero-extendsidi peephole2): New peephole to skip intermediate
17844 general register in SSE zero-extend sequence.
17846 2017-05-15 Jeff Law <law@redhat.com>
17848 * reorg.c (relax_delay_slots): Create a new variable to hold
17849 the temporary target rather than clobbering TARGET_LABEL.
17851 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17852 missing argument to extract_bit_field call.
17853 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17855 2017-05-15 Martin Liska <mliska@suse.cz>
17858 * gcc.c (process_command): Do not allow empty argument of -o option.
17860 2017-05-15 Renlin Li <renlin.li@arm.com>
17862 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17863 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17864 * config/aarch64/constraints.md (Usf): Add long call check.
17865 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17866 (call_value): Likewise.
17867 (sibcall): Likewise.
17868 (sibcall_value): Likewise.
17870 (call_value_insn): New.
17871 (sibcall_insn): Update rtx pattern.
17872 (sibcall_value_insn): Likewise.
17873 (call_internal): Remove.
17874 (call_value_internal): Likewise.
17875 (sibcall_internal): Likewise.
17876 (sibcall_value_internal): Likewise.
17877 (call_reg): Likewise.
17878 (call_symbol): Likewise.
17879 (call_value_reg): Likewise.
17880 (call_value_symbol): Likewise.
17882 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17885 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17887 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17889 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17890 compatible with CCGOCmode and with CCZmode.
17892 2017-05-14 Martin Sebor <msebor@redhat.com>
17894 PR middle-end/77671
17895 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17896 (gimple_fold_builtin_snprintf): Same.
17897 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17898 (gimple_fold_builtin_snprintf): Same.
17899 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17900 of character types.
17901 (is_call_safe): New function.
17902 (try_substitute_return_value): Call it.
17903 (try_simplify_call): New function.
17904 (pass_sprintf_length::handle_gimple_call): Call it.
17906 2017-05-14 Martin Sebor <msebor@redhat.com>
17908 PR middle-end/80669
17909 * builtins.c (expand_builtin_stpncpy): Simplify.
17911 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17913 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17914 * config/i386/i386.h
17915 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17916 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17917 (struct machine_function): Add new members call_ms2sysv,
17918 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17919 (struct machine_frame_state): New fields sp_realigned and
17920 sp_realigned_offset.
17921 * config/i386/i386.c
17922 (enum xlogue_stub): New enum.
17923 (enum xlogue_stub_sets): New enum.
17924 (class xlogue_layout): New class.
17925 (struct ix86_frame): New fields stack_realign_allocate_offset,
17926 stack_realign_offset and outlined_save_offset. Modify comments to
17927 detail stack layout when using out-of-line stubs.
17928 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17929 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17930 -mcall-ms2sysv-xlogues.
17931 (stub_managed_regs): New static variable.
17932 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17933 registers managed by out-of-line stub.
17934 (disable_call_ms2sysv_xlogues): New function.
17935 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17936 m->call_ms2sysv when appropriate and compute frame layout for
17938 (sp_valid_at, fp_valid_at): New inline functions.
17939 (choose_basereg): New function.
17940 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17942 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17943 Use align parameter of choose_baseaddr to generated aligned SSE movs
17945 (pro_epilogue_adjust_stack): Modify to track
17946 machine_frame_state::sp_realigned.
17947 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17948 (ix86_nsaved_sseregs): Likewise.
17949 (ix86_emit_save_regs): Likewise.
17950 (ix86_emit_save_regs_using_mov): Likewise.
17951 (ix86_emit_save_sse_regs_using_mov): Likewise.
17952 (get_scratch_register_on_entry): Likewise.
17953 (gen_frame_set): New function.
17954 (gen_frame_load): Likewise.
17955 (gen_frame_store): Likewise.
17956 (emit_outlined_ms2sysv_save): Likewise.
17957 (emit_outlined_ms2sysv_restore): Likewise.
17958 (ix86_expand_prologue): Modify stack re-alignment code and call
17959 emit_outlined_ms2sysv_save when appropriate.
17960 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17961 parameter rtx_insn *insn, which allows the function to be used to only
17962 generate the notes.
17963 (ix86_expand_epilogue): Modify validity checks of frame and stack
17964 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17965 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17966 * config/i386/predicates.md
17967 (save_multiple): New predicate.
17968 (restore_multiple): Likewise.
17969 * config/i386/sse.md
17970 (save_multiple<mode>): New pattern.
17971 (save_multiple_realign<mode>): Likewise.
17972 (restore_multiple<mode>): Likewise.
17973 (restore_multiple_and_return<mode>): Likewise.
17974 (restore_multiple_leave_return<mode>): Likewise.
17975 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17977 2017-05-14 Julia Koval <julia.koval@intel.com>
17979 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17980 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17981 (__builtin_ia32_xsetbv): New builtins.
17982 * config/i386/i386.c (ix86_expand_special_args_builtin):
17984 (ix86_expand_builtin): Special expand for new intrinsics.
17985 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17986 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17987 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17989 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17991 * cfganal.c (inverted_post_order_compute): Change argument type
17993 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17994 * df-core.c (rest_of_handle_df_initialize): Adjust.
17995 (rest_of_handle_df_finish): Likewise.
17996 (df_analyze_1): Likewise.
17997 (df_analyze): Likewise.
17998 (loop_inverted_post_order_compute): Change argument to be a vec *.
17999 (df_analyze_loop): Adjust.
18000 (df_get_n_blocks): Likewise.
18001 (df_get_postorder): Likewise.
18002 * df.h (struct df_d): Change field to be a vec.
18003 * lcm.c (compute_laterin): Adjust.
18004 (compute_available): Likewise.
18005 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18006 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18007 * tree-ssa-pre.c (compute_antic): Likewise.
18009 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18011 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18012 (depth_first_search::depth_first_search): Change structure init
18013 function to this constructor.
18014 (depth_first_search::add_bb): Rename function to this member.
18015 (depth_first_search::execute): Likewise.
18016 (flow_dfs_compute_reverse_finish): Adjust.
18018 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18020 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18021 (longest_simple_path): Likewise.
18022 * shrink-wrap.c (spread_components): Likewise.
18023 (disqualify_problematic_components): Likewise.
18024 (emit_common_heads_for_components): Likewise.
18025 (emit_common_tails_for_components): Likewise.
18026 (insert_prologue_epilogue_for_components): Likewise.
18028 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18030 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18033 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18035 * df-core.c (df_set_blocks): Start using auto_bitmap.
18036 (df_compact_blocks): Likewise.
18037 * df-problems.c (df_rd_confluence_n): Likewise.
18038 * df-scan.c (df_insn_rescan_all): Likewise.
18039 (df_process_deferred_rescans): Likewise.
18040 (df_update_entry_block_defs): Likewise.
18041 (df_update_exit_block_uses): Likewise.
18042 (df_entry_block_bitmap_verify): Likewise.
18043 (df_exit_block_bitmap_verify): Likewise.
18044 (df_scan_verify): Likewise.
18045 * lra-constraints.c (lra_constraints): Likewise.
18046 (undo_optional_reloads): Likewise.
18047 (lra_undo_inheritance): Likewise.
18048 * lra-remat.c (calculate_gen_cands): Likewise.
18049 (do_remat): Likewise.
18050 * lra-spills.c (assign_spill_hard_regs): Likewise.
18051 (spill_pseudos): Likewise.
18052 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18053 (bitmap_set_subtract_values): Likewise.
18055 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18057 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18058 management with auto_bitmap.
18059 (fix_inter_tick): Likewise.
18060 (fix_recovery_deps): Likewise.
18061 * ira.c (add_store_equivs): Likewise.
18062 (find_moveable_pseudos): Likewise.
18063 (split_live_ranges_for_shrink_wrap): Likewise.
18064 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18065 (rtx_reuse_manager::seen_def_p): Likewise.
18066 (rtx_reuse_manager::set_seen_def): Likewise.
18067 * print-rtl.h (class rtx_reuse_manager): Likewise.
18069 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18071 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18073 (migrate_btr_def): Likewise.
18074 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18075 * df-core.c (loop_post_order_compute): Likewise.
18076 (loop_inverted_post_order_compute): Likewise.
18077 * hsa-common.h: Likewise.
18078 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18079 * init-regs.c (initialize_uninitialized_regs): Likewise.
18080 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18081 (inline_small_functions): Likewise.
18082 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18083 * ira.c (combine_and_move_insns): Likewise.
18084 (build_insn_chain): Likewise.
18085 * loop-invariant.c (find_invariants): Likewise.
18086 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18087 * predict.c (tree_predict_by_opcode): Likewise.
18088 (predict_paths_leading_to): Likewise.
18089 (predict_paths_leading_to_edge): Likewise.
18090 (estimate_loops_at_level): Likewise.
18091 (estimate_loops): Likewise.
18092 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18093 (spread_components): Likewise.
18094 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18095 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18096 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18097 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18098 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18099 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18100 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18101 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18102 (mark_threaded_blocks): Likewise.
18103 (thread_through_all_blocks): Likewise.
18104 * tree-ssa.c (verify_ssa): Likewise.
18105 (execute_update_addresses_taken): Likewise.
18106 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18110 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18112 (post_order_compute): Likewise.
18113 (inverted_post_order_compute): Likewise.
18114 (pre_and_rev_post_order_compute_fn): Likewise.
18116 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18118 * genrecog.c (int_set::int_set): Explicitly construct our
18119 auto_vec base class.
18120 * vec.h (auto_vec::auto_vec): New constructor.
18122 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18124 * bitmap.h (class auto_bitmap): New constructor taking
18125 bitmap_obstack * argument.
18127 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18129 * bitmap.h (class auto_bitmap): Change type of m_bits to
18130 bitmap_head, and adjust ctor / dtor and member operators.
18132 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18134 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18135 when returned register mode doesn't match original mode.
18137 2017-05-12 Jeff Law <law@redhat.com>
18138 Jakub Jelinek <jakub@redhat.com>
18140 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18141 we look for cc setter after the compare-elim changes.
18142 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18143 within the vector to match what compare-elim now expects.
18144 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18145 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18147 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18148 after the compare-elim changes.
18149 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18150 the vector to match what compare-elim now expects.
18151 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18152 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18153 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18154 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18155 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18157 * config/visium/visium.c (single_set_and_flags): Fix where
18158 we look for cc setter after the compare-elim changes.
18159 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18160 with the vector to match what compare-elim now expects.
18161 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18162 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18163 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18164 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18165 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18166 (neg<mode>2_insn_set_overflow): Likewise.
18168 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18170 PR middle-end/79794
18171 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18172 maybe_expand_insn call, set ops[0].target. If still set after call,
18173 set alt_rtl. Add extra arg to recursive calls.
18174 (extract_bit_field): Add alt_rtl argument. Pass to
18176 * expmed.h (extract_bit_field): Fix prototype.
18177 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18178 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18179 to extract_bit_field_calls.
18180 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18181 Pass alt_rtl to extract_bit_field calls.
18182 * calls.c (store_unaligned_arguments_into_psuedos)
18183 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18184 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18186 * optabs.h (struct expand_operand): Add target bitfield.
18188 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18190 * compare-elim.c (try_eliminate_compare): Canonicalize
18191 operation with embedded compare to
18192 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18193 (set (reg) (operation)].
18195 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18197 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18200 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18201 cost of adding a carry flag for ADC instruction.
18202 [case MINUS]: Ignore the cost of subtracting a carry flag
18203 for SBB instruction.
18205 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18207 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18209 * config/rs6000/bmiintrin.h: New file.
18210 * config/rs6000/bmi2intrin.h: New file.
18211 * config/rs6000/x86intrin.h: New file.
18213 2017-05-12 Jeff Law <law@redhat.com>
18215 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18218 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18220 PR middle-end/80707
18221 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18223 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18225 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18226 early expansion of vector divide builtins.
18227 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18228 builtins identified as having unsigned arguments.
18230 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18232 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18233 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18234 expansion of vector logical operations (and, andc, or, xor,
18237 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18239 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18240 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18242 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18245 early GIMPLE expansion of vector multiplies.
18247 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18249 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18250 TARGET_HAVE_MOVT conditional.
18251 (movt splitter): Likewise.
18253 2017-05-12 Richard Biener <rguenther@suse.de>
18255 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18256 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18257 Fold all stmts not inplace.
18259 2017-05-12 Richard Biener <rguenther@suse.de>
18261 PR tree-optimization/80713
18262 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18263 inserted_exprs bit for not removed stmts.
18265 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18267 PR middle-end/69921
18268 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18269 parallelized" attribute for parallelized OpenACC kernels.
18270 * omp-offload.c (execute_oacc_device_lower): Use it.
18272 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18273 Set "oacc kernels" attribute.
18274 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18275 parameter. Adjust all users.
18276 (oacc_fn_attrib_kernels_p): Remove function.
18277 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18278 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18279 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18280 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18281 assert "oacc kernels" attribute is set.
18283 2017-05-11 Carl Love <cel@us.ibm.com>
18285 * config/rs6000/rs6000-c: Add support for built-in functions
18286 vector unsigned char vec_popcnt (vector signed char)
18287 vector unsigned char vec_popcnt (vector unsigned char)
18288 vector unsigned short vec_popcnt (vector signed short)
18289 vector unsigned short vec_popcnt (vector unsigned short)
18290 vector unsigned int vec_popcnt (vector signed int)
18291 vector unsigned int vec_popcnt (vector unsigned int)
18292 vector unsigned long long vec_popcnt (vector signed long long)
18293 vector unsigned long long vec_popcnt (vector unsigned long long)
18294 vector signed long long vec_slo (vector signed long long,
18295 vector signed char)
18296 vector signed long long vec_slo (vector signed long long,
18297 vector unsigned char)
18298 vector unsigned long long vec_slo (vector unsigned long long,
18299 vector signed char)
18300 vector unsigned long long vec_slo (vector unsigned long long,
18301 vector unsigned char)
18302 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18303 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18304 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18305 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18306 * doc/extend.texi: Update the built-in documentation file for the
18307 new built-in functions.
18309 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18311 * attribs.h (sorted_attr_string): Move machine independent
18312 functions for target clone support from the i386 port to common
18313 code. Rename ix86_function_versions to common_function_versions.
18314 Rename make_name to make_unique_name.
18315 (common_function_versions): Likewise.
18316 (make_unique_name): Likewise.
18317 (make_dispatcher_decl): Likewise.
18318 (is_function_default_version): Likewise.
18319 * attribs.c (attr_strcmp): Likewise.
18320 (sorted_attr_string): Likewise.
18321 (common_function_versions): Likewise.
18322 (make_unique_name): Likewise.
18323 (make_dispatcher_decl): Likewise.
18324 (is_function_default_version): Likewise.
18325 * config/i386/i386.c (attr_strcmp): Likewise.
18326 (sorted_attr_string): Likewise.
18327 (ix86_function_versions): Likewise.
18328 (make_name): Likewise.
18329 (make_dispatcher_decl): Likewise.
18330 (is_function_default_version): Likewise.
18331 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18333 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18336 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18337 Account for direct move costs for vec_construct of integer
18340 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18343 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18344 (UNSPEC_STX_ATOMIC): Ditto.
18345 (loaddi_via_sse): New insn.
18346 (storedi_via_sse): Ditto.
18347 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18348 Update corresponding peephole2 patterns.
18349 (atomic_storedi_fpu): Ditto.
18351 2017-05-11 Julia Koval <julia.koval@intel.com>
18353 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18354 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18356 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18357 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18358 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18360 2017-05-11 Nathan Sidwell <nathan@acm.org>
18362 * graphite-poly.c: Include dumpfile.h.
18364 * dumpfle.h (dump_function): Declare here ...
18365 * tree-dump.h (dump_function): ... not here.
18366 * dumpfile.c: #include tree-cfg.h.
18367 (dump_function): Move here from ...
18368 * tree-dump.c (dump_function): ... here.
18369 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18370 * graphite-poly.c: Don't include tree-dump.h.
18371 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18372 * print-tree.c: Likewise.
18373 * stor-layout.c: Likewise.
18374 * tree-nested.c: Likewise.
18376 * dumpfile.c (dump_start): Use TDF_FLAGS.
18377 (dump_enable_all): Fix TDF_KIND check thinko.
18379 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18381 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18382 array entries to represent two legal parameterizations of the
18383 overloaded __builtin_cmpb function, as represented by the
18384 P6_OV_BUILTIN_CMPB constant.
18385 (altivec_resolve_overloaded_builtin): Add special case handling
18386 for the __builtin_cmpb function, as represented by the
18387 P6_OV_BUILTIN_CMPB constant.
18388 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18389 (BU_P6_64BIT_2): New macro.
18390 (BU_P6_OVERLOAD_2): New macro
18391 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18392 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18393 (CMPB): Add overload support to represent both 32-bit and 64-bit
18394 compare-bytes function.
18395 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18396 support for TARGET_CMPB.
18397 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18398 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18399 documentation of the __builtin_cmpb overloaded built-in function.
18401 2017-05-11 Richard Biener <rguenther@suse.de>
18403 PR tree-optimization/80705
18404 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18405 bases are not vectorizable.
18407 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18409 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18410 when counting register pressure.
18412 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18414 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18415 (iv_ca_more_deps): Renamed to ...
18416 (iv_ca_compare_deps): ... this.
18417 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18419 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18421 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18423 (determine_group_iv_costs): ... here.
18424 (find_inv_vars_cb): Record inv var if it's not recorded before.
18426 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18428 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18429 (get_shiftadd_cost): Ditto.
18431 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18433 * tree-ssa-address.c: Include header file.
18434 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18436 (add_to_parts): Refactor.
18437 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18438 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18443 PR tree-optimization/53090
18444 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18446 (extract_cond_operands): Detect condition with IV on both sides
18447 and return COMP_IV_EXPR_2.
18448 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18449 (rewrite_use_compare): Simplify by removing call to function
18450 extract_cond_operands.
18452 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18454 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18455 (extract_cond_operands): Detect condition comparing against non-
18456 invariant bound and return appropriate enum value.
18457 (find_interesting_uses_cond): Update use of extract_cond_operands.
18458 Handle its return value accordingly.
18459 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18461 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18463 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18464 nonlinear iv_use computation in loop invariant sensitive way.
18466 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18468 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18469 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18471 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18473 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18474 (dump_cand): Support iv_cand.inv_exprs.
18475 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18477 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18482 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18484 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18485 as local function. Include necessary header files.
18486 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18488 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18490 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18494 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18495 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18496 RSHIFT_EXPR and BIT_NOT_EXPR.
18498 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18500 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18501 (adjust_setup_cost): New parameter supporting round up adjustment.
18502 (struct address_cost_data): Delete.
18503 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18504 (split_address_cost, ptr_difference_cost): Delete.
18505 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18506 (struct ainc_cost_data): New struct.
18507 (get_address_cost_ainc): New function.
18508 (get_address_cost, get_computation_cost): Reimplement.
18509 (determine_group_iv_cost_address): Record inv_expr for all uses of
18511 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18512 (iv_ca_has_deps): Reimplemented to ...
18513 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18515 (iv_ca_extend): Call iv_ca_more_deps.
18517 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18519 * tree-ssa-address.c (struct mem_address): Move to header file.
18520 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18521 * tree-ssa-address.h (struct mem_address): Move from C file.
18522 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18524 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18526 * tree-affine.h (aff_combination_type): New interface.
18527 (aff_combination_zero_p): Remove static.
18528 (aff_combination_const_p): New interface.
18529 (aff_combination_singleton_var_p): New interfaces.
18531 2017-05-11 Richard Biener <rguenther@suse.de>
18533 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18534 Skip unreachable blocks and destinations.
18535 (eliminate): Move stmt removal and fixup ...
18536 (fini_eliminate): ... here. Skip inserted exprs.
18537 (pass_pre::execute): Move fini_pre after fini_eliminate.
18538 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18539 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18540 PRE to get rid of dead code that has invalid SSA form and
18541 split critical edges again.
18543 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18545 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18547 2017-05-11 Richard Biener <rguenther@suse.de>
18549 * passes.c (execute_function_todo): Verify loops if they are
18550 said to be up-to-date.
18551 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18552 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18554 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18557 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18558 handle calling assemble_external ourself.
18561 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18562 modes with zero size. Enhance comment.
18564 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18566 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18567 built-ins for vec_xl and vec_xst with short and char pointer
18570 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18572 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18573 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18574 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18575 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18576 (_mm_maskz_min_round_ss): New intrinsics.
18577 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18578 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18579 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18580 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18581 (__builtin_ia32_minss_mask_round): New builtins.
18582 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18583 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18584 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18586 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18587 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18589 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18592 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18594 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18595 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18596 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18597 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18598 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18599 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18600 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18601 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18602 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18603 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18604 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18605 (__builtin_ia32_mulss_mask_round): New builtins.
18606 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18607 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18608 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18610 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18611 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18613 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18616 2017-05-10 Julia Koval <julia.koval@intel.com>
18618 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18619 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18620 (_mm256_setr_m128i): New intrinsics.
18622 2017-05-10 Julia Koval <julia.koval@intel.com>
18624 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18625 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18626 (_mm_maskz_rcp14_ss): New intrinsics.
18627 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18628 (__builtin_ia32_rcp14ss_mask): New builtins.
18629 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18631 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18633 PR tree-optimization/51513
18634 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18635 (assert_unreachable_fallthru_edge_p): Use it.
18636 (group_case_labels_stmt): Likewise.
18637 * tree-cfg.h: Prototype it.
18638 * stmt.c: Include cfghooks.h and tree-cfg.h.
18639 (emit_case_dispatch_table) <gap_label>: New local variable.
18640 Use it to fill dispatch table gaps.
18641 Test for default_label before updating probabilities.
18642 (expand_case) <default_label>: Remove unneeded initialization.
18643 Test for unreachable default case statement and remove its edge.
18644 Set default_label accordingly.
18645 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18647 2017-05-10 Carl Love <cel@us.ibm.com>
18649 * config/rs6000/rs6000-c: Add support for built-in functions
18650 vector signed char vec_neg (vector signed char)
18651 vector signed short int vec_neg (vector short int)
18652 vector signed int vec_neg (vector signed int)
18653 vector signed long long vec_neg (vector signed long long)
18654 vector float vec_neg (vector float)
18655 vector double vec_neg (vector double)
18656 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18658 * config/rs6000/altivec.h: Add define for vec_neg
18659 * doc/extend.texi: Update the built-in documentation for the
18660 new built-in functions.
18662 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18664 PR tree-optimization/77644
18665 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18667 2017-05-10 Nathan Sidwell <nathan@acm.org>
18669 * dumpfile.h (TDI_lang_all): New.
18670 (TDF_KIND): New. Renumber others
18671 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18673 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18675 (get_dump_file_name): Adjust suffix generation.
18676 (dump_enable_all): Use TDF_KIND.
18677 * doc/invoke.texi (-fdump-lang-all): Document.
18679 * dumpfile.h: Tabify.
18681 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18684 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18685 Move member access before delete.
18687 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18689 * tree-inline.c (expand_call_inline): Split block at stmt
18692 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18695 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18696 are now unused after splitting mov{sf,sd}_hardfloat.
18697 (f32_lr2): Likewise.
18698 (f32_lm): Likewise.
18699 (f32_lm2): Likewise.
18700 (f32_li): Likewise.
18701 (f32_li2): Likewise.
18702 (f32_lv): Likewise.
18703 (f32_sr): Likewise.
18704 (f32_sr2): Likewise.
18705 (f32_sm): Likewise.
18706 (f32_sm2): Likewise.
18707 (f32_si): Likewise.
18708 (f32_si2): Likewise.
18709 (f32_sv): Likewise.
18710 (f32_dm): Likewise.
18711 (f32_vsx): Likewise.
18712 (f32_av): Likewise.
18713 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18714 For movsf, order stores so the VSX stores occur before the GPR
18715 store which encourages the register allocator to use a traditional
18716 FPR instead of a GPR. For movsd, order the stores so that the GPR
18717 store comes before the VSX stores to allow the power6 to work.
18718 This is due to the power6 not having a 32-bit integer store
18719 instruction from a FPR.
18720 (movsf_hardfloat): Likewise.
18721 (movsd_hardfloat): Likewise.
18723 2017-05-09 Martin Sebor <msebor@redhat.com>
18725 PR translation/80280
18726 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18729 PR translation/80280
18730 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18731 data member added in r247778.
18732 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18734 2017-05-09 Nathan Sidwell <nathan@acm.org>
18736 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18738 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18741 2017-05-09 Marek Polacek <polacek@redhat.com>
18743 * doc/invoke.texi: Fix typo.
18745 2017-05-09 Richard Biener <rguenther@suse.de>
18747 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18748 (vrp_val_is_min): Likewise.
18749 (set_value_range_to_value): Likewise.
18750 (set_value_range_to_nonnegative): Likewise.
18751 (gimple_assign_nonzero_p): Likewise.
18752 (gimple_stmt_nonzero_p): Likewise.
18753 (vrp_int_const_binop): Likewise. Remove unreachable case.
18754 (adjust_range_with_scev): Adjust comments.
18755 (compare_range_with_value): Likewise.
18756 (extract_range_from_phi_node): Likewise.
18757 (test_for_singularity): Likewise.
18759 2017-05-09 Richard Biener <rguenther@suse.de>
18761 * tree-vrp.c (get_single_symbol): Add assert that we don't
18762 get overflowed constants as invariant part.
18763 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18764 checks. Use wi::cmp instead of recursing for integer constants.
18765 (compare_values): Just ignore whether we assumed undefined
18766 overflow instead of failing the compare.
18767 (extract_range_for_var_from_comparison_expr): Add comment before the
18768 TREE_NO_WARNING sets.
18769 (test_for_singularity): Likewise.
18770 (extract_range_from_comparison): Do not disable optimization
18771 when we assumed undefined overflow.
18772 (extract_range_basic): Remove init of unused var.
18774 2017-05-09 Richard Biener <rguenther@suse.de>
18776 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18777 (extract_range_from_multiplicative_op_1): Adjust.
18778 (extract_range_from_binary_expr_1): Use int_const_binop.
18780 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18783 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18784 rs6000_store_data_bypass_p in seven define_bypass directives and
18785 in several comments.
18786 * config/rs6000/rs6000-protos.h: Add prototype for
18787 rs6000_store_data_bypass_p function.
18788 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18789 function implements slightly different (rs6000-specific) semantics
18790 than store_data_bypass_p, returning false rather than aborting
18791 with assertion error when arguments do not satisfy the
18792 requirements of store data bypass.
18793 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18794 rs6000_store_data_bypass_p.
18796 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18798 * config/xtensa/xtensa-protos.h
18799 (xtensa_initial_elimination_offset): New declaration.
18800 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18801 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18802 macro definition, add case for FRAME_POINTER_REGNUM when
18803 FRAME_GROWS_DOWNWARD.
18804 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18805 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18806 xtensa_initial_elimination_offset.
18808 2017-05-08 Nathan Sidwell <nathan@acm.org>
18810 * doc/invoke.texi: Alphabetize -fdump options.
18812 2017-05-08 Martin Sebor <msebor@redhat.com>
18814 PR translation/80280
18815 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18817 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18819 * target.def (compute_frame_layout): New optional target hook.
18820 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18821 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18822 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18824 * reload1.c (verify_initial_elim_offsets): Likewise.
18825 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18826 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18827 (arm_get_frame_offsets): Split up into this ...
18828 (arm_compute_frame_layout): ... and this function.
18830 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18832 * config/aarch64/constraints.md (Usa): New constraint.
18833 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18835 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18837 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18838 with_multilib_list after it has been checked.
18840 2017-05-08 Richard Biener <rguenther@suse.de>
18842 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18843 (bitmap_set_subtract_values): Likewise.
18845 2017-05-08 Richard Biener <rguenther@suse.de>
18847 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18848 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18850 (gimple_stmt_nonzero_warnv_p): Rename to ...
18851 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18853 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18854 (extract_range_basic): Adjust, do not disable propagation on
18855 strict overflow sensitive simplification.
18856 (vrp_visit_cond_stmt): Likewise.
18858 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18860 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18861 body size unconditionally.
18863 2017-05-07 Jeff Law <law@redhat.com>
18866 2017-05-06 Jeff Law <law@redhat.com>
18867 PR tree-optimization/78496
18868 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18871 PR tree-optimization/78496
18872 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18873 (simplify_stmt_using_ranges): Call it.
18874 (vrp_dom_walker::before_dom_children): Extract equivalences
18875 from an ASSERT_EXPR with an equality comparison against a
18878 2017-05-06 Jeff Law <law@redhat.com>
18880 PR tree-optimization/78496
18881 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18884 PR tree-optimization/78496
18885 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18886 (simplify_stmt_using_ranges): Call it.
18887 (vrp_dom_walker::before_dom_children): Extract equivalences
18888 from an ASSERT_EXPR with an equality comparison against a
18891 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18893 * lra-constraints.c (lra_copy_reg_equiv): New function.
18894 (split_reg): Use it to copy equivalence information from the
18895 original register to the spill register.
18897 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18899 PR rtl-optimization/75964
18900 * simplify-rtx.c (simplify_const_relational_operation): Remove
18901 invalid handling of comparisons of integer ABS.
18903 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18905 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18906 initialize to zero.
18907 (init_regs): Remove declaration.
18908 (function_arg_advance_32): Initialize error_p as boolean variable.
18910 2017-05-05 Nathan Sidwell <nathan@acm.org>
18912 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18913 lines. Use for (;;).
18915 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18917 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18918 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18919 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18920 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18921 VF=2 that require versioning.
18923 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18925 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18928 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18930 * diagnostic.h (diagnostic_override_option_index): Convert from
18931 macro to inline function.
18933 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18935 * diagnostic.c (last_module_changed_p): New function.
18936 (set_last_module): New function.
18937 (diagnostic_report_current_module): Convert macro usage to
18938 the above functions.
18939 * diagnostic.h (diagnostic_context::last_module): Strengthen
18940 from const line_map * to const line_map_ordinary *.
18941 (diagnostic_last_module_changed): Delete macro.
18942 (diagnostic_set_last_module): Delete macro.
18944 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18946 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18947 with diagnostic_report_diagnostic.
18948 (diagnostic_n_impl_richloc): Likewise.
18949 * diagnostic.h (report_diagnostic): Delete macro.
18950 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18951 with diagnostic_report_diagnostic.
18952 * substring-locations.c (format_warning_va): Likewise.
18954 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18956 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18957 save/restor of format_spec. Move option-printing code to...
18958 (print_option_information): ...this new function, and
18959 reimplement by simply printing to the pretty_printer,
18960 rather than appending to the format string.
18962 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18964 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18965 handling logic into...
18966 (update_effective_level_from_pragmas): ...this new function.
18968 2017-05-04 Andrew Waterman <andrew@sifive.com>
18970 * config/riscv/riscv.opt (mstrict-align): New option.
18971 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18972 (SLOW_UNALIGNED_ACCESS): Define.
18973 (riscv_slow_unaligned_access): Declare.
18974 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18976 (riscv_slow_unaligned_access): New variable.
18977 (rocket_tune_info): Set slow_unaligned_access to true.
18978 (optimize_size_tune_info): Set slow_unaligned_access to false.
18979 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18980 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18981 (riscv_option_override): Set riscv_slow_unaligned_access.
18982 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18984 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18986 * config/riscv/riscv.md: Unify indentation.
18988 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18993 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18995 (extendsi<mode>2): Add support for doing sign extension via
18996 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18997 don't have ISA 3.0 instructions.
18998 (extendsi<mode>2 splitter): Likewise.
18999 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19000 generate the normal insns since SImode can now go in vector
19001 registers. Disallow the special UNSPECs needed for previous
19002 machines to hide SImode being used. Add new insns
19003 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19004 (fix_trunc<mode>si2_stfiwx): Likewise.
19005 (fix_trunc<mode>si2_internal): Likewise.
19006 (fixuns_trunc<mode>si2): Likewise.
19007 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19008 (fctiw<u>z_<mode>_smallint): Likewise.
19009 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19010 of floating point to 32-bit integer from doing a direct move to
19011 the GPR registers to do a store.
19012 (fctiwz_<mode>): Break long line.
19014 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19016 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19017 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19018 (addr_list, addr_offset_valid_p): New.
19019 (split_address_groups): Check offset validity with above function.
19020 (gt-tree-ssa-loop-ivopts.h): Include header file.
19022 2017-05-05 Nathan Sidwell <nathan@acm.org>
19024 * config.gcc (arm*-*-*): Add missing 'fi'.
19026 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19028 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19029 included in -fopt-info does not matter.
19030 * doc/optinfo.texi (-fopt-info): Fix description of default
19031 behavour. Explicitly say order of options included in -fopt-info
19034 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19036 * config.gcc: Allow combinations of aprofile and rmprofile values for
19037 --with-multilib-list.
19038 * config/arm/t-multilib: New file.
19039 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19040 variables. Remove setting of ISA and floating-point ABI in
19041 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19042 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19043 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19044 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19046 * config/arm/t-rmprofile: Likewise except for the matches changes.
19047 * doc/install.texi (--with-multilib-list): Document the combination of
19048 aprofile and rmprofile values and warn about pitfalls in doing that.
19050 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19052 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19053 (movdi_aarch64): Likewise.
19055 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19057 PR tree-optimization/80632
19058 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19060 (build_arrays): Initialize it for virtual phis.
19061 (fix_phi_nodes): Use it for virtual phis.
19063 PR tree-optimization/80558
19064 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19065 [x, y] op z into [x op, y op z] for op & or | if conditions
19068 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19069 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19072 * config/arm/arm.md (use_literal_pool): Remove.
19073 (64-bit immediate split): No longer takes cost into consideration
19074 if arm_disable_literal_pool is enabled.
19075 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19076 used when arm_disable_literal_pool is enabled.
19077 (arm_max_const_double_inline_cost): Remove use of
19078 arm_disable_literal_pool.
19079 (push_minipool_fix): Add assert.
19080 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19081 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19082 (no_literal_pool_sf_immediate): New.
19084 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19086 PR tree-optimization/80613
19087 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19088 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19090 2017-05-05 Richard Biener <rguenther@suse.de>
19092 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19094 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19096 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19097 of this flag from insn conditions due to removal from r247495.
19099 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19101 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19103 (arm_early_store_addr_dep_ptr): Likewise.
19104 * config/arm/aarch-common-protos.h
19105 (arm_early_load_addr_dep_ptr): Add prototype.
19106 (arm_early_store_addr_dep_ptr): Likewise.
19107 * config/arm/cortex-a53.md: Add new bypasses.
19109 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19111 * tree.c (next_type_uid): Change type to unsigned.
19112 (type_hash_canon): Decrement back next_type_uid if
19113 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19114 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19117 2017-05-04 Martin Sebor <msebor@redhat.com>
19119 * builtins.c: Fix a trivial typo in a comment.
19121 PR middle-end/79234
19122 * builtins.c (check_sizes): Adjust to handle reading past the end.
19123 Avoid printing excessive upper bound of ranges. Use %E to print
19124 tree nodes instead of converting them to %wu.
19125 (expand_builtin_memchr): New function.
19126 (compute_dest_size): Rename...
19127 (compute_objsize): ...to this.
19128 (expand_builtin_memcpy): Adjust.
19129 (expand_builtin_mempcpy): Adjust.
19130 (expand_builtin_strcat): Adjust.
19131 (expand_builtin_strcpy): Adjust.
19132 (check_strncat_sizes): Adjust.
19133 (expand_builtin_strncat): Adjust.
19134 (expand_builtin_strncpy): Adjust and simplify.
19135 (expand_builtin_memset): Adjust.
19136 (expand_builtin_bzero): Adjust.
19137 (expand_builtin_memcmp): Adjust.
19138 (expand_builtin): Handle memcmp.
19139 (maybe_emit_chk_warning): Check strncat just once.
19141 2017-05-04 Martin Sebor <msebor@redhat.com>
19143 PR preprocessor/79214
19144 PR middle-end/79222
19145 PR middle-end/79223
19146 * builtins.c (check_sizes): Add inlining context and issue
19147 warnings even when -Wno-system-headers is set.
19148 (check_strncat_sizes): Same.
19149 (expand_builtin_strncat): Same.
19150 (expand_builtin_memmove): New function.
19151 (expand_builtin_stpncpy): Same.
19152 (expand_builtin): Handle memmove and stpncpy.
19154 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19156 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19157 which is not used any more.
19159 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19161 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19163 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19165 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19166 (cortexa53_tunings): Likewise.
19167 (cortexa57_tunings): Likewise.
19168 (cortexa72_tunings): Likewise.
19169 (cortexa73_tunings): Likewise.
19171 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19173 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19174 Set loop alignment to 8.
19176 2017-05-04 Martin Sebor <msebor@redhat.com>
19178 PR translation/80280
19179 * builtins.c (expand_builtin_object_size): Add missing quoting to
19180 %D and like directives.
19181 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19182 (hsa_type_for_tree_type): Same.
19183 (verify_function_arguments): Same.
19184 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19185 * varasm.c (get_section): Same.
19188 2017-05-04 Martin Sebor <msebor@redhat.com>
19190 PR translation/80280
19191 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19193 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19195 * config/aarch64/aarch64.c (generic_addrcost_table):
19196 Change HI/TI mode setting.
19198 2017-05-04 Martin Jambor <mjambor@suse.cz>
19200 PR tree-optimization/80622
19201 * tree-sra.c (comes_initialized_p): New function.
19202 (build_accesses_from_assign): Only set write lazily when
19203 comes_initialized_p is false.
19204 (analyze_access_subtree): Use comes_initialized_p.
19205 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19206 instead of testing for PARM_DECL.
19208 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19210 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19211 constraint on operand 0 to allow more general addressing modes.
19212 Adjust output template.
19213 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19215 * config/aarch64/aarch64-protos.h
19216 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19217 * config/aarch64/constraints.md (Dp): New address constraint.
19218 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19221 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19223 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19224 update use of estimate_ipcp_clone_size_and_time.
19225 (estimate_local_effects): Update use of
19226 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19227 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19228 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19229 Return nonspecialized time.
19231 2017-05-04 Richard Biener <rguenther@suse.de>
19233 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19234 for the last VUSE which def dominates the PHI. Directly call
19236 (get_continuation_for_phi_1): Remove.
19238 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19240 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19241 to explain the use of truncating division. Cap the number of
19242 iterations to the maximum given by nb_iterations_upper_bound,
19245 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19247 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19248 * configure: Regenerate.
19249 * config.in: Regenerate.
19250 * config/i386/driver-mingw32.c: new file.
19251 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19252 * config.host: Link driver-mingw32.o on MinGW host.
19253 * doc/install.texi: Document new --enable-mingw-wildcard configure
19256 2017-05-04 Marek Polacek <polacek@redhat.com>
19258 PR tree-optimization/80612
19259 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19261 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19262 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19264 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19265 (movt splitter): Likewise.
19266 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19267 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19268 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19269 block for Thumb-1 with MOVT.
19270 (thumb2_legitimate_address_p): Move code block ...
19271 (can_avoid_literal_pool_for_label_p): ... into this new function.
19272 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19274 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19275 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19276 "M-profile targets with the MOVT instruction".
19278 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19280 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19281 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19282 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19284 2017-05-04 Martin Liska <mliska@suse.cz>
19286 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19287 variable cond_code.
19289 2017-05-04 Richard Biener <rguenther@suse.de>
19291 * tree.c (array_at_struct_end_p): Handle arrays at struct
19292 end with flexarrays more conservatively. Refactor and treat
19293 arrays of arrays or aggregates more strict. Fix
19294 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19295 * tree.c (array_at_struct_end_p): Adjust prototype.
19296 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19297 * gimple-fold.c (get_range_strlen): Likewise.
19298 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19300 2017-05-04 Richard Biener <rguenther@suse.de>
19302 PR tree-optimization/31130
19303 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19305 (supports_overflow_infinity): Likewise.
19306 (is_negative_overflow_infinity): Likewise.
19307 (is_positive_overflow_infinity): Likewise.
19308 (is_overflow_infinity): Likewise.
19309 (stmt_overflow_infinity): Likewise.
19310 (overflow_infinity_range_p): Likewise.
19311 (usable_range_p): Remove as always returning true.
19312 (make_overflow_infinity): Remove.
19313 (negative_overflow_infinity): Likewise.
19314 (positive_overflow_infinity): Likewise.
19315 (avoid_overflow_infinity): Likewise.
19316 (set_value_range): Adjust accordingly.
19317 (set_value_range_to_nonnegative): Likewise, remove now unused
19318 overflow_infinity arg.
19319 (vrp_operand_equal_p): Adjust.
19320 (update_value_range): Likewise.
19321 (range_int_cst_singleton_p): Likewise.
19322 (operand_less_p): Likewise.
19323 (compare_values_warnv): Likewise.
19324 (extract_range_for_var_from_comparison_expr): Likewise.
19325 (vrp_int_const_binop): Likewise.
19326 (zero_nonzero_bits_from_vr): Likewise.
19327 (extract_range_from_multiplicative_op_1): Likewise.
19328 (extract_range_from_binary_expr_1): Likewise.
19329 (extract_range_from_unary_expr): Likewise.
19330 (extract_range_from_comparison): Likewise.
19331 (extract_range_basic): Likewise.
19332 (adjust_range_with_scev): Likewise.
19333 (compare_ranges): Likewise.
19334 (compare_range_with_value): Likewise.
19335 (dump_value_range): Likewise.
19336 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19338 (simplify_cond_using_ranges): Adjust.
19340 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19342 * brig-builtins.def: Added a builtin for class_f64.
19343 * builtin-types.def: Added a builtin type needed by class_f64.
19345 2017-05-03 Jason Merrill <jason@redhat.com>
19347 * timevar.def: Add TV_CONSTEXPR.
19349 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19351 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19353 2017-05-03 Martin Jambor <mjambor@suse.cz>
19355 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19356 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19357 * ipa-cp.c (ipcp_propagate_stage): Do not call
19358 ipa_update_after_lto_read.
19359 * ipa-inline.c (ipa_inline): Likewise.
19361 2017-05-03 Martin Jambor <mjambor@suse.cz>
19363 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19364 tag. Added a default constructor and a destructor.
19365 (ipa_edge_args_sum_t): New class;
19366 (ipa_edge_args_sum): Declare.
19367 (ipa_edge_args_vector): Remove declaration.
19368 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19369 (ipa_free_edge_args_substructures): Remove declaration.
19370 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19371 (ipa_edge_args_info_available_for_edge_p): Likewise.
19372 * ipa-prop.c (ipa_edge_args_vector): Removed.
19373 (edge_removal_hook_holder): Likewise.
19374 (edge_duplication_hook_holder): Likewise.
19375 (ipa_edge_args_sum): New variable.
19376 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19377 ipa_edge_args_vector.
19378 (ipa_free_edge_args_substructures): Likewise.
19379 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19380 ipa_edge_args_vector.
19381 (ipa_edge_removal_hook): Turned into method
19382 ipa_edge_args_sum_t::remove.
19383 (ipa_edge_duplication_hook): Turned into method
19384 ipa_edge_args_sum_t::duplicate.
19385 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19386 registering edge hooks.
19387 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19388 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19389 ipa_edge_args_sum instead of ipa_edge_args_vector.
19390 * ipa-profile.c (ipa_profile): Likewise.
19392 2017-05-03 Martin Jambor <mjambor@suse.cz>
19394 * symbol-summary.h (function_summary): New method exists.
19395 (function_summary::symtab_removal): Deallocate through release.
19396 (call_summary): New class.
19397 (gt_ggc_mx): New overload.
19398 (gt_pch_nx): Likewise.
19399 (gt_pch_nx): Likewise.
19401 2017-05-03 Jeff Law <law@redhat.com>
19403 PR tree-optimization/78496
19404 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19405 from simplify_cond_using_ranges. Split off code to walk
19406 backwards through casts into ...
19407 (simplify_cond_using_ranges_2): New function.
19408 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19409 (execute_vrp): After identifying jump threads, call
19410 simplify_cond_using_ranges_2.
19412 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19415 * ipa-inline.h (inline_summary): Add ctor.
19416 (create_ggc): Do not use ggc_cleared_alloc.
19418 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19419 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19421 * gcc.c (handle_braces): Support escaping in switch matching
19423 * doc/invoke.texi (Spec Files): Document it.
19424 Remove superfluous @code markup in items.
19426 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19428 * diagnostic-show-locus.c (struct column_range): New struct.
19429 (get_affected_columns): New function.
19430 (get_printed_columns): New function.
19431 (struct correction): New struct.
19432 (correction::ensure_capacity): New function.
19433 (correction::ensure_terminated): New function.
19434 (struct line_corrections): New struct.
19435 (line_corrections::~line_corrections): New dtor.
19436 (line_corrections::add_hint): New function.
19437 (layout::print_trailing_fixits): Reimplement in terms of the new
19439 (selftest::test_overlapped_fixit_printing): New function.
19440 (selftest::diagnostic_show_locus_c_tests): Call it.
19442 2017-05-03 Nathan Sidwell <nathan@acm.org>
19444 Canonicalize canonical type hashing
19445 * tree.h (type_hash_canon_hash): Declare.
19446 * tree.c (type_hash_list, attribute_hash_list): Move into
19447 type_hash_canon_hash.
19448 (build_type_attribute_qual_variant): Break out hash code calc into
19449 type_hash_canon_hash.
19450 (type_hash_canon_hash): New. Generic type hash computation.
19451 (build_range_type_1, build_array_type_1, build_function_type,
19452 build_method_type_directly, build_offset_type, build_complex_type,
19453 make_vector_type): Call it.
19455 2017-05-03 Richard Biener <rguenther@suse.de>
19457 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19458 When all DRs have unknown misaligned do not always peel
19459 when there is a store but apply the same costing model as if
19460 there were only loads.
19462 2017-05-03 Richard Biener <rguenther@suse.de>
19465 PR tree-optimization/80492
19466 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19467 compare_base_decls returning dont-know properly.
19469 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19471 * config/arm/iterators.md (CCSI): New mode iterator.
19472 (arch): New mode attribute.
19473 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19474 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19475 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19476 code iterator for success result mode.
19477 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19478 the corresponding new insn generators.
19480 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19483 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19484 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19486 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19488 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19489 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19490 (DDR_A): Wrap DDR argument in brackets.
19491 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19492 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19493 (DDR_REVERSED_P): Likewise.
19495 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19497 PR tree-optimization/79472
19498 * tree-switch-conversion.c (struct switch_conv_info): Add
19499 contiguous_range and default_case_nonstandard fields.
19500 (collect_switch_conv_info): Compute contiguous_range and
19501 default_case_nonstandard fields, don't clear final_bb if
19502 contiguous_range and only the default case doesn't have the required
19504 (check_all_empty_except_final): Set default_case_nonstandard instead
19505 of failing if contiguous_range and the default case doesn't have empty
19507 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19508 and only the default case doesn't have the required constants. Skip
19510 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19511 if default_case_nonstandard.
19512 (build_constructors): Build constant 1 just once. Assert that default
19513 values aren't inserted in between cases if contiguous_range. Skip
19515 (build_arrays): Skip virtual phis.
19516 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19517 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19518 Handle virtual phis.
19519 (gen_inbound_check): Handle default_case_nonstandard case.
19520 (process_switch): Adjust check_final_bb caller. Call
19521 gather_default_values with the first non-default case instead of
19522 default case if default_case_nonstandard.
19524 2017-05-02 Nathan Sidwell <nathan@acm.org>
19526 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19527 check. Fix formatting.
19529 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19531 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19532 errors when comparing specialized and unspecialized times.
19534 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19536 * diagnostic-show-locus.c
19537 (layout::should_print_annotation_line_p): Make private.
19538 (layout::print_annotation_line): Make private.
19539 (layout::annotation_line_showed_range_p): Make private.
19540 (layout::show_ruler): Make private.
19541 (layout::print_source_line): Make private. Pass in line and
19542 line_width, rather than calling location_get_source_line. Drop
19544 (layout::print_leading_fixits): New method.
19545 (layout::print_any_fixits): Rename to...
19546 (layout::print_trailing_fixits): ...this, and make private.
19547 Don't print newline fixits.
19548 (diagnostic_show_locus): Move logic for printing one row into...
19549 (layout::print_line): ...this new function. Move the
19550 location_get_source_line call and error-handling from
19551 print_source_line to here. Call print_leading_fixits, and rename
19552 print_any_fixits to print_trailing_fixits.
19553 (selftest::test_fixit_insert_containing_newline): Update now that
19554 newlines are partially supported.
19555 (selftest::test_fixit_insert_containing_newline_2): New test.
19556 (selftest::test_fixit_replace_containing_newline): Update comments.
19557 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19558 * edit-context.c (class added_line): New class.
19559 (class edited_line): Describe newline handling in comment.
19560 (edited_line::actually_edited_p): New method.
19561 (edited_line::print_content): Delete redundant decl.
19562 (edited_line::m_predecessors): New field.
19563 (edited_file::print_content): Call edited_line::print_content.
19564 (edited_file::print_diff): Update to support newlines.
19565 (edited_file::print_diff_hunk): Likewise.
19566 (edited_file::print_run_of_changed_lines): New function.
19567 (edited_file::print_diff_line): Convert to...
19568 (print_diff_line): ...this.
19569 (edited_file::get_effective_line_count): New function.
19570 (edited_line::edited_line): Initialize new field m_predecessors.
19571 (edited_line::~edited_line): Clean up m_predecessors.
19572 (edited_line::apply_fixit): Handle newlines.
19573 (edited_line::get_effective_line_count): New function.
19574 (edited_line::print_content): New function.
19575 (edited_line::print_diff_lines): New function.
19576 (selftest::test_applying_fixits_insert_containing_newline): New
19578 (selftest::test_applying_fixits_replace_containing_newline): New
19580 (selftest::insert_line): New function.
19581 (selftest::test_applying_fixits_multiple_lines): Add example of
19583 (selftest::edit_context_c_tests): Call the new tests.
19585 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19587 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19588 parameter cand. Update dump information.
19589 (get_computation_cost): Update uses.
19591 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19593 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19594 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19595 (get_computation_at, rewrite_use_address): Update use of
19596 get_computation_aff.
19598 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19600 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19601 (get_computation): Delete.
19602 (get_computation_cost): Implement like get_computation_cost_at.
19603 Use get_computation_at.
19604 (get_computation_cost_at): Delete.
19605 (rewrite_use_nonlinear_expr): Use get_computation_at.
19606 (rewrite_use_compare, remove_unused_ivs): Ditto.
19608 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19610 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19612 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19614 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19615 (ivopts_global_cost_for_size): Rename parameter and update uses.
19616 (iv_ca_recount_cost): Update uses.
19617 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19618 candidates seperately in n_invs and n_cands.
19619 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19621 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19623 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19624 (find_inv_vars_cb): New.
19625 (find_depends): Renamed to ...
19626 (find_inv_vars): ... this.
19627 (add_candidate_1, force_var_cost): Call find_inv_vars.
19628 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19630 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19632 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19633 inv_vars. Add inv_exprs.
19634 (struct iv_cand): Rename depends_on to inv_vars.
19635 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19636 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19637 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19638 (dump_cand): Dump inv_vars.
19639 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19640 (record_invariant, find_depends, add_candidate_1): Ditto.
19641 (set_group_iv_cost, force_var_cost): Ditto.
19642 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19643 (get_computation_cost_at, get_computation_cost): Ditto.
19644 (determine_group_iv_cost_generic): Ditto.
19645 (determine_group_iv_cost_address): Ditto.
19646 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19647 (determine_group_iv_costs): Ditto.
19648 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19649 (iv_ca_set_remove_invariants): Renamed to ...
19650 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19651 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19652 (iv_ca_set_add_invariants): Renamed to ...
19653 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19654 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19655 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19656 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19657 (create_new_ivs): Remove useless dump.
19659 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19661 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19663 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19664 (iv_ca_set_no_cp, create_new_iv): Ditto.
19666 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19668 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19670 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19672 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19673 function tree_check2.
19675 2017-05-02 Martin Liska <mliska@suse.cz>
19677 * doc/gcov.texi: Add missing preposition.
19678 * gcov.c (function_info::function_info): Properly fill up
19679 all member variables.
19681 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19683 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19685 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19687 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19689 2017-05-02 Martin Liska <mliska@suse.cz>
19692 * lto-streamer-in.c (lto_read_tree_1): Remove
19693 LTO_STREAMER_DEBUG.
19694 * lto-streamer.c (struct tree_hash_entry): Likewise.
19695 (struct tree_entry_hasher): Likewise.
19696 (tree_entry_hasher::hash): Likewise.
19697 (tree_entry_hasher::equal): Likewise.
19698 (lto_streamer_init): Likewise.
19699 (lto_orig_address_map): Likewise.
19700 (lto_orig_address_get): Likewise.
19701 (lto_orig_address_remove): Likewise.
19702 * lto-streamer.h: Likewise.
19703 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19704 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19706 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19708 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19709 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19710 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19711 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19712 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19713 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19714 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19715 (mm_maskz_sub_ss): New intrinsics.
19716 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19717 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19718 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19719 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19720 (__builtin_ia32_subss_mask_round): New builtins.
19721 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19722 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19723 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19725 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19726 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19728 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19731 2017-05-02 Martin Jambor <mjambor@suse.cz>
19733 PR tree-optimization/78687
19734 * tree-sra.c (access): New field parent.
19735 (process_subtree_disqualification): New function.
19736 (disqualify_candidate): Call it.
19737 (build_accesses_from_assign): Reset write flag if creating an
19739 (build_access_subtree): Fill in parent field and also prpagate
19740 down grp_write flag.
19741 (create_artificial_child_access): New parameter set_grp_write, set
19742 grp_write to its value.
19743 (propagate_subaccesses_across_link): Also propagate grp_write flag
19745 (propagate_all_subaccesses): Push the closest parent back to work
19746 queue if add_access_to_work_queue returned true.
19748 2017-05-02 Richard Biener <rguenther@suse.de>
19750 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19751 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19752 -fstrict-overflow documentation.
19753 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19754 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19755 flag_strict_overflow.
19756 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19757 * lto-opts.c (lto_write_options): Do not stream it.
19758 * lto-wrapper.c (merge_and_complain): Do not handle it.
19759 * opts.c (default_options_table): Do not set -fstrict-overflow.
19760 (finish_options): Likewise do not clear it when sanitizing.
19761 * simplify-rtx.c (simplify_const_relational_operation): Do not
19762 test flag_strict_overflow.
19764 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19766 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19767 using enabled attribute.
19768 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19769 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19770 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19771 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19772 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19773 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19774 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19775 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19776 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19777 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19779 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19781 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19783 2017-05-02 Richard Biener <rguenther@suse.de>
19785 PR tree-optimization/80591
19787 2017-04-10 Richard Biener <rguenther@suse.de>
19789 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19792 2017-05-02 Richard Biener <rguenther@suse.de>
19794 PR tree-optimization/80549
19795 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19796 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19797 headers if they do not have a preheader.
19799 2017-05-02 Martin Liska <mliska@suse.cz>
19802 * common.opt: Fix typo.
19803 * doc/invoke.texi: Likewise.
19805 2017-05-01 Jan Beulich <jbeulich@suse.com>
19807 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19808 swapping, add (x,x,m,x,n) alternative.
19810 2017-05-01 Nathan Sidwell <nathan@acm.org>
19812 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19813 unnecessary unadjusted_alignment check.
19815 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19818 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19820 * gimplify.c (gimplify_cilk_detach): New function.
19821 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19822 * tree-core.h: Document EXPR_CILK_SPAWN.
19823 * tree.h (EXPR_CILK_SPAWN): Define.
19825 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19827 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19828 to use new fixit_hint representation, using the "replace" logic.
19829 (get_line_span_for_fixit_hint): Likewise.
19830 (layout::print_any_fixits): Likewise.
19831 (selftest::test_one_liner_many_fixits): Rename to...
19832 (selftest::test_one_liner_many_fixits_1): ...this, and update
19833 comment and expected output to reflect that the multiple fix-it
19834 hints are now consolidated into one insertion.
19835 (selftest::test_one_liner_many_fixits_2): New test.
19836 (selftest::test_diagnostic_show_locus_one_liner): Update for
19838 (selftest::test_fixit_consolidation): Update for fix-it API
19840 * diagnostic.c (print_parseable_fixits): Likewise.
19841 * edit-context.c (edited_line::m_line_events): Convert from
19842 auto_vec <line_event *> to auto_vec <line_event>.
19843 (class line_event): Convert from abstract base class to a concrete
19844 class, taking over the role of replace_event.
19845 (class insert_event): Delete.
19846 (class replace_event): Rename to class line_event. Convert to
19848 (edit_context::add_fixits): Reimplement.
19849 (edit_context::apply_insert): Delete.
19850 (edit_context::apply_replace): Rename to...
19851 (edit_context::apply_fixit): ...this. Convert to half-open range.
19852 (edited_file::apply_insert): Delete.
19853 (edited_file::apply_replace): Rename to...
19854 (edited_file::apply_fixit): ...this.
19855 (edited_line::~edited_line): Drop deletion of events.
19856 (edited_line::apply_insert): Delete.
19857 (edited_line::apply_replace): Rename to...
19858 (edited_line::apply_fixit): ...this. Convert to half-open range.
19859 Update for change to type of m_line_events.
19860 * edit-context.h (edit_context::apply_insert): Delete.
19861 (edit_context::apply_replace): Rename to...
19862 (edit_context::apply_fixit): ...this.
19864 2017-05-01 Martin Sebor <msebor@redhat.com>
19866 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19869 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19872 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19873 __get_cpuid_max returns 0.
19874 (__get_cpuid_count): Ditto.
19876 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19878 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19879 replacement expression is another instance of one of its arguments.
19881 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19884 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19885 check for stack push/pop autoinc.
19886 * config/i386/i386.c (ix86_agi_dependent): Return false
19887 if the only reason why modified_in_p returned true is that
19888 addr is SP based and set_insn is a push or pop.
19890 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19892 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19895 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19898 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19899 (account_size_time): Use two predicates - exec_pred and
19901 (evaluate_conditions_for_known_args): Compute both clause and
19903 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19904 (inline_summary_t::duplicate): Update.
19905 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19907 (compute_inline_parameters): Likewise.
19908 (estimate_edge_size_and_time): Update caluclation of time.
19909 (estimate_node_size_and_time): Compute both time and nonspecialized
19911 (estimate_ipcp_clone_size_and_time): Update.
19912 (inline_merge_summary): Update.
19913 (do_estimate_edge_time): Update.
19914 (do_estimate_edge_size): Update.
19915 (do_estimate_edge_hints): Update.
19916 (inline_read_section, inline_write_summary): Stream both new predicates.
19917 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19919 (compute_inlined_call_time): Cleanup.
19920 (big_speedup_p): Update.
19921 (edge_badness): Update.
19922 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19923 (size_time_entry): Replace predicate by exec_predicate and
19924 nonconst_predicate.
19925 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19926 (estimate_edge_time): Return also nonspec_time.
19927 (reset_edge_growth_cache): Update.
19929 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19931 PR rtl-optimization/80491
19932 * ifcvt.c (noce_process_if_block): When looking for x setter
19933 with missing else_bb, don't check only the insn right before
19934 cond_earliest, but look for the last insn that x is modified in
19935 within the same bb.
19937 PR rtl-optimization/80491
19938 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19940 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19942 PR tree-optimization/80487
19943 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19945 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19947 PR tree-optimization/79697
19948 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19949 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19950 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19952 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19953 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19955 2017-04-28 Martin Sebor <msebor@redhat.com>
19957 PR tree-optimization/80523
19958 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19959 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19961 (maybe_warn, format_directive, parse_directive): Use new functions.
19962 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19964 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19966 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19968 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19970 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19971 target_header_dir): Set correctly.
19972 * configure: Regenerated.
19973 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19974 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19975 instead of SYSTEM_HEADER_DIR.
19977 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19979 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19980 (estimate_local_effects): Likewise.
19981 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19982 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19983 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19984 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19985 do_estimate_edge_time, estimate_edge_time): Likewise.
19986 * ipa-inline-analysis.c (estimate_node_size_and_time,
19987 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19988 (estimate_time_after_inlining): Remove.
19990 2017-04-28 Martin Liska <mliska@suse.cz>
19992 * doc/gcov.texi: Enhance documentation of gcov.
19994 2017-04-28 Martin Liska <mliska@suse.cz>
19996 * doc/gcov.texi: Sort options in alphabetic order.
19997 * doc/gcov-dump.texi: Likewise.
19998 * doc/gcov-tool.texi: Likewise.
19999 * gcov.c (print_usage): Likewise.
20000 * gcov-dump.c (print_usage): Likewise.
20001 * gcov-tool.c (print_merge_usage_message): Likewise.
20002 (print_rewrite_usage_message): Likewise.
20003 (print_overlap_usage_message): Likewise.
20005 2017-04-28 Martin Liska <mliska@suse.cz>
20007 PR gcov-profile/53915
20008 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20010 2017-04-28 Martin Liska <mliska@suse.cz>
20012 PR gcov-profile/79891
20013 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20014 is marked by compiler as living on a line.
20015 (get_cycles_count): Remove usage of the union.
20016 (output_intermediate_file): Likewise.
20017 (find_source): Fix GNU coding style.
20018 (accumulate_line_counts): Remove old non-all block mode.
20019 (output_lines): Remove usage of the union.
20020 * profile.c (output_location): Include all BBs, even if
20021 belonging to a same line (and file) as a previous BB.
20023 2017-04-28 Martin Liska <mliska@suse.cz>
20025 * gcov.c (process_args): Handle new argument 'w'.
20026 (read_graph_file): Assign ID to BBs.
20027 (output_branch_count): Display BB # if verbose flag is set.
20028 (output_lines): Likewise for arcs.
20029 (print_usage): Add '--verbose' option help.
20030 * doc/gcov.texi: Document --verbose (-w) option.
20032 2017-04-28 Martin Liska <mliska@suse.cz>
20034 * gcov.c (struct block_location_info): New struct.
20035 (process_file): Fill up the new structure.
20036 (read_graph_file): Replace usage of encoding by the newly added
20038 (add_line_counts): Likewise.
20039 (accumulate_line_counts): Remove usage of the union.
20040 (function_info::function_info): New function.
20041 (function_info::~function_info): Likewise.
20042 (process_file): Call delete instead of release_function.
20043 (release_function): Release the function.
20044 (release_structures): Call delete instead of release_function.
20045 (solve_flow_graph): Replace usage of num_blocks.
20046 (find_exception_blocks): Likewise.
20047 (output_lines): Fix GNU coding style.
20049 2017-04-28 Martin Liska <mliska@suse.cz>
20052 * coverage.c (coverage_remove_note_file): New function.
20053 * coverage.h: Declare the function.
20054 * toplev.c (finalize): Clean if an error has been seen.
20056 2017-04-28 Martin Liska <mliska@suse.cz>
20058 PR gcov-profile/80031
20059 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20060 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20061 * gcov.c (read_graph_file): Read just number of blocks.
20062 * profile.c (branch_prob): Do not stream 0 flags per a basic
20065 2017-04-28 Martin Liska <mliska@suse.cz>
20067 * gcov-dump.c (tag_*): Add new argument to declarations.
20068 (dump_gcov_file): Likewise.
20069 (tag_blocks): Add and use new argument depth.
20070 (tag_arcs): Likewise.
20071 (tag_lines): Likewise.
20072 (tag_counters): Likewise.
20073 (tag_summary): Likewise.
20074 (dump_working_sets): Use depth to do a proper indentation.
20076 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20079 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20080 * symtab.c (symtab_node::debug_symtab): Move definition here.
20082 2017-04-28 Richard Biener <rguenther@suse.de>
20084 * lto-streamer.h (LTO_major_version): Bump to 7.
20086 2017-04-28 Richard Biener <rguenther@suse.de>
20088 * tree-vrp.c (assert_info): New struct.
20089 (add_assert_info): New helper.
20090 (register_edge_assert_for_2): Refactor to add asserts to a vector
20092 (register_edge_assert_for_1): Likewise.
20093 (register_edge_assert_for): Likewise.
20094 (finish_register_edge_assert_for): New helper actually registering
20095 asserts where live on edge.
20096 (find_conditional_asserts): Adjust.
20097 (find_switch_asserts): Likewise.
20098 (evrp_dom_walker::try_find_new_range): Generalize.
20099 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20101 2017-04-27 Marek Polacek <polacek@redhat.com>
20104 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20105 arg10 and arg11 to itype.
20107 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20109 * doc/extend.texi (Object Size Checking): Improve grammar.
20111 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20114 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20115 that the logic for permitting reciprocal estimates matches that
20118 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20121 * tree.c (type_cache_hasher::equal): Only compare
20122 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20123 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20124 non-aggregate element types.
20125 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20126 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20128 2017-04-27 Richard Biener <rguenther@suse.de>
20130 PR middle-end/80533
20131 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20132 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20133 keeping a reference to a trailing array.
20135 2017-04-27 Richard Biener <rguenther@suse.de>
20137 PR middle-end/80539
20138 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20139 being in loop-closed SSA form conservatively.
20140 (chrec_fold_multiply_poly_poly): Likewise.
20142 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20144 PR middle-end/79665
20145 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20146 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20148 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20151 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20152 (aarch64_function_arg_alignment): Return unsigned int again, but still
20153 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20154 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20155 Don't emit -Wpsabi note.
20156 (aarch64_function_arg_boundary): Likewise.
20157 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20160 2017-04-26 Nathan Sidwell <nathan@acm.org>
20162 * tree.h (crc32_unsigned_n): Declare.
20163 (crc32_unsigned, crc32_unsigned): Make inline.
20164 * tree.c (crc32_unsigned_bits): Replace with ...
20165 (crc32_unsigned_n): ... this.
20166 (crc32_unsigned, crc32_byte): Remove.
20167 (crc32_string): Remove unnecessary braces.
20169 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20171 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20172 * ipa-inline-analysis.c (MAX_TIME): Remove.
20173 (account_size_time): Use sreal for time.
20174 (dump_inline_summary): Update.
20175 (estimate_function_body_sizes): Update.
20176 (estimate_edge_size_and_time): Update.
20177 (estimate_calls_size_and_time): Update.
20178 (estimate_node_size_and_time): Update.
20179 (inline_merge_summary): Update.
20180 (inline_update_overall_summary): Update.
20181 (estimate_time_after_inlining): Update.
20182 (inline_read_section): Update.
20183 (inline_write_summary): Update.
20184 * ipa-inline.c (compute_uninlined_call_time): Update.
20185 (compute_inlined_call_time): Update.
20186 (recursive_inlining): Update.
20187 (inline_small_functions): Update.
20188 (dump_overall_stats): Update.
20189 * ipa-inline.h: Include sreal.h.
20190 (size_time_entry): Turn time to sreal.
20191 (inline_summary): Turn self_time nad time to sreal.
20193 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20195 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20197 (sreal::stream_out, sreal::stream_in): New.
20198 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20200 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20202 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20205 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20208 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20210 (dimode_scalar_chain::compute_convert_gain): Ditto.
20211 (dimode_scalar_chain::make_vector_copies): Ditto.
20212 (dimode_scalar_chain::convert_reg): Ditto.
20213 (dimode_scalar_chain::convert_insn): Ditto.
20214 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20215 (VI248_AVX512BW_1): New mode iterator.
20216 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20217 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20220 2017-04-25 Martin Sebor <msebor@redhat.com>
20222 PR tree-optimization/80497
20223 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20224 constants are representable in HOST_WIDE_INT.
20225 (parse_directive): Ditto.
20227 2017-04-25 Martin Sebor <msebor@redhat.com>
20230 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20231 (new_zero_array): Adjust signature.
20232 (dom_info::dom_init): Used unsigned rather that size_t.
20233 (dom_info::dom_info): Same.
20235 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20236 Jakub Jelinek <jakub@redhat.com>
20239 * config/arm/arm.c: Include gimple.h.
20240 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20241 returns negative, increment ncrn only if it returned positive.
20242 (arm_needs_doubleword_align): Return int instead of bool,
20243 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20244 members, but if there is any such non-FIELD_DECL
20245 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20246 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20247 returns negative, increment nregs only if it returned positive.
20248 (arm_setup_incoming_varargs): Likewise.
20249 (arm_function_arg_boundary): Emit -Wpsabi note if
20250 arm_needs_doubleword_align returns negative, return
20251 DOUBLEWORD_ALIGNMENT only if it returned positive.
20253 2017-04-25 Marek Polacek <polacek@redhat.com>
20256 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20257 first argument to type.
20259 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20263 type checks to test for compatibility instead of equality.
20265 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20266 Jakub Jelinek <jakub@redhat.com>
20269 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20271 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20272 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20273 the alignment computation, but return their maximum in warn_alignment.
20274 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20275 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20277 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20278 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20281 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20283 * config/arc/simdext.md (dmpyh): Fix typo.
20285 2017-04-25 Richard Biener <rguenther@suse.de>
20287 PR tree-optimization/80492
20288 * alias.c (compare_base_decls): Handle registers with asm
20289 specification conservatively.
20290 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20291 compare_base_decls returning dont-know properly.
20293 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20295 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20296 (legitimate_offset_address_p): New function.
20297 (arc_legitimate_address_p): Use above function.
20299 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20301 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20303 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20305 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20306 ACCH registers whenever they are available.
20308 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20310 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20311 double regs fix when not used.
20313 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20315 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20317 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20318 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20320 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20322 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20323 when emitting short ADD instructions.
20325 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20327 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20329 (cmpsi_cc_c_insn): Likewise.
20330 (cbranchsi4_scratch): Compute proper instruction length using
20331 compact_hreg_operand.
20332 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20334 2017-04-25 Richard Biener <rguenther@suse.de>
20336 PR middle-end/80509
20337 * passes.c (pass_manager::pass_manager): Initialize
20338 m_name_to_pass_map.
20340 2017-04-25 Richard Biener <rguenther@suse.de>
20342 PR tree-optimization/79201
20343 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20345 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20348 * config/s390/vector.md: Split MEM->GPR vector moves for
20349 non-s_operand addresses.
20351 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20354 * config/s390/predicates.md (reload_const_wide_int_operand): New
20356 * config/s390/s390.md ("movti"): Remove d/P alternative.
20357 ("movti_bigconst"): New pattern definition.
20359 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20362 * s390-protos.h (s390_expand_cs_hqi): Removed.
20363 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20364 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20365 modes as well as CCZ1mode and CCZmode.
20366 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20367 signature of s390_emit_compare_and_swap.
20368 (s390_expand_cs_hqi): Likewise, make static.
20369 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20370 compare-and-swap, in some cases.
20371 (s390_expand_cs): Wrapper function.
20372 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20374 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20375 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20376 patterns for small and large integers. Forbid symref memory operands.
20377 Move expander to s390.c. Require cc register.
20378 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20379 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20380 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20381 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20382 symref memory operands. Remove CC mode and call s390_match_ccmode
20384 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20386 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20388 * config/s390/s390.md (define_peephole2): New peephole to help
20389 combining the load-and-test pattern with volatile memory.
20391 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20393 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20394 with CCZmode for TARGET_Z196.
20396 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20398 PR rtl-optimization/80501
20399 * combine.c (make_compound_operation_int): Set subreg_code to SET
20400 even for AND with mask of the sign bit of mode.
20402 PR rtl-optimization/80500
20403 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20404 sum's initial value.
20406 2017-04-25 Julian Brown <julian@codesourcery.com>
20407 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20409 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20411 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20413 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20415 2017-04-25 Julian Brown <julian@codesourcery.com>
20416 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20418 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20419 (thunderx2t99_sha): New Reservation.
20421 2017-04-25 Julian Brown <julian@codesourcery.com>
20422 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20424 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20425 type for 1-element load.
20427 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20429 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20431 2017-04-24 Martin Jambor <mjambor@suse.cz>
20433 PR tree-optimization/80293
20434 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20435 char arrays not totally scalarizable if it is false.
20436 (analyze_all_variable_accesses): Pass correct value in the new
20437 parameter. Add a statistics counter.
20439 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20441 PR middle-end/79931
20442 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20444 2017-04-24 Richard Biener <rguenther@suse.de>
20446 PR tree-optimization/80494
20447 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20448 out for complex types.
20450 2017-04-24 Richard Biener <rguenther@suse.de>
20452 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20453 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20454 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20455 (DFS): Adjust and never fail.
20456 (sccvn_dom_walker::fail): Remove.
20457 (sccvn_dom_walker::before_dom_children): Adjust.
20458 (run_scc_vn): Likewise and never fail.
20459 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20460 (pass_fre::execute): Likewise.
20462 2017-04-24 Richard Biener <rguenther@suse.de>
20464 PR tree-optimization/79725
20465 * tree-ssa-sink.c (statement_sink_location): Return whether
20466 failure reason was zero uses. Move that check later.
20467 (sink_code_in_bb): Deal with zero uses by removing the stmt
20470 2017-04-24 Richard Biener <rguenther@suse.de>
20473 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20474 pointer-based references.
20476 2017-04-24 Richard Biener <rguenther@suse.de>
20479 * pass_manager.h (pass_manager::operator new): Remove.
20480 (pass_manager::operator delete): Likewise.
20481 * passes.c (pass_manager::operator new): Remove.
20482 (pass_manager::operator delete): Likewise.
20483 (pass_manager::pass_manager): Zero individual pass members.
20485 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20488 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20489 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20490 Check "XEXP (src, 1)" operand here.
20491 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20492 Check "XEXP (src, 1)" operand here.
20493 (dimode_scalar_chain::make_vector_copies): Detect count register
20494 of a shift instruction. Zero extend count register from QImode
20495 to DImode to satisfy vector shift pattern count operand predicate.
20496 Substitute vector shift count operand with a DImode copy.
20497 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20500 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20502 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20503 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20504 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20505 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20506 (UNSPEC_NOREX_MEM): Remove definition.
20508 2017-04-21 Richard Biener <rguenther@suse.de>
20510 PR tree-optimization/79547
20511 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20512 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20513 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20514 without any constraints.
20516 2017-04-21 Richard Biener <rguenther@suse.de>
20518 PR tree-optimization/78847
20519 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20521 2017-04-21 Richard Biener <rguenther@suse.de>
20523 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20524 (build_distinct_type_copy): Likewise.
20525 (build_variant_type_copy): Likewise.
20526 * tree.c (build_qualified_type): Pass down mem-stat info.
20527 (build_distinct_type_copy): Likewise.
20528 (build_variant_type_copy): Likewise.
20530 2017-04-21 Richard Biener <rguenther@suse.de>
20532 PR tree-optimization/80237
20533 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20535 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20536 for a simplified result.
20538 2016-04-21 Richard Biener <rguenther@suse.de>
20540 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20541 sth as strict as a simple_iv but a chrec without symbols and an
20542 operand defined in the loop we are peeling (and not some subloop).
20543 (propagate_constants_for_unrolling): Propagate all constants.
20545 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20548 * config/i386/i386.c (print_reg): Remove assert for disalowed
20549 regno values, call output_operand_lossage instead.
20551 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20554 * config/i386/constraints.md (Yc): New register constraint.
20555 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20556 Use Yc constraint for alternative 2 of operand 0. Remove
20557 preferred_for_speed attribute.
20559 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20561 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20562 lastprivate clauses in SIMT case.
20564 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20566 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20568 2017-04-20 Richard Biener <rguenther@suse.de>
20570 PR tree-optimization/57796
20571 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20572 as N scalar stores.
20573 (vect_model_load_cost): Cost gathers as N scalar loads.
20575 2017-04-20 Richard Biener <rguenther@suse.de>
20577 * ggc-page.c (ggc_allocated_p): Rename to ...
20578 (safe_lookup_page_table_entry): ... this and return the lookup
20580 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20582 2017-04-20 Richard Biener <rguenther@suse.de>
20584 PR tree-optimization/80453
20585 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20586 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20587 from the conditions.
20588 (vn_phi_eq): Pass them down.
20589 (vn_phi_lookup): Record them.
20590 (vn_phi_insert): Likewise.
20592 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20594 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20595 uninitialized variable warning to avoid buffer overrun.
20597 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20600 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20601 is suppressed for '{ 0 }' in C.
20603 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20605 * BASE-VER: Set to 8.0.0.
20607 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20609 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20610 priority .init_array and .fini_array section with SECTION_NOTYPE
20613 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20615 PR middle-end/80423
20616 * tree.h (build_array_type): Add typeless_storage default argument.
20617 * tree.c (type_cache_hasher::equal): Also compare
20618 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20619 (build_array_type): Add typeless_storage argument, set
20620 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20622 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20623 (build_array_type): Likewise. Add typeless_storage argument.
20625 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20626 Jakub Jelinek <jakub@redhat.com>
20628 PR tree-optimization/80426
20629 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20630 operation on symbolic operands, also compute the overflow for the
20631 invariant part when the operation degenerates into a negation.
20633 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20636 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20637 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20640 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20642 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20645 * config/avr/avr.c (tree.h): Include it.
20646 (cgraph.h): Include it.
20647 (avr_encode_section_info): Don't warn for uninitialized progmem
20648 variable if it's just an alias.
20650 2017-04-19 Richard Biener <rguenther@suse.de>
20653 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20654 when needed by AutoPGO.
20656 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20659 * doc/lto.texi: Remove an extra 'that'.
20661 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20663 PR rtl-optimization/80429
20664 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20665 are only used in debug insns.
20667 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20668 Vladimir Makarov <vmakarov@redhat.com>
20670 * config/sparc/predicates.md (input_operand): Add comment. Return
20671 true for any memory operand when LRA is in progress.
20672 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20674 2017-04-18 Jeff Law <law@redhat.com>
20677 * mips.md ({return,simple_return}_internal): Do not overwrite
20680 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20682 PR tree-optimization/80443
20683 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20684 instead of adding 1, subtract -1 and similarly instead of subtracting
20687 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20689 PR rtl-optimization/80357
20690 * haifa-sched.c (tmp_bitmap): New variable.
20691 (model_recompute): Handle duplicate use records.
20692 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20693 (free_global_sched_pressure_data): Free it.
20695 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20698 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20699 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20700 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20701 instead of SYSTEM_HEADER_DIR.
20703 2017-04-18 Jeff Law <law@redhat.com>
20705 PR middle-end/80422
20706 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20707 predecessors after walking up the insn chain.
20709 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20712 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20713 sizetype type into debug info.
20715 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20718 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20719 unneeded test for TARGET_UPPER_REGS_SF.
20720 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20722 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20725 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20726 instead of gsi_after_labels.
20728 2017-04-18 Jeff Law <law@redhat.com>
20730 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20734 2017-04-13 Jeff Law <law@redhat.com>
20735 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20736 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20738 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20741 * config/avr/avr.c (intl.h): Include it.
20742 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20744 2017-04-18 Martin Liska <mliska@suse.cz>
20746 PR gcov-profile/78783
20747 * gcov-tool.c (gcov_output_files): Validate that destination
20748 file is either removed by the tool or by a user.
20750 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20751 Guy Benyei <guybe@mellanox.com>
20753 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20754 block, and do not negate it, the stored id is already negative.
20756 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20758 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20760 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20763 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20764 masks of options that should be turned off if the VSX vector
20765 options are turned off.
20766 (OTHER_P8_VECTOR_MASKS): Likewise.
20767 (OTHER_VSX_VECTOR_MASKS): Likewise.
20768 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20769 rs6000_disable_incompatible_switches to validate no type switches
20771 (rs6000_incompatible_switch): New function to disallow turning on
20772 other vector options if -mno-vsx, -mno-power8-vector, or
20773 -mno-power9-vector are specified.
20775 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20777 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20779 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20781 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20782 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20783 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20784 (ARG_POINTER_CFA_OFFSET): Likewise.
20786 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20788 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20789 conditions to take advantage of various optimizations.
20791 2017-04-13 Jeff Law <law@redhat.com>
20793 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20794 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20795 (zero_extendsidi2_dext): Likewise.
20797 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20800 * fold-const.c (fold_ternary_loc): Revert
20801 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20804 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20806 PR rtl-optimization/80343
20807 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20808 new scratch pseudo.
20810 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20813 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20814 to ubsan_encode_value.
20816 2017-04-13 Jeff Law <law@redhat.com>
20818 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20819 appearing in DEBUG_INSNs.
20821 2017-04-13 Martin Liska <mliska@suse.cz>
20823 PR gcov-profile/80413
20824 * gcov-io.c (gcov_write_string): Copy to buffer just when
20825 allocated size is greater than zero.
20827 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20830 * dwarf2out.c (decls_for_scope): Ignore declarations of
20831 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20833 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20836 * ipa-visibility.c (non_local_p): Fix typos.
20837 (localize_node): When localizing symbol in same comdat group,
20838 dissolve the group only when we know external symbols are going
20840 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20842 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20844 PR tree-optimization/79390
20845 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20846 order does not result in usable sequence, retry with reversed operand
20852 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20853 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20854 op0 instead of fold_convert_loc (loc, type, arg0).
20856 2017-04-12 Jeff Law <law@redhat.com>
20858 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20859 has a delay slot in the generated code.
20861 * config/cris/cris.md (cris_preferred_reload_class): Return
20862 GENNONACR_REGS rather than GENERAL_REGS.
20864 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20867 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20868 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20869 signedness of the result type.
20871 2017-04-12 Richard Biener <rguenther@suse.de>
20872 Jeff Law <law@redhat.com>
20874 PR tree-optimization/80359
20875 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20876 trim stores to TARGET_MEM_REFs.
20878 2017-04-12 Richard Biener <rguenther@suse.de>
20880 PR tree-optimization/79390
20881 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20882 threading case even more.
20884 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20887 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20888 for quad_address_p for TImode, instead of just not indexed_address.
20890 2017-04-12 Richard Biener <rguenther@suse.de>
20891 Bernd Edlinger <bernd.edlinger@hotmail.de>
20893 PR middle-end/79671
20894 * alias.c (component_uses_parent_alias_set_from): Handle
20895 TYPE_TYPELESS_STORAGE.
20896 (get_alias_set): Likewise.
20897 * tree-core.h (tree_type_common): Add typeless_storage flag.
20898 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20899 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20900 for types containing members with TYPE_TYPELESS_STORAGE.
20901 (place_field): Likewise.
20902 (layout_type): Likewise for ARRAY_TYPE.
20903 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20904 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20905 TYPE_TYPELESS_STORAGE.
20906 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20908 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20911 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20912 first argument to type.
20914 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20918 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20919 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20920 (rs6000_expand_binop_builtin): Likewise.
20921 (rs6000_expand_ternop_builtin): Likewise; also add missing
20922 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20923 vshasigma built-ins.
20924 * doc/extend.texi: Document that vec_xxpermdi's third argument
20925 must be a constant.
20927 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20929 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20930 Use shift_const cost parameter when calculating gain of STV shifts.
20932 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20934 PR rtl-optimization/70478
20935 * lra-constraints.c (process_alt_operands): Check memory for
20936 disfavoring memory insn operand.
20938 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20940 PR middle-end/80100
20941 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20942 left shift in unsigned HOST_WIDE_INT type.
20944 PR rtl-optimization/80385
20945 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20946 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20949 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20950 if they have any depend clauses.
20952 2017-04-11 Martin Liska <mliska@suse.cz>
20955 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20956 * ipa-split.c (split_function): Create a local comdat symbol
20957 if caller is in a comdat group.
20959 2017-04-11 Martin Liska <mliska@suse.cz>
20962 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20965 2017-04-11 Martin Sebor <msebor@redhat.com>
20967 PR middle-end/80364
20968 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20969 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20971 (directive::set_width, directive::set_precision, format_character):
20973 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20976 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20979 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20980 conflict, set target->arch_name instead of target->cpu_name.
20982 2017-04-11 Richard Biener <rguenther@suse.de>
20984 PR tree-optimization/80374
20985 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20986 build_zero_cst, remove fold_convertible_p check again.
20988 2017-04-11 Martin Liska <mliska@suse.cz>
20991 * ubsan.c (instrument_object_size): Do not instrument register
20994 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20997 * config/i386/i386-builtin-types.def
20998 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20999 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21000 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21001 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21002 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21003 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21004 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21005 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21006 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21007 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21008 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21009 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21010 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21011 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21012 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21013 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21014 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21015 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21016 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21017 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21018 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21019 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21020 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21021 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21022 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21023 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21024 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21025 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21026 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21027 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21028 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21029 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21030 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21031 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21032 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21033 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21034 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21035 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21036 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21037 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21038 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21039 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21040 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21041 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21042 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21043 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21045 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21046 flag to second_arg_count, handle 4 argument function type _COUNT
21047 aliases, handle second_arg_count on second argument rather than last.
21049 2017-04-10 Jeff Law <law@redhat.com>
21051 PR tree-optimization/80374
21052 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21053 record anything if we can not convert integer_zero_node to the
21056 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21059 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21060 Enhance special handling given to the TARGET_P9_MINMAX option in
21061 relation to certain other options.
21063 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21065 PR tree-optimization/80153
21066 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21067 remove POINTER_PLUS_EXPR's base part directly, rather than through
21070 2017-04-10 Richard Biener <rguenther@suse.de>
21071 Bin Cheng <bin.cheng@arm.com>
21073 PR tree-optimization/80153
21074 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21075 the first element of pointer type aff_tree. Build result expr in
21077 (add_elt_to_tree): Convert to type unconditionally. Remove other
21078 fold_convert calls.
21079 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21080 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21082 2017-04-10 Richard Biener <rguenther@suse.de>
21084 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21087 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21089 PR rtl-optimization/70478
21090 * lra-constraints.c (curr_small_class_check): New.
21091 (update_and_check_small_class_inputs): New.
21092 (process_alt_operands): Update curr_small_class_check. Disfavor
21093 alternative insn memory operands. Check available regs for small
21096 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21099 * config/mips/mips.opt (-mvirt): Update description.
21100 * doc/invoke.texi (-mvirt): Likewise.
21102 2017-04-10 Richard Biener <rguenther@suse.de>
21104 PR middle-end/80362
21105 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21106 looking for NEGATE_EXPR in -A / -B to A / B folding.
21108 2017-04-10 Martin Liska <mliska@suse.cz>
21110 PR gcov-profile/80224
21111 * gcov.c (print_usage): Fix usage string.
21112 (get_gcov_intermediate_filename): Remove.
21113 (output_gcov_file): Use both for normal and intermediate format.
21114 (generate_results): Do not initialize special file for
21115 intermediate format.
21117 2017-04-10 Richard Biener <rguenther@suse.de>
21119 PR tree-optimization/80304
21120 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21123 2017-04-10 Nathan Sidwell <nathan@acm.org>
21126 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21127 (rs6000_init_builtins): Use it.
21129 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21131 * config/arm/arm.md (<mrc>): Add mode to SET source.
21132 (<mrrc>): Likewise.
21134 2017-04-10 Richard Biener <rguenther@suse.de>
21136 PR middle-end/80344
21137 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21139 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21142 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21143 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21144 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21145 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21146 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21147 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21148 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21149 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21150 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21151 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21152 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21153 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21154 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21155 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21156 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21157 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21158 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21159 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21160 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21161 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21162 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21163 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21164 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21166 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21168 PR rtl-optimization/70478
21169 * lra-constraints.c: Reverse the last patch.
21171 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21173 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21174 Add comment for WCHAR_T.
21176 2017-04-08 Martin Liska <mliska@suse.cz>
21179 2017-04-07 Martin Liska <mliska@suse.cz>
21182 * ipa-split.c (split_function): Add function part to a same comdat
21185 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21188 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21190 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21192 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21193 (vec_pairing): ...to this.
21194 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21195 (rs6000_sched_init): Adjust for name change.
21196 (struct rs6000_sched_context): Likewise.
21197 (rs6000_init_sched_context): Likewise.
21198 (rs6000_set_sched_context): Likewise.
21200 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21206 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21208 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21209 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21210 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21212 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21214 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21216 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21218 PR rtl-optimization/70703
21219 * ira-color.c (update_conflict_hard_regno_costs): Use
21220 int64_t instead of HOST_WIDE_INT.
21222 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21224 PR rtl-optimization/70478
21225 * lra-constraints.c (process_alt_operands): Disfavor alternative
21226 insn memory operands.
21228 2017-04-07 Jeff Law <law@redhat.com>
21230 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21231 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21233 2017-04-07 Martin Liska <mliska@suse.cz>
21236 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21237 Show error message instead of an ICE.
21239 2017-04-07 Martin Liska <mliska@suse.cz>
21242 * ipa-split.c (split_function): Add function part to a same comdat
21245 2017-04-07 Richard Biener <rguenther@suse.de>
21247 PR middle-end/80341
21248 * tree.c (get_unwidened): Also handle ! for_type case for
21250 * convert.c (do_narrow): Split out from ...
21251 (convert_to_integer_1): ... here. Do not pass final truncation
21252 type to get_unwidened for TRUNC_DIV_EXPR.
21254 2017-04-07 Richard Biener <rguenther@suse.de>
21256 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21258 (aff_combination_const): Adjust.
21259 (aff_combination_scale): Likewise.
21260 (aff_combination_add_elt): Likewise.
21261 (aff_combination_add_cst): Likewise.
21262 (aff_combination_convert): Likewise.
21263 (add_elt_to_tree): Likewise. Remove unused argument.
21264 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21266 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21268 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21270 * config/arm/arm.c (arm_default_short_enums): Use
21271 ARM_DEFAULT_SHORT_ENUMS.
21272 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21274 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21277 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21278 members with redundant out-of-class redeclaration.
21280 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21283 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21284 * config/i386/i386.md (*zero_extendsidi2):
21285 Add (?*x,*x) and (?*v,*v) alternatives.
21287 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21290 * config/i386/i386.c (ix86_expand_builtin)
21291 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21292 mode from insn data. Convert operands to insn operand mode.
21293 Copy operands that don't satisfy insn predicate to a register.
21295 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21297 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21300 2017-04-06 Richard Biener <rguenther@suse.de>
21302 PR tree-optimization/80334
21303 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21304 preserve alignment of accesses.
21306 2017-04-06 Richard Biener <rguenther@suse.de>
21308 PR tree-optimization/80262
21309 * tree-sra.c (build_ref_for_offset): Preserve address-space
21311 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21312 Drop useless address-space information on MEM_REF offsets.
21314 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21316 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21318 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21320 PR rtl-optimization/70703
21321 * ira-color.c (update_conflict_hard_regno_costs): Use
21322 HOST_WIDE_INT instead of long.
21324 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21327 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21328 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21330 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21331 for x86_64 target. Handle -m3dnowa option.
21333 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21335 PR rtl-optimization/70703
21336 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21337 (update_conflict_hard_regno_costs): Use long instead of unsigned
21338 arithmetic for cost calculation.
21340 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21341 Bernd Edlinger <bernd.edlinger@hotmail.de>
21344 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21347 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21350 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21351 ptr_mode with Pmode throughout.
21352 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21353 into probe_stack_range and use DImode.
21355 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21358 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21359 call_eh_return is true.
21361 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21363 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21364 Initialize last_match_fntype_index.
21366 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21369 * tree-nvr.c: Include internal-fn.h.
21370 (pass_return_slot::execute): Ignore internal calls without
21373 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21374 Richard Biener <rguenther@suse.de>
21377 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21378 captures used multiple times, except for the last use.
21379 * generic-match-head.c: Include gimplify.h.
21381 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21383 PR tree-optimization/79390
21384 * target.h (struct noce_if_info): Declare.
21385 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21386 * target.def (noce_conversion_profitable_p): New target hook.
21387 * ifcvt.h (struct noce_if_info): New type, moved from ...
21388 * ifcvt.c (struct noce_if_info): ... here.
21389 (noce_conversion_profitable_p): Renamed to ...
21390 (default_noce_conversion_profitable_p): ... this. No longer
21392 (noce_try_store_flag_constants, noce_try_addcc,
21393 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21394 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21395 instead of noce_conversion_profitable_p.
21396 * config/i386/i386.c: Include ifcvt.h.
21397 (ix86_option_override_internal): Don't override
21398 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21399 (ix86_noce_conversion_profitable_p): New function.
21400 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21401 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21402 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21403 * doc/tm.texi: Regenerated.
21405 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21407 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21410 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21413 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21414 instructions for small multiply cores.
21416 2017-04-04 Jeff Law <law@redhat.com>
21418 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21420 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21421 that are not set by the loop over the elements.
21423 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21426 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21427 int mode, convert_modes it to mode as unsigned, otherwise use
21428 lowpart_subreg to mode rather than SImode.
21429 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21430 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21431 Use DImode instead of SImode for the shift count operand.
21432 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21435 2017-04-04 Richard Biener <rguenther@suse.de>
21437 PR middle-end/80281
21438 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21439 arithmetic done for the negate or the plus. Simplify.
21440 (A - (-B) -> A + B): Likewise.
21441 * fold-const.c (split_tree): Make sure to not negate pointers.
21443 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21445 PR rtl-optimization/60818
21446 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21447 a compare of comparisons with the thing compared if this results
21448 in a different machine mode.
21450 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21452 * alias.c (base_alias_check): Fix typo in comment.
21453 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21454 * cgraphunit.c (symbol_table::compile): Likewise.
21455 * collect2.c (maybe_run_lto_and_relink): Likewise.
21456 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21457 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21458 * config/avr/avr.c (avr_map_op_t): Likewise.
21459 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21460 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21461 * config/epiphany/epiphany.md (movcc): Likewise.
21462 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21463 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21465 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21466 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21467 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21468 * config/sh/sh.c (sh_rtx_costs): Likewise.
21469 * fold-const.c (fold_truth_andor): Likewise.
21470 * genautomata.c (collapse_flag): Likewise.
21471 * gengtype.h (struct type::u::s): Likewise.
21472 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21473 * input.c (FORMAT_AMOUNT): Likewise.
21474 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21475 (known_aggs_to_agg_replacement_list): Likewise.
21476 * ipa-inline-analysis.c: Likewise.
21477 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21478 * ipa-polymorphic-call.c
21479 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21480 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21481 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21483 * modulo-sched.c (apply_reg_moves): Likewise.
21484 * omp-expand.c (build_omp_regions_1): Likewise.
21485 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21486 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21487 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21488 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21489 * value-prof.c: Likewise.
21490 * var-tracking.c (val_reset): Likewise.
21492 2017-04-03 Richard Biener <rguenther@suse.de>
21494 PR tree-optimization/80275
21495 * fold-const.c (split_address_to_core_and_offset): Handle
21498 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21500 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21501 descriptors is at least equal to that of functions.
21503 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21505 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21507 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21510 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21511 (mov<IMOD4:mode>): New expander.
21512 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21514 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21516 PR rtl-optimization/79405
21517 * fwprop.c (propagations_left): New variable.
21518 (forward_propagate_into): Decrement it.
21519 (fwprop_init): Initialize it.
21520 (fw_prop): If the variable has reached zero, stop propagating.
21521 (fwprop_addr): Ditto.
21523 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21526 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21527 a FUNCTION_DECL, pass it as decl instead of origin to
21530 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21532 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21535 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21538 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21539 TARGET_VSX_SMALL_INTEGER.
21541 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21543 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21544 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21546 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21548 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21549 extraction from odd-numbered MSA register.
21551 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21553 PR middle-end/80173
21554 * expmed.c (store_bit_field_1): Don't attempt to create
21555 a word subreg out of hard registers wider than word if they
21556 have HARD_REGNO_NREGS of 1 for their mode.
21558 PR middle-end/80163
21559 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21560 conversions to integer types wider than word and pointer.
21563 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21564 (rtx_equal_for_cselib_p): Pass 0 to it.
21565 * cselib.c (cselib_hasher::equal): Likewise.
21566 (rtx_equal_for_cselib_1): Add depth argument. If depth
21567 is 128, don't look up VALUE locs and punt. Increment
21568 depth in recursive calls when walking VALUE locs.
21570 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21572 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21573 (make_gcov_file_name): Use the canonical path name for generating
21575 (read_line): Fix handling of files with ascii null bytes.
21577 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21579 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21580 to initialise a vector register instead
21581 of using a const_int.
21583 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21585 PR translation/80189
21586 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21587 diagnostic messages.
21589 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21592 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21593 (dfp_diex_<mode>): Update mode of operand 1.
21594 * doc/extend.texi (dxex, dxexq): Document change to return type.
21595 (diex, diexq): Document change to argument type.
21597 2017-03-30 Martin Jambor <mjambor@suse.cz>
21600 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21601 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21602 it reflects the signature changes performed at the callee side.
21603 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21604 to cgraph_build_function_type_skip_args.
21605 (build_function_decl_skip_args): Adjust call to the above function.
21607 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21610 * config/i386/sse.md
21611 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21612 register as dest whenever it is a MEM not rtx_equal_p to the
21613 corresponding dup operand, and when forcing into reg move the
21614 reg into the memory afterwards.
21615 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21616 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21617 for the force_reg mode.
21618 (avx512vl_vextractf128<mode>): Use register as dest either
21619 always when a MEM, or when it is a MEM not rtx_equal_p to the
21620 corresponding dup operand, or even not when it is a CONST_VECTOR
21621 depending on the mode and lo vs. hi.
21622 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21624 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21625 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21626 Likewise. Require that operands[2] is even.
21627 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21628 Remove extraneous parens. Require that operands[2] is a multiple
21630 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21631 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21632 disallow memory then.
21634 2017-03-30 Richard Biener <rguenther@suse.de>
21636 PR tree-optimization/77498
21637 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21638 to non-constants over backedges.
21640 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21642 PR rtl-optimization/80233
21643 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21644 as last_combined_insn. Do not test for BARRIER_P separately.
21646 2017-03-29 Andreas Schwab <schwab@suse.de>
21649 * calls.c (prepare_call_address): Convert funexp to Pmode before
21650 copying to temp reg.
21652 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21654 PR tree-optimization/80158
21655 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21656 Handle possible future case of more than one alternate
21658 (replace_rhs_if_not_dup): Likewise.
21659 (replace_one_candidate): Likewise.
21661 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21663 PR rtl-optimization/80193
21664 * ira.c (ira): Do not check allocation for LRA.
21666 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21668 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21669 (nvptx_output_simt_exit): Declare.
21670 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21671 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21672 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21673 (nvptx_declare_function_name): ...here. Emit declaration of local
21674 memory space buffer for omp_simt_enter insn.
21675 (nvptx_output_unisimt_switch): New.
21676 (nvptx_output_softstack_switch): New.
21677 (nvptx_output_simt_enter): New.
21678 (nvptx_output_simt_exit): New.
21679 * config/nvptx/nvptx.h (struct machine_function): New fields
21680 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21681 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21682 (UNSPECV_SIMT_EXIT): Ditto.
21683 (omp_simt_enter_insn): New insn.
21684 (omp_simt_enter): New expansion.
21685 (omp_simt_exit): New insn.
21686 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21688 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21689 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21690 (expand_GOMP_SIMT_EXIT): New.
21691 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21692 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21693 (GOMP_SIMT_EXIT): Ditto.
21694 * target-insns.def (omp_simt_enter): New insn.
21695 (omp_simt_exit): Ditto.
21696 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21698 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21699 (lower_rec_input_clauses): Likewise.
21700 (lower_lastprivate_clauses): Handle SIMT privatization.
21702 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21703 (ompdevlow_adjust_simt_enter): New.
21704 (find_simtpriv_var_op): New.
21705 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21706 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21708 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21709 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21710 (copy_decl_for_dup_finish): Ditto.
21712 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21714 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21717 * config/i386/i386.c (ix86_option_override_internal): Always
21718 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21720 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21722 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21724 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21726 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21727 mark new edge's irreducible flag accordign to it.
21728 (vect_do_peeling): Check loop preheader edge's irreducible flag
21729 and pass it to function slpeel_add_loop_guard.
21731 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21733 PR tree-optimization/80218
21734 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21735 Update block frequencies and counts.
21737 2017-03-28 Richard Biener <rguenther@suse.de>
21739 PR tree-optimization/78644
21740 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21741 of a simplification result we may not use it at all.
21743 2017-03-28 Richard Biener <rguenther@suse.de>
21746 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21747 without arguments, generate default definition of a SSA name.
21749 2017-03-28 Richard Biener <rguenther@suse.de>
21751 PR middle-end/80222
21752 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21753 TYPE_REF_CAN_ALIAS_ALL references.
21754 * fold-const.c (fold_indirect_ref_1): Likewise.
21756 2017-03-28 Martin Liska <mliska@suse.cz>
21759 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21760 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21762 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21763 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21765 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21766 (EXTRA_SPECS): Define.
21767 (SUBTARGET_EXTRA_SPECS): Likewise.
21768 (SUBTARGET_CPP_SPEC): Likewise.
21769 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21770 SUBTARGET_EXTRA_SPECS.
21771 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21773 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21775 * config/arc/simdext.md (vst64_insn): Update pattern.
21776 (vld32wh_insn): Likewise.
21777 (vld32wl_insn): Likewise.
21778 (vld64_insn): Likewise.
21779 (vld32_insn): Likewise.
21781 2017-03-28 Marek Polacek <polacek@redhat.com>
21784 * fold-const.c (fold_comparison): Use protected_set_expr_location
21785 instead of SET_EXPR_LOCATION.
21787 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21789 * tree.c (add_expr): Avoid name lookup warning.
21791 2017-03-27 Jeff Law <law@redhat.com>
21793 PR tree-optimization/80216
21794 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21795 function name. Limit recursion depth.
21796 (record_temporary_equivalences): Corresponding changes.
21798 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21800 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21803 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21806 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21808 * cfgcleanup.c (reg_note_cfa_p): New array.
21809 (insns_have_identical_cfa_notes): New function.
21810 (old_insns_match_p): Don't cross-jump in between /f
21811 and non-/f instructions. If both i1 and i2 are frame related,
21812 verify all CFA notes, their order and content.
21814 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21817 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21818 HImode and SImode with zero extend to DImode to one insn.
21819 (bswap<mode>2_extenddi): Likewise.
21820 (bswapsi2_extenddi): Likewise.
21821 (bswaphi2_extendsi): Likewise.
21822 (bswaphi2): Combine bswap HImode and SImode into one insn.
21823 Separate memory insns from swapping register.
21824 (bswapsi2): Likewise.
21825 (bswap<mode>2): Likewise.
21826 (bswaphi2_internal): Delete, no longer used.
21827 (bswapsi2_internal): Likewise.
21828 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21829 store, and gpr<-gpr swap insns.
21830 (bswap<mode>2_store): Likewise.
21831 (bswaphi2_reg): Register only splitter, combine with the splitter.
21832 (bswaphi2 splitter): Likewise.
21833 (bswapsi2_reg): Likewise.
21834 (bswapsi2 splitter): Likewise.
21835 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21836 the insns into load, store, and register/register insns.
21837 (bswapdi2_ldbrx): Likewise.
21838 (bswapdi2_load): Likewise.
21839 (bswapdi2_store): Likewise.
21840 (bswapdi2_reg): Likewise.
21842 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21844 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21845 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21847 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21850 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21852 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21853 special handling for target option conflicts between dform
21854 options (-mpower9-dform, -mpower9-dform-vector,
21855 -mpower9-dform-scalar) and -mno-direct-move.
21857 2017-03-27 Richard Biener <rguenther@suse.de>
21859 PR tree-optimization/80181
21860 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21862 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21864 * config/arc/predicates.md (move_double_src_operand): Replace the
21865 call to move_double_src_operand with a call to address_operand.
21867 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21869 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21870 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21871 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21873 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21875 * config/arc/predicates.md (long_immediate_loadstore_operand):
21876 Consider scaled addresses cases.
21878 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21880 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21881 restored when in interrupt.
21882 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21883 doesn't have delay slot.
21885 2017-03-27 Richard Biener <rguenther@suse.de>
21888 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21889 inlined thunk clones.
21891 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21894 * asan.c (instrument_derefs): Copy over last operand from
21895 original COMPONENT_REF to the new COMPONENT_REF with
21896 DECL_BIT_FIELD_REPRESENTATIVE.
21897 * ubsan.c (instrument_object_size): Likewise.
21899 2017-03-27 Richard Biener <rguenther@suse.de>
21901 PR tree-optimization/80170
21902 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21903 sure DR/SCEV didnt fold in constants we do not see when looking
21904 at the reference base alignment.
21906 2017-03-27 Richard Biener <rguenther@suse.de>
21908 PR middle-end/80171
21909 * gimple-fold.c (fold_ctor_reference): Properly guard against
21910 NULL return value from canonicalize_constructor_val.
21912 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21915 * config/i386/i386.c (ix86_expand_builtin)
21916 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21917 flags reg setting and flags reg using instructions.
21918 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21919 clobbering instructions to zero extend op2.
21921 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21923 * doc/install.texi (Configuration) <--with-aix-soname>:
21924 Update link to AIX ld.
21926 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21928 PR rtl-optimization/80160
21929 PR rtl-optimization/80159
21930 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21931 reg_alternate_class into account.
21933 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21936 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21937 to consider in curr_insn_transform.
21939 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21941 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21942 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21943 and emit_mode_inner.
21945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21947 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21948 argument to the overloaded builtin variants. Use the new flag to
21949 deprecate certain builtin variants.
21950 * config/s390/s390-builtin-types.def: Add new builtin types.
21951 * config/s390/s390-builtins.h: Support new flags field for
21952 overloaded builtins.
21953 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21954 (s390_macro_to_expand): Enable vector float data type.
21955 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21956 builtins by incrementing the __VEC__ version number.
21957 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21959 (s390_resolve_overloaded_builtin): Emit error messages depending
21960 on the builtin flags.
21961 * config/s390/s390.c (s390_expand_builtin): Support additional
21962 flags argument. Change error message to match the messages
21963 emitted in s390-c.c.
21964 * config/s390/s390.md: New UNSPEC_* constants.
21965 (op_type): Add new instruction types.
21966 * config/s390/vecintrin.h: Add new builtins and test data class
21968 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21969 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21970 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21971 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21972 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21973 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21975 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21976 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21977 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21978 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21980 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21981 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21982 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21983 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21985 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21986 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21987 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21988 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21989 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21990 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21991 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21993 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21994 ("vec_scatter_element<V_HW_4:mode>_DI")
21995 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21996 ("vec_fpint<mode>", "vflls")
21997 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21998 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21999 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22000 ("*vec_cmphe<mode>_cc"): ... these.
22002 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22003 mode constant instead of magic value.
22005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22007 * config/s390/s390.c (s390_expand_vec_compare): Support other
22008 vector floating point modes than just V2DF.
22009 (s390_expand_vcond): Likewise.
22010 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22011 (s390_cannot_change_mode_class): Prevent mode changes between TF
22012 and V1TF in vector registers.
22013 * config/s390/s390.md (DF, SF): New mode attributes.
22014 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22015 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22016 SFmode support for VRs.
22017 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22019 (VFT, VF_HW): New mode iterators.
22020 (vw, sdx): New mode attributes.
22021 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22022 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22023 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22024 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22025 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22026 also the new vector floating point modes. Renaming to ...
22028 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22029 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22030 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22031 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22032 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22033 ("vec_unordered<mode>"): ... these.
22035 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22036 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22037 ("*vec_extendv2df"): New insn definitions.
22039 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22041 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22042 ("mulditi3_2", "*muldi3_sign"): New patterns.
22043 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22044 rename the pattern definition.
22046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22048 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22050 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22054 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22055 instruction if possible.
22056 * config/s390/vector.md (vec_halfnumelts): New mode
22058 ("*vec_vllezlf<mode>"): New pattern.
22060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22062 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22063 ("popcountv4si2", "popcountv2di2"): Rename to ...
22064 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22065 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22067 ("popcount<mode>2_vxe"): New pattern.
22069 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22071 * common/config/s390/s390-common.c (processor_flags_table): Add
22073 * config.gcc: Add arch12.
22074 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22075 Default to arch12 for unknown CPU model numbers.
22076 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22077 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22078 PROCESSOR_max sanity check.
22079 * config/s390/s390-opts.h (enum processor_type): Add
22081 * config/s390/s390.c (processor_table): Add arch12.
22082 (s390_expand_builtin): Add check for B_VXE flag.
22083 (s390_issue_rate): Add PROCESSOR_ARCH12.
22084 (s390_get_sched_attrmask): Likewise.
22085 (s390_get_unit_mask): Likewise.
22086 (s390_sched_score): Enable z13 scheduling for arch12.
22087 (s390_sched_reorder): Likewise.
22088 (s390_sched_variable_issue): Likewise.
22089 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22091 (s390_tune_attr): Use z13 scheduling also for arch12.
22092 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22093 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22094 (TARGET_VXE_P): New macros.
22095 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22096 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22097 * config/s390/s390.opt: Add arch12 as processor_type.
22099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22101 * config/s390/s390.md
22102 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22103 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22104 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22106 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22107 Rename expanders to ...
22109 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22110 ("fixuns_truncdddi2_emu"): ... these.
22112 ("fixuns_trunc<mode>si2_emu"): New expander.
22114 ("*fixuns_truncdfdi2_z13"): Rename to ...
22115 ("*fixuns_truncdfdi2_vx"): ... this.
22117 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22119 * config/s390/2964.md: Remove the single element vector compare
22120 instructions which are no longer used.
22121 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22123 (s390_canonicalize_comparison): Remove handling of DFmode
22125 (s390_expand_vec_compare_scalar): Remove function.
22126 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22127 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22129 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22131 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22133 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22135 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22136 will anyway by matched by mov<mode>_64dfp.
22138 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22140 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22141 vlef/vstef. Add missing operand to vleif.
22143 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22145 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22146 pair for all vector types with 64 bit elements.
22147 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22148 * config/s390/vector.md (V_HW_64): ... here.
22149 (V_128_NOSINGLE): New mode iterator.
22150 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22151 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22152 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22153 ("*vec_load_pairv2di"): Change to ...
22154 ("*vec_load_pair<mode>"): ... this one.
22156 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22158 * config/s390/constraints.md: Add comments.
22159 (jKK): Reject element sizes > 8 bytes.
22160 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22162 * config/s390/s390.md: Add the s_operand checks formerly in
22163 s390_split_ok_p to various splitters where they are still
22165 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22166 for 128 bit vectors. Plus two splitters.
22168 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22170 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22173 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22176 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22177 error if the boundary argument is not constant.
22179 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22181 PR rtl-optimization/80112
22182 * loop-doloop.c (doloop_condition_get): Don't check condition
22183 if cmp isn't SET with IF_THEN_ELSE src.
22185 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22187 PR tree-optimization/80158
22188 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22189 replacing a candidate statement, also replace it for the
22190 candidate's alternate interpretation.
22191 (replace_rhs_if_not_dup): Likewise.
22192 (replace_one_candidate): Likewise.
22194 2017-03-24 Richard Biener <rguenther@suse.de>
22196 PR tree-optimization/80167
22197 * graphite-isl-ast-to-gimple.c
22198 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22200 (translate_isl_ast_to_gimple::get_rename): Likewise.
22202 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22204 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22205 handling of certain combinations of target options, including the
22206 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22207 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22209 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22212 * config/arm/arm.md (*load_multiple): Add reload_completed to
22213 matching condition.
22215 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22216 Richard Biener <rguenth@suse.de>
22218 PR tree-optimization/79908
22219 PR tree-optimization/80136
22220 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22221 been cast away, gimplify_and_add suffices.
22223 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22225 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22227 2017-03-23 Richard Biener <rguenther@suse.de>
22229 PR tree-optimization/80032
22230 * gimplify.c (gimple_push_cleanup): Forced unconditional
22231 cleanups still have to go to the conditional_cleanups
22234 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22236 PR tree-optimization/80072
22237 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22239 (next_operand_entry_id): Change type to unsigned int.
22240 (sort_by_operand_rank): Make sure to return the right return value
22241 even if unsigned fields are bigger than INT_MAX.
22242 (struct oecount): Change cnt and id type to unsigned int.
22243 (oecount_hasher::equal): Formatting fix.
22244 (oecount_cmp): Make sure to return the right return value
22245 even if unsigned fields are bigger than INT_MAX.
22246 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22249 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22250 TREE_READONLY on result if writing it more than once.
22253 * doc/invoke.texi (-fsanitize=thread): Document that with
22254 -fnon-call-exceptions atomics are not able to throw
22258 * tsan.c: Include tree-eh.h.
22259 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22260 maybe_clean_or_replace_eh_stmt where needed.
22261 (instrument_memory_accesses): Add cfg_changed argument.
22262 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22263 if it returned true.
22264 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22266 PR rtl-optimization/63191
22267 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22268 wrapper function, moved the whole old content into ...
22269 (ix86_delegitimize_address_1): ... this. New inline function.
22270 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22271 true as last argument instead of ix86_delegitimize_address.
22273 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22275 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22276 cortexa57_branch_cost.
22278 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22280 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22282 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22285 * doc/md.texi (Constraints): Document wA constraint.
22286 * config/rs6000/constraints.md (wA): New.
22287 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22288 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22289 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22290 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22292 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22295 * gimplify.c (is_oacc_declared): New function.
22296 (oacc_default_clause): Use it to set default flags for acc declared
22297 variables inside parallel regions.
22298 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22299 declared variables.
22300 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22301 declare attribute to any decl as necessary.
22303 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22306 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22307 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22308 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22309 (arm_arch_lpae): This.
22310 * config/arm/arm.c (arm_arch7ve): Rename into ...
22311 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22312 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22315 2017-03-22 Martin Liska <mliska@suse.cz>
22318 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22319 error message instead of an ICE.
22321 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22323 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22325 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22327 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22332 2017-03-21 Martin Sebor <msebor@redhat.com>
22334 * doc/extend.texi: Use "cannot" instead of "can't."
22335 * doc/hostconfig.texi: Same.
22336 * doc/install.texi: Same.
22337 * doc/invoke.texi: Same.
22338 * doc/loop.texi: Same.
22339 * doc/md.texi: Same.
22340 * doc/objc.texi: Same.
22341 * doc/rtl.texi: Same.
22342 * doc/tm.texi: Same.
22343 * doc/tm.texi.in: Same.
22344 * doc/trouble.texi: Same.
22346 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22349 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22350 (collect_checksum_attributes): Set it.
22351 (die_checksum_ordered): Use it.
22353 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22355 PR tree-optimization/79908
22356 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22357 change: For a VA_ARG whose LHS has been cast away, use
22358 force_gimple_operand to construct the side effects.
22360 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22362 PR translation/80001
22363 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22364 more amenable to translation.
22365 (oacc_loop_auto_partitions): Likewise.
22367 2017-03-21 Marek Polacek <polacek@redhat.com>
22368 Martin Sebor <msebor@redhat.com>
22370 PR tree-optimization/80109
22371 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22372 on INTEGRAL_TYPE_P.
22374 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22375 Segher Boessenkool <segher@kernel.crashing.org>
22378 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22379 check reg_used_between_p between insn and one of succ or succ2
22380 depending on if succ is artificial insn not inserted into insn
22383 2017-03-21 Martin Liska <mliska@suse.cz>
22385 PR gcov-profile/80081
22386 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22387 * doc/gcc.texi: Include gcov-dump stuff.
22388 * doc/gcov-dump.texi: New file.
22390 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22392 PR rtl-optimization/79150
22393 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22394 conditional jump, if the jump is the last insn of the loop.
22396 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22397 Richard Biener <rguenth@suse.de>
22399 PR tree-optimization/79908
22400 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22401 been cast away, use force_gimple_operand to construct the side
22404 2017-03-21 Martin Liska <mliska@suse.cz>
22406 PR libfortran/79956
22407 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22410 2017-03-21 Brad Spengler <spender@grsecurity.net>
22413 * plugin.c (htab_hash_plugin): New function.
22414 (add_new_plugin): Use it and adjust.
22415 (parse_plugin_arg_opt): Adjust.
22416 (init_one_plugin): Likewise.
22418 2017-03-21 Richard Biener <rguenther@suse.de>
22420 PR tree-optimization/80032
22421 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22422 if set force the cleanup to happen unconditionally.
22423 (gimplify_target_expr): Push inserted clobbers with force_uncond
22424 to avoid them being removed by control-dependent DCE.
22426 2017-03-21 Richard Biener <rguenther@suse.de>
22428 PR tree-optimization/80122
22429 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22430 va_arg_pack_len when the inlined call stmt requires pack
22432 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22434 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22437 * tsan.c (instrument_builtin_call): If the memory model argument
22438 is not a constant, assume it is valid.
22441 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22444 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22446 PR rtl-optimization/79910
22447 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22448 if its dest is used by an insn before I2 (other than the combined
22449 insns themselves, which are properly handled already).
22451 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22454 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22456 * combine.c (record_used_regs): New static function.
22457 (try_combine): Handle situations where there is an additional
22458 instruction between I2 and I3 which needs to have a LOG_LINK
22462 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22464 * combine.c (try_combine): Delete redundant i1 test. Call
22465 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22467 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22470 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22471 alternatives 13/14.
22473 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22475 PR tree-optimization/80054
22476 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22477 the optimization if a PHI or any of its arguments is not dominated
22478 by the candidate's basis. Use gphi* rather than gimple* as
22480 (replace_profitable_candidates): Clean up a gimple* variable that
22481 should be a gphi* variable.
22483 2017-03-20 Martin Sebor <msebor@redhat.com>
22486 * doc/extend.texi (attribute constructor): Document present limitation.
22488 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22491 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22492 __POWER9_VECTOR__ #ifdef control, change template definition to
22493 use Power9-specific built-in function.
22494 (vec_any_eq): Likewise.
22495 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22496 to control outcomes from this test.
22497 (vector_ae_<mode>p): For VEC_F modes, likewise.
22499 2017-03-20 Ian Lance Taylor <iant@google.com>
22501 * config/i386/i386.c (ix86_function_regparm): Save an extra
22502 register for -fsplit-stack with DECL_STATIC_CHAIN.
22504 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22507 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22508 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22510 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22512 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22514 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22517 2017-03-20 Marek Polacek <polacek@redhat.com>
22520 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22522 2017-03-20 Richard Biener <rguenther@suse.de>
22524 PR tree-optimization/80113
22525 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22526 allocate extra SSA name for PHI def.
22527 (add_close_phis_to_outer_loops): Likewise.
22528 (add_close_phis_to_merge_points): Likewise.
22529 (copy_loop_close_phi_args): Likewise.
22530 (copy_cond_phi_nodes): Likewise.
22532 2017-03-20 Martin Liska <mliska@suse.cz>
22534 PR middle-end/79753
22535 * tree-chkp.c (chkp_build_returned_bound): Do not build
22536 returned bounds for a LHS that's not a BOUNDED_P type.
22538 2017-03-20 Martin Liska <mliska@suse.cz>
22542 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22543 COMPLEX_CST and VECTOR_CST.
22545 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22548 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22549 target operand. A new splitter adds the clobber statement in case
22550 the target operand is dead anyway.
22552 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22554 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22555 to age-old versions of binutils and glibc.
22557 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22559 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22561 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22563 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22565 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22567 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22568 requirement for binutils 2.13.
22570 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22572 * combine.c (try_combine): Delete redundant i1 test. Call
22573 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22575 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22577 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22578 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22580 <riscv64-*-elf>: Re-arrange section
22581 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22582 <riscv32-*-linux>: Likewise.
22583 <riscv64-*-elf>: Likewise
22584 <riscv64-*-linux>: Likewise.
22586 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22589 * aarch64.opt(verbose-cost-dump): Fix typo.
22591 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22594 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22595 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22597 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22599 * reload.c (find_reloads): When reloading a nonoffsettable address,
22600 use RELOAD_OTHER for it and its address reloads.
22602 PR rtl-optimization/79910
22603 * combine.c (record_used_regs): New static function.
22604 (try_combine): Handle situations where there is an additional
22605 instruction between I2 and I3 which needs to have a LOG_LINK
22608 2017-03-17 Jeff Law <law@redhat.com>
22610 PR tree-optimization/71437
22611 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22612 conditional in the hash table first.
22613 (vrp_dom_walker::before_dom_children): Extract condition from
22614 ASSERT_EXPR. Record condition, its inverion and any implied
22615 conditions as well.
22617 2017-03-17 Marek Polacek <polacek@redhat.com>
22618 Markus Trippelsdorf <markus@trippelsdorf.de>
22620 PR tree-optimization/80079
22621 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22624 2017-03-17 Richard Biener <rguenther@suse.de>
22626 PR middle-end/80075
22627 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22628 Properly verify the LHS before the RHS possibly claims to be
22630 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22633 2017-03-17 Martin Jambor <mjambor@suse.cz>
22635 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22636 (List of -O2 options): Likewise.
22637 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22640 2017-03-17 Tom de Vries <tom@codesourcery.com>
22642 * gcov-dump.c (print_usage): Print bug_report_url.
22644 2017-03-17 Richard Biener <rguenther@suse.de>
22646 PR middle-end/80050
22647 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22648 (parser::peek): Likewise.
22650 2017-03-17 Richard Biener <rguenther@suse.de>
22652 PR tree-optimization/80048
22653 * sese.c (free_sese_info): Properly release rename_map and
22654 copied_bb_map elements.
22656 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22658 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22659 Add linked-list forward and backlinks. Insert on
22660 construction, remove on destruction.
22661 (class pass_store_merging): Add m_stores_head field.
22662 (pass_store_merging::terminate_and_process_all_chains):
22663 Iterate over m_stores_head list.
22664 (pass_store_merging::terminate_all_aliasing_chains):
22666 (pass_store_merging::execute): Check for debug stmts first.
22667 Push new chains onto the m_stores_head stack.
22669 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22672 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22673 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22674 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22676 2017-03-16 Jeff Law <law@redhat.com>
22678 PR tree-optimization/71437
22679 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22680 member function. Implementation moved into after_dom_children
22681 member function and into the threader's thread_outgoing_edges
22683 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22684 some code into new thread_outgoing_edges.
22685 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22686 definition. Simplify marker handling (do it here). Assume we always
22687 have the available expression and the const/copies tables.
22688 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22690 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22691 * tree-vrp.c (equiv_stack): No longer file scoped.
22692 (vrp_dom_walker): New class.
22693 (vrp_dom_walker::before_dom_children): New member function.
22694 (vrp_dom_walker::after_dom_children): Likewise.
22695 (identify_jump_threads): Setup domwalker. Use it rather than
22696 walking edges in a random order by hand. Simplify setup/finalization.
22697 (finalize_jump_threads): Remove.
22698 (vrp_finalize): Do not call identify_jump_threads here.
22699 (execute_vrp): Do it here instead and call thread_through_all_blocks
22702 PR tree-optimization/71437
22703 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22705 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22707 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22708 (dom_opt_dom_walker::thread_across_edge): Remove
22709 handle_dominating_asserts argument. All callers changed.
22710 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22711 changes. Remove calls to lhs_of_dominating_assert. Other
22712 uses of handle_dominating_asserts turn into unconditional code
22713 (simplify_control_stmt_condition_1): Likewise.
22714 (simplify_control_stmt_condition): Likewise.
22715 (thread_through_normal_block, thread_across_edge): Likewise.
22716 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22717 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22718 object if it is not an SSA_NAME.
22719 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22720 before calling into the VRP specific simplifiers.
22721 (identify_jump_threads): Remove handle_dominating_asserts
22724 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22727 * tree-diagnostic.c (default_tree_printer): No longer static.
22728 * tree-diagnostic.h (default_tree_printer): New prototype.
22730 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22732 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22733 Change ins into fmov.
22735 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22737 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22738 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22739 Use h_con constraint for operand 1.
22740 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22741 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22743 2017-03-15 Jeff Law <law@redhat.com>
22745 PR tree-optimization/71437
22746 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22747 (record_temporary_equivalences): Use it.
22749 PR tree-optimization/71437
22750 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22751 tree-ssa-scopedtables.
22752 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22753 (record_conditions, record_cond, vuse_eq): Likewise.
22754 (record_edge_info): Adjust to API tweak of record_conditions.
22755 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22756 (record_temporary_equivalences, optimize_stmt): Likewise.
22757 (eliminate_redundant_computations): Likewise.
22758 (record_equivalences_from_stmt): Likewise.
22759 * tree-ssa-scopedtables.c: Include options.h and params.h.
22760 (vuse_eq): New function, moved from tree-ssa-dom.c
22761 (build_and_record_new_cond): Likewise.
22762 (record_conditions): Likewise. Accept vector of conditions rather
22763 than edge_equivalence structure for first argument.
22764 for the first argument.
22765 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22766 from tree-ssa-dom.c.
22767 (avail_exprs_stack::record_cond): Likewise.
22768 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22769 from tree-ssa-dom.c.
22770 (avail_exprs_stack): Add new member functions lookup_avail_expr
22772 (record_conditions): Declare.
22774 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22777 * lra-constraints.c (process_alt_operands): Increase reject for
22778 reloading an input/output operand.
22780 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22783 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22784 insns to convert from signed/unsigned char/short to IEEE 128-bit
22786 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22788 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22791 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22792 subreg of inner mode for values already in registers.
22794 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22796 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22797 iteration reg is used after the loop.
22799 2017-03-14 Martin Sebor <msebor@redhat.com>
22801 PR tree-optimization/79800
22802 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22803 precision in negative-positive range.
22804 (format_floating): Call non-const overload with adjusted precision.
22806 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22809 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22812 2017-03-14 Martin Sebor <msebor@redhat.com>
22814 PR middle-end/80020
22815 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22816 * builtins.def (aligned_alloc): Use it.
22819 * Makefile.in (GTFILES): Add calls.c.
22820 * calls.c: Include "gt-calls.h".
22822 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22824 PR rtl-optimization/79728
22825 * regs.h (struct target_regs): New field
22826 x_contains_allocatable_regs_of_mode.
22827 (contains_allocatable_regs_of_mode): New macro.
22828 * reginfo.c (init_reg_sets_1): Initialize it, and change
22829 contains_reg_of_mode so it includes global regs as well.
22830 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22831 rather than contains_regs_of_mode.
22833 2017-03-14 Martin Liska <mliska@suse.cz>
22835 * doc/invoke.texi: Document options that can't be combined with
22836 -fcheck-pointer-bounds.
22838 2017-03-14 Martin Liska <mliska@suse.cz>
22840 PR middle-end/79831
22841 * doc/invoke.texi (-Wchkp): Document the option.
22843 2017-03-14 Martin Liska <mliska@suse.cz>
22845 * Makefile.in: Install gcov-dump.
22847 2017-03-14 Martin Liska <mliska@suse.cz>
22849 * multiple_target.c (expand_target_clones): Bail out for
22850 an invalid attribute.
22852 2017-03-14 Richard Biener <rguenther@suse.de>
22854 * alias.c (struct alias_set_entry): Pack properly.
22855 * cfgloop.h (struct loop): Likewise.
22856 * cse.c (struct set): Likewise.
22857 * ipa-utils.c (struct searchc_env): Likewise.
22858 * loop-invariant.c (struct invariant): Likewise.
22859 * lra-remat.c (struct cand): Likewise.
22860 * recog.c (struct change_t): Likewise.
22861 * rtl.h (struct address_info): Likewise.
22862 * symbol-summary.h (function_summary): Likewise.
22863 * tree-loop-distribution.c (struct partition): Likewise.
22864 * tree-object-size.c (struct object_size_info): Likewise.
22865 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22866 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22867 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22868 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22869 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22870 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22871 (struct _stmt_vec_info): Likewise.
22873 2017-03-14 Martin Liska <mliska@suse.cz>
22876 * multiple_target.c (create_dispatcher_calls): Check that
22877 a target can create a function dispatcher.
22879 2017-03-14 Martin Liska <mliska@suse.cz>
22882 * multiple_target.c (expand_target_clones): Drop local.local
22883 flag for default implementation.
22885 2017-03-14 Richard Biener <rguenther@suse.de>
22887 PR tree-optimization/80030
22888 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22890 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22892 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22893 gcc_fallthrough() instead of __attribute__((fallthrough));
22895 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22897 * doc/gcc.texi: Remove "up" link to (DIR).
22898 * doc/gccint.texi: Ditto.
22900 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22902 * doc/install.texi (Specific) <avr>: Remove reference to
22905 2017-03-13 Jeff Law <law@redhat.com>
22907 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22908 attribute rather than comments.
22910 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22911 match_scratch operand is highest.
22913 2017-03-13 Martin Liska <mliska@suse.cz>
22915 PR middle-end/78339
22916 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22917 is a CHKP clone, use original declaration.
22919 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22921 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22922 (arc_conditional_register_usage): Use a different allocation order
22923 when optimizing for size.
22924 * common/config/arc/arc-common.c (arc_option_optimization_table):
22925 Section anchors default on when optimizing for size.
22927 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22929 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22931 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22933 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22934 * config/arc/arc.md (cpu_facility): Add cd variant.
22935 (*movqi_insn): Add code density variant.
22936 (*movhi_insn): Likewise.
22937 (*movqi_insn): Likewise.
22938 (*addsi3_mixed): Likewise.
22939 (subsi3_insn): Likewise.
22941 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22943 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22945 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22947 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22948 expressions with MINUS and UNARY ops.
22950 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22953 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22955 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22956 between vec_select and vector argument.
22957 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22958 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22959 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22960 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22961 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22962 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22964 2017-03-13 Richard Biener <rguenther@suse.de>
22967 * params.def (vect-max-peeling-for-alignment): Fix typo.
22969 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22971 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22972 issue that only occurred with binutils below 2.18.
22974 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22976 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22977 refer to binutils 2.11/2.12 minimum.
22979 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22981 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22982 ftp.kernel.org and simplify binutils requirement.
22984 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22986 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22988 (Optimize Options): Ditto. Also remove redundancy.
22990 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22992 PR translation/79848
22993 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22995 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22996 to G_ to avoid double translation.
22998 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23000 PR translation/79923
23001 * auto-profile.c (get_combined_location): Convert leading
23002 character of diagnostics to lower case and remove trailing period.
23003 (read_profile): Likewise for various diagnostics.
23004 * config/arm/arm.c (arm_option_override): Remove trailing period
23005 from various diagnostics.
23006 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23007 (msp430_expand_delay_cycles): Likewise.
23009 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23012 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23013 full command-line argument, rather than just "str".
23014 (aarch64_validate_march): Likewise.
23015 (aarch64_validate_mtune): Likewise.
23017 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23019 PR rtl-optimization/78911
23020 * lra-assigns.c (must_not_spill_p): New function.
23021 (spill_for): Use it.
23023 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23025 PR tree-optimization/79981
23026 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23027 ATOMIC_COMPARE_EXCHANGE ifn result.
23028 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23029 IFN_ATOMIC_COMPARE_EXCHANGE.
23031 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23034 * opts.c (parse_sanitizer_options): Add missing question mark to
23035 "did you mean" message.
23037 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23039 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23041 (VMULEUH_UNS): Likewise.
23042 (VMULOUB_UNS): Likewise.
23043 (VMULOUH_UNS): Likewise.
23044 * config/rs6000/rs6000.c (builtin_function_type): Remove
23045 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23047 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23050 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23051 x with result of extra_parsing_for_operand_code_0.
23052 (function_reader::extra_parsing_for_operand_code_0): Convert
23053 return type from void to rtx, returning x. When reading
23054 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23055 larger size containing struct block_symbol.
23057 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23059 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23060 -mfloat128-hardware without -m64.
23062 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23065 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23066 entries to the case statement that marks unsigned arguments to
23067 overloaded functions.
23069 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23071 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23072 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23074 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23077 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23078 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23080 2017-03-10 Martin Liska <mliska@suse.cz>
23084 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23085 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23088 2017-03-10 Olivier Hainque <hainque@adacore.com>
23090 * tree-switch-conversion (array_value_type): Start by resetting
23091 candidate type to it's main variant.
23093 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23095 PR rtl-optimization/79909
23096 * combine.c (try_combine): Use simplify_replace_rtx on individual
23097 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23098 of the whole CALL_INSN_FUNCTION_USAGE.
23100 PR tree-optimization/79972
23101 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23102 get_range_info on SSA_NAMEs. Formatting fixes.
23104 2017-03-10 Richard Biener <rguenther@suse.de>
23105 Jakub Jelinek <jakub@redhat.com>
23107 PR tree-optimization/77975
23108 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23109 edge to be constant.
23110 (get_val_for): For constant x return it. Formatting fix.
23111 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23112 would use the same bases as the current one.
23114 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23116 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23117 instead of vec_select for V1TImode.
23118 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23120 (VSX_LE_128): Add V1TI to this mode iterator.
23121 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23122 (*vsx_le_perm_store_<mode>): Likewise.
23123 (pre-reload splitter for VSX stores): Likewise.
23124 (post-reload splitter for VSX stores): Likewise.
23125 (*vsx_xxpermdi2_le_<mode>): Likewise.
23126 (*vsx_lxvd2x2_le_<mode>): Likewise.
23127 (*vsx_stxvd2x2_le_<mode>): Likewise.
23129 2017-03-09 Michael Eager <eager@eagercon.com>
23131 Correct failures with --enable-checking=yes,rtl.
23133 * config/microblaze/microblaze.c (microblaze_expand_shift):
23134 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23135 test for const0_rtx.
23136 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23137 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23139 2017-03-09 Richard Biener <rguenther@suse.de>
23141 PR tree-optimization/79977
23142 * graphite-scop-detection.c (scop_detection::merge_sese):
23143 Handle the case of extra exits to blocks dominating the entry.
23145 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23147 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23150 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23152 PR rtl-optimization/79949
23153 * lra-constraints.c (process_alt_operands): Check memory when
23154 trying to predict a cycle. Print about the overall increase.
23156 2017-03-09 Richard Biener <rguenther@suse.de>
23158 PR middle-end/79971
23159 * gimple-expr.c (useless_type_conversion_p): Preserve
23160 TYPE_SATURATING for fixed-point types.
23162 2017-03-09 Richard Biener <rguenther@suse.de>
23165 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23166 alignment of BLKmode params.
23168 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23171 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23172 (VALL_NO_V2Q): Likewise.
23174 * config/aarch64/aarch64-simd.md
23175 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23177 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23178 VALL_NO_V2Q mode iterator.
23179 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23181 2017-03-09 Martin Liska <mliska@suse.cz>
23183 PR tree-optimization/79631
23184 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23185 tree_int_cst_sign_bit just for INTEGER constants.
23187 2017-03-09 Martin Liska <mliska@suse.cz>
23191 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23194 2017-03-09 Marek Polacek <polacek@redhat.com>
23197 * tree.c (inchash::add_expr): Handle TREE_VEC.
23199 2017-03-09 Martin Liska <mliska@suse.cz>
23202 (chkp_narrow_size_and_offset): New function.
23203 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23204 (void chkp_parse_bit_field_ref): New function.
23205 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23206 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23208 2017-03-09 Martin Liska <mliska@suse.cz>
23211 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23212 (chkp_find_bounds_1): Remove gcc_unreachable.
23214 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23217 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23218 BUILT_IN_SYNC*, determine the access type from the size suffix and
23219 always build a MEM_REF with that type. Handle forgotten
23220 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23223 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23224 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23225 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23226 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23227 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23228 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23229 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23230 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23231 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23232 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23233 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23234 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23235 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23236 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23237 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23238 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23239 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23240 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23241 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23242 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23243 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23244 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23245 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23246 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23247 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23248 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23249 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23250 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23251 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23252 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23253 definitions outside of __OPTIMIZE__ guarded section.
23256 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23257 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23258 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23259 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23262 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23264 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23265 ("vfenez<mode>"): Add missing constraints.
23267 2017-03-08 Martin Sebor <msebor@redhat.com>
23270 * config/nds32/nds32.c (nds32_option_override):
23271 Fix misspelled diagnostic.
23273 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23276 * gimplify.c (gimplify_omp_for): Replace index var in outer
23277 taskloop statement with an artificial variable and add
23278 OMP_CLAUSE_PRIVATE clause for it.
23280 2017-03-08 Richard Biener <rguenther@suse.de>
23282 PR tree-optimization/79955
23283 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23284 for accesses that are completely outside of the variable.
23286 2017-03-08 Andrew Haley <aph@redhat.com>
23288 PR tree-optimization/79943
23289 * tree-ssa-loop-split.c (compute_new_first_bound): When
23290 calculating the new upper bound, (END-BEG) should be added, not
23293 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23295 * config/avr/avr.md (setmemhi): Make sure match_dup
23296 operand number comes before match_scratch.
23298 2017-03-08 Richard Biener <rguenther@suse.de>
23300 PR tree-optimization/79920
23301 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23302 with ncopies == 1 to ...
23303 (vect_transform_slp_perm_load): ... here. Properly compute
23304 all element loads by iterating VF times over the group. Do
23305 not handle ncopies (computed in a broken way) in
23306 vect_create_mask_and_perm.
23308 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23311 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23312 is a uniform vector, use uniform_vector_p return value instead of
23313 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23315 2017-03-07 Marek Polacek <polacek@redhat.com>
23317 PR middle-end/79809
23318 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23319 (alloca_call_type): Likewise.
23321 2017-03-07 Martin Liska <mliska@suse.cz>
23323 * gcov.c (process_args): Put comment to correct location.
23325 2017-03-07 Martin Liska <mliska@suse.cz>
23327 PR middle-end/68270
23328 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23329 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23330 (chkp_narrow_bounds_for_field): Likewise.
23331 (chkp_parse_array_and_component_ref): Pass one more argument to
23334 2017-03-07 Richard Biener <rguenther@suse.de>
23336 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23339 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23341 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23342 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23344 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23347 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23348 to end of description.
23349 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23351 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23353 PR rtl-optimization/79901
23354 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23356 (*avx512f_<code><mode>3<mask_name>): ... this.
23357 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23358 iterator instead of VI8_AVX2_AVX512BW.
23360 PR rtl-optimization/79901
23361 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23362 min/max expander, expand it using expand_vec_cond_expr.
23365 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23368 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23371 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23372 to void * for PCH reasons.
23373 * dwarf2out.c (output_loc_operands, output_die): Cast
23374 v.val_vec.array to unsigned char *.
23376 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23379 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23382 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23384 PR rtl-optimization/79571
23385 * lra-constraints.c (process_alt_operands): Calculate static
23386 reject and subtract it from overall when only addresses will be
23389 2017-03-06 Julia Koval <julia.koval@intel.com>
23392 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23393 incoming stack boundary to 128 for 64-bit targets.
23395 2017-03-06 Richard Biener <rguenther@suse.de>
23397 PR tree-optimization/79894
23398 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23399 to NULL after folding it.
23401 2017-03-06 Richard Biener <rguenther@suse.de>
23403 PR tree-optimization/79824
23404 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23405 check disabling peeling for gaps.
23407 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23409 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23410 attributes): Document gettimeofday.
23412 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23414 * config/s390/s390.c (s390_option_override_internal): Set
23415 PARAM_MIN_VECT_LOOP_BOUND
23417 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23419 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23420 * config/s390/s390.md: Likewise.
23422 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23425 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23426 (<avx2_avx512>_perm<mode>): Rename to ...
23427 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23429 (<avx512>_perm<mode>_mask): Rename to ...
23430 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23432 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23433 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23434 instead of VI8F_256_512.
23435 (avx512f_perm<mode>): New define_expand.
23436 (avx512f_perm<mode>_mask): Likewise.
23437 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23438 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23440 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23442 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23443 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23445 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23447 2017-03-06 Martin Liska <mliska@suse.cz>
23450 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23451 when having a SSA NAME w/o VAR_DECL assigned to it.
23453 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23455 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23456 msa_dpsub_<su>_d): Fix MODE for vec_select.
23458 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23460 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23462 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23464 2017-03-06 Richard Biener <rguenther@suse.de>
23466 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23467 * plugin.c (register_plugin_info): Likewise.
23468 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23470 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23472 * config/i386/sse.md (sse_storehps, sse_storelps,
23473 avx_<castmode><avxsizesuffix>_<castmode>,
23474 avx512f_<castmode><avxsizesuffix>_<castmode>,
23475 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23476 in condition that at least one operand is not a MEM.
23478 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23480 PR middle-end/79805
23481 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23482 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23484 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23485 gimple_call_nothrow_p flag based on whether original builtin can throw.
23486 If it can, emit following stmts on the fallthrough edge.
23487 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23488 don't create new bb if inserting just debug stmts on the edge, try to
23489 insert them on the fallthru bb or just reset debug stmts.
23491 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23494 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23495 restore recog_data (including the operand rtxes inside it) around
23496 the call to get_insn_template.
23498 2017-03-03 Martin Sebor <msebor@redhat.com>
23500 PR tree-optimization/79699
23501 * context.c (context::~context): Free MPFR caches to avoid
23502 a memory leak on program exit.
23504 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23506 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23507 Use wide_int::ulow () instead of .elt (0).
23509 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23511 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23512 (*pushxf): Limit oF constraint to 32bit targets and add oC
23513 constraint for 64bit targets.
23514 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23515 (*pushdf): Change rmF constraint to rmC.
23517 2017-03-03 Martin Liska <mliska@suse.cz>
23519 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23520 Remove unused variable.
23522 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23525 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23526 is a memory operand, increase num_memory.
23527 (ix86_expand_args_builtin): Likewise.
23529 2017-03-03 Jan Hubicka <jh@suse.cz>
23532 * ipa-devirt.c (maybe_record_node): Properly handle
23533 __cxa_pure_virtual visibility.
23535 2017-03-03 Martin Liska <mliska@suse.cz>
23537 PR tree-optimization/79803
23538 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23540 (pass_loop_prefetch::execute): Disabled optimization if an
23541 assumption about L1 cache size is not met.
23543 2017-03-03 Martin Liska <mliska@suse.cz>
23545 PR rtl-optimization/79574
23546 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23547 (hash_scan_set): Likewise.
23548 (dump_hash_table): Likewise.
23549 (hoist_code): Likewise.
23551 2017-03-03 Richard Biener <rguenther@suse.de>
23553 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23554 in place of uhigh ().
23555 (fixed_convert_from_real): Likewise.
23557 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23560 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23562 2017-03-03 Richard Biener <rguenther@suse.de>
23564 PR middle-end/79818
23565 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23566 TYPE_OVERFLOW_UNDEFINED check.
23568 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23570 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23572 (vector_ae_<mode>_p): Likewise.
23573 (vector_nez_<mode>_p): Likewise.
23574 (vector_ne_v2di_p): Likewise.
23575 (vector_ae_v2di_p): Likewise.
23576 (vector_ne_<mode>_p): Likewise.
23577 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23579 (vsx_tsqrt<mode>2_fe): Likewise.
23581 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23584 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23586 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23588 PR rtl-optimization/79780
23589 * cprop.c (one_cprop_pass): When second and further conditional trap
23590 in a single basic block is turned into an unconditional trap, turn it
23591 into a deleted note to avoid RTL verification failures.
23593 2017-03-02 Richard Biener <rguenther@suse.de>
23595 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23597 2017-03-02 Richard Biener <rguenther@suse.de>
23599 PR tree-optimization/79345
23601 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23602 param and abort the walk, returning -1 if it is hit.
23603 (walk_aliased_vdefs): Take a limit param and pass it on.
23604 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23605 defaulting to 0 and return a signed int.
23606 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23607 (check_defs): New helper.
23608 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23609 about uninitialized memory.
23610 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23611 bogus uninitialized warning.
23612 (fixed_convert_from_real): Likewise.
23614 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23616 PR tree-optimization/66768
23617 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23618 iv_use if base object can't be determined.
23620 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23622 PR tree-optimization/79345
23623 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23624 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23625 (get_pattern_stats): Initialize it.
23626 * genemit.c (gen_expand): Verify match_scratch numbers come after
23627 match_operand/match_dup numbers.
23628 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23629 match_scratch numbers.
23630 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23632 * config/s390/s390.md (trunctdsd2): Likewise.
23634 2017-03-02 Richard Biener <rguenther@suse.de>
23636 * wide-int.h (wide_int_storage::operator=): Implement in terms
23639 2017-03-02 Richard Biener <rguenther@suse.de>
23641 PR tree-optimization/79777
23642 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23643 the to insert expression to sth existing.
23645 2017-03-01 Martin Sebor <msebor@redhat.com>
23647 PR middle-end/79692
23648 * gimple-ssa-sprintf.c
23649 (directive::known_width_and_precision): New function.
23650 (format_integer): Use it.
23651 (get_mpfr_format_length): Consider the full range of precision
23652 when computing %g output with the # flag. Set the likely byte
23653 count to 3 rather than 1 when precision is indeterminate.
23654 (format_floating): Correct the lower bound of precision.
23656 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23658 * doc/invoke.texi: Document default code model for 64-bit Linux.
23660 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23663 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23664 udiv rather than div since input pattern is unsigned.
23666 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23668 * config/i386/i386.c (print_reg): Warn for values of
23669 unsupported size in integer register.
23671 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23674 * config/rs6000/predicates.md (current_file_function_operand): Do
23675 not allow self calls to be local if the function is replaceable.
23677 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23680 * config/rs6000/altivec.h (vec_ctz and others): Change the
23681 preprocessor macro that controls conditional compilation from
23682 _ARCH_PWR9 to __POWER9_VECTOR__.
23683 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23684 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23685 control (instead of _ARCH_PWR9 control) so that template
23686 definition uses power9-specific function.
23687 (vec_any_eq): Likewise.
23688 (vec_all_ne): Change macro definition to use a power9-specific
23689 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23690 _ARCH_PWR9 control).
23691 (vec_any_eq) Likewise.
23692 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23693 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23694 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23695 support for xvcmpnedp instruction.
23696 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23697 macro expansion so that Power9 implementation of vec_all_ne does
23698 not use the AltiVec predicate framework.
23699 (VCMPNEH_P): Likewise.
23700 (VCMPNEW_P): Likewise.
23701 (VCMPNED_P): Likewise.
23702 (VCMPNEFP_P): Likewise.
23703 (VCMPNEDP_P): Likewise.
23704 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23705 implementation of vec_any_eq to not use AltiVec predicate
23707 (VCMPAEH_P): Likewise.
23708 (VCMPAEW_P): Likewise.
23709 (VCMPAED_P): Likewise.
23710 (VCMPAEFP_P): Likewise.
23711 (VCMPAEDP_P): Likewise.
23712 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23713 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23714 not use the AltiVec predicate framework.
23715 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23716 of vec_any_eq to not use AltiVec predicate framework.
23717 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23718 support for predefined __POWER9_VECTOR__ macro to indicate that
23719 Power9 instruction selection is enabled.
23720 (altivec_overloaded_builtins): Remove extraneous
23721 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23722 function argument types RS6000_BTI_bool_V16QI and
23723 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23724 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23725 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23726 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23727 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23728 Power9 for implementations of vec_cmpne. Change the signature for
23729 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23730 (representing vec_all_ne) to remove the previously described first
23731 argument of type RS6000_BTI_INTSI, as this was an artifact of
23732 reliance on the AltiVec predicate framework, which is no longer
23733 used in the implementation of these functions. Add
23734 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23735 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23736 since, unlike the AltiVec predicate framework implementation, we
23737 do not share function descriptors between vec_alle and vec_anyeq.
23738 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23739 set of modes that receive special treatment even when
23740 TARGET_P9_VECTOR is true. The special treatment emits code that
23741 does not depend on Power9 instructions.
23742 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23743 define_expand to not rely on AltiVec predicate framework.
23744 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23746 (vector_ne_v2di_p): Change this define_expand to not rely on
23747 AltiVec predicate framework.
23748 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23750 (vector_ne_<mode>_p): Change this define_expand to not rely on
23751 AltiVec predicate framework.
23752 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23754 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23755 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23756 define_insn pattern.
23757 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23758 define_insn pattern because the xvcmpne<VSs>. instruction is not
23760 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23761 instruction is not supported.
23763 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23765 * config/nvptx/nvptx.c: Include intl.h.
23767 2017-03-01 Martin Jambor <mjambor@suse.cz>
23770 * ipa-prop.h (ipa_bits): Removed field known.
23771 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23772 to pointers. Adjusted their comments to warn about their sharing.
23773 (ipcp_transformation_summary): Change bits to a vector of pointers.
23774 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23775 (ipa_get_ipa_bits_for_value): Declare.
23776 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23777 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23778 (ipa_bits_hash_table): Likewise.
23779 (ipa_vr_ggc_hash_traits): Likewise.
23780 (ipa_vr_hash_table): Likewise.
23781 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23782 being pointers and vr_known being removed.
23783 (ipa_set_jf_unknown): Likewise.
23784 (ipa_get_ipa_bits_for_value): New function.
23785 (ipa_set_jfunc_bits): Likewise.
23786 (ipa_get_value_range): New overloaded functions.
23787 (ipa_set_jfunc_vr): Likewise.
23788 (ipa_compute_jump_functions_for_edge): Use the above functions to
23789 construct bits and vr parts of jump functions.
23790 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23791 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23793 (ipcp_grow_transformations_if_necessary): Also allocate
23794 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23796 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23797 them. Fix too long lines.
23798 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23799 vr_known being removed.
23800 (ipa_read_jump_function): Use new setter functions to construct bits
23801 and vr parts of jump functions or set them to NULL.
23802 (write_ipcp_transformation_info): Adjust for bits being pointers.
23803 (read_ipcp_transformation_info): Likewise.
23804 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23806 Include gt-ipa-prop.h.
23807 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23809 (ipcp_store_bits_results): Likewise.
23810 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23811 Do not write to existing jump functions but use a temporary instead.
23813 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23816 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23817 attempt to use its first operand as BIT_FIELD_REF base.
23819 2017-03-01 Richard Biener <rguenther@suse.de>
23821 PR middle-end/79721
23822 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23823 interpolating formula in wrapping arithmetic.
23824 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23826 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23828 PR tree-optimization/79734
23829 * tree-vect-generic.c (expand_vector_condition): Optimize
23830 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23831 Handle VEC_COND_EXPR where comparison has different inner width from
23832 type's inner width.
23834 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23836 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23837 markup, and similar issues. Remove @opindex entries for things
23838 that aren't options. Add missing -mmpy-option entries.
23840 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23842 PR tree-optimization/79737
23843 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23844 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23845 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23846 instead of byte_size. Formatting fix.
23847 (shift_bytes_in_array_right): Formatting fix.
23849 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23852 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23853 condition on optimize for the leaf function test.
23855 2017-02-28 Martin Liska <mliska@suse.cz>
23858 * read-rtl-function.c (function_reader::handle_unknown_directive):
23859 Bail out when one uses -flto.
23861 2017-02-28 Martin Liska <mliska@suse.cz>
23863 * common.opt: Replace space with tabular for options of <number>
23865 * config/i386/i386.opt: Show <number> value for
23866 -mlarge-data-threshold.
23867 * opts.c (print_filtered_help): Do not display number in hexadecimal
23870 2017-02-28 Martin Liska <mliska@suse.cz>
23872 * common.opt: Fix --help=option -Q for options which are of
23875 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23877 * config/i386/i386.c (print_reg): Error out for values
23878 of 8-bit size in invalid integer register.
23880 2017-02-28 Martin Sebor <msebor@redhat.com>
23882 PR tree-optimization/79691
23883 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23885 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23888 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23889 gcc_unreachable with output_operand_lossage.
23891 2017-02-28 Richard Biener <rguenther@suse.de>
23893 PR tree-optimization/79740
23894 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23896 (visit_nary_op): Insert the nary into the hashtable if we
23897 pattern-matched sth.
23898 * tree-ssa-pre.c (eliminate_insert): Robustify.
23900 2017-02-28 Richard Biener <rguenther@suse.de>
23902 PR middle-end/79731
23903 * fold-const.c (decode_field_reference): Reject out-of-bound
23906 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23908 * config/i386/i386.c: Include intl.h.
23909 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23910 instead of just cond ? "..." : "...".
23911 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23912 * coverage.c (read_counts_file): Likewise.
23913 * omp-offload.c: Include intl.h.
23914 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23915 of just cond ? "..." : "...".
23916 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23917 of just cond ? "..." : "...".
23919 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23922 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23924 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23925 'tune for' CPU name.
23926 * config/arm/arm-cpu-data.h: Regenerated.
23928 2017-02-28 Richard Biener <rguenther@suse.de>
23930 PR tree-optimization/79732
23931 * tree-inline.c (expand_call_inline): Do not shadow var.
23933 2017-02-28 Richard Biener <rguenther@suse.de>
23935 PR tree-optimization/79723
23936 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23937 address-space properly.
23939 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23941 * doc/optinfo.texi (Optimization groups): Fix option used for
23943 * doc/invoke.texi (-fopt-info): Document "omp".
23944 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23945 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23946 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23947 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23948 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23950 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23952 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23953 OPTGROUP_OMP with "omp".
23955 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23958 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23959 for arithmetic shift of unsigned V2DI.
23961 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23963 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23964 arc/linux.h headers.
23965 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23966 (LINK_SPEC): Likewise.
23967 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23968 (EXTRA_SPECS): Likewise.
23969 (STARTFILE_SPEC): Likewise.
23970 (ENDFILE_SPEC): Likewise.
23971 (LIB_SPEC): Likewise.
23972 (TARGET_SDATA_DEFAULT): Likewise.
23973 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23974 (MULTILIB_DEFAULTS): Likewise.
23975 (DWARF2_UNWIND_INFO): Likewise.
23976 * config/arc/big.h: New file.
23977 * config/arc/elf.h: Likewise.
23978 * config/arc/linux.h: Likewise.
23979 * config/arc/t-uClibc: Remove.
23981 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23983 PR tree-optimization/77536
23984 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23985 (tree_transform_and_unroll_loop): Use above function to compute the
23986 estimated niter of unrolled loop and use it when scaling profile.
23987 Also use count info rather than frequency if it's non-zero.
23988 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23989 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23990 (vect_transform_loop): Call above function.
23992 2017-02-27 Richard Biener <rguenther@suse.de>
23994 PR tree-optimization/45397
23995 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23996 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23997 (visit_nary_op): Add pattern matching for CSEing sign-changed
23998 or truncated operations with wider ones.
24000 2017-02-27 Richard Biener <rguenther@suse.de>
24002 PR tree-optimization/79690
24003 * tree-vect-stmts.c (vectorizable_store): Use vector type
24004 built from the DR with address-space.
24006 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24008 * doc/invoke.texi (Optimize Options): Refine the description
24009 of asan-use-after-return.
24011 2017-02-25 Alan Modra <amodra@gmail.com>
24013 PR rtl-optimization/79584
24014 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24015 base, not ad->base_term, the reg within base. Remove assertion
24016 that ad->base == ad->base_term. Replace gen_int_mode using
24017 bogus mode with const0_rtx.
24019 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24021 PR middle-end/79396
24022 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24023 FMA_EXPR like tcc_binary or tcc_unary.
24025 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24028 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24030 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24031 (output_loc_operands): Handle DW_OP_call_ref and
24032 DW_OP_GNU_variable_value.
24033 (struct variable_value_struct): New type.
24034 (struct variable_value_hasher): Likewise.
24035 (variable_value_hash): New variable.
24036 (string_types): Remove.
24037 (copy_loc_descr): New function.
24038 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24039 (prepend_loc_descr_to_each): New function.
24040 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24041 instead of add_loc_descr_to_each if the first argument is single
24042 location list and the second has multiple.
24043 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24044 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24045 when looking for variable value which doesn't have other location info.
24046 (loc_list_from_tree): Formatting fix.
24047 (gen_array_type_die): Simplify DW_AT_string_length handling.
24048 (adjust_string_types): Remove.
24049 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24050 string_types. Call resolve_variable_values.
24051 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24052 (resolve_addr_in_expr): Likewise. Add A argument.
24053 (copy_deref_exprloc): Remove deref argument. Adjust for the
24054 original expression being DW_OP_GNU_variable_value with optionally
24055 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24056 optionally after it.
24057 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24058 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24059 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24060 (variable_value_hasher::hash, variable_value_hasher::equal): New
24062 (resolve_variable_value_in_expr, resolve_variable_value,
24063 resolve_variable_values, note_variable_value_in_expr,
24064 note_variable_value): New functions.
24065 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24068 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24071 * opts.h (handle_generated_option): Add GENERATED_P argument.
24072 * opts-common.c (handle_option): Adjust function comment.
24073 (handle_generated_option): Add GENERATED_P argument, pass it to
24075 (control_warning_option): Pass false to handle_generated_option
24077 * opts.c (maybe_default_option): Pass true to handle_generated_option
24079 * optc-gen.awk: Likewise.
24081 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24083 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24084 a REG, look at the REG it is a SUBREG of.
24085 (splitter for cmpeqsi_t): Ditto.
24087 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24089 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24090 the special USEs with the pattern of the insn, not the insn itself.
24092 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24095 * doc/invoke.texi: Document -mload-store-pairs.
24097 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24098 Sandra Loosemore <sandra@codesourcery.com>
24100 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24101 argument isn't a CONST_INT.
24102 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24103 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24104 (nios2_validate_compare): Bypass alternate compare logic if *op2
24105 is not a CONST_INT.
24106 (ldstwm_operation_p): Return false if first_base is not a REG or
24107 if first_offset is not a CONST_INT.
24109 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24111 * config/cris/cris.md: Use correct operand in a define_peephole2.
24113 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24115 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24117 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24119 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24120 this_insn if it is an INSN or JUMP_INSN.
24121 (force_offsettable): Look at base, not at addr.
24122 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24123 on things that aren't necessarily CONST_INTs.
24125 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24127 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24128 -mfpmath=sse is the default also for x86-32 targets with SSE2
24129 instruction set when @option{-ffast-math} is enabled
24131 2017-02-24 Jeff Law <law@redhat.com>
24133 PR rtl-optimizatoin/79286
24134 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24137 2017-02-24 Richard Biener <rguenther@suse.de>
24139 PR tree-optimization/79389
24140 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24143 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24145 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24146 function comment to reflect reality.
24147 (loop_exits_before_overflow): Fix typo in function description.
24149 2017-02-24 Richard Biener <rguenther@suse.de>
24151 PR tree-optimization/79389
24152 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24153 properly that a threading opportunity exists. Detect conditional
24154 copy/constant propagation opportunities.
24156 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24158 * config/visium/visium.md (type): Add trap.
24159 (b): New mode attribute.
24160 (*btst): Rename into...
24161 (*btst<mode>): ...this and adjust.
24162 (*cbranchsi4_btst_insn): Rename into...
24163 (*cbranch<mode>4_btst_insn): ...this and adjust.
24164 (trap): New define_insn.
24166 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24168 PR tree-optimization/79389
24169 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24170 (noce_reversed_cond_code): New function.
24171 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24172 reversed_comparison_code. Formatting fix.
24173 (noce_try_store_flag): Test rev_cond != NULL in addition to
24174 reversed_comparison_code.
24175 (noce_try_store_flag_constants): Likewise.
24176 (noce_try_store_flag_mask): Likewise.
24177 (noce_try_addcc): Use rev_cond if non-NULL instead of
24178 reversed_comparison_code.
24179 (noce_try_cmove_arith): Likewise. Formatting fixes.
24180 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24181 (noce_find_if_block): Initialize rev_cond.
24182 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24183 instead of false as last argument never attempt to reverse it
24186 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24188 PR tree-optimization/79663
24189 * tree-predcom.c (combine_chains): Process refs in reverse order
24190 only for ZERO length chains, and add explaining comment.
24192 2017-02-23 Jeff Law <law@redhat.com>
24194 PR tree-optimization/79578
24195 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24196 in call to operand_equal_p.
24198 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24201 * config/i386/cpuid.h: Fix another undefined behavior.
24203 2017-02-23 Richard Biener <rguenther@suse.de>
24205 PR tree-optimization/79683
24206 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24207 vector types for data-refs.
24209 2017-02-23 Martin Liska <mliska@suse.cz>
24211 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24213 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24215 PR middle-end/79665
24216 * internal-fn.c (get_range_pos_neg): Moved to ...
24217 * tree.c (get_range_pos_neg): ... here. No longer static.
24218 * tree.h (get_range_pos_neg): New prototype.
24219 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24220 are known to be in between 0 and signed maximum inclusive, try to
24221 expand both unsigned and signed divmod and use the cheaper one from
24224 2017-02-22 Jeff Law <law@redhat.com>
24226 PR tree-optimization/79578
24227 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24228 to compare base operands.
24230 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24233 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24234 gpc_reg_operand instead of fpr_reg_operand.
24236 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24238 * config/mips/mips.c (mips_return_in_memory): Force FP
24239 vector types to be returned in memory for o32 ABI.
24241 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24243 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24244 instead of DW_TAG_member for static data member declarations and don't
24245 set no_linkage_name for static inline data members.
24246 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24249 2017-02-22 Martin Liska <mliska@suse.cz>
24251 * doc/invoke.texi: Replace inequality signs with square brackets
24254 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24256 PR tree-optimization/68644
24257 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24259 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24262 * lra-constraints.c (simplify_operand_subreg): Handle
24263 WORD_REGISTER_OPERATIONS targets.
24265 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24268 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24269 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24270 elimination by swapping fld*.
24272 2017-02-22 Richard Biener <rguenther@suse.de>
24274 PR tree-optimization/79673
24275 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24276 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24277 irrelevant address-space qualifiers and avoiding a
24278 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24280 2017-02-22 Richard Biener <rguenther@suse.de>
24282 PR tree-optimization/79666
24283 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24284 to not symbolically negate if that may introduce undefined
24287 2017-02-22 Martin Liska <mliska@suse.cz>
24290 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24291 * data-streamer-out.c (streamer_write_gcov_count_stream):
24293 * value-prof.c (stream_out_histogram_value): Make assert more
24294 precise based on type of counter.
24296 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24299 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24300 Use nonimmediate_operand instead of memory_operand for operand 1.
24301 (float-extend standard_x87sse_constant_load splitter): Ditto.
24303 2017-02-21 Jeff Law <law@redhat.com>
24305 PR tree-optimization/79621
24306 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24307 blocks with edges to themselves.
24309 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24312 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24313 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24314 Use gimple_call_builtin_p.
24317 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24318 on temporarily removed DEBUG_INSNs.
24320 PR tree-optimization/79649
24321 * tree-loop-distribution.c (classify_partition): Give up on
24322 non-generic address space loads/stores.
24324 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24326 * doc/loop.texi (Loop manipulation): Remove nonexistent
24327 tree_ssa_loop_version from the documentation.
24328 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24330 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24333 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24334 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24335 * config/rs6000/rs6000.c: Include except.h.
24336 (rs6000_expand_split_stack_prologue): Call
24337 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24339 2017-02-21 Martin Jambor <mjambor@suse.cz>
24342 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24343 have been analyzed.
24345 2017-02-21 Martin Jambor <mjambor@suse.cz>
24347 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24348 for backward compatibility only.
24349 * doc/invoke.texi (Option Summary): Remove all references to
24350 -fipa-cp-alignment.
24352 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24356 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24358 * lra-constraints.c (curr_insn_transform): Handle
24359 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24361 2017-02-21 Martin Liska <mliska@suse.cz>
24363 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24365 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24367 PR translation/79638
24368 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24370 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24373 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24374 (arm_function_ok_for_sibcall): Return false for an indirect call by
24375 descriptor if all the argument registers are used.
24376 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24377 alignment of the function.
24379 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24381 PR tree-optimization/61441
24382 * simplify-rtx.c (simplify_const_unary_operation): For
24383 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24384 the sNaN unmodified.
24386 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24388 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24389 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24390 instead of SYSTEM_HEADER_DIR.
24392 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24393 Martin Liška <mliska@suse.cz>
24395 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24396 Fix typos and grammar, use active voice, and clarify.
24398 2017-02-20 Marek Polacek <polacek@redhat.com>
24400 PR middle-end/79537
24401 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24404 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24406 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24409 * config/i386/i386.c (ix86_expand_builtin): Handle
24410 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24411 ix86_builtins_isa[fcode].isa as a requirement of those
24412 flags and any other flag in the bitmask.
24413 (ix86_init_mmx_sse_builtins): Use 0 instead of
24414 ~OPTION_MASK_ISA_64BIT as mask.
24415 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24416 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24417 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24418 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24420 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24423 * lra-constraints.c (split_reg): Check requested split mode
24424 is supported by the register.
24426 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24428 * lra-constraints.c (simplify_operand_subreg): Remove early
24431 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24434 * lra-constraints.c (curr_insn_transform): Tighten condition
24435 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24437 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24440 * lra-constraints.c (curr_insn_transform): Handle
24441 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24443 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24446 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24448 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24450 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24453 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24456 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24458 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24461 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24463 * final.c (last_columnnum, override_columnnum): New variables.
24464 (final_start_function): Set last_columnnum, pass it to begin_prologue
24465 hook and pass 0 to dwarf2out_begin_prologue.
24466 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24467 to source_line debug hook.
24468 (notice_source_line): Compute last_columnnum and for debug_column_info
24469 return true on column changes.
24470 * debug.h (struct gcc_debug_hooks): Add column argument to
24471 source_line and begin_prologue hooks.
24472 (debug_nothing_int_charstar_int_bool): Remove prototype.
24473 (debug_nothing_int_int_charstar,
24474 debug_nothing_int_int_charstar_int_bool): New prototypes.
24475 (dwarf2out_begin_prologue): Add column argument.
24476 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24477 begin_prologue hooks.
24478 (debug_nothing_int_charstar_int_bool): Remove.
24479 (debug_nothing_int_int_charstar,
24480 debug_nothing_int_int_charstar_int_bool): New functions.
24481 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24482 through to dwarf2out_source_line.
24483 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24484 (dwarf2out_source_line): Add column argument, emit it if requested.
24485 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24487 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24488 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24489 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24490 through to dwarf2out_begin_prologue.
24491 (vmsdbgout_source_line): Add column argument, pass it through to
24492 dwarf2out_source_line.
24493 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24494 dbxout_source_line caller.
24495 (dbxout_source_line): Add column argument.
24497 * common.opt (gno-column-info, gcolumn-info): New options.
24498 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24499 (check_die): Also test for multiple DW_AT_decl_column attributes.
24500 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24501 DW_AT_decl_column if requested.
24502 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24504 (gen_variable_die): Likewise.
24505 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24506 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24509 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24510 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24511 (ix86_handle_option): Handle OPT_m3dnowa.
24512 * doc/invoke.texi (-m3dnowa): Document.
24513 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24514 -m3dnowa instead of -m3dnow -march=athlon.
24517 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24518 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24520 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24523 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24524 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24525 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24526 generator for vsx_xxpermdi_<mode>_be.
24527 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24528 force big-endian semantics.
24529 (vsx_xxpermdi_<mode>_be): New define_expand with same
24530 implementation as previous version of vsx_xxpermdi_<mode>.
24532 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24534 PR tree-optimization/79327
24535 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24536 variable, its initialization and use.
24538 2017-02-17 Julia Koval <julia.koval@intel.com>
24540 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24541 (OPTION_MASK_ISA_PKU_UNSET): New.
24542 (ix86_handle_option): Handle -mrdpid.
24543 * config/i386/cpuid.h (bit_RDPID): New.
24544 * config/i386/driver-i386.c (host_detect_local_cpu):
24545 Detect RDPID feature.
24546 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24547 * config/i386/i386-c.c (ix86_target_macros_internal):
24549 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24550 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24551 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24552 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24553 * config/i386/i386.md (define_insn "rdpid"): New.
24554 * config/i386/i386.opt Add -mrdpid.
24555 * config/i386/immintrin.h (_rdpid_u32): New.
24557 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24559 PR rtl-optimization/79541
24560 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24561 instead of transforming it into USE.
24563 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24565 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24566 If HONOR_SNANS (SFmode) force the input to a register.
24567 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24568 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24569 an frsp or similar insn.
24571 2017-02-17 Martin Liska <mliska@suse.cz>
24573 PR rtl-optimization/79577
24574 * params.def (selsched-max-sched-times): Increase minimum to 1.
24576 2017-02-17 Martin Liska <mliska@suse.cz>
24578 PR rtl-optimization/79574
24579 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24581 2017-02-17 Martin Liska <mliska@suse.cz>
24583 PR tree-optimization/79529
24584 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24585 ssa_defined_default_def_p to handle cases which are implicitly
24587 * tree-ssa.c (ssa_defined_default_def_p): New function.
24588 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24589 which are implicitly defined.
24590 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24592 2017-02-17 Richard Biener <rguenther@suse.de>
24594 PR middle-end/79576
24595 * params.def (max-ssa-name-query-depth): Limit to 10.
24597 2017-02-17 Richard Biener <rguenther@suse.de>
24599 PR tree-optimization/79552
24600 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24603 2017-02-17 Richard Biener <rguenther@suse.de>
24606 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24608 2017-02-17 Marek Polacek <polacek@redhat.com>
24610 PR middle-end/79536
24611 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24612 (fold_negate_expr): New wrapper.
24614 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24616 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24617 Correct terminology and de-emphasize pre-standard behavior.
24619 2017-02-16 Alan Modra <amodra@gmail.com>
24621 PR rtl-optimization/79286
24622 * ira.c (def_dominates_uses): New function.
24623 (update_equiv_regs): Don't create an equivalence for insns that
24624 may trap where the register def does not dominate the use.
24626 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24628 PR rtl-optimization/78127
24629 * lra.c (lra): Call lra_eliminate before finish the loop after
24632 2017-02-16 Richard Biener <rguenther@suse.de>
24634 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24636 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24637 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24638 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24639 (isl_val_int_from_wi): New function.
24640 (extract_affine_gmp): Rename to ...
24641 (extract_affine_wi): ... this, take a widest_int.
24642 (extract_affine_int): Just wrap extract_affine_wi.
24643 (add_param_constraints): Use isl_val_int_from_wi.
24644 (add_loop_constraints): Likewise, and extract_affine_wi.
24646 2017-02-15 Jeff Law <law@redhat.com>
24648 PR middle-end/79521
24649 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24650 ira_init_register_move_cost_if_necessary.
24652 2017-02-15 Martin Sebor <msebor@redhat.com>
24654 PR middle-end/32003
24655 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24656 removed in a prior commit.
24658 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24660 PR tree-optimization/79347
24661 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24662 counters during peeling.
24664 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24666 * Makefile.in (site.exp): Remove "set ISLVER".
24668 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24671 * real.c (real_from_integer): Call real_convert even for decimal.
24673 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24676 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24678 2017-02-14 Andrew Pinski <apinski@cavium.com>
24680 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24681 cores and change the partno/implementer to be correct.
24682 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24683 the 'B" as the implementer.
24684 * config/aarch64/aarch64-tune.md: Regenerate.
24686 2017-02-14 Carl Love <cel@us.ibm.com>
24688 * config/rs6000/rs6000.c: Add case statement entry to make the
24689 xvcvuxdsp built-in argument unsigned.
24690 * config/rs6000/vsx.md: Fix the source and return operand types so they
24691 match the instruction definitions from the ISA document. Fix typo
24692 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24695 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24698 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24699 member early_clobber_alts.
24700 * lra-lives.c (reg_early_clobber_p): New.
24701 (process_bb_lives): Use it.
24702 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24703 (debug_operand_data): Initialize early_clobber_alts.
24704 (setup_operand_alternative): Set up early_clobber_alts.
24705 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24706 alternatives to new_insn_reg.
24707 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24709 (lra_update_insn_regno_info): Pass the new arg.
24711 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24713 PR middle-end/79505
24714 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24715 (new_oacc_loop_raw): Don't clear already cleared fields.
24718 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24719 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24720 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24722 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24725 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24727 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24730 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24731 the extra instruction to the right place to store 128-bit constant
24734 2017-02-14 Martin Sebor <msebor@redhat.com>
24736 PR middle-end/79448
24737 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24738 warning for strings of unknown length.
24740 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24742 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24744 2017-02-14 Jeff Law <law@redhat.com>
24747 * ira-costs.c (scan_one_insn): Initialize register move costs
24748 for pseudos seen in USE/CLOBBER insns.
24750 PR tree-optimization/79095
24751 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24752 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24753 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24754 if the operands are known to be not equal, then the resulting range
24756 (intersect_ranges): If the new range is ~[0,0] and the old range is
24757 wide, then prefer ~[0,0].
24758 * tree-vrp.c (overflow_comparison_p_1): New function.
24759 (overflow_comparison_p): New function.
24760 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24761 if NAME is used in an overflow test.
24762 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24763 overflow check that can be expressed as an equality test, then adjust
24764 ops to be that equality test.
24766 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24768 * config/s390/s390-builtin-types.def: Remove flags argument.
24769 * config/s390/s390.c (s390_init_builtins): Likewise.
24771 2017-02-14 Martin Liska <mliska@suse.cz>
24773 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24774 vector. Fix trailing white spaces.
24776 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24778 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24781 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24783 PR rtl-optimization/68664
24784 * config/arm/arm.c (arm_sched_can_speculate_insn):
24785 New function. Declare prototype.
24786 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24788 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24790 PR rtl-optimization/68664
24791 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24793 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24795 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24797 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24798 max skip bytes for function, loop and jump.
24800 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24802 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24803 ABS_EXPR for gimple dump.
24805 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24808 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24810 PR tree-optimization/79408
24811 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24812 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24813 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24814 also if rhs1 is INTEGER_CST.
24816 2017-02-14 Richard Biener <rguenther@suse.de>
24818 PR middle-end/79432
24819 * tree-into-ssa.c (insert_phi_nodes): When the function can
24820 have abnormal edges rewrite SSA names with broken use-def
24821 dominance out of SSA and register them for PHI insertion.
24823 2017-02-13 Martin Sebor <msebor@redhat.com>
24825 PR middle-end/79496
24826 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24827 clearing info.nowrite flag when snprintf size argument is a range.
24829 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24831 * cprop.c (cprop_jump): Add missing space in string literal.
24832 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24833 (get_constraint_for_component_ref): Likewise.
24834 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24835 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24836 * lra-constraints.c (process_alt_operands): Likewise.
24837 * ipa-inline.c (inline_small_functions): Likewise.
24838 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24839 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24840 * trans-mem.c (diagnose_tm_1_op): Likewise.
24841 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24842 (grid_parallel_clauses_gridifiable): Likewise.
24844 * config/nvptx/mkoffload.c (process): Add space in between
24847 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24848 "MOD4_SSE_REGS" and "ALL_REGS".
24850 * spellcheck.c (test_data): Add , in between "foo" and "food".
24852 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24855 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24856 boundary crossing check and subsequent code generation agree.
24858 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24860 * config/aarch64/aarch64.c (has_memory_op): Delete.
24861 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24864 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24866 PR rtl-optimization/79388
24867 PR rtl-optimization/79450
24868 * combine.c (distribute_notes): When removing TEM_INSN for which
24869 corresponding dest has last value recorded, invalidate that last
24872 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24874 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24875 of explicit '@'. Add missing assembly comment marker on branch costs
24878 2017-02-13 Nathan Sidwell <nathan@acm.org>
24880 * gengtype-lex.l (<in_struct>): Add '/'.
24882 2017-02-13 Martin Liska <mliska@suse.cz>
24885 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24887 2017-02-13 Richard Biener <rguenther@suse.de>
24889 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24891 * configure: Re-generate.
24892 * config.in: Likewise.
24893 * graphite-dependences.c: Simplify as if
24894 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24895 * graphite-isl-ast-to-gimple.c: Likewise.
24896 * graphite-optimize-isl.c: Likewise.
24897 * graphite-poly.c: Likewise.
24898 * graphite-sese-to-poly.c: Likewise.
24899 * graphite.h: Likewise.
24900 * toplev.c: Include isl/version.h and use isl_version () for
24901 printing the ISL version.
24902 * doc/install.texi: Update ISL requirement.
24904 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24906 * doc/standards.texi (Standards): Update reference to
24909 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24911 * doc/extend.texi (Named Address Spaces): sourceware.org now
24913 * doc/install.texi (Binaries): Ditto.
24916 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24918 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24919 standard terminology "stringize"/"stringizing" throughout.
24920 * doc/cppinternals.texi: Likewise.
24922 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24924 * doc/extend.texi: Fix some spelling mistakes and typos.
24925 * doc/invoke.texi: Likewise.
24927 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24930 * params.def (inline-min-speedup) Change from 10 to 8.
24932 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24934 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24937 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24940 * ipa-inline-analysis.c (get_minimal_bb): New function.
24941 (record_modified): Use it.
24942 (remap_edge_change_prob): Handle also ancestor functions.
24944 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24946 * doc/contrib.texi (Contributors): Remove broken link into
24947 the Mauve CVS repository.
24949 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24951 PR middle-end/79454
24952 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24953 result computation whenever lhs doesn't have vector mode, not
24954 just when it has BLKmode.
24956 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24958 * doc/makefile.texi (profiledbootstrap): Refer to the
24959 installation instructions only in textual form.
24961 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24964 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24966 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24968 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24969 (Specific): Update mingw-w64 reference.
24971 (Specific): Remove broken link to Renesas RX processor.
24973 2017-02-10 Richard Biener <rguenther@suse.de>
24975 * toplev.c (process_options): Do not mention obsolete graphite
24976 options when printing sorry message about missing graphite support.
24977 Mention -floop-nest-optimize.
24979 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24981 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24982 (vtst_p16): Likewise.
24983 (vtstq_p8): Likewise.
24984 (vtstq_p16): Likewise.
24986 (vtstq_p64): Likewise.
24987 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24988 (vset_lane_p64): New.
24989 (vsetq_lane_p64): New.
24991 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24993 PR tree-optimization/79411
24994 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24995 stmt operands are SSA_NAMEs used in abnormal phis.
24996 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24999 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25002 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25005 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25007 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25009 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25011 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25015 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25016 not arbitrary TREE_CONSTANT.
25019 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25020 "omp declare target link" attribute unless is_global_var.
25021 * omp-offload.c (find_link_var_op): Likewise.
25023 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25024 Chung-Lin Tang <cltang@codesourcery.com>
25026 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25028 (gimplify_adjust_omp_clauses): Don't delete TILE.
25029 (gimplify_omp_for): Deal with TILE.
25030 * internal-fn.c (expand_GOACC_TILE): New function.
25031 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25033 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25034 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25036 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25037 avoid DIV for outermost collapse var.
25038 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25039 Remove out of date comments, fix whitespace.
25040 * omp-general.c (omp_extract_for_data): Deal with tiling.
25041 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25042 adjust OLF_DIM_BASE value.
25043 (struct omp_for_data): Add tiling field.
25044 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25045 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25046 for auto loops. Remove default auto determining, moved to
25047 oacc_loop_fixed_partitions.
25048 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25049 stmts, add e_mask field.
25050 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25051 (oacc_thread_numbers): Use oacc_dim_call.
25052 (oacc_xform_tile): New.
25053 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25054 (finish_oacc_loop): Adjust for ifns vector.
25055 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25056 add case for GOACC_TILE fns.
25057 (oacc_loop_xform_loop): Delete.
25058 (oacc_loop_process): Iterate over call list directly, and add
25059 handling for GOACC_TILE fns.
25060 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25062 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25063 vector partitioning to outer loops. Assign 2 partitions to loops
25064 when available. Add TILE handling.
25065 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25066 (execite_oacc_device_lower): Process GOACC_TILE fns,
25067 ignore unknown specs.
25068 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25069 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25070 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25072 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25074 * configure.ac (ACX_BUGURL): Update.
25075 * configure: Regenerate.
25077 2017-02-09 Richard Biener <rguenther@suse.de>
25079 PR tree-optimization/69823
25080 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25081 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25083 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25085 * config/arc/arc-c.def: Add __NPS400__ definition.
25086 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25087 (TARGET_NPS400): Define.
25089 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25091 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25093 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25094 pointer, arch_info.
25095 (arc_cpu_types): Fill the arch_info field with a pointer into the
25096 arc_arch_types table.
25097 (arc_selected_cpu): Declare.
25098 * config/arc/arc.c (arc_selected_cpu): Make global.
25099 (arc_selected_arch): Delete.
25100 (arc_base_cpu): Delete.
25101 (arc_override_options): Remove references to deleted variables,
25102 update access to arch information.
25103 (ARC_OPT): Update access to arch information.
25104 (ARC_OPTX): Likewise.
25105 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25106 (TARGET_ARC600): Update access to arch information.
25107 (TARGET_ARC601): Likewise.
25108 (TARGET_ARC700): Likewise.
25109 (TARGET_EM): Likewise.
25110 (TARGET_HS): Likewise.
25111 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25114 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25117 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25118 condition/operands for integer GE/LE/GEU/LEU operations.
25120 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25122 PR translation/79397
25123 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25126 2017-02-08 Martin Jambor <mjambor@suse.cz>
25129 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25130 whether allocation happened.
25131 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25132 nothing was allocated.
25134 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25136 PR tree-optimization/79408
25137 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25138 constant, but SSA_NAME with a known integer range, use the minimum
25139 of that range instead of op1 to determine if modulo can be replaced
25140 with its first operand.
25142 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25144 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25146 2017-02-08 Richard Biener <rguenther@suse.de>
25148 PR tree-optimization/71824
25149 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25150 Check all loops contained in the merged region.
25152 2017-02-07 Andrew Pinski <apinski@cavium.com>
25154 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25156 2017-02-07 Andrew Pinski <apinski@cavium.com>
25158 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25159 (thunderxt88): Likewise.
25160 (thunderxt81): Disable LSE and change v8.1 to v8.
25161 (thunderxt83): Likewise.
25163 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25164 Richard Biener <rguenther@suse.de>
25166 PR middle-end/79399
25167 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25168 type from int to size_t.
25169 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25171 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25173 PR rtl-optimization/79386
25174 * cprop.c (bypass_conditional_jumps): Initialize
25175 bypass_last_basic_block already before splitting bbs after
25176 unconditional traps...
25177 (bypass_conditional_jumps): ... rather than here.
25180 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25181 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25182 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25183 fix -masm=intel patterns.
25185 2017-02-07 Richard Biener <rguenther@suse.de>
25187 PR tree-optimization/79256
25188 PR middle-end/79278
25189 * builtins.c (get_object_alignment_2): Use min_align_of_type
25190 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25191 and ADJUST_FIELD_ALIGN.
25193 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25195 * doc/tm.texi: Regenerate.
25196 * stor-layout.c (layout_decl): Adjust.
25197 (update_alignment_for_field): Likewise.
25198 (place_field): Likewise.
25199 (min_align_of_type): Likewise.
25200 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25201 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25202 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25203 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25204 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25205 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25206 * config/i386/i386.c (x86_field_alignment): Likewise.
25207 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25208 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25209 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25210 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25211 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25212 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25216 2017-01-30 Richard Biener <rguenther@suse.de>
25218 PR tree-optimization/79256
25219 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25220 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25223 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25225 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25226 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25227 builtins to SImode and emit a zero-extend, if necessary.
25229 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25231 * docs/invoke.texi (RISC-V Options): Alphabetize.
25233 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25235 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25238 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25240 * config/riscv/riscv.c: New file.
25241 * common/config/riscv/riscv-common.c: Likewise.
25242 * config.gcc: Likewise.
25243 * config/riscv/constraints.md: Likewise.
25244 * config/riscv/elf.h: Likewise.
25245 * config/riscv/generic.md: Likewise.
25246 * config/riscv/linux.h: Likewise.
25247 * config/riscv/multilib-generator: Likewise.
25248 * config/riscv/peephole.md: Likewise.
25249 * config/riscv/pic.md: Likewise.
25250 * config/riscv/predicates.md: Likewise.
25251 * config/riscv/riscv-builtins.c: Likewise.
25252 * config/riscv/riscv-c.c: Likewise.
25253 * config/riscv/riscv-ftypes.def: Likewise.
25254 * config/riscv/riscv-modes.def: Likewise.
25255 * config/riscv/riscv-opts.h: Likewise.
25256 * config/riscv/riscv-protos.h: Likewise.
25257 * config/riscv/riscv.h: Likewise.
25258 * config/riscv/riscv.md: Likewise.
25259 * config/riscv/riscv.opt: Likewise.
25260 * config/riscv/sync.md: Likewise.
25261 * config/riscv/t-elf-multilib: Likewise.
25262 * config/riscv/t-linux: Likewise.
25263 * config/riscv/t-linux-multilib: Likewise.
25264 * config/riscv/t-riscv: Likewise.
25265 * configure.ac: Likewise.
25266 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25267 Waterman as RISC-V maintainers.
25268 * doc/install.texi: Add RISC-V entries.
25269 * doc/invoke.texi: Add RISC-V options section.
25270 * doc/md.texi: Add RISC-V constraints section.
25271 * configure: Regenerated.
25273 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25276 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25277 false values to be constant vectors with all 0 or all 1 bits set.
25278 (vcondu<mode><mode>): Likewise.
25279 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25281 (fpmask_comparison_operator): Update comment.
25282 (vecint_comparison_operator): New predicate.
25283 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25284 vector conditionals when the true and false values are constant
25285 vectors with all 0 bits or all 1 bits set.
25287 2017-02-06 Martin Sebor <msebor@redhat.com>
25289 PR tree-optimization/79376
25290 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25292 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25294 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25295 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25296 to simplify split condition.
25298 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25300 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25301 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25304 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25306 PR rtl-optimization/68664
25307 * target.def (can_speculate_insn): New hook.
25308 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25309 * doc/tm.texi: Regenerate.
25310 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25311 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25312 (rs6000_sched_can_speculate_insn): New function.
25314 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25316 PR tree-optimization/79284
25317 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25318 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25319 vectorizable_mask_load_store, vectorizable_operation,
25320 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25321 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25322 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25323 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25324 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25325 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25326 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25327 is_gimple_assign (stmt). Replace another such test with
25328 is_gimple_assign (stmt).
25330 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25333 * config/avr/avr.c (rtl-iter.h): Include it.
25334 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25335 (avr_legitimate_combined_insn): ...and implementation.
25337 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25339 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25340 * config/s390/s390.c (s390_const_operand_ok)
25341 (s390_canonicalize_comparison, s390_extract_part)
25342 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25343 (s390_contiguous_bitmask_p, s390_rtx_costs)
25344 (legitimize_pic_address): Likewise.
25345 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25346 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25347 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25348 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25349 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25351 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25353 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25354 REGNO($0) == REGNO($1).
25356 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25358 * config/s390/linux.h(SIZE_TYPE): Add comment.
25360 2017-02-06 Julian Brown <julian@codesourcery.com>
25361 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25362 Virendra Pathak <virendra.pathak@broadcom.com>
25364 * config/aarch64/aarch64-cores.def: Change the scheduler
25366 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25367 * config/aarch64/thunderx2t99.md: New file.
25369 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25371 * doc/standards.texi (Go Language): Update link to language
25374 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25376 * tree-eh.c (lower_resx): Sanitize profile.
25377 (cleanup_empty_eh_move_lp): Likewise.
25379 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25382 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25384 * cfgloopmanip.h (loop_version): Update prototype.
25385 * modulo-sched.c (sms_schedule): Update call of loop_version.
25386 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25387 * tree-parloops.c (gen_parallel_loop): Likewise.
25388 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25389 * tree-ssa-loop-split.c (split_loop): Likewise.
25390 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25391 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25393 2017-02-05 Martin Liska <mliska@suse.cz>
25396 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25398 (print_operand_address): Initialize a struct to zero.
25400 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25402 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25403 garbage collector only in textual form.
25405 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25407 * doc/extend.texi (x86 specific memory model extensions for
25408 transactional memory): Simplify a phrase.
25410 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25413 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25414 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25415 (atomic_storedi_1): Likewise.
25417 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25419 PR tree-optimization/79338
25420 * tree-parloops.c (gather_scalar_reductions): Don't call
25421 vect_analyze_loop_form for loop->inner before destroying loop's
25424 2017-02-03 Martin Sebor <msebor@redhat.com>
25426 PR tree-optimization/79327
25427 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25428 when precision has resulted in leading zeros.
25429 (format_integer): Adjust the likely counter to assume an unknown
25430 argument that may be zero is non-zero.
25432 2017-02-03 Jason Merrill <jason@redhat.com>
25435 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25436 avoid copying non-taken branch.
25438 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25440 PR tree-optimization/79340
25441 * tree-vect-loop.c (vectorizable_reduction): Release
25442 vec_defs elements after safe_splicing them into other vectors.
25445 PR tree-optimization/79327
25446 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25447 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25449 (format_integer): Use wide_int_to_tree instead of build_int_cst
25450 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25451 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25452 of shortest and longest sequence.
25454 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25456 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25457 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25459 2017-02-03 Walter Lee <walt@tilera.com>
25462 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25463 after initial stackframe link reg save.
25464 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25466 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25469 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25470 wu for stxssp alternative.
25472 2017-02-03 Martin Sebor <msebor@redhat.com>
25474 PR tree-optimization/79352
25475 * gimple-fold.c (get_range_strlen): Add argument.
25476 (get_range_strlen): Change return type to bool.
25477 (get_maxval_strlen): Pass in a dummy argument.
25478 * gimple-fold.h (get_range_strlen): Change return type to bool.
25479 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25480 * tree.h (array_at_struct_end_p): Add argument.
25481 * tree.c (array_at_struct_end_p): Handle it.
25483 2017-02-03 Martin Liska <mliska@suse.cz>
25486 * multiple_target.c (create_dispatcher_calls): Redirect edge
25487 from a caller of a dispatcher.
25488 (expand_target_clones): Make the clones local.
25489 (ipa_target_clone): Do both target clones and resolvers.
25490 (ipa_dispatcher_calls): Remove the pass.
25491 (pass_dispatcher_calls::gate): Likewise.
25492 (make_pass_dispatcher_calls): Likewise.
25493 * passes.def (pass_target_clone): Put as very first IPA early
25496 2017-02-03 Martin Liska <mliska@suse.cz>
25498 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25499 in case of a function with ifunc attribute.
25501 2017-02-03 Martin Liska <mliska@suse.cz>
25503 * cgraph.c (cgraph_node::dump): Dump function version info.
25504 * symtab.c (symtab_node::dump_base): Add missing new line.
25506 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25508 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25509 (ifcombine_ifandif): Use it.
25511 2017-02-03 Martin Liska <mliska@suse.cz>
25513 * doc/invoke.texi: Document default value for
25514 use-after-scope-direct-emission-threshold.
25516 2017-02-03 Martin Liska <mliska@suse.cz>
25518 PR tree-optimization/79339
25519 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25520 (format_floating): Likewise.
25522 2017-02-03 Martin Liska <mliska@suse.cz>
25525 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25527 (ipa_node_params_t::remove): Likewise.
25528 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25529 initialization from removed ipa_node_params_t::insert.
25530 (ipa_node_params::~ipa_node_params): Move from removed
25531 ipa_node_params_t::release.
25532 * symbol-summary.h (symbol_summary::m_released): New member.
25533 Do not release a summary twice. Do not allow to call finalizer
25534 for types of a summary that live in GGC memory.
25536 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25538 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25541 2017-02-02 Martin Sebor <msebor@redhat.com>
25543 PR middle-end/79275
25544 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25545 (format_string): Tighten up the range of output for non-constant
25546 strings and correct the expected range for wide non-constant strings.
25548 2017-02-02 Martin Sebor <msebor@redhat.com>
25550 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25552 PR middle-end/32003
25553 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25555 (-fdump-tree-@var): Add to index and document how to come up
25556 with pass-specific option and dump file names.
25557 (-fdump-passes): Clarify where to look for output.
25559 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25561 PR middle-end/77445
25562 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25563 statistics of the analyzed path; allow threading for speed when
25564 any of BBs along the path are optimized for speed.
25566 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25568 PR middle-end/78468
25569 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25570 settings of the virtual registers.
25573 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25575 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25576 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25579 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25581 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25582 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25584 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25586 * config/s390/s390.md: Add missing comments with the expanded
25588 * config/s390/vector.md: Likewise.
25589 * config/s390/vx-builtins.md: Likewise.
25591 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25594 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25595 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25596 conditions on a single line.
25598 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25600 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25601 __S390_VX__ to __VX__.
25603 2017-02-01 Andrew Pinski <apinski@cavium.com>
25605 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25606 stmt_info to record_stmt_cost.
25607 (vect_get_known_peeling_cost): Pass stmt_info if known to
25609 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25610 cpu_vector_cost field into
25611 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25612 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25613 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25614 splitting of scalar_stmt_cost and vec_stmt_cost.
25615 (thunderx_vector_cost): Likewise.
25616 (cortexa57_vector_cost): LIkewise.
25617 (exynosm1_vector_cost): Likewise.
25618 (xgene1_vector_cost): Likewise.
25619 (thunderx2t99_vector_cost): Improve after the splitting of the two
25621 (aarch64_builtin_vectorization_cost): Update for the splitting of
25622 scalar_stmt_cost and vec_stmt_cost.
25624 2017-02-01 Torvald Riegel <triegel@redhat.com>
25625 Richard Henderson <rth@redhat.com>
25627 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25628 conditional on existance of a fast atomic load.
25629 * optabs-query.c (can_atomic_load_p): New function.
25630 * optabs-query.h (can_atomic_load_p): Declare it.
25631 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25632 no fast atomic load is available for the particular size of access.
25633 (expand_atomic_compare_and_swap): Likewise.
25634 (expand_atomic_load): Likewise.
25635 (expand_atomic_store): Likewise.
25636 (expand_atomic_fetch_op): Likewise.
25637 * testsuite/lib/target-supports.exp
25638 (check_effective_target_sync_int_128): Remove x86 because it provides
25639 no fast atomic load.
25640 (check_effective_target_sync_int_128_runtime): Likewise.
25642 2017-02-01 Richard Biener <rguenther@suse.de>
25644 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25645 (graphite_transform_loops): Provide opt-info for optimized nests.
25646 * tree-parloop.c (parallelize_loops): Provide opt-info for
25647 parallelized loops.
25649 2017-02-01 Richard Biener <rguenther@suse.de>
25651 PR middle-end/79315
25652 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25653 was not set before.
25655 2017-02-01 Richard Biener <rguenther@suse.de>
25657 PR tree-optimization/71824
25658 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25659 Verify the loops are valid in the merged SESE region.
25660 (scop_detection::can_represent_loop_1): Check analyzing the
25661 evolution of the number of iterations in the region succeeds.
25663 2017-01-31 Ian Lance Taylor <iant@golang.org>
25665 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25666 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25668 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25670 PR preprocessor/79210
25671 * input.c (get_substring_ranges_for_loc): Replace line_width
25672 assertion with error-handling.
25674 2017-01-31 Richard Biener <rguenther@suse.de>
25676 PR tree-optimization/77318
25677 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25678 (create_pw_aff_from_tree): Take loop parameter.
25679 (add_condition_to_pbb): Pass loop of the condition to
25680 create_pw_aff_from_tree.
25682 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25684 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25685 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25687 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25691 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25693 (convert_int_to_float128): Likewise.
25694 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25695 (convert_int_to_float128): Likewise.
25696 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25697 (UNSPEC_IEEE128_CONVERT): Likewise.
25698 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25699 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25700 Use local variables for IBM extended format.
25701 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25702 (fix_trunc<mode>si2_fprs): Likewise.
25703 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25704 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25705 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25706 to know that we can now have integers of all sizes in vector
25708 (fix<uns>_<mode>di2_hw): Likewise.
25709 (float<uns>_<mode>si2_hw): Likewise.
25710 (fix_<mode>si2_hw): Likewise.
25711 (fixuns_<mode>si2_hw): Likewise.
25712 (float<uns>_<mode>di2_hw): Likewise.
25713 (float_<mode>di2_hw): Likewise.
25714 (float_<mode>si2_hw): Likewise.
25715 (floatuns_<mode>di2_hw): Likewise.
25716 (floatuns_<mode>si2_hw): Likewise.
25717 (xscvqp<su>wz_<mode>): Delete, no longer used.
25718 (xscvqp<su>dz_<mode>): Likewise.
25719 (xscv<su>dqp_<mode>): Likewise.
25720 (ieee128_mfvsrd_64bit): Likewise.
25721 (ieee128_mfvsrd_32bit): Likewise.
25722 (ieee128_mfvsrwz): Likewise.
25723 (ieee128_mtvsrw): Likewise.
25724 (ieee128_mtvsrd_64bit): Likewise.
25725 (ieee128_mtvsrd_32bit): Likewise.
25727 2017-01-31 Martin Liska <mliska@suse.cz>
25730 * ipa-prop.c (ipa_free_all_node_params): Call release method
25731 instead of ~sumbol_summary to not to trigger double times
25734 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25736 PR tree-optimization/71691
25737 * bitmap.h (class auto_bitmap): New.
25738 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25739 is_maybe_undefined instead of ssa_undefined_value_p.
25741 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25743 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25744 __S390_ARCH_LEVEL__ to __ARCH__.
25746 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25748 PR tree-optimization/79267
25749 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25750 if should_remove_lhs_p is true.
25752 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25755 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25756 (add_alignment_attribute): New.
25757 (base_type_die): Add alignment attribute.
25758 (subrange_type_die): Likewise.
25759 (modified_type_die): Likewise.
25760 (gen_array_type_die): Likewise.
25761 (gen_descr_array_type_die: Likewise.
25762 (gen_enumeration_type_die): Likewise.
25763 (gen_subprogram_die): Likewise.
25764 (gen_variable_die): Likewise.
25765 (gen_field_die): Likewise.
25766 (gen_ptr_to_mbr_type_die): Likewise.
25767 (gen_struct_or_union_type_die): Likewise.
25768 (gen_subroutine_type_die): Likewise.
25769 (gen_typedef_die): Likewise.
25770 (base_type_cmp): Compare alignment attribute.
25772 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25775 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25776 (setb_unsigned) New pattern for setb with CCUNS.
25777 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25778 subfc./subfe sequence to avoid overflow problems. Generate a
25779 shorter sequence with cmpld/setb for power9.
25780 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25781 for generating subfc. instruction.
25782 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25783 now uses this instruction.
25785 2017-01-30 Ian Lance Taylor <iant@google.com>
25788 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25789 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25791 2017-01-30 Martin Sebor <msebor@redhat.com>
25793 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25794 Move constant to the right of a relational operator.
25795 (get_mpfr_format_length, format_character, format_string): Ditto.
25796 (should_warn_p, maybe_warn): Same.
25798 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25800 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25803 * asan.c (get_translation_unit_decl): Remove function.
25804 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25806 2017-01-30 Martin Liska <mliska@suse.cz>
25808 PR gcov-profile/79259
25809 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25810 -fprofile-generate.
25812 2017-01-30 Martin Liska <mliska@suse.cz>
25815 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25816 Initialize variables with NULL value.
25818 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25821 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25823 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25825 2017-01-30 Richard Biener <rguenther@suse.de>
25827 PR tree-optimization/79276
25828 * tree-vrp.c (process_assert_insertions): Properly adjust common
25829 when removing a duplicate.
25831 * gcc.dg/torture/pr79276.c: New testcase.
25833 2017-01-30 Richard Biener <rguenther@suse.de>
25835 PR tree-optimization/79256
25836 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25837 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25839 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25841 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25844 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25845 ("*r<noxa>sbg_<mode>_sll_bitmask")
25846 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25847 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25848 Use contiguous_bitmask_nowrap_operand.
25850 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25853 * config/rs6000/altivec.h (vec_xl): Revise #define.
25854 (vec_xst): Likewise.
25856 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25858 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25860 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25862 PR rtl-optimization/79194
25863 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25864 traps before call to bypass_conditional_jumps.
25866 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25868 PR tree-optimization/71374
25869 * lra-constraints.c (check_conflict_input_operands): New.
25870 (match_reload): Use it.
25872 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25875 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25876 account to calculate conflict_set.
25878 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25880 PR rtl-optimization/78559
25881 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25882 other_insn in combine.
25884 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25886 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25887 uint16_type_node for BT_UINT16.
25889 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25891 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25892 "RTL Tests" to menu.
25893 (GIMPLE Tests): New node.
25894 (RTL Tests): New node.
25896 2017-01-27 Richard Biener <rguenther@suse.de>
25898 PR tree-optimization/79245
25899 * tree-loop-distribution.c (distribute_loop): Apply cost
25900 modeling also to detected patterns.
25902 2017-01-27 Richard Biener <rguenther@suse.de>
25904 PR tree-optimization/71433
25905 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25906 (compare_assert_loc): New function.
25907 (process_assert_insertions): Sort and optimize assert locations
25908 to remove duplicates and push down identical assertions on
25909 edges to their destination block.
25911 2017-01-27 Richard Biener <rguenther@suse.de>
25913 PR tree-optimization/79244
25914 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25915 out SSA names even if abnormal.
25917 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25919 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25920 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25921 instead of MPFR_RNDN.
25923 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25926 * arm.c (arm_option_override): Don't call build_target_option_node
25927 until after doing all option overrides.
25928 (arm_valid_target_attribute_tree): Likewise.
25930 2017-01-27 Martin Liska <mliska@suse.cz>
25932 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25933 for {cd}tors and C++ {cd}tors.
25935 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25937 * config/s390/s390.md ("*setmem_long_and")
25938 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25940 2017-01-26 Martin Sebor <msebor@redhat.com>
25942 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25945 2017-01-26 Martin Sebor <msebor@redhat.com>
25947 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25948 HAVE_DFmode before using XFmode or DFmode.
25949 (parse_directive): Avoid using the z length modifier to avoid
25950 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25952 PR middle-end/78703
25953 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25954 to accept adjustment as an array.
25955 (get_int_range): New function.
25956 (struct directive): Make width and prec arrays.
25957 (directive::set_width, directive::set_precision): Call get_int_range.
25958 (format_integer, format_floating): Handle width and precision ranges.
25959 (format_string, parse_directive): Same.
25961 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25964 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25965 comdat_type_p set, just clone them, but keep the children in the
25969 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25970 which have direct callers with -fvar-tracking-assignments enabled
25972 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25973 inside of type units.
25975 2017-01-26 Martin Sebor <msebor@redhat.com>
25977 PR middle-end/78703
25978 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25980 (struct format_result): Replace number_chars, number_chars_min,
25981 and number_chars_max with a single member of struct result_range.
25983 (format_result::operator+=): Adjust.
25984 (struct fmtresult): Remove bounded. Handle likely and unlikely
25986 (fmtresult::adjust_for_width_or_precision): New function.
25987 (fmtresult:type_max_digits): New function.
25988 (bytes_remaining): Handle likely and unlikely counters.
25989 (min_bytes_remaining): Remove.
25990 (format_percent): Simplify.
25991 (format_integer, format_floating): Set likely and unlikely counters.
25992 (get_string_length, format_character, format_string): Same.
25993 (format_plain, should_warn_p): New function.
25994 (maybe_warn): Call should_warn_p. Update diagnostic messages
25995 and handle those for all directives, including plain strings.
25996 (format_directive): Handle likely and unlikely counters.
25997 Remove unnecessary quoting from diagnostics. Add an informational
25999 (add_bytes): Remove.
26000 (pass_sprintf_length::compute_format_length): Simplify.
26001 (try_substitute_return_value): Handle likely and unlikely counters.
26003 2017-01-26 Carl Love <cel@us.ibm.com>
26005 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26006 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26008 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26011 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26012 endianess for subregs into account.
26013 * lra-constraints.c (lra_constraints): Do risky transformations
26014 always on the first iteration.
26015 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26016 last_call_used_reg_set.
26017 (process_bb_lives): Define and use last_call_used_reg_set.
26018 * lra.c (lra): Always continue after lra_constraints on the first
26021 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26023 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26025 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26027 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26029 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26030 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26031 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26032 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26033 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26034 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26035 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26036 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26037 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26039 2017-01-26 Marek Polacek <polacek@redhat.com>
26042 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26043 for the third operand.
26045 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26047 PR middle-end/79236
26048 * omp-low.c (struct omp_context): Add simt_stmt field.
26049 (scan_omp_for): Return omp_context *.
26050 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26051 context to the _simt_ SIMD stmt.
26052 (lower_omp_for): For combined SIMD with sibling _simt_
26053 SIMD, make sure to use the same decls in _looptemp_
26054 clauses as in the sibling.
26056 2017-01-26 David Sherwood <david.sherwood@arm.com>
26058 PR middle-end/79212
26059 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26062 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26065 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26066 emit fld b; fld a; if possible.
26068 * brig-builtins.def: Update copyright years.
26069 * config/arm/arm_acle_builtins.def: Update copyright years.
26071 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26074 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26075 constraint instead of o for the stxsd instruction.
26077 2017-01-25 Carl Love <cel@us.ibm.com>
26079 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26080 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26082 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26084 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26086 2017-01-25 Richard Biener <rguenther@suse.de>
26088 PR tree-optimization/69264
26089 * target.def (vector_alignment_reachable): Improve documentation.
26090 * doc/tm.texi: Regenerate.
26091 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26093 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26094 earlier changes with respect to TYPE_USER_ALIGN.
26095 (vector_alignment_reachable_p): Likewise. Improve dumping.
26097 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26100 * config/arm/arm.md (xordi3): Force constant operand into a register
26103 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26105 * doc/invoke.texi (-fstore-merging): Correct default optimization
26106 levels at which it is enabled.
26107 (-O): Move -fstore-merging from list to...
26110 2017-01-25 Richard Biener <rguenther@suse.de>
26113 * omp-expand.c: Include debug.h.
26114 (expand_omp_taskreg): Make sure to generate early debug before
26115 outlining anything from a function.
26116 (expand_omp_target): Likewise.
26117 (grid_expand_target_grid_body): Likewise.
26119 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26122 * asan.c (get_translation_unit_decl): New function.
26123 (asan_add_global): Extract modules file name from globals
26124 TRANSLATION_UNIT_DECL name.
26126 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26129 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26130 for long calls with APCS frame and VFP.
26132 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26134 * cfg.c (original_copy_tables_initialized_p): New function.
26135 * cfg.h (original_copy_tables_initialized_p): New decl.
26136 * cfgrtl.c (relink_block_chain): Guard the call to
26137 free_original_copy_tables with a call to
26138 original_copy_tables_initialized_p.
26139 * cgraph.h (symtab_node::native_rtl_p): New decl.
26140 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26141 (symtab_node::needed_p): Don't assert for early assembly output
26142 for __RTL functions.
26143 (cgraph_node::finalize_function): Set "force_output" for __RTL
26145 (cgraph_node::analyze): Bail out early for __RTL functions.
26146 (analyze_functions): Update assertion to support __RTL functions.
26147 (cgraph_node::expand): Bail out early for __RTL functions.
26148 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26150 * function.h (struct function): Update comment for field
26152 * gimple-expr.c: Include "tree-pass.h".
26153 (gimple_has_body_p): Return false for __RTL functions.
26154 * Makefile.in (OBJS): Add run-rtl-passes.o.
26155 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26157 (gcc::pass_manager::get_clean_slate): New accessor.
26158 * passes.c: Include "insn-addr.h".
26159 (should_skip_pass_p): Add logging. Update logic for running
26160 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26161 property-provider override so it is only done for gimple passes.
26163 (skip_pass): New function.
26164 (execute_one_pass): Call skip_pass when skipping passes.
26165 * read-md.c (md_reader::read_char): Support filtering
26166 the input to a subset of line numbers.
26167 (md_reader::md_reader): Initialize fields
26168 m_first_line and m_last_line.
26169 (md_reader::read_file_fragment): New function.
26170 * read-md.h (md_reader::read_file_fragment): New decl.
26171 (md_reader::m_first_line): New field.
26172 (md_reader::m_last_line): New field.
26173 * read-rtl-function.c (function_reader::create_function): Only
26174 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26175 curr_properties. Set DECL_INITIAL to a dummy block.
26176 (read_rtl_function_body_from_file_range): New function.
26177 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26179 * run-rtl-passes.c: New file.
26180 * run-rtl-passes.h: New file.
26182 2017-01-24 Jeff Law <law@redhat.com>
26184 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26187 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26189 PR tree-optimization/79159
26190 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26191 (record_nonwrapping_iv): Improve boundary using above function if no
26192 value range information.
26194 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26195 Martin Jambor <mjambor@suse.cz>
26197 * brig-builtins.def: New file.
26198 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26199 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26200 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26201 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26202 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26203 * builtin-types.def (BT_INT8): New.
26204 (BT_INT16): Likewise.
26205 (BT_UINT8): Likewise.
26206 (BT_UINT16): Likewise.
26207 (BT_FN_ULONG): Likewise.
26208 (BT_FN_UINT_INT): Likewise.
26209 (BT_FN_UINT_ULONG): Likewise.
26210 (BT_FN_UINT_LONG): Likewise.
26211 (BT_FN_UINT_PTR): Likewise.
26212 (BT_FN_ULONG_PTR): Likewise.
26213 (BT_FN_INT8_FLOAT): Likewise.
26214 (BT_FN_INT16_FLOAT): Likewise.
26215 (BT_FN_UINT32_FLOAT): Likewise.
26216 (BT_FN_UINT16_FLOAT): Likewise.
26217 (BT_FN_UINT8_FLOAT): Likewise.
26218 (BT_FN_UINT64_FLOAT): Likewise.
26219 (BT_FN_UINT16_UINT32): Likewise.
26220 (BT_FN_UINT32_UINT16): Likewise.
26221 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26222 (BT_FN_INT_PTR_INT): Likewise.
26223 (BT_FN_UINT_PTR_UINT): Likewise.
26224 (BT_FN_LONG_PTR_LONG): Likewise.
26225 (BT_FN_ULONG_PTR_ULONG): Likewise.
26226 (BT_FN_VOID_UINT64_UINT64): Likewise.
26227 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26228 (BT_FN_INT8_INT8_INT8): Likewise.
26229 (BT_FN_INT16_INT16_INT16): Likewise.
26230 (BT_FN_INT_INT_INT): Likewise.
26231 (BT_FN_UINT_FLOAT_UINT): Likewise.
26232 (BT_FN_FLOAT_UINT_UINT): Likewise.
26233 (BT_FN_ULONG_UINT_UINT): Likewise.
26234 (BT_FN_ULONG_UINT_PTR): Likewise.
26235 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26236 (BT_FN_UINT_UINT_UINT): Likewise.
26237 (BT_FN_VOID_UINT_PTR): Likewise.
26238 (BT_FN_UINT_UINT_PTR: Likewise.
26239 (BT_FN_UINT32_UINT64_PTR): Likewise.
26240 (BT_FN_INT_INT_UINT_UINT): Likewise.
26241 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26242 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26243 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26244 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26245 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26246 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26247 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26248 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26249 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26250 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26251 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26252 * doc/frontends.texi: List BRIG FE.
26253 * doc/install.texi (Testing): Add BRIG tesring requirements.
26254 * doc/invoke.texi (Overall Options): Mention BRIG.
26255 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26257 2017-01-24 Richard Biener <rguenther@suse.de>
26259 PR translation/79208
26260 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26262 2017-01-24 Martin Jambor <mjambor@suse.cz>
26265 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26266 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26267 and known_contexts.
26269 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26271 PR middle-end/79123
26272 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26273 casts from signed to unsigned really don't have a range.
26275 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26277 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26278 GMP_RNDx for compatiblity.
26280 2017-01-24 Martin Liska <mliska@suse.cz>
26283 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26284 that would prevent us to call alloca with -1 as argument.
26286 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26288 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26289 Avoid -Wformat-security warning.
26291 2017-01-23 Andrew Pinski <apinski@cavium.com>
26293 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26296 2017-01-23 Martin Sebor <msebor@redhat.com>
26298 PR middle-end/78703
26299 * gimple-ssa-sprintf.c (warn_level): New global.
26300 (format_integer): Use it here and throughout the rest of the file.
26301 Use the same switch to compute sign as base.
26302 (maybe_warn): New function.
26303 (format_directive): Factor out warnings into maybe_warn.
26304 Add debugging output. Use warn_level.
26305 (add_bytes): Use warn_level.
26306 (pass_sprintf_length::compute_format_length): Add debugging output.
26307 (try_substitute_return_value): Same.
26308 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26310 PR middle-end/78703
26311 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26312 (struct fmtresult, format_integer, format_floating): Adjust.
26313 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26314 (get_string_length, format_string,format_directive): Same.
26315 (pass_sprintf_length::compute_format_length): Same.
26316 (try_substitute_return_value): Simplify slightly.
26318 PR middle-end/78703
26319 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26320 (fmtresult::operator+=): Outlined.
26321 (struct fmtresult): Add ctors.
26322 (struct conversion_spec): Rename...
26323 (struct directive): ...to this. Add and remove data members.
26324 (directive::set_width, directive::set_precision): New functions.
26325 (format_percent): Use fmtresult ctor.
26326 (get_width_and_precision): Remove.
26327 (format_integer): Make naming changes. Avoid computing width and
26329 (format_floating): Same. Adjust indentation.
26330 (format_character, format_none): New functions.
26331 (format_string): Moved character handling to format_character.
26332 (format_directive): Remove arguments, change return type.
26333 (parse_directive): New function.
26334 (pass_sprintf_length::compute_format_length): Move directive
26335 parsing to parse_directive.
26337 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26339 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26340 (assign_assembler_name_if_needed): ... this.
26341 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26342 (assign_assembler_name_if_needed): ... this.
26343 (free_lang_data_in_cgraph): Adjust callers.
26344 * cgraphunit.c (cgraph_node::analyze): Likewise.
26345 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26348 2017-01-23 Richard Biener <rguenther@suse.de>
26350 PR tree-optimization/79088
26351 PR tree-optimization/79188
26352 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26353 resetting loop bounds after last path deletion. Reset loop
26354 bounds of the target loop, make code match the comments.
26355 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26356 Make sure loops need no fixups.
26358 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26360 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26361 exponent support with double type for first argument.
26362 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26363 type returned by __builtin_vec_extract_sig,
26364 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26365 functions from "vector int" to "vector unsigned int" or from
26366 "vector long long int" to "vector unsigned long long int".
26367 Changed type returned by __builtin_vec_extract_exp,
26368 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26369 functions from "vector int" to "vector unsigned int" or from
26370 "vector long long int" to "vector unsigned long long int".
26371 Changed return type of __builtin_vec_test_data_class,
26372 __builtin_vec_test_data_class_sp, and
26373 __builtin_vec_test_data_class_dp from "vector int" to
26374 "vector bool int" or from "vector long long int" to "vector bool
26375 long long int" and changed second argument type from "unsigned
26376 int" to "int". Added new overloaded function forms "vector float
26377 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26378 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26379 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26380 double, vector unsigned long long int)" and "vector double
26381 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26382 long int)". Changed return type of
26383 __builtin_scalar_test_data_class and
26384 __builtin_scalar_test_data_class_sp and
26385 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26386 int" and changed second argument from "unsigned int" to "int".
26387 Changed type returned by __builtin_scalar_test_neg,
26388 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26389 from "int" to "bool int". Added new overloaded function form
26390 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26391 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26392 exponent double-precision with floating point first argument.
26393 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26394 documentation of scalar_test_data_class, scalar_test_neg,
26395 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26396 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26397 vec_test_data_class built-in functions to reflect refinements in
26398 their type signatures.
26400 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26402 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26404 (aarch64_elf_asm_destructor): Likewise.
26406 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26408 PR rtl-optimization/78634
26409 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26410 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26411 * ifcvt.c (noce_try_cmove): Add missing cost check.
26413 PR rtl-optimization/71724
26414 * combine.c (if_then_else_cond): Look for situations where it is
26415 beneficial to undo the work of one of the recursive calls.
26417 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26419 PR tree-optimization/70754
26420 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26421 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26422 combined stmt before it if not NULL.
26423 (combine_chains): Process refs reversely and compute dominance point
26426 2017-01-23 Martin Liska <mliska@suse.cz>
26428 PR tree-optimization/79196
26429 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26430 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26432 (strlen_optimize_stmt): Call the renamed function.
26434 2017-01-23 Michael Matz <matz@suse.de>
26436 PR tree-optimization/78384
26437 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26439 2017-01-23 Richard Biener <rguenther@suse.de>
26441 PR tree-optimization/79186
26442 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26443 both incoming edges before moving an assert.
26445 2017-01-23 Martin Jambor <mjambor@suse.cz>
26447 * ipa-prop.c (load_from_param_1): Removed.
26448 (load_from_unmodified_param): Bits from load_from_param_1 put back
26450 (load_from_param): Removed.
26451 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26452 with stmt. Reverted back to use of load_from_unmodified_param.
26454 2017-01-23 Martin Jambor <mjambor@suse.cz>
26457 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26458 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26459 field a pointer to garbage collected vector, mark lattices and
26460 ipcp_orig_node with GTY((skip)).
26461 (ipa_get_param_count): Adjust to descriptors being a pointer.
26462 (ipa_get_param): Likewise.
26463 (ipa_get_type): Likewise.
26464 (ipa_get_param_move_cost): Likewise.
26465 (ipa_set_param_used): Likewise.
26466 (ipa_get_controlled_uses): Likewise.
26467 (ipa_set_controlled_uses): Likewise.
26468 (ipa_is_param_used): Likewise.
26469 (ipa_node_params_t): Move into garbage collector. New methods insert
26471 (ipa_node_params_sum): Annotate wth GTY(()).
26472 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26474 (ipa_load_from_parm_agg): Adjust declaration.
26475 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26476 * ipa-profile.c (ipa_profile): Likewise.
26477 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26478 (ipa_populate_param_decls): Make descriptors parameter garbage
26480 (ipa_dump_param): Adjust to descriptors being a pointer.
26481 (ipa_alloc_node_params): Likewise.
26482 (ipa_initialize_node_params): Likewise.
26483 (load_from_param_1): Make descriptors parameter garbage collected.
26484 (load_from_unmodified_param): Likewise.
26485 (load_from_param): Likewise.
26486 (ipa_load_from_parm_agg): Likewise.
26487 (ipa_node_params::~ipa_node_params): Removed.
26488 (ipa_free_all_node_params): Remove call to delete operator.
26489 (ipa_node_params_t::insert): New.
26490 (ipa_node_params_t::remove): Likewise.
26491 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26492 copy known_csts and known_contexts vectors.
26493 (ipa_read_node_info): Adjust to descriptors being a pointer.
26494 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26496 (ipcp_transform_function): Make descriptors variable garbage
26499 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26501 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26502 * config/i386/avx512dqintrin.h: Ditto.
26503 * config/i386/avx512fintrin.h: Ditto.
26504 * config/i386/i386.c: Handle new builtins.
26505 * config/i386/i386-builtin.def: Add new builtins.
26506 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26507 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26509 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26510 Martin Liska <mliska@suse.cz>
26512 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26513 * asan.c (asan_expand_poison_ifn): Support stores and use
26514 appropriate ASAN report function.
26515 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26516 * internal-fn.def (ASAN_POISON_USE): Declare.
26517 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26518 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26519 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26520 ASAN_POISON calls w/o LHS.
26521 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26522 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26523 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26524 * gimplify.c (asan_poison_variables): Add attribute
26525 use_after_scope_memory to variables that really needs to live
26527 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26528 having the attribute.
26530 2017-01-23 Martin Liska <mliska@suse.cz>
26532 * asan.c (create_asan_shadow_var): New function.
26533 (asan_expand_poison_ifn): Likewise.
26534 * asan.h (asan_expand_poison_ifn): New declaration.
26535 * internal-fn.c (expand_ASAN_POISON): Likewise.
26536 * internal-fn.def (ASAN_POISON): New builtin.
26537 * sanopt.c (pass_sanopt::execute): Expand
26538 asan_expand_poison_ifn.
26539 * tree-inline.c (copy_decl_for_dup_finish): Make function
26541 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26542 * tree-ssa.c (is_asan_mark_p): New function.
26543 (execute_update_addresses_taken): Rewrite local variables
26544 (identified just by use-after-scope as addressable) into SSA.
26546 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26548 * doc/install.texi (Specific): opensource.apple.com uses https
26549 now. Remove trailing slash.
26551 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26553 * README.Portability: Remove note on an Irix compatibility issue.
26555 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26557 * gcov.c (INCLUDE_ALGORITHM): Define.
26558 (INCLUDE_VECTOR): Define.
26559 No longer include <vector> and <algorithm> directly.
26561 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26563 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26565 * doc/invoke.texi (Code Gen Options): Ditto.
26567 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26570 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26572 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26574 rtl-optimization/79125
26575 * cprop.c (local_cprop_pass): Handle cases where we make an
26576 unconditional trap.
26578 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26582 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26583 read from, for big endian.
26585 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26587 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26588 register pauth builtins for LP64 only.
26590 2017-01-20 Marek Polacek <polacek@redhat.com>
26593 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26596 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26598 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26600 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26601 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26602 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26604 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26607 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26608 in big-endian mode when they are not a single duplicated value.
26610 2017-01-20 Richard Biener <rguenther@suse.de>
26612 * BASE-VER: Bump to 7.0.1.
26614 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26616 * omp-low.c (omplow_simd_context): New struct. Use it...
26617 (lower_rec_simd_input_clauses): ...here and...
26618 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26619 references to idx, lane, max_vf, is_simt.
26621 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26623 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26626 2017-01-20 Martin Jambor <mjambor@suse.cz>
26628 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26629 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26631 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26632 (GTFILES): Rename hsa.c to hsa-common.c.
26633 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26634 * hsa-dump.c: Likewise.
26635 * hsa-gen.c: Likewise.
26636 * hsa-regalloc.c: Likewise.
26637 * ipa-hsa.c: Likewise.
26638 * omp-expand.c: Likewise.
26639 * omp-low.c: Likewise.
26640 * toplev.c: Likewise.
26642 2017-01-20 Marek Polacek <polacek@redhat.com>
26645 * doc/invoke.texi: Document -Wduplicated-branches.
26646 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26647 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26648 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26649 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26650 return 0 only when not OEP_LEXICOGRAPHIC.
26651 (fold_build_cleanup_point_expr): Use the expression
26652 location when building CLEANUP_POINT_EXPR.
26653 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26654 * tree.c (add_expr): Handle error_mark_node.
26656 2017-01-20 Martin Liska <mliska@suse.cz>
26659 * tree-profile.c (init_ic_make_global_vars): Do not call
26661 (gimple_init_gcov_profiler): Likewise.
26663 2017-01-20 Martin Liska <mliska@suse.cz>
26666 * cgraph.h (maybe_create_reference): Remove argument and
26668 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26670 * ipa-cp.c (create_specialized_node): Likewise.
26671 * symtab.c (symtab_node::maybe_create_reference): Handle
26672 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26674 2017-01-20 Martin Liska <mliska@suse.cz>
26676 * read-rtl-function.c (function_reader::create_function): Use
26677 build_decl instread of build_decl_stat.
26679 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26681 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26682 * config/i386/avx512dqintrin.h: Ditto.
26683 * config/i386/avx512fintrin.h: Ditto.
26684 * config/i386/i386-builtin-types.def: Add new types.
26685 * config/i386/i386.c: Handle new types.
26686 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26687 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26688 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26689 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26690 (__builtin_ia32_kshiftridi): New.
26691 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26693 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26697 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26698 define to rs6000_init_stack_protect_guard.
26699 (rs6000_init_stack_protect_guard): New function.
26701 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26702 Yunqiang Su <yunqiang.su@imgtec.com>
26704 * config.gcc (supported_defaults): Add madd4.
26705 (with_madd4): Add validation.
26706 (all_defaults): Add madd4.
26707 * config/mips/mips.opt (mmadd4): New option.
26708 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26710 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26712 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26713 (ISA_HAS_FUSED_MADD4): Likewise.
26714 * doc/invoke.texi (-mmadd4): Document the new option.
26715 * doc/install.texi (--with-madd4): Document the new option.
26717 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26719 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26720 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26721 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26722 (aarch64_init_pauth_hint_builtins): New.
26723 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26724 (aarch64_expand_builtin): Expand new builtins.
26726 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26728 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26729 * combine-stack-adj.c (no_unhandled_cfa): Handle
26730 REG_CFA_TOGGLE_RA_MANGLE.
26731 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26732 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26733 info for return address signing.
26734 (aarch64_expand_epilogue): Likewise.
26736 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26738 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26739 * config/aarch64/aarch64-protos.h
26740 (aarch64_return_address_signing_enabled): New declaration.
26741 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26743 (aarch64_expand_prologue): Sign return address before it's pushed onto
26745 (aarch64_expand_epilogue): Authenticate return address fetched from
26747 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26748 (aarch64_attributes): New function attributes for "sign-return-address".
26749 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26750 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26751 ("*do_return"): Generate combined instructions according to key index.
26752 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26753 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26755 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26756 * config/aarch64/aarch64.opt (msign-return-address=): New.
26757 * doc/extend.texi (AArch64 Function Attributes): Documents
26758 "sign-return-address=".
26759 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26761 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26763 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26764 overall option summary.
26766 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26768 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26769 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26770 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26771 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26773 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26775 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26776 -mpower9-minmax by default for -mcpu=power9.
26777 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26778 128-bit floating point.
26780 2017-01-20 Alan Modra <amodra@gmail.com>
26782 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26783 optimizing for size.
26785 2017-01-20 Alan Modra <amodra@gmail.com>
26788 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26789 for strcmp and strncmp from corresponding builtin decl.
26791 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26793 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26794 instead of i386/rtems-64.h.
26795 * config/i386/rtems-64.h: Remove.
26797 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26801 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26803 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26805 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26807 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26808 Change int to HOST_WIDE_INT.
26809 * config/aarch64/aarch64-protos.h
26810 (aarch64_simd_gen_const_vector_dup): Likewise.
26811 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26813 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26815 * langhooks-def.h (lhd_type_for_size): New decl.
26816 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26817 * langhooks.c (lhd_type_for_size): New function, taken from
26820 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26822 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26823 define_bypass for CR latency.
26824 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26825 (power9-alu2): Add define_bypass for CR latency.
26827 (power9-mul): Update insn latency.
26828 (power9-mul-compare): Update insn latency, bypass latency and remove
26831 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26833 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26835 * config/aarch64/aarch64.md
26836 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26837 aarch64_nopcrelative_literal_loads.
26838 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26840 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26842 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26843 TARGET_LOONGSON_3A.
26844 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26846 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26849 * config.gcc (supported_defaults): Add lxc1-sxc1.
26850 (with_lxc1_sxc1): Add validation.
26851 (all_defaults): Add lxc1-sxc1.
26852 * config/mips/mips.opt (mlxc1-sxc1): New option.
26853 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26855 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26856 __mips_no_lxc1_sxc1.
26857 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26858 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26859 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26861 2017-01-19 Richard Biener <rguenther@suse.de>
26863 PR tree-optimization/72488
26864 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26865 sure to restore SSA info.
26866 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26868 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26870 PR rtl-optimization/79121
26871 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26872 of the inner type when shifting an extended value.
26874 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26877 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26878 interposable aliases.
26880 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26883 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26884 Use the evmergelohi instruction.
26885 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26886 (mov_sitf_e500_subreg8_2_be): Likewise.
26887 (mov_sitf_e500_subreg12_2_le): Likewise.
26888 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26889 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26890 (mov_sitf_e500_subreg8_2_le): Likewise.
26891 (mov_sitf_e500_subreg12_2_be): Likewise.
26893 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26895 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26896 attribute from vecsimple to vecperm.
26897 (altivec_vbpermq2): Likewise.
26899 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26902 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26904 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26905 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26906 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26907 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26908 case where N arg is SIZE_MAX.
26909 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26910 (cmpstrsi): Add pattern.
26912 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26914 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26915 __builtin_vec_revb builtins.
26916 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26917 built-in functions to support generation of the ISA 3.0 XXBR<x>
26918 vector byte reverse instructions.
26919 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26920 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26921 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26922 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26923 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26924 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26925 (P9V_BUILTIN_VEC_REVB): Likewise.
26926 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26927 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26928 (p9_xxbrq_v16qi): Likewise.
26929 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26930 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26931 (p9_xxbrh_v8hi): Likewise.
26932 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26933 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26934 vec_revb built-in functions.
26936 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26938 PR rtl-optimization/78952
26939 * config/i386/i386.md (any_extract): New code iterator.
26940 (*insvqi_2): Use any_extract for source operand.
26941 (*insvqi_3): Use any_shiftrt for source operand.
26943 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26945 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26947 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26949 2017-01-18 Matthias Klose <doko@ubuntu.com>
26951 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26953 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26955 * config/rs6000/altivec.h (vec_bperm): Change #define.
26956 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26957 (altivec_vbpermq2): New define_insn.
26958 (altivec_vbpermd): Likewise.
26959 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26960 function interface.
26961 (VBPERMD): Likewise.
26962 (VBPERM): New polymorphic function interface.
26963 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26964 Add entries for P9V_BUILTIN_VEC_VBPERM.
26965 * doc/extend.texi: Add interfaces for vec_bperm.
26967 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26969 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26970 first letter of error messages.
26971 (s390_resolve_overloaded_builtin): Likewise.
26972 * config/s390/s390.c (s390_expand_builtin): Likewise.
26973 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26974 (s390_valid_target_attribute_inner_p): Likewise.
26975 * config/s390/s390.md ("tabort"): Likewise.
26977 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26979 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26980 (ISA_AVOID_DIV_HILO): New macro.
26981 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26982 (ISA_HAS_DDIV): Likewise.
26984 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26986 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26988 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26990 * doc/invoke.texi (fabi-version): Spelling fix.
26992 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26995 * doc/invoke.texi (fabi-version): Mention mangling fix for
26998 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27001 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27003 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27006 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27007 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27009 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27011 (stack_protect_test): Ditto.
27012 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27013 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27015 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27016 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27017 -mstack-protector-guard-offset=.
27018 (RS/6000 and PowerPC Options): Ditto.
27020 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27022 * config/i386/i386.h (MASK_CLASS_P): New define.
27023 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27024 there are no registers from different register sets also when
27025 mask registers are used. Update function comment.
27026 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27027 to (*k/*r) and (*k/*km) alternatives.
27029 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27031 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27032 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27033 (EH_RETURN_HANDLER_RTX): New define.
27034 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27035 Force frame pointer in EH return functions.
27036 (aarch64_expand_epilogue): Add barrier for eh_return.
27037 (aarch64_final_eh_return_addr): Remove.
27038 (aarch64_eh_return_handler_rtx): New function.
27039 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27041 (aarch64_eh_return_handler_rtx): New prototype.
27043 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27045 * config/rs6000/altivec.h (vec_rlmi): New #define.
27046 (vec_vrlnm): Likewise.
27047 (vec_rlnm): Likewise.
27048 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27049 (UNSPEC_VRLNM): Likewise.
27050 (VIlong): New mode iterator.
27051 (altivec_vrl<VI_char>mi): New define_insn.
27052 (altivec_vrl<VI_char>nm): Likewise.
27053 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27055 (VRLDNM): Likewise.
27056 (RLNM): New polymorphic function entry.
27057 (VRLWMI): New monomorphic function entry.
27058 (VRLDMI): Likewise.
27059 (RLMI): New polymorphic function entry.
27060 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27061 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27062 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27065 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27068 * dwarf2out.c (field_byte_offset): Restore the
27069 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27070 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27071 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27074 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27077 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27079 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27082 * dwarf2out.c (add_data_member_location_attribute): For constant
27083 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27084 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27085 DW_AT_byte_size attributes.
27087 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27089 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27090 after forcing to constant memory when the code model is medium.
27092 2017-01-17 Julia Koval <julia.koval@intel.com>
27095 * config/i386/avx512fintrin.h
27096 (_mm512_i32gather_ps): Change __addr type to void const*.
27097 (_mm512_mask_i32gather_ps): Ditto.
27098 (_mm512_i32gather_pd): Ditto.
27099 (_mm512_mask_i32gather_pd): Ditto.
27100 (_mm512_i64gather_ps): Ditto.
27101 (_mm512_mask_i64gather_ps): Ditto.
27102 (_mm512_i64gather_pd): Ditto.
27103 (_mm512_mask_i64gather_pd): Ditto.
27104 (_mm512_i32gather_epi32): Ditto.
27105 (_mm512_mask_i32gather_epi32): Ditto.
27106 (_mm512_i32gather_epi64): Ditto.
27107 (_mm512_mask_i32gather_epi64): Ditto.
27108 (_mm512_i64gather_epi32): Ditto.
27109 (_mm512_mask_i64gather_epi32): Ditto.
27110 (_mm512_i64gather_epi64): Ditto.
27111 (_mm512_mask_i64gather_epi64): Ditto.
27112 (_mm512_i32scatter_ps): Change __addr type to void*.
27113 (_mm512_mask_i32scatter_ps): Ditto.
27114 (_mm512_i32scatter_pd): Ditto.
27115 (_mm512_mask_i32scatter_pd): Ditto.
27116 (_mm512_i64scatter_ps): Ditto.
27117 (_mm512_mask_i64scatter_ps): Ditto.
27118 (_mm512_i64scatter_pd): Ditto.
27119 (_mm512_mask_i64scatter_pd): Ditto.
27120 (_mm512_i32scatter_epi32): Ditto.
27121 (_mm512_mask_i32scatter_epi32): Ditto.
27122 (_mm512_i32scatter_epi64): Ditto.
27123 (_mm512_mask_i32scatter_epi64): Ditto.
27124 (_mm512_i64scatter_epi32): Ditto.
27125 (_mm512_mask_i64scatter_epi32): Ditto.
27126 (_mm512_i64scatter_epi64): Ditto.
27127 (_mm512_mask_i64scatter_epi64): Ditto.
27128 * config/i386/avx512pfintrin.h
27129 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27130 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27131 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27132 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27133 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27134 (_mm512_prefetch_i32scatter_ps): Ditto.
27135 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27136 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27137 (_mm512_prefetch_i64scatter_pd): Ditto.
27138 (_mm512_prefetch_i64scatter_ps): Ditto.
27139 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27140 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27141 * config/i386/avx512vlintrin.h
27142 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27143 (_mm_mmask_i32gather_ps): Ditto.
27144 (_mm256_mmask_i32gather_pd): Ditto.
27145 (_mm_mmask_i32gather_pd): Ditto.
27146 (_mm256_mmask_i64gather_ps): Ditto.
27147 (_mm_mmask_i64gather_ps): Ditto.
27148 (_mm256_mmask_i64gather_pd): Ditto.
27149 (_mm_mmask_i64gather_pd): Ditto.
27150 (_mm256_mmask_i32gather_epi32): Ditto.
27151 (_mm_mmask_i32gather_epi32): Ditto.
27152 (_mm256_mmask_i32gather_epi64): Ditto.
27153 (_mm_mmask_i32gather_epi64): Ditto.
27154 (_mm256_mmask_i64gather_epi32): Ditto.
27155 (_mm_mmask_i64gather_epi32): Ditto.
27156 (_mm256_mmask_i64gather_epi64): Ditto.
27157 (_mm_mmask_i64gather_epi64): Ditto.
27158 (_mm256_i32scatter_ps): Change __addr type to void*.
27159 (_mm256_mask_i32scatter_ps): Ditto.
27160 (_mm_i32scatter_ps): Ditto.
27161 (_mm_mask_i32scatter_ps): Ditto.
27162 (_mm256_i32scatter_pd): Ditto.
27163 (_mm256_mask_i32scatter_pd): Ditto.
27164 (_mm_i32scatter_pd): Ditto.
27165 (_mm_mask_i32scatter_pd): Ditto.
27166 (_mm256_i64scatter_ps): Ditto.
27167 (_mm256_mask_i64scatter_ps): Ditto.
27168 (_mm_i64scatter_ps): Ditto.
27169 (_mm_mask_i64scatter_ps): Ditto.
27170 (_mm256_i64scatter_pd): Ditto.
27171 (_mm256_mask_i64scatter_pd): Ditto.
27172 (_mm_i64scatter_pd): Ditto.
27173 (_mm_mask_i64scatter_pd): Ditto.
27174 (_mm256_i32scatter_epi32): Ditto.
27175 (_mm256_mask_i32scatter_epi32): Ditto.
27176 (_mm_i32scatter_epi32): Ditto.
27177 (_mm_mask_i32scatter_epi32): Ditto.
27178 (_mm256_i32scatter_epi64): Ditto.
27179 (_mm256_mask_i32scatter_epi64): Ditto.
27180 (_mm_i32scatter_epi64): Ditto.
27181 (_mm_mask_i32scatter_epi64): Ditto.
27182 (_mm256_i64scatter_epi32): Ditto.
27183 (_mm256_mask_i64scatter_epi32): Ditto.
27184 (_mm_i64scatter_epi32): Ditto.
27185 (_mm_mask_i64scatter_epi32): Ditto.
27186 (_mm256_i64scatter_epi64): Ditto.
27187 (_mm256_mask_i64scatter_epi64): Ditto.
27188 (_mm_i64scatter_epi64): Ditto.
27189 (_mm_mask_i64scatter_epi64): Ditto.
27190 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27191 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27192 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27193 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27194 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27195 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27196 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27197 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27198 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27199 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27200 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27201 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27202 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27203 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27204 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27205 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27206 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27207 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27208 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27209 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27210 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27211 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27212 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27213 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27214 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27215 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27216 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27217 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27218 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27219 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27220 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27221 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27222 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27223 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27224 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27225 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27226 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27227 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27228 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27229 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27230 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27231 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27232 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27233 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27234 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27235 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27236 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27237 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27238 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27239 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27240 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27241 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27242 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27243 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27244 definitions accordingly.
27246 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27247 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27250 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27253 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27256 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27257 hard regs for inner regno.
27259 2017-01-17 Martin Liska <mliska@suse.cz>
27262 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27263 assumption and add comment.
27265 2017-01-17 Nathan Sidwell <nathan@acm.org>
27267 * ipa-visibility.c (localize_node): New function, broken out of ...
27268 (function_and_variable_visibility): ... here. Call it.
27270 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27272 PR middle-end/77445
27273 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27274 correctly set frequency of oudgoing edge.
27275 (duplicate_thread_path): Fix profile updating.
27277 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27280 * configure.ac: Add GCC_BASE_VER.
27281 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27282 version from BASE-VER file.
27283 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27284 (gcc.o): Depend on $(BASEVER).
27285 * common.opt (dumpfullversion): New option.
27286 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27287 * doc/invoke.texi: Document -dumpfullversion.
27288 * doc/install.texi: Document --with-gcc-major-version-only.
27289 * configure: Regenerated.
27291 2017-01-17 Richard Biener <rguenther@suse.de>
27293 PR tree-optimization/71433
27294 * tree-vrp.c (register_new_assert_for): Merge same asserts
27295 on all incoming edges.
27296 (process_assert_insertions_for): Handle insertions at the
27299 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27301 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27302 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27304 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27307 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27310 2017-01-17 Alan Modra <amodra@gmail.com>
27313 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27314 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27315 symbolic stack limit when pic.
27317 2017-01-16 Martin Sebor <msebor@redhat.com>
27319 PR tree-optimization/78608
27320 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27322 2017-01-16 Jeff Law <law@redhat.com>
27325 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27326 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27327 for several include directories that may be relative to sysroot.
27328 * config/i386/x-mingw32 (gplus_includedir): Define.
27329 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27330 (native_system_includedir): Likewise.
27331 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27332 override if TARGET_SYSTEM_ROOT is defined.
27333 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27335 PR tree-optimization/79090
27336 PR tree-optimization/33562
27337 PR tree-optimization/61912
27338 PR tree-optimization/77485
27339 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27340 and computed trims into the dump file.
27342 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27344 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27346 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27349 * gimplify.c (gimplify_init_constructor): If want_value and
27350 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27354 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27355 sequence. Formatting fixes.
27356 (doloop_optimize): Formatting fixes.
27359 * gcc.c (debug_level_greater_than_spec_func): New function.
27360 (static_spec_functions): Add debug-level-gt spec function.
27361 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27363 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27364 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27365 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27366 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27367 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27368 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27370 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27372 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27373 QImode fixups to general and mask registers only.
27375 2017-01-16 Carl Love <cel@us.ibm.com>
27377 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27378 for built-in functions
27379 vector signed char vec_nabs (vector signed char)
27380 vector signed short vec_nabs (vector signed short)
27381 vector signed int vec_nabs (vector signed int)
27382 vector signed long long vec_nabs (vector signed long long)
27383 vector float vec_nabs (vector float)
27384 vector double vec_nabs (vector double)
27385 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27387 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27388 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27389 * doc/extend.texi: Update the documentation file for the new built-in
27392 2017-01-16 Martin Sebor <msebor@redhat.com>
27394 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27397 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27399 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27400 UNSPEC_VSX__XXSPLTD to require special splat handling.
27402 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27405 * system.h: Poison strndup.
27407 2017-01-16 Alan Modra <amodra@gmail.com>
27410 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27413 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27415 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27417 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27419 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27420 call recog here. Assert that INSN_CODE (insn) is non-negative.
27422 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27425 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27427 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27428 in the currently scheduled RTL fragment.
27430 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27432 PR rtl-optimization/78751
27433 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27436 2017-01-14 Jeff Law <law@redhat.com>
27438 PR tree-optimization/79090
27439 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27440 variable length stores.
27441 (compute_trims): Delete dead assignment to *trim_tail.
27442 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27445 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27447 PR rtl-optimization/78626
27448 PR rtl-optimization/78727
27449 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27450 of a block, and split such blocks after everything else is finished.
27452 2017-01-14 Alan Modra <amodra@gmail.com>
27455 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27456 target legitimate_combined_insn.
27457 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27458 (rs6000_legitimate_combined_insn): New function.
27459 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27461 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27462 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27463 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27465 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27467 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27469 2017-01-13 Jeff Law <law@redhat.com>
27471 PR tree-optimization/33562
27472 PR tree-optimization/61912
27473 PR tree-optimization/77485
27474 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27476 (delete_dead_assignment): Likewise.
27477 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27478 statement to delete_dead_call and delete_dead_assignment.
27480 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27483 * substring-locations.c (format_warning_va): Strengthen case 1 so
27484 that both endpoints of the substring must be within the format
27485 range for just the substring to be printed.
27487 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27489 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27490 * config/i386/i386.c (ix86_target_string): Add missing options
27491 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27492 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27493 flags_other and ix86_target_other to flags2_other. Display unknown
27495 (ix86_valid_target_attribute_inner_p): Add missing options and
27496 reorder options by implied ISAs, as in ix86_target_string.
27498 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27500 * hash-table.h (hash_table::too_empty_p): New function.
27501 (hash_table::expand): Use it.
27502 (hash_table::traverse): Likewise.
27503 (hash_table::empty_slot): Use sizeof (value_type) instead of
27504 sizeof (PTR) to convert bytes to elements. Shrink the table
27505 if the current size is excessive for the current number of
27508 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27510 * ira-costs.c (record_reg_classes): Break from the inner loop
27511 early once alt_fail is known to be true. Update outer loop
27512 handling accordingly.
27514 2017-01-13 Jeff Law <law@redhat.com>
27516 * tree-ssa-dse.c (decrement_count): New function.
27517 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27518 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27519 when we know the partially dead statement is a mem* function.
27521 PR tree-optimization/61912
27522 PR tree-optimization/77485
27523 * tree-ssa-dse.c: Include expr.h.
27524 (maybe_trim_constructor_store): New function.
27525 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27527 PR tree-optimization/33562
27528 PR tree-optimization/61912
27529 PR tree-optimization/77485
27530 * doc/invoke.texi: Document new dse-max-object-size param.
27531 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27532 * tree-ssa-dse.c: Include params.h.
27533 (dse_store_status): New enum.
27534 (initialize_ao_ref_for_dse): New, partially extracted from
27536 (valid_ao_ref_for_dse, normalize_ref): New.
27537 (setup_live_bytes_from_ref, compute_trims): Likewise.
27538 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27539 (maybe_trim_partially_dead_store): Likewise.
27540 (maybe_trim_complex_store): Likewise.
27541 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27542 Track what bytes live from the original store. Return tri-state
27543 for dead, partially dead or live.
27544 (dse_dom_walker): Add constructor, destructor and new private members.
27545 (delete_dead_call, delete_dead_assignment): New extracted from
27547 (dse_optimize_stmt): Make a member of dse_dom_walker.
27548 Use initialize_ao_ref_for_dse.
27550 PR tree-optimization/33562
27551 PR tree-optimization/61912
27552 PR tree-optimization/77485
27553 * sbitmap.h (bitmap_count_bits): Prototype.
27554 (bitmap_clear_range, bitmap_set_range): Likewise.
27555 * sbitmap.c (bitmap_clear_range): New function.
27556 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27558 2017-01-13 Martin Liska <mliska@suse.cz>
27561 * function.c (set_cfun): Add new argument force.
27562 * function.h (set_cfun): Likewise.
27563 * ipa-inline-transform.c (inline_call): Use the function when
27564 strict alising from is dropped for function we inline to.
27566 2017-01-13 Richard Biener <rguenther@suse.de>
27568 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27569 for dumping GIMPLE INTEGER_CSTs.
27571 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27573 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27574 to 201112L since C++17.
27576 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27579 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27580 if -fsanitize=kernel-address is present.
27582 2017-01-13 Richard Biener <rguenther@suse.de>
27584 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27585 as _Literal ( type ) number in case usual suffixes do not
27586 preserve all information.
27588 2017-01-13 Richard Biener <rguenther@suse.de>
27590 PR tree-optimization/77283
27591 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27592 and ssa-iterators.h.
27593 (is_feasible_trace): Implement a cost model based on joiner
27596 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27599 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27600 char or short to __float128/_Float128 directly.
27602 2017-01-12 Martin Sebor <msebor@redhat.com>
27604 to -Wformat-overflow.
27605 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27606 (min_bytes_remaining): Same.
27607 (get_string_length): Same.
27608 (format_string): Same.
27609 (format_directive): Same.
27611 (pass_sprintf_length::handle_gimple_call): Same.
27613 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27615 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27616 info.nowrite calls with no lhs that can't throw. Return bool
27617 whether gsi_remove has been called or not.
27618 (pass_sprintf_length::handle_gimple_call): Return bool whether
27619 try_substitute_return_value called gsi_remove. Formatting fix.
27620 (pass_sprintf_length::execute): Don't use gsi_remove if
27621 handle_gimple_call returned true.
27624 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27625 be removed due to side-effects, don't remove following barrier nor
27626 turn the successor edge into fallthru edge.
27628 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27631 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27632 element-reversing loads and stores as not swappable.
27634 2017-01-12 Nathan Sidwell <nathan@acm.org>
27635 Nicolai Stange <nicstange@gmail.com>
27637 * combine.c (try_combine): Don't ignore result of overlap checking
27638 loop. Combine overlap & asm check into single loop.
27640 2017-01-12 Richard Biener <rguenther@suse.de>
27642 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27643 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27645 2017-01-12 Richard Biener <rguenther@suse.de>
27647 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27648 and TS_TARGET_OPTION directly derive from TS_BASE.
27649 * tree-core.h (tree_optimization_option): Derive from tree_base.
27650 (tree_target_option): Likewise.
27652 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27654 * config/i386/i386.c (memory_address_length): Increase len
27655 only when rip_relative_addr_p returns false.
27657 2017-01-11 Julia Koval <julia.koval@intel.com>
27659 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27660 (OPTION_MASK_ISA_SGX_SET): New.
27661 (ix86_handle_option): Handle OPT_msgx.
27662 * config.gcc: Added sgxintrin.h.
27663 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27664 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27665 * config/i386/i386.c (ix86_target_string): Add -msgx.
27667 (ix86_option_override_internal): Handle new options.
27668 (ix86_valid_target_attribute_inner_p): Add sgx.
27669 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27670 * config/i386/i386.opt: Add msgx.
27671 * config/i386/sgxintrin.h: New file.
27672 * config/i386/x86intrin.h: Add sgxintrin.h.
27674 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27677 * fold-const.c (maybe_nonzero_address): Return 1 for function
27679 (tree_single_nonzero_warnv_p): Don't handle function local objects
27683 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27686 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27689 * opts.c: Include "spellcheck.h"
27690 (struct string_fragment): New struct.
27691 (struct edit_distance_traits<const string_fragment &>): New
27693 (get_closest_sanitizer_option): New function.
27694 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27696 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27698 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27700 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27701 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27702 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27703 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27704 for initial die_offset if dwarf_split_debug_info.
27705 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27706 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27708 (output_skeleton_debug_sections): Formatting fix. Use
27709 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27710 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27712 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27714 * config/arm/cortex-a53.md: Add bypasses for
27715 cortex_a53_r2f_cvt.
27716 (cortex_a53_r2f): Only use for transfers.
27717 (cortex_a53_f2r): Likewise.
27718 (cortex_a53_r2f_cvt): Add reservation for conversions.
27719 (cortex_a53_f2r_cvt): Likewise.
27721 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27723 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27724 to all inlined functions, change static to extern.
27726 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27729 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27731 (arm_assemble_integer): Likewise.
27733 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27735 * config.gcc: Use new awk script to check CPU, FPU and architecture
27736 parameters for --with-... options.
27737 * config/arm/parsecpu.awk: New file
27738 * config/arm/arm-cpus.in: New file.
27739 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27741 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27743 * config/arm/t-arm: Update dependency rules.
27744 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27745 of processing .def files.
27746 * config/arm/genopt.sh: Deleted.
27747 * config/arm/gentune.sh: Deleted.
27748 * config/arm/arm-cores.def: Deleted.
27749 * config/arm/arm-arches.def: Deleted.
27750 * config/arm/arm-fpus.def: Deleted.
27751 * config/arm/arm-tune.md: Regenerated.
27752 * config/arm/arm-tables.opt: Regenerated.
27753 * config/arm/arm-cpu.h: New generated file.
27754 * config/arm/arm-cpu-data.h: New generated file.
27755 * config/arm/arm-cpu-cdata.h: New generated file.
27757 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27760 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27762 (input_varpool_node): Unpack dynamically_initialized bit.
27764 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27766 PR rtl-optimization/79032
27767 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27768 the alignment of the adjusted memory reference against that of MODE,
27769 instead of the alignment of the original memory reference.
27771 2017-01-11 Martin Jambor <mjambor@suse.cz>
27773 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27775 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27776 decorated functions.
27778 2017-01-11 Richard Biener <rguenther@suse.de>
27780 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27781 set range/nonnull info for PHI results. Do not set it on
27782 stmts marked for removal.
27784 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27786 * expr.c (store_field): In the bitfield case, fetch the return value
27787 from the registers before applying a single big-endian adjustment.
27788 Always do a final load for a BLKmode value not larger than a word.
27790 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27793 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27794 that we correctly handle column numbers greater than
27795 LINE_MAP_MAX_COLUMN_NUMBER.
27797 2017-01-10 Martin Sebor <msebor@redhat.com>
27799 PR middle-end/78245
27800 * gimple-ssa-sprintf.c (get_destination_size): Call
27801 {init,fini}object_sizes.
27802 * tree-object-size.c (addr_object_size): Adjust.
27803 (pass_through_call): Adjust.
27804 (pass_object_sizes::execute): Adjust.
27805 * tree-object-size.h (fini_object_sizes): Declare.
27807 2017-01-10 Martin Sebor <msebor@redhat.com>
27809 PR tree-optimization/78775
27810 * builtins.c (get_size_range): Move...
27811 * calls.c: ...to here.
27812 (alloc_max_size): Accept zero argument.
27813 (operand_signed_p): Remove.
27814 (maybe_warn_alloc_args_overflow): Call get_size_range.
27815 * calls.h (get_size_range): Declare.
27817 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27819 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27820 from TI's devices.csv file as of September 2016.
27821 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27823 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27825 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27826 * doc/invoke.texi: Likewise.
27827 * doc/md.texi: Likewise.
27828 * doc/objc.texi: Likewise.
27830 2017-01-10 Joshua Conner <joshconner@google.com>
27832 * config/arm/fuchsia-elf.h: New file.
27833 * config/fuchsia.h: New file.
27834 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27835 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27837 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27839 2016-01-10 Richard Biener <rguenther@suse.de>
27841 PR tree-optimization/79034
27842 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27843 Propagate out degenerate PHIs in the joiner.
27845 2017-01-10 Martin Liska <mliska@suse.cz>
27847 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27848 (sort_congruence_classes_by_decl_uid): Likewise.
27849 (sort_congruence_class_groups_by_decl_uid): Likewise.
27850 (sem_item_optimizer::merge_classes): Sort class, groups in these
27851 classes and members in the groups by DECL_UID of declarations.
27852 This would make merge operations stable.
27854 2017-01-10 Martin Liska <mliska@suse.cz>
27856 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27857 usage of m_classes_vec.
27858 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27859 (sem_item_optimizer::get_group_by_hash): Likewise.
27860 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27861 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27862 (sem_item_optimizer::verify_classes): Likewise.
27863 (sem_item_optimizer::process_cong_reduction): Likewise.
27864 (sem_item_optimizer::dump_cong_classes): Likewise.
27865 (sem_item_optimizer::merge_classes): Likewise.
27866 * ipa-icf.h (congruence_class_hash): Rename from
27867 congruence_class_group_hash. Remove declaration of m_classes_vec.
27869 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27871 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27872 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27873 * config.gcc: Add avx512vpopcntdqintrin.h.
27874 * config/i386/avx512vpopcntdqintrin.h: New.
27875 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27876 * config/i386/i386-builtin-types.def: Add new types.
27877 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27878 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27879 __builtin_ia32_vpopcountq_v8di_mask): New.
27880 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27881 __AVX512VPOPCNTDQ__.
27882 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27883 (PTA_AVX512VPOPCNTDQ): Define.
27884 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27885 TARGET_AVX512VPOPCNTDQ_P): Define.
27886 * config/i386/i386.opt: Add mavx512vpopcntdq.
27887 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27888 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27890 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27892 PR middle-end/77484
27893 * predict.def (PRED_CALL): Set to 67.
27895 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27897 * expr.c (store_field): In the bitfield case, if the value comes from
27898 a function call and is of an aggregate type returned in registers, do
27899 not modify the field mode; extract the value in all cases if the mode
27900 is BLKmode and the size is not larger than a word.
27902 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27905 * config/i386/cpuid.h: Fix undefined behavior.
27907 2017-01-04 Jeff Law <law@redhat.com>
27909 PR tree-optimization/79007
27910 PR tree-optimization/67955
27911 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27912 conservative for pt.null when flag_non_call_exceptions is on.
27914 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27916 PR translation/79019
27917 PR translation/79020
27918 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27919 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27920 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27922 * config/avr/avr.opt (maccumulate-args): Likewise.
27923 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27924 * common.opt (freport-bug): Likewise.
27925 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27926 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27927 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27928 translatable string.
27929 * config/i386/i386.c (function_value_32): Likewise.
27930 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27931 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27933 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27934 * common/config/msp430/msp430-common.c (msp430_handle_option):
27936 * symtab.c (symtab_node::verify_base): Likewise.
27937 * opts.c (set_debug_level): Likewise.
27938 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27939 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27940 missing whitespace to translatable strings.
27941 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27942 * config/sh/superh.h: Likewise.
27943 * config/i386/xopintrin.h: Likewise.
27944 * config/i386/znver1.md: Likewise.
27945 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27946 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27947 * double-int.h (struct double_int): Likewise.
27948 * double-int.c (div_and_round_double): Likewise.
27949 * wide-int.cc: Likewise.
27950 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27951 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27952 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27953 (crossjumps_occurred): ... this.
27954 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27957 PR tree-optimization/78899
27958 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27959 returning bool return struct loop *, NULL for failure and the new
27961 (versionable_outer_loop_p): Don't version outer loop if it has
27962 dont_vectorized bit set.
27963 (tree_if_conversion): When versioning outer loop, ensure
27964 tree_if_conversion is performed also on the inner loop of the
27965 non-vectorizable outer loop copy.
27966 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27967 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27968 prevent vectorization of it.
27969 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27970 the outer loop vectorization of the non-scalar version is attempted
27971 before vectorization of the inner loop in scalar version. If
27972 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27973 vectorization of its inner loop.
27974 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27975 has 2 inner loops, rename also on edges from bb whose single pred
27976 is outer_loop->header. Fix typo in function comment.
27978 2017-01-09 Martin Sebor <msebor@redhat.com>
27981 * asan.c (asan_emit_stack_protection): Increase local buffer size
27982 to avoid snprintf truncation warning.
27984 2017-01-09 Andrew Pinski <apinski@cavium.com>
27986 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27987 to reference thunderx2t99 for the tuning structure
27988 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27990 (thunderx2t99_extra_costs): This.
27991 * config/aarch64/aarch64-tune.md: Regenerate.
27992 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27993 (vulcan_addrcost_table): This.
27994 (vulcan_regmove_cost): Rename to ...
27995 (thunderx2t99_regmove_cost): This.
27996 (vulcan_vector_cost): Rename to ...
27997 (thunderx2t99_vector_cost): this.
27998 (vulcan_branch_cost): Rename to ...
27999 (thunderx2t99_branch_cost): This.
28000 (vulcan_tunings): Rename to ...
28001 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28002 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28004 2017-01-09 Martin Jambor <mjambor@suse.cz>
28008 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28009 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28010 (propagate_vr_accross_jump_function): Use the above function for all
28011 value range computations for pass-through jump functions and type
28012 converasion from explicit value range values.
28013 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28014 parameters from TYPE_ARG_TYPES.
28015 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28016 (ipa_write_node_info): Stream type of the actual argument.
28017 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28019 2017-01-09 Martin Liska <mliska@suse.cz>
28022 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28023 (lookup_compiler): Do not show error message with have_E.
28025 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28027 PR tree-optimization/78938
28028 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28029 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28030 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28031 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28034 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28036 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28039 2017-01-09 Richard Biener <rguenther@suse.de>
28041 PR tree-optimization/78997
28042 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28043 name condition properly.
28045 2017-01-09 Richard Biener <rguenther@suse.de>
28048 * dwarf2out.c (is_cxx): New overload with context.
28049 (is_naming_typedef_decl): Use it.
28051 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28053 * invoke.texi (Option Summary): Correct spacing in option lists
28054 and add line breaks to fix over-long lines.
28056 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28058 PR middle-end/17660
28060 * extend.texi (Common Variable Attributes): Add xref to GCC
28061 Internals manual to explain mode attribute keywords.
28063 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28066 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28067 and Preprocessor Options.
28068 (Options for Linking): Document -pthread here....
28069 (RS/6000 and PowerPC Options): ...not here.
28070 (Solaris 2 Options): ...or here.
28071 * doc/cppopts.texi: Document -pthread.
28073 2017-01-08 Martin Sebor <msebor@redhat.com>
28075 PR middle-end/77708
28076 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28077 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28078 New member functions.
28079 (format_directive): Used them.
28081 (pass_sprintf_length::handle_gimple_call): Same.
28082 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28083 to avoid truncation for any argument.
28084 (extract_affine_mul): Same.
28085 * tree.c (get_file_function_name): Same.
28087 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28089 PR middle-end/77484
28090 * predict.def (PRED_INDIR_CALL): Set to 86.
28092 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28094 PR preprocessor/54124
28095 * doc/cppopts.texi: Reformat -d subtable to list the full name
28096 of the options. Add cross-reference to the docs for the general
28097 compiler -d options.
28098 * doc/invoke.texi (Developer Options): Add cross-reference to the
28099 preprocessor-specific -d option documentation.
28101 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28103 PR preprocessor/13498
28104 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28105 redudant material, and reflect new command-line options.
28106 (System Headers): Likewise.
28108 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28110 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28111 -isystem, and -idirafter. Copy-edit.
28112 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28113 default for -ftrack-macro-expansion. Delete obsolete and
28114 badly-formatted implementation details about -fdebug-cpp output.
28115 * doc/cppwarnopts.texi: Copy-edit.
28117 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28120 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28121 that the transition from a max line width >= 1<<10 to narrower
28122 lines works correctly.
28124 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28126 * doc/options.texi (PerFunction): New.
28127 * opt-functions.awk (switch_flags): Map both Optimization and
28128 PerFunction to CL_OPTIMIZATION.
28129 * opth-gen.awk: Test for PerFunction flag along with
28131 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28132 it only when the latter is present. Skip those that don't in
28133 the hash function generator.
28134 * common.opt (fvar-tracking): Mark as PerFunction instead of
28136 (fvar-tracking-assignments): Likewise.
28137 (fvar-tracking-assignments-toggle): Likewise.
28138 (fvar-tracking-uninit): Likewise.
28140 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28142 PR translation/79018
28143 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28146 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28149 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28150 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28151 TARGET_LONG_JUMP_TABLE_OFFSETS.
28152 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28153 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28154 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28155 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28156 * config/m68k/m68k.md (tablejump expander): Likewise.
28157 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28158 TARGET_LONG_JUMP_TABLE_OFFSETS.
28159 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28160 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28162 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28163 David Holsgrove <david.holsgrove@xilinx.com>
28165 * common/config/microblaze/microblaze-common.c
28166 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28167 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28169 * config/microblaze/microblaze.c (microblaze_must_save_register)
28170 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28172 (microblaze_eh_return): New function.
28173 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28174 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28175 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28176 * config/microblaze/microblaze.md (eh_return): New pattern.
28178 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28180 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28181 GCC_DIAGNOSTIC_STRINGIFY): Define.
28183 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28185 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28187 * config/arm/arm.md (<mcrr>): New.
28189 * config/arm/arm.c (arm_arch5te): New.
28190 (arm_option_override): Set arm_arch5te.
28191 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28193 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28194 (arm_mcrr_qualifiers): ... this. New.
28195 (MRRC_QUALIFIERS): Define to...
28196 (arm_mrrc_qualifiers): ... this. New.
28197 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28199 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28200 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28201 (MRRCI, mrrc, MRRC): New.
28202 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28203 VUNSPEC_MRRC2): New.
28205 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28207 * config/arm/arm.md (<mcr>): New.
28209 * config/arm/arm.c (arm_coproc_builtin_available): Add
28210 support for mcr, mrc, mcr2 and mrc2.
28211 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28212 (arm_mcr_qualifiers): ... this. New.
28213 (MRC_QUALIFIERS): Define to ...
28214 (arm_mrc_qualifiers): ... this. New.
28215 (MCR_QUALIFIERS): Define to ...
28216 (arm_mcr_qualifiers): ... this. New.
28217 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28219 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28220 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28221 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28222 VUNSPEC_MRC2): New.
28224 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28226 * config/arm/arm.md (*ldc): New.
28230 * config/arm/arm.c (arm_coproc_builtin_available): Add
28231 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28232 (arm_coproc_ldc_stc_legitimate_address): New.
28233 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28234 'qualifier_const_pointer'.
28235 (LDC_QUALIFIERS): Define to...
28236 (arm_ldc_qualifiers): ... this. New.
28237 (STC_QUALIFIERS): Define to...
28238 (arm_stc_qualifiers): ... this. New.
28239 * config/arm/arm-protos.h
28240 (arm_coproc_ldc_stc_legitimate_address): New.
28241 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28242 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28243 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28244 stc2, stcl, stc2l): New.
28245 * config/arm/constraints.md (Uz): New.
28246 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28247 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28248 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28249 VUNSPEC_STC2L): New.
28251 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28253 * config/arm/arm.md (<cdp>): New.
28254 * config/arm/arm.c (neon_const_bounds): Rename this ...
28255 (arm_const_bounds): ... this.
28256 (arm_coproc_builtin_available): New.
28257 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28258 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28259 (CDP_QUALIFIERS): Define to...
28260 (arm_cdp_qualifiers): ... this. New.
28262 (arm_expand_builtin_args): Add case for 6 arguments.
28263 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28264 (arm_const_bounds): ... this.
28265 (arm_coproc_builtin_available): New.
28266 * config/arm/arm_acle.h (__arm_cdp): New.
28268 * config/arm/arm_acle_builtins.def (cdp): New.
28270 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28271 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28272 'arm_const_bounds'.
28273 * config/arm/types.md (coproc): New.
28274 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28275 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28276 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28277 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28279 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28281 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28282 (UBINOP_QUALIFIERS): New.
28284 (acle_builtin_data): New. Change comment.
28285 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28286 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28287 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28288 arm_acle_builtins.def.
28289 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28290 (arm_init_acle_builtins): New.
28291 (CRC32_BUILTIN): Remove.
28292 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28293 crc32cb, crc32ch and crc32cw.
28294 (arm_init_crc32_builtins): Remove.
28295 (arm_init_builtins): Use arm_init_acle_builtins rather
28296 than arm_init_crc32_builtins.
28297 (arm_expand_acle_builtin): New.
28298 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28299 * config/arm/arm_acle_builtins.def: New.
28301 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28303 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28304 (arm_builtin_datum): ... this.
28305 (arm_init_neon_builtin): Rename to ...
28306 (arm_init_builtin): ... this. Add a new parameters PREFIX
28307 and USE_SIG_IN_NAME.
28308 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28309 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28310 'arm_builtin_datum'.
28311 (arm_init_vfp_builtins): Likewise.
28312 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28313 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28314 (arm_expand_neon_args): Rename to ...
28315 (arm_expand_builtin_args): ... this. Rename builtin_arg
28316 enum values and differentiate between ARG_BUILTIN_MEMORY
28317 and ARG_BUILTIN_NEON_MEMORY.
28318 (arm_expand_neon_builtin_1): Rename to ...
28319 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28320 values, arm_expand_builtin_args and add bool parameter NEON.
28321 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28322 (arm_expand_vfp_builtin): Likewise.
28323 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28325 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28327 PR middle-end/77484
28328 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28329 * predict.c (tree_estimate_probability_bb): Reverse direction of
28330 polymorphic call predictor.
28332 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28334 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28335 (determine_pass_name_match): ...this new function and...
28336 (should_skip_pass_p): ...this new function.
28338 2017-01-06 Nathan Sidwell <nathan@acm.org>
28340 * ipa-visibility.c (function_and_variable_visibility): Reformat
28341 comments and long lines. Remove extrneous if.
28342 * symtab.c (symtab_node::make_decl_local): Fix code format.
28343 (symtab_node::set_section_for_node): Fix comment typo.
28345 2017-01-06 Martin Liska <mliska@suse.cz>
28348 * lra-constraints.c: Rename invariant to lra_invariant.
28349 * predict.c (set_even_probabilities): Initialize e to NULL.
28351 2017-01-05 Martin Sebor <msebor@redhat.com>
28353 PR tree-optimization/78910
28354 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28355 (format_integer): Correct off-by-one error in the handling
28356 of precision with negative numbers in signed conversions..
28358 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28360 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28362 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28364 PR tree-optimization/71016
28365 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28366 factor_out_conditional_conversion. Formatting fix.
28367 (factor_out_conditional_conversion): Add cond_stmt argument.
28368 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28369 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28372 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28374 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28375 read-rtl-function.o, and selftest-rtl.o.
28376 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28377 (selftest::aarch64_test_loading_full_dump): New function.
28378 (selftest::aarch64_run_selftests): New function.
28379 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28380 selftest::aarch64_run_selftests.
28381 * config/i386/i386.c
28382 (selftest::ix86_test_loading_dump_fragment_1): New function.
28383 (selftest::ix86_test_loading_call_insn): New function.
28384 (selftest::ix86_test_loading_full_dump): New function.
28385 (selftest::ix86_test_loading_unspec): New function.
28386 (selftest::ix86_run_selftests): Call the new functions.
28387 * emit-rtl.c (maybe_set_max_label_num): New function.
28388 * emit-rtl.h (maybe_set_max_label_num): New decl.
28389 * function.c (instantiate_decls): Guard call to
28390 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28391 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28393 * gensupport.c (gen_reader::gen_reader): Pass "false"
28394 for new "compact" param of rtx_reader.
28395 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28396 rather than an empty string for NULL strings.
28397 * read-md.c: Potentially include config.h rather than bconfig.h.
28398 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28399 (have_error): New global, copied from errors.c.
28400 (md_reader::read_name): Rename to...
28401 (md_reader::read_name_1): ...this, adding "out_loc" param,
28402 and converting "missing name or number" to returning false, rather
28404 (md_reader::read_name): Reimplement in terms of read_name_1.
28405 (md_reader::read_name_or_nil): New function.
28406 (md_reader::read_string): Handle "(nil)" by returning NULL.
28407 (md_reader::md_reader): Add new param "compact".
28408 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28409 (md_reader::read_file): New method.
28410 * read-md.h (md_reader::md_reader): Add new param "compact".
28411 (md_reader::read_file): New method.
28412 (md_reader::is_compact): New accessor.
28413 (md_reader::read_name): Convert return type from void to file_location.
28414 (md_reader::read_name_or_nil): New decl.
28415 (md_reader::read_name_1): New decl.
28416 (md_reader::m_compact): New field.
28417 (noop_reader::noop_reader): Pass "false" for new "compact" param
28419 (rtx_reader::rtx_reader): Add new "compact" param.
28420 (rtx_reader::read_rtx_operand): Make virtual and convert return
28421 type from void to rtx.
28422 (rtx_reader::read_until): New decl.
28423 (rtx_reader::handle_any_trailing_information): New virtual function.
28424 (rtx_reader::postprocess): New virtual function.
28425 (rtx_reader::finalize_string): New virtual function.
28426 (rtx_reader::m_in_call_function_usage): New field.
28427 (rtx_reader::m_reuse_rtx_by_id): New field.
28428 * read-rtl-function.c: New file.
28429 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28430 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28431 (selftest::verify_three_block_rtl_cfg): New decl.
28432 * read-rtl-function.h: New file.
28433 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28434 For host, include function.h, memmodel.h, and emit-rtl.h.
28435 (one_time_initialization): New function.
28436 (struct compact_insn_name): New struct.
28437 (compact_insn_names): New array.
28438 (find_code): Handle insn codes in compact dumps.
28439 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28440 (bind_subst_iter_and_attr): Likewise.
28441 (add_condition_to_string): Likewise.
28442 (add_condition_to_rtx): Likewise.
28443 (apply_attribute_uses): Likewise.
28444 (add_current_iterators): Likewise.
28445 (apply_iterators): Likewise.
28446 (initialize_iterators): Guard usage of apply_subst_iterator with
28447 #ifdef GENERATOR_FILE.
28448 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28449 (md_reader::read_mapping): Likewise.
28450 (add_define_attr_for_define_subst): Likewise.
28451 (add_define_subst_attr): Likewise.
28452 (read_subst_mapping): Likewise.
28453 (check_code_iterator): Likewise.
28454 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28456 (one_time_initialization): New function.
28457 (rtx_reader::read_until): New method.
28458 (read_flags): New function.
28459 (parse_reg_note_name): New function.
28460 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28461 Handle reuse_rtx ids.
28462 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28463 Add parsing support for RTL dumps, mirroring the special-cases in
28464 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28465 values, and calling handle_any_trailing_information.
28466 (rtx_reader::read_rtx_operand): Convert return type from void
28467 to rtx, returning return_rtx. Handle case 'e'. Call
28468 finalize_string on XSTR and XTMPL fields.
28469 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28470 "(nil)" values were omitted. Call the postprocess vfunc on the
28472 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28473 class ctor. Initialize m_in_call_function_usage. Call
28474 one_time_initialization.
28475 * rtl-tests.c (selftest::test_uncond_jump): Call
28476 set_new_first_and_last_insn.
28477 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28478 * selftest-rtl.c: New file.
28479 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28480 (selftest::get_insn_by_uid): New decl.
28481 * selftest-run-tests.c (selftest::run_tests): Call
28482 read_rtl_function_c_tests.
28483 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28484 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28487 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28489 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28490 operands in a special way. Assert that pos+len <= mode precision.
28492 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28494 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28495 3 argument Alias with unlimited for the negative form.
28496 (fno-vect-cost-model): Removed.
28498 2017-01-05 Martin Liska <mliska@suse.cz>
28500 * hsa-gen.c (gen_hsa_divmod): New function.
28501 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28503 2017-01-05 Martin Liska <mliska@suse.cz>
28506 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28509 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28511 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28512 small constant length operands.
28514 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28516 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28517 between loop iterations.
28519 2017-01-05 Martin Liska <mliska@suse.cz>
28522 * gimplify.c (gimplify_decl_expr): Compare to
28523 asan_poisoned_variables instread of checking flags.
28524 (gimplify_target_expr): Likewise.
28525 (gimplify_expr): Likewise.
28526 (gimplify_function_tree): Conditionally initialize
28527 asan_poisoned_variables.
28529 2017-01-04 Jeff Law <law@redhat.com>
28531 PR tree-optimizatin/78812
28532 * rtl.h (contains_mem_rtx_p): Prototype.
28533 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28534 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28535 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28536 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28537 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28539 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28541 * input.c (assert_char_at_range): Default-initialize actual_range.
28543 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28545 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28546 to match the caller.
28548 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28550 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28551 insns after final jump in test to emit dummy move.
28553 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28555 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28556 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28558 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28560 * multiple_target.c (create_dispatcher_calls): Init e_next.
28561 * tree-ssa-loop-split.c (split_loop): Init border.
28562 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28565 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28570 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28571 (altivec_register_operand): Do not return true if the operand
28572 contains a SUBREG mixing SImode and SFmode.
28573 (vsx_register_operand): Likewise.
28574 (vsx_reg_sfsubreg_ok): New predicate.
28575 (vfloat_operand): Do not return true if the operand contains a
28576 SUBREG mixing SImode and SFmode.
28577 (vint_operand): Likewise.
28578 (vlogical_operand): Likewise.
28579 (gpc_reg_operand): Likewise.
28580 (int_reg_operand): Likewise.
28581 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28582 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28583 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28585 (rs6000_emit_move_si_sf_subreg): New helper function.
28586 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28587 fixup SUBREGs involving SImode and SFmode.
28588 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28589 numbers for the new peephole2 optimization.
28590 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28591 the GLIBC math library that do AND/IOR/XOR operations on single
28592 precision floating point.
28593 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28594 target macros to say whether we need to avoid SUBREGs mixing
28596 (TARGET_ALLOW_SF_SUBREG): Likewise.
28597 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28598 (UNSPEC_SI_FROM_SF): Likewise.
28599 (iorxor): Change spacing.
28600 (and_ior_xor): New iterator for AND, IOR, and XOR.
28601 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28602 (movdi_from_sf_zero_ext): Likewise.
28603 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28604 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28605 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28606 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28607 (fms<mode>4): Likewise.
28608 (fnma<mode>4): Likewise.
28609 (fnms<mode>4): Likewise.
28610 (nfma<mode>4): Likewise.
28611 (nfms<mode>4): Likewise.
28613 2017-01-04 Marek Polacek <polacek@redhat.com>
28616 * doc/invoke.texi: Document -Wpointer-compare.
28618 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28620 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28623 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28624 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28627 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28630 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28631 documentation of the powerpc_popcntb_ok attribute.
28632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28633 code to issue warning messages if a requested CPU configuration is
28634 not supported by the binary (assembler and loader) toolchain.
28635 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28636 made to define a built-in function that has been disabled.
28637 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28638 made to define a built-in function that has been disabled.
28639 (altivec_init_builtins): Add comment explaining why definition
28640 of the DST built-in functions is not preceded by an assertion
28641 check. Add assertions to prevent ICE if attempts are made to
28642 define an altivec predicate or an abs* built-in function that has
28644 (htm_init_builtins): Add comment explaining why definition of the
28645 htm built-in functions is not preceded by an assertion check.
28647 2017-01-04 Jeff Law <law@redhat.com>
28649 PR tree-optimizatin/67955
28650 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28651 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28652 the points-to solution does not include pt_null. Use DECL_PT_UID
28655 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28657 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28658 Use gen_int_mode instead of gen_lopwart for const_int operands.
28660 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28662 PR tree-optimization/71563
28663 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28664 out of range value - has low bits known to be zero.
28666 2017-01-04 Alan Modra <amodra@gmail.com>
28668 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28669 * configure: Regenerate.
28670 * config.in: Regenerate.
28672 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28675 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28676 a substring of the message, but strcmp with the whole message. Ifdef
28677 ENABLE_NLS, translate the message first using dgettext.
28679 2017-01-03 Jeff Law <law@redhat.com>
28681 PR tree-optimizatin/78856
28682 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28683 (mark_threaded_blocks): Remove code to truncate thread paths that
28684 cross multiple loop headers. Instead invalidate the cached loop
28685 iteration information and handle case of a thread path walking
28686 into an irreducible region.
28688 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28691 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28692 assertions. Add support for doing the signbit if the IEEE 128-bit
28693 floating point value is in a GPR.
28694 * config/rs6000/rs6000.md (Fsignbit): Delete.
28695 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28696 Update the length attribute if the value is in a GPR.
28697 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28698 the sign or zero extension instruction, since the value is always 0/1.
28699 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28702 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28703 extracting SImode to a GPR register so that we can generate a
28704 store, limit the vector to be in a traditional Altivec register
28705 for the vextuwrx instruction.
28707 2017-01-03 Ian Lance Taylor <iant@google.com>
28709 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28711 2017-01-03 Martin Sebor <msebor@redhat.com>
28713 PR tree-optimization/78696
28714 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28715 precision. Use MPFR for %f for greater fidelity. Correct handling
28717 (pass_sprintf_length::compute_format_length): Set width and precision
28718 specified by asrerisk to void_node for vararg functions.
28719 (try_substitute_return_value): Adjust dump output.
28721 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28723 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28725 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28727 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28728 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28729 -mlra/-mno-lra was passed to the compiler.
28731 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28733 PR rtl-optimization/65618
28734 * emit-rtl.c (try_split): Move initialization of "before" and
28735 "after" to just before the call to emit_insn_after_setloc.
28737 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28739 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28741 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28743 * dwarf2out.c (gen_enumeration_type_die): When
28744 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28746 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28748 PR tree-optimization/78965
28749 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28750 Change first argument from const call_info & to call_info &. For %n
28751 set info.nowrite to false.
28753 PR middle-end/78901
28754 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28755 possibly throwing calls.
28757 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28758 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28759 and fns handling, rather than in a separate case SSA_NAME.
28761 2017-01-02 Jeff Law <law@redhat.com>
28763 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28764 fixes for first_period and second_period variables.
28766 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28769 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28770 (*insvqi_1): New insn pattern.
28771 (*insvqi_1_mem_rex64): Ditto.
28772 (*insvqi_2): Ditto.
28773 (*insvqi_3): Rename from *insvqi.
28775 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28777 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28779 * doc/cfg.texi (Edges): Remove reference to Java.
28780 (Maintaining the CFG): Ditto.
28782 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28784 PR middle-end/77674
28785 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28786 transparent aliases.
28788 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28790 PR middle-end/77484
28791 * predict.def (PRED_CALL): Update hitrate.
28792 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28793 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28794 into direct/indirect/polymorphic variants.
28796 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28798 Update copyright years.
28800 * gcc.c (process_command): Update copyright notice dates.
28801 * gcov-dump.c (print_version): Ditto.
28802 * gcov.c (print_version): Ditto.
28803 * gcov-tool.c (print_version): Ditto.
28804 * gengtype.c (create_file): Ditto.
28805 * doc/cpp.texi: Bump @copying's copyright year.
28806 * doc/cppinternals.texi: Ditto.
28807 * doc/gcc.texi: Ditto.
28808 * doc/gccint.texi: Ditto.
28809 * doc/gcov.texi: Ditto.
28810 * doc/install.texi: Ditto.
28811 * doc/invoke.texi: Ditto.
28813 Copyright (C) 2017 Free Software Foundation, Inc.
28815 Copying and distribution of this file, with or without modification,
28816 are permitted in any medium without royalty provided the copyright
28817 notice and this notice are preserved.